Login Register

Access the GIFSQ Portal

Select your user type to log in or register a new account.

Student Portal

Access your food safety courses and exams.

Instructor Portal

Manage courses, view student submissions, and grade quizzes.

Company Portal

Manage corporate setup, view employee logs, and access QA services.

GIFSQ ID: FTN-0099

← All subjects

Precision Fermentation

Unit 1: What Precision Fermentation Is

Note

Unit 1: What Precision Fermentation Is

A question to start with

A baker adds yeast to dough and the dough rises. A cheese maker adds a culture to milk and the milk sets into curd. Both are fermentation, and both are thousands of years old. Now imagine the same yeast, but changed so that instead of making bread rise, it makes the whey protein of milk, or a sweet protein that tastes like sugar, or a red colour for a drink. The yeast is the same kind of microbe, the tank is the same kind of tank, but the product is chosen in advance, made to order, molecule by molecule. That is precision fermentation: using microbes as small factories to make one chosen food ingredient. This unit explains what it is, how it differs from the fermentation in bread and beer, and why it has become one of the fastest growing fields in food science.

Fermentation: the old idea

Fermentation is the use of microbes to change a food. In bread, yeast eats sugars in the dough and releases CO₂, which makes the dough rise, and C₂H₅OH, which bakes away in the oven. In yoghurt, lactic acid bacteria eat lactose, the sugar in milk, and make lactic acid, which thickens the milk and gives it a tangy taste. In soy sauce, moulds and yeasts break down soy proteins over months into savoury compounds. In each case, the microbe eats a feed, grows, and leaves useful products behind.

Traditional fermentation uses the microbe as it is found in nature. The maker picks a good strain, feeds it well, and keeps the conditions right, but the microbe decides what it makes. A yeast makes alcohol and CO₂ because that is what yeast does. A lactic acid bacterium makes lactic acid because that is what it does. The maker can guide the process, but cannot ask the microbe to make a different molecule.

The foods made this way are whole foods: bread, cheese, beer, yoghurt, kimchi. The microbe is part of the food or is removed from it, but the food itself is the product. Precision fermentation is different in one clear way: the product is a single ingredient, made pure, and then added to a food. The microbe is the factory, and the ingredient is what leaves the factory gate. That difference, whole food versus single ingredient, is the line between traditional fermentation and precision fermentation.

What precision means

Precision fermentation starts with a choice: which molecule do we want? It may be a milk protein such as beta lactoglobulin, the main protein in whey. It may be a sweet protein such as thaumatin, which tastes sweet at tiny amounts. It may be an enzyme such as chymosin, which sets milk into curd for cheese. It may be a colour such as leghemoglobin, which gives a plant based burger its meaty red colour, or a fat, or a vitamin. The choice is made first, and the rest of the work serves it.

The next step is to give a microbe the instructions for making that molecule. The instructions are written in DNA, the molecule that carries the code for building proteins. Scientists take the gene, the stretch of DNA that codes for the chosen protein, from the organism that makes it naturally, and place that gene into a host microbe, usually a yeast or a fungus. The host then reads the gene and builds the protein, alongside its own proteins. The host is called a production strain, and the gene is the recipe it follows.

That is the precision in the name: the microbe is told exactly what to make, and it makes that one thing, in large amounts, while the rest of its work stays the same. The process is precise in a second sense too: the conditions in the tank, the temperature, the pH, the feed, the O₂, are measured and controlled to a fine degree, so that each batch makes the same product at the same quality. Traditional fermentation is a craft with wide variation. Precision fermentation is a craft turned into a controlled manufacturing process.

The host microbes

The hosts used in precision fermentation are working microbes with long histories in food. The most common is the yeast Saccharomyces cerevisiae, the same yeast that makes bread and beer, which is valued because it is safe, well understood, and easy to grow. Another common yeast is Komagataella phaffii, known in older books as Pichia pastoris, which is valued because it can make large amounts of a foreign protein and release it outside the cell, where it is easy to collect.

Filamentous fungi are also used, including Aspergillus oryzae, the mould that makes soy sauce and miso, and Trichoderma reesei, a fungus that naturally makes large amounts of enzymes. Bacteria are used for some products too, including Escherichia coli for certain proteins and Bacillus species for enzymes, though bacteria are less often used for food proteins because their products need more cleaning to be safe for eating.

The choice of host is a technical decision with real consequences. A yeast that releases the protein outside the cell makes the harvest easy: the liquid is collected and the protein is taken out of it. A host that keeps the protein inside the cell needs an extra step to break the cells open. A host that adds sugar chains to the protein, a step called glycosylation, may change how the protein behaves, and that change may help or hurt, depending on the product. The maker tests several hosts and picks the one that gives the most protein, at the best quality, at the lowest cost.

From gene to ingredient: the steps in order

The work of precision fermentation follows a clear order. First, the target is chosen and its gene is found, either by copying it from the organism that makes it naturally or by writing the DNA code from scratch. Second, the gene is placed into the host microbe, along with the control signals that tell the host when to read it. Third, the new strain is tested in small flasks: does it make the protein, how much, and is the protein the right shape and the right size?

Fourth, the strain is grown in larger tanks, step by step, from a flask to a pilot tank to a production tank, and the conditions are tuned at each step. Fifth, the protein is harvested and purified: the cells are removed, the liquid is filtered, and the protein is separated from the other molecules in the liquid by steps such as filtration, chromatography and drying. Sixth, the finished ingredient is tested for identity, purity, safety and function, and only then is it sold to a food maker.

A worked example shows the scale of the growth. A yeast cell doubles in about two hours. A production tank may hold 100,000 litres. Starting from a small flask, the culture is stepped up through larger tanks over several days, each step about ten times larger than the last, until the production tank is full. The protein is then harvested from that tank, and a single tank can yield tens or hundreds of kilograms of pure protein, depending on the strain and the product. That is the arithmetic that makes a microbe a factory.

How it differs from genetic modification of crops

People often ask whether precision fermentation is the same as growing a genetically modified crop, and the answer is no, in two ways that matter. First, the microbe is changed, not a crop, and the microbe is grown in a closed tank, not in a field. Second, the finished ingredient is purified away from the microbe: the cells are removed, the protein is taken out and cleaned, and the final ingredient holds no living microbe and, in most processes, no DNA from the microbe at all. The food that the shopper eats holds the protein, not the organism that made it.

That separation shapes how the product is judged. A food safety authority looks at the finished ingredient: what it is, how pure it is, whether it matches the natural molecule, and whether it is safe to eat. The authority also looks at the process: which host was used, how the protein was purified, and whether any trace of the host remains. The two checks together, product and process, are what decide whether the ingredient can be sold, and they are covered in full in Unit 6.

The distinction also matters for labelling. In many countries, a food made with a purified ingredient from precision fermentation is labelled by what the ingredient is, such as whey protein, not by how it was made, while in other countries the production method must be stated. The rules differ, and they are changing, so a maker who sells in several countries must follow each one. This subject describes the general principles and tells the reader to check the current rule in their own country, because no single country's rule is the world's rule.

Why now, and why it matters

Precision fermentation is not a new idea: the enzyme chymosin, used to set milk for cheese, has been made this way since the 1990s, and it is now used in most cheese making around the world. What is new is the range of products and the fall in cost. Reading and writing DNA has become far cheaper and faster, so a new strain can be built in months rather than years. Fermentation tanks, filters and dryers have become cheaper and better, so a small company can now do what once needed a large corporation. And the demand for animal free dairy proteins, sweet proteins and colours has grown, so there are buyers waiting for the ingredients.

The promise is plain: a milk protein without a cow, a sweet protein without sugar cane fields, an egg protein without a hen. Each one cuts the land, the feed and the animals needed for that ingredient, while keeping the molecule itself the same. A whey protein made by yeast is the same molecule as the whey protein in milk, with the same amino acids in the same order, so it behaves the same way in a recipe: it foams, it gels, it browns, and it feeds the body in the same way.

The limits are plain too. Precision fermentation makes ingredients, not whole foods: it cannot make a steak, only a protein that goes into a food. It needs sugar as a feed, and that sugar comes from a crop, so the footprint of the ingredient depends on how the sugar was grown. It needs sterile tanks, clean water and energy, and it needs a safety file before it can be sold. The units that follow take each of these points in turn, starting with the microbes themselves.

How much protein can one cell make: a worked example

A fermentation tank is judged by its yield: how much protein it makes per litre of broth, and how much sugar it eats to make it. The yield is worked out from two numbers that the operator measures every run: the cell density, how many grams of dry cells grow per litre, and the protein share, how much of the cell is the target protein. A yeast that grows to 50 g of dry cells per litre, and that holds 10 percent of its dry weight as the target protein, makes 5 g of protein per litre. That is the yield, and it is the number the strain builder tries to lift.

The sugar side is worked out the same way. If the tank was fed 100 g of sugar per litre to reach that yield, the sugar to protein ratio is 100 divided by 5, which is 20 kg of sugar per kilogram of protein. That ratio decides the cost, because the sugar is bought by the kilogram and the protein is sold by the kilogram. A strain that lifts the yield from 5 g to 10 g per litre, on the same sugar, halves the sugar per kilogram of protein, and halves the largest cost in the tank.

The worked example shows why small improvements matter so much. A yield of 5 g per litre in a 100,000 litre tank gives 500 kg of protein per run. A yield of 6 g per litre gives 600 kg, a fifth more, from the same tank, the same sugar and the same week. The strain builder who lifts the yield by a fifth has built a fifth more factory without building anything. That is why strain engineering, covered in Unit 8, is the most valuable work in precision fermentation, and why the yield per litre is the first number any maker quotes.

Questions to think about

  • Why is chymosin, made by fermentation since the 1990s, a precision fermentation product, while the cheese it makes is a traditional fermented food?
  • A whey protein made by yeast has the same amino acids in the same order as the whey protein in milk. Why does that sameness matter for how it behaves in a recipe?
  • Why must the finished ingredient be purified away from the microbe, and what would remain if that step were skipped?
Think about the chymosin question before you read on

Chymosin is a single ingredient, one enzyme, made to order by a microbe that was given the gene for it, then purified and sold. The cheese is a whole food made by a traditional process that uses that enzyme as one input. The difference is the product: a pure ingredient made by a chosen microbe versus a whole food made by a mixed culture doing what it naturally does.

From gene to ingredient

Choose the target molecule and find its gene
↓
Place the gene into a host yeast or fungus
↓
Test the new strain in small flasks
↓
Grow it up step by step to a production tank
↓
Harvest the protein and purify it away from the cells
↓
Test identity, purity and safety, then sell the ingredient

The diagram shows the precision fermentation chain in order. The gene is the recipe, the host is the factory, and the purification step is what separates the ingredient from the microbe that made it.

Practical note: In a practical session, compare two cheeses: one made with traditional animal rennet and one made with fermentation produced chymosin, if both can be found in a shop. Read the labels and note how each one names the coagulant. Then set a small pot of milk with a vegetarian rennet tablet, following the packet directions, and watch the milk set into curd: note the time, the temperature, and the firmness of the curd when you cut it. Record how long the set took and how the curd felt, and compare your notes with a classmate who used a different milk.

Quick self-check

Question: What is the difference between traditional fermentation and precision fermentation?

Traditional fermentation uses a microbe as it is found in nature to make a whole food, such as bread or yoghurt, and the microbe decides what it makes. Precision fermentation gives a microbe the gene for one chosen ingredient, grows it in a tank, and purifies that single ingredient for sale.

Question: Why is the host microbe usually a yeast or a fungus with a history in food?

Because those microbes are safe, well understood, and easy to grow, and because a long history of use in food makes the safety file easier to build. A new host with no food history would need far more testing before its product could be sold.

Question: A yeast grows to 50 g of dry cells per litre and holds 10 percent target protein. What is the yield per litre?

50 multiplied by 0.10, which is 5 g of protein per litre. In a 100,000 litre tank that gives 500 kg of protein per run.

Try it yourself

Go to a supermarket and find three foods that list an enzyme or a fermentation made ingredient on the label, such as cheese, bread, or a dairy free product. Write down the exact words the label uses for that ingredient. Then find a product that says animal free dairy or similar words on the front and read its ingredient list: does it name a whey or casein protein, and does it say how that protein was made? Write down what the label tells you and what it leaves out, and answer: could a shopper tell from the label alone how the protein was made?

Key ideas to carry forward

  • Fermentation uses microbes to change a food, and traditional fermentation lets the microbe decide what it makes.
  • Precision fermentation chooses one molecule first, then gives a host microbe the gene for it, so the microbe makes that one thing to order.
  • Common hosts are food yeasts and fungi, chosen because they are safe, well understood and easy to grow.
  • The finished ingredient is purified away from the cells, so the food holds the protein, not the living microbe.
  • A whey protein made by yeast has the same amino acids in the same order as the whey protein in milk, so it behaves the same way.
  • Chymosin for cheese has been made this way since the 1990s, and falling DNA costs have now widened the range of products.
  • Precision fermentation makes ingredients, not whole foods, and it still needs sugar, sterile tanks, energy and a safety file.

  • Yield per litre equals cell density multiplied by the protein share, and sugar per kilogram of protein decides the cost.

Glossary

Chymosin

The enzyme that sets milk into curd for cheese making. Food tech example: most cheese today is made with chymosin produced by fermentation rather than taken from calves.

DNA

The molecule that carries the code for building proteins. Food tech example: the gene for a milk protein is a stretch of DNA placed into a yeast.

Fermentation

The use of microbes to change a food or to make a product. Food tech example: yeast fermenting dough releases CO₂ that makes bread rise.

Gene

A stretch of DNA that codes for one protein. Food tech example: the gene for beta lactoglobulin codes for the main protein in whey.

Glycosylation

The adding of sugar chains to a protein by a host cell. Food tech example: a yeast host may add sugar chains to a protein, changing how it behaves.

Host microbe

The microbe that is given a gene and makes the chosen protein. Food tech example: a food yeast is the host microbe for a fermentation made whey protein.

Production strain

A host microbe that has been given the gene for a product and tested for making it. Food tech example: the production strain is grown in tanks to make the ingredient.

Beta lactoglobulin

The main protein in whey, the liquid left after milk sets into curd. Food tech example: beta lactoglobulin made by fermentation foams and gels like the whey protein in milk.

Thaumatin

A sweet tasting protein from a West African fruit, now made by fermentation. Food tech example: thaumatin tastes sweet at tiny amounts, so a little goes a long way in a drink.

Leghemoglobin

A red protein from soy roots that gives a meaty colour and flavour note. Food tech example: leghemoglobin made by fermentation colours a plant based burger pink like raw beef.

Purification

Separating the chosen protein from the cells and the other molecules in the liquid. Food tech example: purification by filtration takes the whey protein out of the fermentation liquid.

Lactic acid bacteria

Bacteria that turn milk sugar into lactic acid. Food tech example: lactic acid bacteria thicken milk into yoghurt.

Saccharomyces cerevisiae

The yeast used for bread and beer, and a common host for precision fermentation. Food tech example: Saccharomyces cerevisiae is the host for several fermentation made food proteins.

Chromatography

A separation method that sorts molecules by size or by how they stick to a material. Food tech example: chromatography is one step that purifies a fermentation made protein.

Lactose

The sugar in milk. Food tech example: lactic acid bacteria eat lactose and make lactic acid, which sets yoghurt.

Cell density

How many grams of dry cells grow per litre of broth. Food tech example: a yeast at 50 g of dry cells per litre makes 5 g of protein per litre at 10 percent protein.

Yield

How much product a process makes per litre or per kilogram of feed. Food tech example: lifting the yield from 5 g to 6 g per litre gives a fifth more protein from the same tank.

References

  • Stanbury, P. F., Whitaker, A. and Hall, S. J., Principles of Fermentation Technology, 3rd edition, 2017.
  • Bamforth, C. W., Food, Fermentation and Micro-organisms, 2005.
  • Waites, M. J., Morgan, N. L., Rockey, J. S. and Higton, G., Industrial Microbiology: An Introduction, 2001.
  • Demain, A. L. and Davies, J. E., Manual of Industrial Microbiology and Biotechnology, 2nd edition, 1999.
  • Damodaran, S., Parkin, K. L. and Fennema, O. R., Fennema's Food Chemistry, 4th edition, 2007.
  • Fellows, P. J., Food Processing Technology, 4th edition, 2017.

Unit 2: The Microbes: Yeast, Fungi and Bacteria as Factories

Note

Unit 2: The Microbes: Yeast, Fungi and Bacteria as Factories

A question to start with

A yeast cell is smaller than the dot at the end of this sentence, yet it holds a full factory inside it: it takes in sugar, builds proteins, folds them into shape, and sends some of them out through its wall. A fungus does the same on a larger scale, growing as threads that pour enzymes into the world around them. A bacterium does it faster still, doubling every twenty minutes when fed well. This unit looks inside those three kinds of microbe and asks the practical question: which one makes the best factory for a given food ingredient, and what must the maker do to keep that factory running well?

What a microbe is, and what it needs

A microbe is a living cell too small to see without a microscope. It needs food, water, the right temperature, and the right acidity to grow. Its food is usually a sugar such as glucose, plus a source of nitrogen such as ammonia, plus minerals such as phosphate, potassium and magnesium, plus a few vitamins. Given those, it eats, grows, and divides, making two cells from one, then four, then eight, in a chain of doubling that can fill a tank in a day.

Growth follows a pattern. First comes a lag phase, where the cells adjust to the new tank and grow slowly. Then comes the exponential phase, where they double at a steady rate and grow fast. Then comes the stationary phase, where the feed runs low or a waste product builds up, and growth stops. Then comes the death phase, where cells begin to die. A maker who wants protein harvests during the late exponential or early stationary phase, when the cells are many and still healthy, and the product is at its highest level.

Oxygen matters for yeasts and fungi, which breathe O₂ to grow well, while many bacteria can grow with or without it. The tank must supply O₂ by bubbling air through the liquid and stirring it, so that every cell gets its share. Without enough O₂, a yeast will ferment sugar into C₂H₅OH instead of growing, which is useful for beer and useless for making protein. The balance of feed, air and temperature is the daily work of running a fermentation.

Yeasts: the familiar factory

Yeasts are single celled fungi, round or oval cells that divide by budding, where a small daughter cell grows out of the mother and then breaks away. The best known yeast, Saccharomyces cerevisiae, has been used for bread, beer and wine for thousands of years, and its biology is mapped in detail: scientists know its genes, its pathways and its habits, which makes it easy to work with and easy to explain to a safety authority.

Saccharomyces cerevisiae is a good host for many proteins because it folds proteins well, adds sugar chains to them in a way that is usually acceptable, and keeps the product inside the cell or releases it, depending on the design. Its weakness is yield: it makes less of a foreign protein than some other yeasts, and it can add sugar chains that change the protein's behaviour. For a protein where the sugar chains matter, that change must be tested, not assumed.

Komagataella phaffii, known in older books as Pichia pastoris, is the second great yeast host. It grows to very high cell densities, it releases foreign proteins outside the cell in large amounts, and it adds smaller, simpler sugar chains than Saccharomyces. Those traits make it the first choice for many food proteins, including several milk proteins and sweet proteins. Its history in food is shorter, so its safety file needs more work, but its performance in the tank has made it the workhorse of the field.

Filamentous fungi: the enzyme pourers

Filamentous fungi grow as threads, called hyphae, that branch and tangle into a mass called mycelium. They are the great enzyme makers of nature: a fungus that rots wood must pour enzymes out of its cells to break the wood down, and that habit makes it a natural factory for proteins that must be released outside the cell. Aspergillus oryzae, the mould of soy sauce and miso, and Trichoderma reesei, a fungus that breaks down plant fibre, are the two most used fungal hosts.

A fungal host suits products that must be made in large amounts and released into the liquid, where they are easy to collect. The fungus grows as a thick, tangled mass, and the tank must stir it without breaking the threads too much, and must supply O₂ through a mass that resists mixing. That is the engineering challenge of fungal fermentation: the biology is generous, and the mixing is hard. Tanks for fungi use special stirrers and air spargers, and the feed is added slowly so the mass does not grow too thick to stir.

Fungi also bring a caution: some species make toxins, and a production strain must be proven free of toxin genes or proven not to make the toxin under production conditions. The strains used in food are chosen from species with long safe histories, and each batch is tested. A fungus that makes a useful enzyme and a toxin at the same time is not a food factory, whatever its yield.

Bacteria: fast, but harder to clean

Bacteria are the fastest growers in this unit. Escherichia coli, the best studied bacterium, can double every twenty minutes under ideal conditions, and it is the host for many medicines and for some food enzymes. Bacillus species, including Bacillus subtilis, are used for enzymes because they release proteins outside the cell and they form spores that survive harsh conditions, which makes them easy to store and to handle.

Bacteria have two limits for food proteins. First, they do not fold complex proteins as well as yeasts and fungi, and a protein that needs careful folding may come out wrong, clumped into lumps called inclusion bodies that must be unfolded and refolded, an extra step that costs time and yield. Second, some bacteria, including Escherichia coli, make a toxin called endotoxin in their outer wall, and that toxin must be removed completely before the product can be eaten. The cleaning steps exist and they work, but they add cost and they must be proven for every batch.

A worked example shows the speed. A bacterium doubling every twenty minutes doubles three times in an hour. Starting from one gram, after ten hours there would be thirty doublings, and one gram would become about a billion grams if nothing limited growth. In a real tank, feed and space limit growth long before that, and the harvest is a steady stream. The arithmetic shows why bacteria are tempting and why their limits, folding and cleaning, decide whether the temptation is worth it.

Building a production strain

Building a production strain is a sequence of laboratory steps. First, the gene for the target protein is obtained, either by copying it from the organism that makes it naturally or by writing the DNA code from a known sequence. Second, the gene is placed into a small circle of DNA called a plasmid, or into the host's own chromosome, together with a promoter, a control signal that tells the host when to read the gene, and a signal that tells the host where to send the protein.

Third, the plasmid or the changed chromosome is put into the host cells, a step called transformation, and the cells that took it up are picked out by growing them on a medium where only those cells survive. Fourth, the new strain is tested in small flasks: does it make the protein, how much, is the protein the right size and the right shape, and does it behave as it should in a simple test, such as foaming or setting a gel?

Fifth, the best strain is kept, and the rest are discarded. A maker may test dozens or hundreds of strains before picking one, because small differences in the gene, the promoter and the host change the yield by large amounts. The chosen strain is then stored frozen, in many identical vials, so that every production run starts from the same cells. That frozen store, called a cell bank, is the master copy of the factory, and it is guarded like one.

Keeping the factory running: feed, air and control

A production run starts from a vial of the cell bank, grown up through a chain of tanks, each about ten times larger than the last, until the production tank is full. The feed is usually a sugar syrup, made from maize, wheat or sugar cane, plus ammonia or another nitrogen source, plus minerals and vitamins. The sugar is the energy and the carbon, the nitrogen is the building block for protein, and the minerals are the tools the cell needs to work.

Control is the difference between a good run and a failed one. Temperature is held steady, usually near 30 degrees Celsius for yeasts, by cooling jackets around the tank, because the cells release heat as they grow. The pH is held near the target, often near 5 to 6 for yeasts, by adding acid or alkali automatically. Dissolved O₂ is held above a minimum by stirring faster or by adding more air. Foam is controlled by adding a small amount of an anti foam agent, because a tank that foams over loses product and makes a mess.

Contamination is the constant risk. A single unwanted bacterium or mould spore that gets into a tank can outgrow the production strain and spoil the run, because it eats the same feed and it makes nothing useful. Tanks are therefore sterilised with steam before use, the feed and the air are filtered or heated, and samples are taken during the run and checked under the microscope and on plates. A run that shows contamination is stopped, the tank is emptied and cleaned, and the loss is counted. Prevention, not cure, is the rule, and it is why fermentation plants look more like pharmacies than like kitchens.

Keeping the strain safe: the cell bank

A production strain is too valuable to keep in one flask, so it is stored in a cell bank: a set of frozen vials, each holding the same strain, kept at very low temperature, usually below minus 70 degrees Celsius, where the cells sleep and do not change. One vial is the master bank, opened only to make more vials. The others are the working bank, used to start each production run. Every run starts from a vial of the working bank, so every batch starts from the same cells, and the product stays the same.

The bank is built in steps. A single colony of the production strain is grown in a flask, tested for protein output and for identity, and then grown in a larger flask until there is enough broth to fill hundreds of small vials. Each vial is filled with the same broth, mixed with a cryoprotectant, a substance such as glycerol that stops ice crystals from tearing the cells as they freeze. The vials are frozen slowly, labelled with the strain, the date and the tests they passed, and stored in two places, so a freezer failure in one place does not lose the strain.

The bank is tested before it is used. A vial is thawed, grown, and tested for protein output, for identity, and for contamination, and the result is compared with the original strain. If the thawed cells make the same protein at the same rate, the bank is good. If they do not, the bank is discarded and a new one is built from the master. The bank is also counted: the maker knows how many vials remain, and builds a new working bank before the old one runs out, so production never waits for a strain. The cell bank is the strain's memory, and a factory without a bank has no factory.

Questions to think about

  • Why is Komagataella phaffii often chosen over Saccharomyces cerevisiae for a food protein, and what does it do better?
  • A bacterium doubles every 20 minutes and a yeast every 2 hours. Why does the faster microbe not always win as a factory?
  • Why is a cell bank kept frozen in many identical vials, and what would happen if a factory kept only one vial?
Think about the bacterium question before you read on

Speed is only one part of the choice. Bacteria fold complex proteins less well, so the product may come out as clumps that need refolding, and some bacteria make endotoxin that must be removed completely before the food can be eaten. A slower yeast that folds the protein correctly and releases it cleanly can give more usable product per tank than a fast bacterium that needs extra cleaning steps.

Building a production strain

Obtain the gene for the target protein
↓
Place the gene with a promoter into the host
↓
Transform host cells and pick the ones that took up the gene
↓
Test the new strains in small flasks for yield and quality
↓
Keep the best strain frozen as a cell bank
↓
Grow the cell bank up through larger tanks to production

The diagram shows how a production strain is built and stored. Testing many strains and keeping the best one frozen is what makes every production run start from the same cells.

Practical note: In a practical session, grow baker's yeast in two small jars to see the growth pattern: mix a teaspoon of dried yeast with warm water and a teaspoon of sugar in each jar, keep one jar warm near 30 degrees Celsius and one jar cool near 15 degrees, and mark the level of foam and the smell each hour for four hours. Note which jar grows faster and how the smell changes. Then repeat with twice the sugar in a third jar and note whether more sugar gives more growth or whether the yeast stops sooner. Record your times, temperatures and observations in a table.

Quick self-check

Question: What are the four phases of microbial growth, and when is the protein harvested?

Lag phase, where cells adjust and grow slowly; exponential phase, where they double at a steady rate; stationary phase, where feed runs low and growth stops; and death phase, where cells begin to die. Protein is harvested during the late exponential or early stationary phase, when cells are many and still healthy.

Question: Why must a fungal production strain be proven free of toxin production?

Because some fungi make toxins alongside useful enzymes, and a toxin in a food ingredient would make the product unsafe. The strain must be proven free of toxin genes or proven not to make the toxin under production conditions, and each batch is tested.

Question: Why is the master bank opened only to make more vials, and never to start a production run?

Because the master bank is the original store of the strain, and opening it often risks losing it. The working bank is made from the master and used for runs, so the master stays safe and the strain is never lost.

Try it yourself

Buy a small pack of dried baker's yeast and a pack of fresh yeast from a shop. Make two small doughs with the same flour, water and salt, one with each yeast, and keep both in the same warm place. Note the time each dough takes to double in size, and the smell of each one. Then bake both into small rolls and compare the crumb and the taste. Write down which yeast worked faster and whether the difference mattered in the finished roll. Answer in writing: what does the doubling time of a yeast tell you about its use as a factory?

Key ideas to carry forward

  • A microbe needs sugar, nitrogen, minerals, water, the right temperature and the right pH, and it grows by doubling.
  • Yeasts are single celled fungi that divide by budding, and Saccharomyces cerevisiae is the best known and best mapped host.
  • Komagataella phaffii grows dense and releases protein outside the cell, which makes harvest easy and yields high.
  • Filamentous fungi pour enzymes out of their threads, but their tangled mass is hard to stir and to supply with O₂.
  • Bacteria grow fastest but fold complex proteins less well and may need endotoxin removed before the product can be eaten.
  • A production strain is built by placing a gene with a promoter into a host, testing many strains, and freezing the best one as a cell bank.
  • Contamination can spoil a whole run, so tanks, feed and air are sterilised and samples are checked during every run.

  • A cell bank is a set of frozen vials holding the same strain, with a master bank for making more vials and a working bank for starting runs.

Glossary

Cell bank

A frozen store of identical vials of a production strain. Food tech example: every production run starts from one vial of the cell bank, so each run uses the same cells.

Endotoxin

A toxin in the outer wall of some bacteria that must be removed from a food product. Food tech example: a protein made in Escherichia coli must be tested to prove its endotoxin is gone.

Exponential phase

The phase of growth where cells double at a steady rate. Food tech example: yeast in exponential phase doubles every two hours in a well fed tank.

Glycosylation

The adding of sugar chains to a protein by a host cell. Food tech example: a yeast host may add sugar chains that change how a protein foams.

Hyphae

The fine threads that make up a filamentous fungus. Food tech example: the hyphae of Trichoderma reesei pour enzymes out into the liquid around them.

Inclusion bodies

Clumps of wrongly folded protein inside a bacterium. Food tech example: a protein made in Escherichia coli may form inclusion bodies that must be unfolded and refolded.

Lag phase

The first phase of growth, where cells adjust to a new tank and grow slowly. Food tech example: a fresh culture spends its first hours in lag phase before doubling fast.

Mycelium

The network of hyphae that makes up most of a fungus. Food tech example: the mycelium of a production fungus fills the tank as a tangled mass.

Plasmid

A small circle of DNA used to carry a gene into a host cell. Food tech example: the gene for a milk protein is carried into yeast on a plasmid.

Promoter

A control signal in DNA that tells a cell when to read a gene. Food tech example: a strong promoter tells the yeast to make a large amount of the target protein.

Stationary phase

The phase where growth stops because feed runs low or waste builds up. Food tech example: protein is harvested in late exponential or early stationary phase.

Transformation

The step of putting new DNA into a host cell. Food tech example: transformation puts the plasmid carrying the target gene into yeast cells.

Komagataella phaffii

A yeast host, known in older books as Pichia pastoris, valued for high yields. Food tech example: Komagataella phaffii releases milk proteins outside the cell in large amounts.

Aspergillus oryzae

A mould used for soy sauce and miso, and a host for enzyme production. Food tech example: Aspergillus oryzae pours enzymes out of its cells into the liquid.

Saccharomyces cerevisiae

The yeast of bread, beer and wine, and a common host for precision fermentation. Food tech example: Saccharomyces cerevisiae is the host for several fermentation made food proteins.

Bacillus subtilis

A bacterium used for enzyme production that releases proteins outside the cell. Food tech example: Bacillus subtilis makes enzymes used in food processing.

Cell bank

A set of frozen vials holding the same strain, used to start each run. Food tech example: every production run starts from a vial of the working cell bank.

Cryoprotectant

A substance that stops ice crystals from tearing cells as they freeze. Food tech example: glycerol is added to vials before the strain is frozen.

References

  • Stanbury, P. F., Whitaker, A. and Hall, S. J., Principles of Fermentation Technology, 3rd edition, 2017.
  • Bamforth, C. W., Food, Fermentation and Micro-organisms, 2005.
  • Waites, M. J., Morgan, N. L., Rockey, J. S. and Higton, G., Industrial Microbiology: An Introduction, 2001.
  • Demain, A. L. and Davies, J. E., Manual of Industrial Microbiology and Biotechnology, 2nd edition, 1999.
  • Damodaran, S., Parkin, K. L. and Fennema, O. R., Fennema's Food Chemistry, 4th edition, 2007.
  • Fellows, P. J., Food Processing Technology, 4th edition, 2017.

Unit 3: Making Dairy Proteins Without Animals: Whey and Casein

Note

Unit 3: Making Dairy Proteins Without Animals: Whey and Casein

A question to start with

Pour milk into a glass and you see a white liquid. Leave it to set with rennet and it splits into curd and whey: the curd holds the casein, the solid protein that makes cheese, and the whey holds the soluble proteins, the lactose and the minerals. Those two protein families, casein and whey, give milk its behaviour in every dairy food: casein sets, stretches and melts, whey foams, gels and feeds. This unit explains how both are now made by fermentation, without a cow, and why making them is harder than it looks, especially casein, which needs four proteins, not one, to behave like milk.

The proteins of milk

Milk holds about 3 to 4 percent protein, split into two families. Casein makes up about 80 percent of the protein in cow milk. It is not one protein but four: alpha s1 casein, alpha s2 casein, beta casein and kappa casein. Together they form tiny balls called casein micelles, which float in the milk and scatter light, which is why milk looks white. When acid or rennet is added, the micelles join together and the milk sets into a curd, which is the first step of cheese making.

Whey holds the other 20 percent. Its main proteins are beta lactoglobulin, about half of the whey protein, and alpha lactalbumin, about a fifth, plus smaller amounts of others. Whey proteins stay dissolved in the liquid when milk sets, and they are the proteins that foam in a meringue, gel in a yoghurt, and give a sports drink its protein. Unlike casein, they are single proteins with a set shape, and each one can be made alone by a microbe.

The two families behave differently when heated. Whey proteins unfold and link together when heated above about 70 degrees Celsius, which is why heated milk forms a skin and why a whey drink thickens when it is pasteurised. Casein is far more heat stable: it survives boiling without unfolding, which is why milk can be boiled and still set into curd afterwards. A maker who replaces dairy proteins must copy both behaviours, and that is why whey came first and casein is still catching up.

Whey by fermentation: the first success

Beta lactoglobulin was the first dairy protein made at scale by precision fermentation, because it is a single protein with a known gene, and because a yeast host can make it and release it in large amounts. The gene is placed into a yeast such as Komagataella phaffii, the yeast grows in a tank on sugar, and the protein is released into the liquid, where it is collected, filtered and dried into a powder. The powder looks and behaves like whey protein from milk: it dissolves in water, foams when whipped, gels when heated, and carries the same amino acids in the same order.

The test of sameness is strict. Scientists compare the fermentation made protein with the milk protein by mass, by shape and by function: does it weigh the same, does it fold the same way, does it foam to the same height, does it gel at the same temperature? For beta lactoglobulin, the answers are yes, and that is why the ingredient is accepted as whey protein, not as a copy of whey protein. The molecule is the same; only the maker is different.

A worked example shows the use. A sports drink maker wants 10 g of protein in a 250 ml bottle. Fermentation made whey protein isolate is about 90 percent protein, so the recipe needs about 11 g of powder per bottle. In a run of 10,000 bottles, the maker needs about 110 kg of powder. That powder can be stirred into water, flavoured, and bottled, and the drink will foam a little when shaken, just as a dairy whey drink does, because the protein is the same.

Casein: four proteins and a micelle

Casein is harder, and the reason is structure. A single casein protein made by a yeast is a loose, floppy chain with little shape of its own. In milk, the four caseins join with calcium phosphate to form micelles, and it is the micelle, not the single protein, that gives milk its behaviour: the micelle scatters light, it falls apart in acid to make yoghurt, and it joins into a network under rennet to make cheese curd. Making one casein is not enough: the maker must make the set, and then get them to assemble into something like a micelle.

Progress is real but partial. Makers now produce individual caseins by fermentation, including beta casein and kappa casein, and they are learning to mix them with calcium and phosphate so that they form particles that behave like micelles. A cheese made from fermentation casein can melt and stretch, and a milk made from it can look white and set with acid, but the match is not yet as close as the whey match, and the cost is higher, because four fermentations and an assembly step cost more than one.

The prize is large, because casein is what makes cheese, and cheese is the dairy food that plant proteins imitate least well. A plant based cheese that melts, stretches and browns is still rare, while a cheese made from fermentation casein can do all three, because the protein is the same one that does them in milk. That is why companies keep working on casein even though it is harder and dearer than whey: the food it unlocks is the one shoppers miss most.

From powder to product: ice cream, cheese and yoghurt

A fermentation made dairy protein is sold as a powder, and the food maker uses it the way they would use a dairy powder. In an ice cream, whey protein adds body, holds air, and keeps the texture smooth. In a protein bar, it binds the mix and adds protein without adding lactose. In a sports drink, it dissolves clear and stays stable in acid. In each case the recipe is the same as the dairy version, because the ingredient is the same molecule.

Cheese is the test that matters most. A cheese maker mixes fermentation casein with water, fat, calcium and salt, adds rennet, and watches the mix set into a curd. The curd is cut, heated, drained and pressed, just like a dairy curd. The result can be sliced, grated and melted, and a pizza cheese made this way can stretch in strings when hot. The remaining gap is flavour: much of the flavour of a matured cheese comes from the cultures and the ripening, not from the casein alone, so a fermentation cheese still needs the right cultures and the right ripening time to taste like a dairy cheese.

Yoghurt is a simpler case. A yoghurt needs protein that sets into a gel when the cultures make acid, and both whey and casein can do that. A yoghurt made from fermentation proteins can be thick, tangy and smooth, and it can be made without lactose, because the fermentation made proteins carry no milk sugar with them. That last point matters for people who cannot digest lactose: a dairy free yoghurt that still sets and tastes like yoghurt is a food they could not buy before.

Lactose, allergy and the label question

A fermentation made dairy protein carries no lactose, because lactose is the sugar of milk and the fermentation makes only the protein. That is good news for people with lactose intolerance, who can eat a whey protein made by yeast without the sugar that upsets them. It is not, however, good news for everyone with a milk allergy, and the difference must be stated plainly.

Milk allergy is an allergy to the proteins, mainly casein and beta lactoglobulin, and a fermentation made protein is the same molecule, so a person allergic to milk protein can react to it just as they would to milk. The label must therefore say that the product contains milk protein, even though no cow was involved, because the allergy is to the molecule, not to the animal. A product that hides that fact would put allergic shoppers at risk, and the rules in most countries require the allergen to be named.

The naming of the product is still being settled in many countries. Some makers call it animal free dairy, some call it fermentation made whey, and some use the plain protein name. The rules differ by country and are changing, so a maker must check the current rule in each market where the product is sold. What does not change is the allergen rule: the protein must be named as a milk protein, whatever else the front of the pack says.

Cost, scale and the road ahead

Fermentation made whey protein still costs more per kilogram than whey from milk, because the tanks, the sugar and the purification all cost money, and because milk whey is a by-product of cheese making, sold cheaply. The gap is closing as tanks grow larger, strains make more protein per litre, and purification gets cheaper. A tank that makes twice as much protein per litre halves the cost of the tank per kilogram, and strain improvement is the fastest way to close the gap.

Scale is the second driver. A dairy protein plant that makes thousands of tonnes a year can spread its costs across a large output, while a fermentation plant that makes tens of tonnes cannot. That is why the first products are high value ones, such as sports drinks and premium ice creams, where the shopper will pay more for an animal free label, and why commodity products such as plain milk powder will come later, when the cost falls further.

The road ahead for casein is longer than for whey, but the direction is the same: better strains, better assembly of micelles, larger tanks, lower cost. The test at each step is the food itself: does the cheese melt, does the milk look white, does the yoghurt set, and does the shopper like it? A protein that passes those tests at a price the shopper will pay is a protein that has arrived, and whey has already arrived while casein is still on the road.

Testing whey and casein in the kitchen laboratory: function tests

A fermentation made whey protein is only useful if it behaves like whey from milk, and the behaviour is tested with simple function tests that any food laboratory can run. The first test is foaming: a weighed amount of protein is dissolved in water, whipped for a set time, and the volume of the foam is measured, then measured again after 30 minutes to see how much foam is left. A good whey protein doubles the volume and holds most of it after half an hour. A poor one makes little foam, or the foam collapses at once.

The second test is gelling: the protein solution is heated in a water bath to a set temperature, cooled, and the gel is pressed with a finger or a probe to judge its firmness. A good whey protein sets into a firm, springy gel. A good casein sets into a soft, creamy curd, not a rubbery one. The third test is solubility: the protein is stirred into water at pH 7 and at pH 4, spun, and the protein in the clear liquid is measured. A whey protein should dissolve well at both pH values, while a casein should dissolve poorly at pH 4, near its isoelectric point, which is why it curdles in acid.

A worked example shows how the tests are used. A buyer tests two whey powders, each at 10 g in 100 ml of water. Powder A whips to 180 ml of foam and holds 150 ml after 30 minutes. Powder B whips to 120 ml and holds 60 ml. Both hold the same protein, but A foams better and holds its foam, so A is bought for a mousse and B is bought for a drink, where foam does not matter. The function tests, not the protein number, decide which powder goes into which food, and a maker who skips them buys blind.

Questions to think about

  • Why is a single casein protein not enough to copy milk, and what structure must the caseins form together?
  • A fermentation made whey protein carries no lactose but is still a milk allergen. Why are those two facts both true?
  • Why did whey reach the market before casein, even though cheese is the food shoppers miss most?
Think about the lactose and allergy question before you read on

Lactose is a sugar, and the fermentation makes only the protein, so no lactose is present. Allergy is a reaction to the protein molecule itself, and the fermentation made protein is the same molecule as the milk protein, with the same amino acids in the same order. A person who cannot digest lactose can eat it safely, while a person allergic to milk protein can react to it, so the label must name it as a milk protein.

How fermentation made whey is produced

Place the beta lactoglobulin gene into a yeast host
↓
Grow the yeast in a tank on sugar feed
↓
Collect the liquid with the protein released into it
↓
Filter out the cells and purify the protein
↓
Dry into a powder and test identity and purity
↓
Sell the powder to food makers for drinks, bars and ice cream

The diagram shows the whey route in order. The protein is released into the liquid, which makes harvest easy, and the drying step turns it into a powder a food maker can use like dairy whey.

Practical note: In a practical session, compare two protein powders if available: a dairy whey isolate and a fermentation made whey, or a dairy whey and a pea protein. Dissolve 10 g of each in 100 ml of water, shake for 30 seconds, and measure the foam height at once and after 5 minutes. Then heat each solution gently to 80 degrees Celsius and note whether it thickens or gels. Record the foam heights, the gel, and the taste, and answer: which properties match between the two powders, and which ones differ?

Quick self-check

Question: What is a casein micelle, and why does it matter for cheese?

A micelle is a tiny ball formed by the four caseins joined with calcium phosphate, which floats in milk and scatters light, making milk look white. When acid or rennet is added, micelles join into a network that sets milk into curd, the first step of cheese making.

Question: Why is beta lactoglobulin easier to make by fermentation than casein?

Because it is a single protein with a set shape that a yeast can make and release alone, while casein needs four proteins that must assemble with calcium phosphate into micelles to behave like milk. One fermentation gives whey; casein needs several fermentations plus an assembly step.

Question: Two whey powders hold the same protein, but one whips to 180 ml of foam and holds 150 ml, while the other whips to 120 ml and holds 60 ml. Which is bought for a mousse, and why?

The first one, because it makes more foam and holds it longer, and a mousse needs a stable foam. The second is bought for a drink, where foam does not matter.

Try it yourself

Buy a tub of ice cream or a protein drink that is labelled animal free dairy or fermentation made, if one is sold near you, and buy the dairy version of the same product. Read both ingredient lists and write down the protein ingredient named in each one. Taste both and score each out of 10 for creaminess, sweetness and aftertaste. Write down whether you could tell which one was made without a cow, and what gave it away if you could. If no such product is sold near you, write down the three shops you checked and the closest animal free product you found.

Key ideas to carry forward

  • Milk protein splits into casein, about 80 percent, which sets into curd, and whey, about 20 percent, which foams and gels.
  • Beta lactoglobulin is a single protein that yeast can make and release, so fermentation made whey reached the market first.
  • Casein needs four proteins that assemble with calcium phosphate into micelles, and that assembly is the hard step.
  • A fermentation made whey protein is the same molecule as the milk protein, so it behaves the same way in a recipe.
  • Fermentation made dairy protein carries no lactose, but it is still a milk allergen, because the allergy is to the protein.
  • Cheese is the hardest test for fermentation casein, and its flavour still needs the right cultures and ripening time.
  • Fermentation whey still costs more than dairy whey, and the gap closes as strains yield more and tanks grow larger.

  • Function tests, foaming, gelling and solubility, decide which powder goes into which food, not the protein number alone.

Glossary

Alpha lactalbumin

A whey protein that makes up about a fifth of whey protein. Food tech example: alpha lactalbumin in a sports drink adds protein that dissolves easily.

Beta lactoglobulin

The main protein in whey, about half of whey protein. Food tech example: beta lactoglobulin made by fermentation foams and gels like dairy whey.

Casein

The family of proteins that makes up about 80 percent of milk protein and sets into curd. Food tech example: casein is what makes cheese hold together and stretch when melted.

Casein micelle

A tiny ball of the four caseins joined with calcium phosphate in milk. Food tech example: micelles scatter light so milk looks white, and they join into curd when rennet is added.

Curd

The solid part formed when milk sets, holding casein and fat. Food tech example: cheese making starts by cutting the curd after rennet sets the milk.

Whey

The liquid left after milk sets into curd, holding the soluble proteins. Food tech example: whey is dried into a powder used in sports drinks and protein bars.

Kappa casein

One of the four caseins, sitting on the outside of the micelle. Food tech example: rennet cuts kappa casein, which lets micelles join into curd.

Beta casein

One of the four caseins, a major part of the micelle. Food tech example: beta casein made by fermentation is mixed with other caseins to build micelle like particles.

Lactose intolerance

Difficulty digesting lactose, the sugar in milk. Food tech example: a fermentation made whey protein carries no lactose, so it suits people with lactose intolerance.

Milk allergy

An allergy to milk proteins, mainly casein and beta lactoglobulin. Food tech example: a fermentation made whey protein must be labelled as a milk protein because it can trigger milk allergy.

Rennet

A preparation holding chymosin that sets milk into curd. Food tech example: rennet is added to milk in cheese making to form the curd.

Animal free dairy

A dairy food made from fermentation made proteins without an animal. Food tech example: an animal free ice cream uses fermentation made whey instead of dairy whey.

Calcium phosphate

A mineral that joins caseins into micelles. Food tech example: calcium phosphate holds the four caseins together in a micelle.

Fermentation made whey

Whey protein produced by a microbe given the gene for it. Food tech example: fermentation made whey powder dissolves and foams like dairy whey powder.

Foaming

Making a stable foam by whipping air into a protein solution. Food tech example: whey protein foams well, which suits a mousse.

Gelling

Setting into a firm network when heated or cooled. Food tech example: whey protein sets into a firm gel when heated.

References

  • Stanbury, P. F., Whitaker, A. and Hall, S. J., Principles of Fermentation Technology, 3rd edition, 2017.
  • Bamforth, C. W., Food, Fermentation and Micro-organisms, 2005.
  • Waites, M. J., Morgan, N. L., Rockey, J. S. and Higton, G., Industrial Microbiology: An Introduction, 2001.
  • Demain, A. L. and Davies, J. E., Manual of Industrial Microbiology and Biotechnology, 2nd edition, 1999.
  • Damodaran, S., Parkin, K. L. and Fennema, O. R., Fennema's Food Chemistry, 4th edition, 2007.
  • Fellows, P. J., Food Processing Technology, 4th edition, 2017.

Unit 4: Enzymes, Sweet Proteins, Colours and Fats by Fermentation

Note

Unit 4: Enzymes, Sweet Proteins, Colours and Fats by Fermentation

A question to start with

A cheese maker adds a few drops of rennet to a vat of milk and the milk sets. A baker adds a pinch of enzyme to dough and the bread stays soft for days. A drink maker adds a speck of sweet protein and a whole bottle tastes sweet with almost no sugar. Each of these is a small amount of a powerful molecule, and each one can now be made by fermentation. This unit looks at the ingredients that work in tiny amounts: enzymes that cut and build, sweet proteins that replace sugar, colours that replace synthetic dyes, and fats that carry flavour. They are the quiet ingredients of the food industry, and precision fermentation has changed how they are made.

Enzymes: the workers of the food industry

An enzyme is a protein that speeds up a chemical change without being used up itself. A small amount of enzyme can change a large amount of food, which is why enzymes are used in tiny doses. Chymosin sets milk into curd. Amylase cuts starch into sugars, which is why it is used in bread and in brewing. Lactase cuts lactose into glucose and galactose, which is how lactose free milk is made. Lipase cuts fats, protease cuts proteins, and pectinase clears the haze from fruit juice.

Most food enzymes are already made by fermentation, and have been for decades. The microbe, often a fungus such as Aspergillus or a bacterium such as Bacillus, is grown in a tank, and the enzyme it releases is collected, purified and sold as a liquid or a powder. Precision fermentation has improved this old trade in two ways: it lets makers use a host that yields more enzyme, and it lets them change the enzyme itself, by altering its gene so that it works better at a higher temperature, in acid, or in the presence of salt.

A worked example shows the dose. A cheese maker uses chymosin at about 0.02 percent of the milk, which is 200 ml of a standard liquid rennet per 1,000 litres of milk. That small dose sets the whole vat in about 30 to 40 minutes. The enzyme is not eaten as a nutrient: it is a tool, and it is used up in the sense that it stays in the cheese in tiny amounts, where it keeps working slowly during ripening, adding flavour over weeks and months.

Sweet proteins: sweetness without sugar

Sweet proteins are proteins that taste sweet, some of them hundreds or thousands of times sweeter than sugar by weight. Thaumatin, from the fruit of a West African plant, is about 2,000 times sweeter than sugar. Brazzein, from an African fruit, is about 500 to 2,000 times sweeter. Monellin, from another African fruit, is about 1,500 to 3,000 times sweeter. Because they are so sweet, they are used in milligrams, not grams, and they add almost no energy to a food.

Sweet proteins taste different from sugar. They come on slowly, they linger, and some carry a liquorice note that must be masked. They are also proteins, so they break down when heated strongly, which limits their use in baking, and they are digested like any protein, so they are not a problem for people who must limit sugar. Their strength is in cold foods and drinks, where a tiny amount gives sweetness without sugar, and where the slow onset is less noticed.

Making them by fermentation solves the supply problem. The plants that make them grow in small regions, fruit rarely, and give little protein per fruit, so the natural supply could never fill a factory. A yeast given the gene for thaumatin can make it by the kilogram in a tank, and the purified protein is the same molecule as the one in the fruit. The safety file for a sweet protein is strict, because it will be eaten by many people, including children, and the tests cover allergy, digestion and long term feeding, as Unit 6 explains.

Colours from fermentation

Food colours made by fermentation are replacing synthetic dyes in many products, because shoppers prefer colours that come from nature. The best known is leghemoglobin, the red protein from soy roots, which gives a plant based burger its meaty colour and a note of meaty flavour. It is made by a yeast given the gene for it, then purified and added to the burger, where it turns pink when raw and brown when cooked, copying beef.

Other colours come from microbes that make pigments naturally. Red yeast rice, made by growing a mould on rice, gives a red colour used in foods in Asia. Astaxanthin, the pink pigment that colours salmon, is made by a yeast and by algae, and is used in fish feed and in supplements. Beta carotene, the orange pigment of carrots, is made by a fungus and by algae, and is used to colour drinks, margarine and cheese. Each pigment is a molecule the microbe makes for its own reasons, and the maker harvests it for colour.

A colour must do more than look right on the day it is made: it must stay stable on the shelf, in light, in acid, and in heat. A red that fades in a week is useless, and a pink that turns grey in acid cannot go into a drink. Makers test stability in the actual food, over the full shelf life, and they name the colour on the label by its proper name. A colour that is new to the diet is tested for safety before sale, just like a new protein.

Fats and oils by fermentation

Fats made by fermentation are a newer field. Some yeasts and algae store large amounts of oil inside their cells, up to half their dry weight, and that oil can be harvested and used as a food fat. The oil of a single celled alga can copy the fat of cocoa butter, melting near body temperature, which suits chocolate. The oil of another yeast can be rich in oleic acid, the main fat of olive oil, and can be used in cooking. The attraction is a fat that is made in a tank, without a plantation or an animal, and that can be tuned by choosing the microbe and the feed.

The process is simple in outline and costly in detail. The microbe is grown until it has stored oil, the cells are broken open, the oil is taken out with a solvent or by pressing, and the oil is cleaned and refined, just like a vegetable oil. The cost is high because the cells must be grown, dried and broken open, and because the oil yield per litre is modest. That is why fermentation fats are used today in high value foods, such as a cocoa butter alternative for chocolate, rather than in bulk cooking oil.

A second route to fats is through enzymes rather than through oil itself. Lipases made by fermentation can rearrange the fats in a vegetable oil, moving fatty acids between positions on the fat molecule to change how the fat melts. That rearranged fat can copy the melting of butter or of cocoa butter, without any new oil being made. The enzyme is the tool, the vegetable oil is the starting material, and the product is a fat with a new behaviour, made by a fermentation product rather than by a fermentation itself.

Vitamins and small molecules

Precision fermentation also makes vitamins and other small molecules that foods need. Vitamin B2, riboflavin, is made by a fungus and by bacteria in tanks, and it is used to fortify foods and to colour them yellow. Vitamin B12, which plants do not make, is made by bacteria in tanks and is added to plant based milks and to supplements. Vitamin C is made by a fermentation step in its production, and several amino acids, including lysine and glutamic acid, are made by bacteria in large tanks and used in foods and feeds.

Glutamic acid deserves a note, because it is the source of monosodium glutamate, the flavour enhancer used in savoury foods around the world. It is made by a bacterium called Corynebacterium glutamicum, grown on sugar, and the bacterium releases glutamic acid into the liquid in large amounts. The purified product is the same molecule as the glutamic acid in tomatoes and cheese, and it gives the same umami taste. The scale is large: amino acid fermentation is one of the biggest fermentation industries in the world, and it shows what precision fermentation can do when the product, the strain and the market are all mature.

These small molecules matter for alternative proteins too. A plant based food that lacks vitamin B12 can be fortified with fermentation made B12. A food that lacks lysine can be balanced with fermentation made lysine. The fermentation tank is therefore not only a source of proteins and colours: it is a source of the small pieces that complete a food, and a maker who understands both uses the tank twice.

Choosing the right ingredient for the job

A maker who needs a sweet taste, a red colour, or a setting enzyme faces the same choice every time: make it by fermentation, take it from a plant or an animal, or make it by chemistry. The choice weighs cost, purity, supply, label and function. A fermentation made enzyme is usually purer and more consistent than one taken from an animal, and its supply does not depend on a slaughterhouse. A fermentation made colour is usually dearer than a synthetic dye, but its label reads better to a shopper who prefers natural colours.

Function decides many choices. An enzyme that must work at 80 degrees Celsius needs a heat stable variant, and precision fermentation can make one by changing the gene. A sweet protein that must survive pasteurisation needs a heat stable sweet protein, and only some of them qualify. A colour that must survive a year on a shelf needs a stable pigment, and only some pigments pass. The maker tests the candidate in the real food, under real conditions, and picks the one that survives.

The future of these small ingredients is wider use at lower cost. As more strains are built and more tanks are filled, the price of a sweet protein or a fermentation fat will fall, and foods that today use sugar or palm oil may use them instead. The change will be quiet, because these ingredients work in milligrams and in drops, and the shopper will see a shorter ingredient list and a cleaner label, not a new food. That quiet change is the real promise of precision fermentation: better ingredients, not louder claims.

Sweetness by numbers: comparing sweet proteins

A sweet protein is judged by its potency: how much sweeter it is than sugar, gram for gram. The potency is measured by tasting: a panel tastes a solution of the sweet protein and a solution of sugar, and the concentrations are adjusted until the two taste equally sweet. If 1 g of the sweet protein in a litre tastes as sweet as 2,000 g of sugar in a litre, the potency is 2,000 times sugar. That number is the whole business case, because a protein that is 2,000 times sweeter than sugar is used in milligrams, not grams, and a small tank can sweeten a large amount of food.

A worked example shows the use. A drink maker wants a drink as sweet as one holding 50 g of sugar per litre. With a sweet protein of potency 2,000, the drink needs 50 divided by 2,000, which is 0.025 g per litre, or 25 mg per litre. A 10,000 litre batch of drink therefore needs 250 g of sweet protein, less than a bag of sugar, to sweeten the whole batch. The maker checks the number by tasting, because potency changes with temperature, with pH, and with the other flavours in the drink, and a potency measured in water may be lower in a sour drink.

The limits matter as much as the potency. A sweet protein may taste sweet slowly, arriving after the sugar would have, and it may leave a lingering taste that some eaters dislike. It may lose its sweetness when heated, because the protein unfolds and the shape that tastes sweet is lost. It may cost far more per kilogram than sugar, so the saving in weight must outweigh the price per kilogram. A sweet protein that is potent, heat stable and clean tasting is worth a fortune. A sweet protein that is potent only in cold water is worth little, and the tasting panel, not the potency number, decides which one it is.

Questions to think about

  • Why is a tiny dose of chymosin enough to set a whole vat of milk?
  • Thaumatin is about 2,000 times sweeter than sugar. Why is it still hard to use in a baked cake?
  • Why might a maker choose a fermentation made colour over a cheaper synthetic dye, and what must the colour prove first?
Think about the thaumatin question before you read on

Thaumatin is a protein, and proteins unfold and break down when heated strongly, so baking can destroy its sweetness. It also comes on slowly and lingers, with a liquorice note, so it suits cold drinks where a tiny amount gives sweetness without sugar, but it does not copy sugar in a hot oven. Heat stability and taste timing, not sweetness alone, decide where it can be used.

From microbe to small ingredient

Choose the target: enzyme, sweet protein, colour, fat or vitamin
↓
Give the host microbe the gene or use a natural producer
↓
Grow the microbe in a tank on sugar feed
↓
Harvest the liquid or break the cells open
↓
Purify the ingredient and test it in the real food
↓
Sell it in milligrams and drops, not in kilograms

The diagram shows the small ingredient route in order. The last panel is the point: these ingredients work in tiny amounts, so a small tank can supply many factories.

Practical note: In a practical session, test lactase, the enzyme that cuts lactose: buy a lactase tablet from a pharmacy and a carton of milk. Stir a crushed tablet into a glass of milk, keep it in the fridge overnight, and taste it the next day beside a glass of untreated milk. Note which one tastes sweeter and why: lactase cuts lactose into glucose and galactose, and glucose tastes sweeter than lactose. Record the time, the temperature and the taste difference, and answer: why is lactose free milk sweeter than ordinary milk even though no sugar was added?

Quick self-check

Question: What is an enzyme, and why is a small amount enough?

An enzyme is a protein that speeds up a chemical change without being used up itself, so a small amount can change a large amount of food. A few drops of chymosin set a whole vat of milk because each enzyme molecule acts again and again.

Question: Why are sweet proteins used in milligrams rather than grams?

Because they are hundreds or thousands of times sweeter than sugar by weight, so a tiny amount gives the sweetness of a large amount of sugar, with almost no energy added to the food.

Question: A drink needs the sweetness of 50 g of sugar per litre, and a sweet protein is 2,000 times sweeter than sugar. How much sweet protein is needed per litre?

50 divided by 2,000, which is 0.025 g per litre, or 25 mg per litre. A 10,000 litre batch needs 250 g in total.

Try it yourself

Buy three foods from a shop: a lactose free milk, a cheese, and a fruit juice that says no added colour or a natural colour on the label. Read the ingredient list of each and find the enzyme, the colour or the culture named. Write down the exact words used and answer: can you tell from the label whether the enzyme or colour was made by fermentation, taken from a plant, or made by chemistry? Write one sentence for each product about what the label tells you and what it leaves out.

Key ideas to carry forward

  • Enzymes are proteins that speed up a change without being used up, so a tiny dose can set a vat of milk or clear a juice.
  • Precision fermentation improves enzymes by using a higher yielding host and by changing the gene for heat or acid tolerance.
  • Sweet proteins such as thaumatin are thousands of times sweeter than sugar, work in milligrams, and break down in strong heat.
  • Fermentation made colours such as leghemoglobin copy the look of meat, but they must prove stability on the shelf.
  • Some yeasts and algae store oil that can copy cocoa butter or olive oil, but the cost still limits them to high value foods.
  • Vitamins B2 and B12 and amino acids such as lysine and glutamic acid are made by fermentation at large scale.
  • Small ingredients are chosen on cost, purity, supply, label and function, and they are tested in the real food before use.

  • Potency, how many times sweeter a protein is than sugar, decides the dose, and the dose is checked by tasting in the real drink.

Glossary

Amylase

An enzyme that cuts starch into sugars. Food tech example: amylase in bread making turns starch into sugars that feed the yeast.

Astaxanthin

A pink pigment made by yeast and algae that colours salmon. Food tech example: astaxanthin is added to fish feed so farmed salmon flesh turns pink.

Beta carotene

An orange pigment from carrots, fungi and algae used as a colour. Food tech example: beta carotene colours margarine and drinks orange.

Brazzein

A sweet protein from an African fruit, hundreds of times sweeter than sugar. Food tech example: brazzein made by fermentation sweetens a drink at tiny amounts.

Chymosin

The enzyme that sets milk into curd for cheese making. Food tech example: chymosin made by fermentation sets a vat of milk in about 30 minutes.

Lactase

An enzyme that cuts lactose into glucose and galactose. Food tech example: lactase is added to milk to make lactose free milk.

Leghemoglobin

A red protein from soy roots that colours a plant based burger like beef. Food tech example: leghemoglobin made by fermentation turns a raw burger pink.

Lipase

An enzyme that cuts fats. Food tech example: lipase made by fermentation rearranges vegetable oil fats to change how they melt.

Monellin

A sweet protein from an African fruit, thousands of times sweeter than sugar. Food tech example: monellin made by fermentation is tested as a sweetener for cold drinks.

Protease

An enzyme that cuts proteins. Food tech example: protease is used to soften meat and to clear haze from beer.

Riboflavin

Vitamin B2, made by fermentation and used to fortify foods. Food tech example: riboflavin made by a fungus is added to foods and colours them yellow.

Thaumatin

A sweet protein about 2,000 times sweeter than sugar. Food tech example: thaumatin made by fermentation sweetens a drink with almost no energy.

Umami

The savoury taste found in mushrooms, soy sauce and yeast extract. Food tech example: glutamic acid made by fermentation gives umami to savoury foods.

Corynebacterium glutamicum

A bacterium that makes glutamic acid in large tanks. Food tech example: Corynebacterium glutamicum is the source of most monosodium glutamate.

Pectinase

An enzyme that breaks down pectin and clears haze from fruit juice. Food tech example: pectinase made by fermentation clears apple juice.

Potency

How many times sweeter a substance is than sugar, gram for gram. Food tech example: a sweet protein with a potency of 2,000 is used in milligrams per litre.

Lingering taste

A taste that stays in the mouth after swallowing. Food tech example: a sweet protein with a lingering taste suits a drink less well than one that tastes clean.

References

  • Stanbury, P. F., Whitaker, A. and Hall, S. J., Principles of Fermentation Technology, 3rd edition, 2017.
  • Bamforth, C. W., Food, Fermentation and Micro-organisms, 2005.
  • Waites, M. J., Morgan, N. L., Rockey, J. S. and Higton, G., Industrial Microbiology: An Introduction, 2001.
  • Demain, A. L. and Davies, J. E., Manual of Industrial Microbiology and Biotechnology, 2nd edition, 1999.
  • Damodaran, S., Parkin, K. L. and Fennema, O. R., Fennema's Food Chemistry, 4th edition, 2007.
  • Fellows, P. J., Food Processing Technology, 4th edition, 2017.

Unit 5: From Lab to Factory: Bioreactors and Scale-Up

Note

Unit 5: From Lab to Factory: Bioreactors and Scale-Up

A question to start with

In a laboratory, a scientist grows a yeast in a flask the size of a coffee cup and makes a few grams of protein. In a factory, the same yeast grows in a steel tank the size of a house and makes tonnes. Between those two sits the hardest step in precision fermentation: scale up. A flask is shaken by hand, a tank is stirred by a motor, and the physics changes with size. This unit follows the product from the bench to the factory: the tanks, the steps that grow larger, the problems that appear only at scale, and the checks that prove each batch is the same as the last.

The bioreactor: a tank that breathes

A bioreactor, also called a fermenter, is a steel tank built to grow microbes under controlled conditions. It holds the liquid, called the broth, where the cells live. It has a stirrer, a shaft with blades that turns and mixes the broth. It has a sparger, a ring that bubbles air into the broth from below. It has a cooling jacket, a layer around the tank where cold water flows to carry away the heat the cells release. And it has ports for adding feed, for taking samples, and for probes that measure temperature, pH and dissolved O₂.

The size range is wide. A laboratory tank holds 1 to 10 litres. A pilot tank holds 100 to 1,000 litres. A production tank holds 10,000 to 200,000 litres. The largest tanks are as tall as a house and hold as much liquid as a small swimming pool. Each size needs its own design: a small tank can be shaken or stirred simply, while a large tank needs a powerful motor, a strong shaft, and a cooling system that can remove the heat of millions of cells growing at once.

Sterility is built into the design. The tank is made of smooth stainless steel with no corners where microbes can hide. It is sterilised with steam before each run, at about 121 degrees Celsius for 20 to 30 minutes, which kills any microbe left from the last run. Every pipe, valve and seal is built to keep the inside sterile while feed flows in and samples flow out. A tank that cannot be sterilised cannot be used for food, whatever its size.

Scale up: growing larger in steps

Scale up never jumps from a flask to a production tank in one step. The culture is grown through a chain of tanks, each about ten times larger than the last: a flask, then a 10 litre tank, then a 100 litre tank, then a 1,000 litre tank, then the production tank. Each step takes a day or less, and at each step the maker checks that the cells are growing well, that the product is being made, and that no contamination has appeared. A failure at a small step costs a little. A failure at the production step costs a whole tank of feed and a week of time.

The reason for the steps is that a microbe's world changes with size. In a flask, O₂ reaches every cell quickly, heat escapes through the glass, and mixing is easy. In a large tank, O₂ must travel through metres of liquid, heat builds up in the centre, and mixing takes time, so some cells sit in a rich zone near the feed inlet while others sit in a poor zone near the wall. The maker must adjust the stirring speed, the air flow and the feed rate at each size to keep the conditions even.

A common rule is to keep one factor constant while scaling up, often the power used for stirring per litre, or the amount of O₂ supplied per litre per minute. Keeping one factor constant lets the others shift, and the maker watches the product to see whether the shift matters. The test at each scale is the same: does the strain make the same amount of protein, of the same quality, at the same rate? If the answer is yes, the scale up is working. If the answer is no, the maker changes the stirring, the air or the feed and tries again.

Feeding the tank: batch, fed batch and continuous

A tank can be fed in three ways. In a batch run, all the feed goes in at the start, the cells grow until the feed runs out, and the tank is then emptied. Batch is simple and easy to clean, but the cells spend part of the run starving, and the product is made in a burst. In a fed batch run, the feed is added slowly during the run, so the cells keep growing steadily and the product keeps being made. Fed batch is the most common way of running a food fermentation, because it gives high yields and good control.

In a continuous run, fresh feed flows in all the time and broth flows out all the time, at the same rate, so the tank holds a steady level and the cells stay in the same state for days or weeks. Continuous running gives a steady product and uses the tank well, but it is harder to keep sterile for a long time, and a contamination can spoil weeks of output. Mycoprotein is made continuously, and most precision fermentation proteins are made by fed batch, because fed batch balances yield against risk.

A worked example shows the feed arithmetic. A tank holds 100,000 litres of broth, and the cells need 20 g of sugar per litre to grow and make protein. The total sugar needed is 100,000 multiplied by 20 g, which is 2,000 kg. If the sugar syrup holds 50 percent sugar, the tank needs 4,000 kg of syrup. That syrup must be bought, stored, sterilised and pumped, and its price is a large part of the cost of the finished protein, which is why the choice of sugar source matters so much, as Unit 7 explains.

Downstream processing: taking the product out

Downstream processing is everything that happens after the tank: separating the product from the cells and the broth, and turning it into a clean, dry ingredient. The steps depend on where the product sits. If the protein is released outside the cell, the first step is to remove the cells, by spinning the broth in a centrifuge or by passing it through a filter. If the protein stays inside the cell, the cells must be broken open first, by a homogeniser that forces them through a narrow gap at high pressure, or by an enzyme that dissolves the cell wall.

Next comes purification. The liquid is passed through filters that sort molecules by size, a step called ultrafiltration, which keeps the protein and lets water and small molecules pass. It may then pass through chromatography, where the protein sticks to a material in a column and is washed off clean, leaving other molecules behind. Each step removes more of what is not wanted, and each step loses a little of what is wanted, so the maker balances purity against yield: a purer product costs more, because more of it is lost on the way.

The last step is drying. Spray drying blows the liquid into a hot chamber as a fine mist, and the water evaporates, leaving a powder. Freeze drying freezes the liquid and removes the water as vapour, which is gentler but dearer. The powder is then tested, packed in sealed bags, and stored cool and dry. A powder that picks up moisture in storage will clump, lose function, and grow microbes, so the packing and the warehouse matter as much as the tank.

What goes wrong at scale

Scale up brings problems that a flask never shows. The first is O₂ starvation: in a large tank, the cells near the top may get plenty of air while the cells near the bottom get little, and the starved cells make less protein or make the wrong products. The fix is stronger stirring and more air, but stronger stirring can damage fragile cells and costs more power. The second is heat: a large tank releases far more heat than a small one, and the cooling jacket must remove it, or the temperature will rise and the cells will slow down or die.

The third is foam. Proteins and cells make foam when air is bubbled through the broth, and a foaming tank can push broth out of the top, losing product and making a mess. Anti foam agents are added in small amounts to break the foam, but too much anti foam can slow O₂ transfer, so the dose is tuned carefully. The fourth is contamination, which is more costly at scale: a contaminated flask loses a day, while a contaminated production tank loses a week and a large amount of feed.

The fifth is variation between batches. A strain that makes 5 g of protein per litre in a flask may make 3 g in a production tank, because the conditions are not the same. The maker must find why, change the feed, the air or the stirring, and test again. That cycle, test, measure, change, test again, is the real work of scale up, and it is why a product that works in a laboratory may take years to reach a factory. A company that skips the pilot steps and jumps straight to a large tank usually learns this lesson at the highest price.

Quality checks on every batch

Every batch of a fermentation made ingredient is tested before it is sold. The tests check identity: is the protein the right molecule, with the right amino acids in the right order? They check purity: how much of the powder is the target protein, and how much is other protein, salt, or water? They check safety: are there microbes, toxins, or traces of the host left in the product? And they check function: does the protein foam, gel, or set as it should in a simple test?

The tests use instruments that a food laboratory will know well. Mass spectrometry weighs the protein and confirms its identity. Chromatography separates the molecules and measures how much of each is present. Electrophoresis sorts proteins by size on a gel, showing whether the product is one clean band or a smear of many proteins. Microbial tests count bacteria and check for specific pathogens. Each test has a limit, written on the specification sheet, and a batch that fails any limit is held, retested, or discarded.

Records tie the whole chain together. Each batch carries a number that links it to the cell bank vial it came from, the tank it grew in, the feed lot, the date, and the people who ran it. If a problem appears in a finished food, the maker can trace it back to the batch, the tank and the feed, and can find whether other batches share the problem. That traceability is not paperwork for its own sake: it is what lets a factory recall a faulty batch and prove that the next batch is sound.

Oxygen transfer: a step-by-step derivation

A large tank can starve its cells of oxygen, and the starvation is described by a simple equation that every fermentation engineer knows: the oxygen transfer rate equals kLa multiplied by the difference between the oxygen the liquid could hold and the oxygen it does hold. The symbols are worth learning one by one. OTR is the oxygen transfer rate, how much oxygen moves into the liquid each hour. kLa is the transfer coefficient, which measures how well the tank moves oxygen, and it rises with stronger stirring and more air. C* is the saturation level, the most oxygen the liquid can hold, and C is the level it does hold, so C* minus C is the gap that drives oxygen in.

The derivation is a balance. Oxygen moves into the liquid because the gap drives it, and the cells eat it as fast as it arrives. When the cells eat faster than the tank can supply, C falls towards zero, the gap grows to its largest, and the transfer rate reaches its limit, which is kLa multiplied by C*. That limit is the most oxygen the tank can ever give, whatever the cells demand. A tank with a small kLa has a low limit, and its cells will starve when they grow many and demand much. A tank with a large kLa has a high limit, and it can feed more cells.

A worked example shows the use. A tank has a kLa of 100 per hour and a saturation level of 8 mg per litre, so its limit is 100 times 8, which is 800 mg per litre per hour. If the cells demand 1,000 mg per litre per hour, the tank cannot feed them, and the cells will starve, make less protein, or make the wrong products. The maker has three fixes: raise the stirring to lift kLa, raise the air flow to lift kLa, or slow the feed so the cells demand less. The equation tells the maker which fix to try, and how far it must go, before the tank is ever built.

Questions to think about

  • Why is scale up done in steps of about ten times, rather than jumping from a flask to a production tank?
  • A protein released outside the cell is easier to harvest than one kept inside. Why, and what extra step does the inside protein need?
  • Why does a large tank run out of O₂ in some places while a flask does not, and what does the maker do about it?
Think about the O₂ question before you read on

In a flask, air reaches every cell quickly because the liquid is shallow and well mixed. In a large tank, O₂ must travel through metres of broth, and cells near the bottom or far from the air inlet can starve while cells near the top have plenty. The maker fixes it with stronger stirring, more air through the sparger, and a slower feed rate so the cells demand less O₂ at once.

From flask to factory

Grow the strain in a flask and test it
↓
Step up to a 10 litre tank and check growth and yield
↓
Step up to a 100 litre pilot tank and tune stirring and air
↓
Step up to a 1,000 litre tank and test the full process
↓
Run the production tank by fed batch
↓
Harvest, purify, dry and test every batch

The diagram shows scale up in steps of about ten times. Each step is checked before the next begins, because a failure at a small step is cheap and a failure at production scale is dear.

Practical note: In a practical session, compare mixing in a small and a large container: put a drop of food colouring into a small glass of water and into a large jug of water, stir each once with a spoon, and count how many stirs each one needs before the colour is even. Then add a spoonful of honey to each and repeat: note how the thick liquid mixes more slowly. Record the number of stirs for each container and answer: why does a large tank need a stronger stirrer than a small flask, and what happens to cells in the parts of a tank that are mixed last?

Quick self-check

Question: What is the difference between batch, fed batch and continuous running?

In a batch run all the feed goes in at the start and the tank is emptied when it runs out. In a fed batch run feed is added slowly during the run so cells keep growing steadily. In a continuous run fresh feed flows in and broth flows out all the time, holding a steady state for days or weeks.

Question: Why is downstream processing a balance between purity and yield?

Because each purification step removes more of what is not wanted but also loses a little of the product. A purer ingredient costs more, because more of it is lost on the way, so the maker balances how pure the product must be against how much can be afforded to lose.

Question: A tank has a kLa of 100 per hour and a saturation level of 8 mg per litre. What is the most oxygen it can transfer per litre per hour?

100 times 8, which is 800 mg per litre per hour. If the cells demand more than that, they will starve, and the maker must raise the stirring, raise the air, or slow the feed.

Try it yourself

Visit a brewery, a yoghurt plant, or a bakery that ferments on site, if one near you offers a tour or an open day, and ask three questions: how large are the tanks, how do they keep them sterile, and how do they check each batch? If no tour is available, find a video or a written tour of a fermentation plant and write down the tank sizes, the sterilisation step, and the batch tests you find. Answer in writing: which step in the plant most resembles the laboratory flask, and which step exists only at factory scale?

Key ideas to carry forward

  • A bioreactor is a steel tank with a stirrer, a sparger, a cooling jacket and probes, built to be sterilised with steam.
  • Scale up moves in steps of about ten times, checking growth and yield at each size before moving on.
  • Fed batch feeding, adding sugar slowly during the run, is the most common way to run a food fermentation.
  • Downstream processing removes cells, purifies the protein by filtration and chromatography, and dries it into a powder.
  • Large tanks face O₂ starvation, heat build up, foam and contamination, and each one is managed by stirring, air and feed control.
  • Every batch is tested for identity, purity, safety and function, and the batch number traces it back to the cell bank, the tank and the feed.
  • Skipping pilot steps and jumping to a large tank is the dearest way to learn that scale up changes the process.

  • Oxygen transfer rate equals kLa multiplied by the gap between saturation and the level held, and its limit decides how many cells a tank can feed.

Glossary

Bioreactor

A tank where microbes are grown under controlled conditions. Food tech example: a bioreactor holds 100,000 litres of yeast broth for making whey protein.

Broth

The liquid in a bioreactor where the cells live and grow. Food tech example: the broth holds sugar, minerals and yeast cells during a fermentation run.

Centrifuge

A machine that spins a liquid fast to separate cells from liquid. Food tech example: a centrifuge spins fermentation broth to remove yeast cells from the protein liquid.

Chromatography

A separation method that sorts molecules by size or by how they stick to a material. Food tech example: chromatography purifies a fermentation made protein in a column.

Downstream processing

All the steps after the tank that turn broth into a clean ingredient. Food tech example: downstream processing filters, purifies and dries a fermentation made protein.

Fed batch

A run where feed is added slowly during the run. Food tech example: most food proteins are made by fed batch, which gives high yields and good control.

Homogeniser

A machine that forces cells through a narrow gap at high pressure to break them open. Food tech example: a homogeniser breaks yeast cells to release a protein kept inside.

Pilot tank

A medium sized tank used to test a process before production scale. Food tech example: a 100 litre pilot tank tests stirring and air settings before the production run.

Scale up

Growing a process larger, from flask to pilot tank to production tank. Food tech example: scale up moves in steps of about ten times, checking yield at each size.

Sparger

A ring in a bioreactor that bubbles air into the broth. Food tech example: the sparger supplies O₂ to yeast cells growing in the tank.

Spray drying

Drying by blowing a liquid as a fine mist into a hot chamber. Food tech example: spray drying turns a purified protein liquid into a powder.

Freeze drying

Drying by freezing a liquid and removing water as vapour. Food tech example: freeze drying is gentler than spray drying but costs more.

Ultrafiltration

Filtering through a membrane that keeps large molecules and lets water pass. Food tech example: ultrafiltration concentrates a protein liquid before drying.

Mass spectrometry

An instrument that weighs molecules to confirm their identity. Food tech example: mass spectrometry confirms a fermentation made protein is the right molecule.

Traceability

The ability to follow a batch back to its cells, tank, feed and date. Food tech example: a batch number traces a protein powder to the tank it grew in.

Oxygen transfer rate

How much oxygen moves into the liquid each hour. Food tech example: a tank with a low oxygen transfer rate starves its cells.

Saturation level

The most oxygen a liquid can hold. Food tech example: the gap between saturation and the level held drives oxygen into the broth.

References

  • Stanbury, P. F., Whitaker, A. and Hall, S. J., Principles of Fermentation Technology, 3rd edition, 2017.
  • Bamforth, C. W., Food, Fermentation and Micro-organisms, 2005.
  • Waites, M. J., Morgan, N. L., Rockey, J. S. and Higton, G., Industrial Microbiology: An Introduction, 2001.
  • Demain, A. L. and Davies, J. E., Manual of Industrial Microbiology and Biotechnology, 2nd edition, 1999.
  • Damodaran, S., Parkin, K. L. and Fennema, O. R., Fennema's Food Chemistry, 4th edition, 2007.
  • Fellows, P. J., Food Processing Technology, 4th edition, 2017.

Unit 6: Safety, Regulation and Labelling

Note

Unit 6: Safety, Regulation and Labelling

A question to start with

A new ingredient cannot be sold on a promise. Before a fermentation made protein reaches a shop, a file of evidence must convince a food safety authority that it is safe, that it is what it claims to be, and that its label tells the truth. That file is long, the questions are sharp, and the answers must be measured, not asserted. This unit explains how that file is built, what the tests are, how different countries handle the decision, and what the label must say. It is the unit that separates a good idea in a laboratory from a food on a shelf.

The safety file: what it holds

A safety file, also called a dossier, is the set of documents a maker sends to a food safety authority before a new ingredient is sold. It starts with identity: what the ingredient is, its amino acid sequence for a protein, its structure, and how it compares with the natural molecule it copies. Then comes the process: which host microbe was used, how the gene was placed, how the protein was purified, and what remains of the host in the finished ingredient, if anything.

Next comes composition: the purity of the ingredient, the amounts of other proteins, salts, water and any by-products, and the results of tests for toxins, heavy metals and microbes. Then comes the allergy check: a comparison of the protein's structure with known allergens, and tests of whether blood from allergic people reacts to it. Then comes the feeding study: animals fed the ingredient for weeks or months, watched for any harmful effect, with the doses set far above what a person would eat.

Finally comes the intake estimate: how much of the food people will eat, how much of the ingredient that means per day, and how that compares with the doses that were safe in the animals. If the expected intake is far below any level that caused harm, and the allergy check is clear, the authority may accept the ingredient. The file is not a formality: authorities ask questions, ask for more tests, and may refuse, and a maker who cannot answer a question does not get to sell.

Allergy: the sharpest question

Allergy is the sharpest question in the file, because a protein that is new to the diet may resemble a protein that already causes allergy. The check starts on paper: the amino acid sequence of the new protein is compared with the sequences of known allergens, and any close match is flagged. A close match does not prove the new protein causes allergy, but it means the maker must test further, and must be ready to label the product with a warning.

The next check is in the laboratory: blood serum from people allergic to the matched allergen is mixed with the new protein, and the test looks for a reaction. If the serum reacts strongly, the protein is treated as a risk, and the maker must either change the process, lower the use level, or label the product clearly. If the serum does not react, the risk is lower, but the file must still record the test and its result.

A special case is a protein that copies a known allergen, such as a fermentation made milk protein. There the answer is already known: the protein is the same molecule as the milk protein, so a person allergic to milk can react to it, and the label must name it as a milk protein. No test can change that fact, and no marketing name can hide it. The label is the protection, and it must be plain.

How countries decide: general principles

Countries decide about new foods in different ways, but the principles are shared. Every country has a food safety authority that reads the file, and every country asks the same core questions: what is it, how is it made, is it safe, and does the label tell the truth? The names of the authorities differ, the procedures differ, and the time taken differs, but the core questions do not. A maker who sells in several countries must satisfy each authority separately, which is why a product is often launched in one market first.

Some countries treat a fermentation made ingredient as a new food and ask for a full file before sale. Others treat it as a familiar ingredient when the molecule is identical to one already eaten, and ask for a shorter file that proves the identity and the purity. Still others have a system where the maker notifies the authority and the authority can object within a set time. The differences matter for business, because they set how long a launch takes and how much it costs, but they do not change the science: the tests are the same, and a protein that fails a safety test in one country will fail it in another.

No single country's rule is the world's rule, and this subject does not present any one country's procedure as universal. The reader who needs the current rule for their own country should check the food safety authority of that country, because rules change, and a rule that was true last year may have been revised. The general principles in this unit, identity, purity, allergy, feeding safety and honest labelling, are the stable core that every system shares.

Labelling: what the pack must say

A label for a food made with a fermentation made ingredient must do three things: name the ingredient, name the allergen, and not mislead the shopper. The ingredient is named by what it is, such as whey protein or thaumatin, in the ingredient list, in descending order of weight. The allergen is named plainly where the law asks, such as milk protein for a fermentation made whey, so that an allergic shopper can avoid it. The front of the pack must not suggest something the food is not: a drink made with fermentation made whey and oat milk cannot be called milk without qualification, and a product with no cow in it cannot hide that fact from a shopper who cares.

The words animal free, made without animals, and fermentation made are all used on packs, and their meaning is being settled in different countries at different speeds. Some authorities allow them, some ask for a clearer statement, and some have not decided. A maker must check the current rule in each market, and must be ready to change the wording. What the maker cannot do is use a word that suggests the product is something else: a protein that is a milk protein cannot be sold as if it were not one.

Claims about nutrition and about the planet follow the same rule as in the rest of this subject: a claim must meet a legal definition and be backed by measurement. A claim of high in protein must meet the threshold for that claim. A claim of lower carbon must be backed by a life cycle assessment of the whole product, not only of the ingredient. A vague claim such as greener or better for the planet, without numbers, is not a claim a careful maker should print, and it is increasingly treated as misleading by authorities.

Inside the factory: checks that never stop

Approval is not the end of safety: it is the start of a routine that runs every day the factory operates. Each delivery of sugar, minerals and other feeds is checked against its specification sheet. Each run is watched for temperature, pH, O₂ and contamination, and samples are taken and tested. Each batch of finished ingredient is tested for identity, purity, microbes and function, and a batch that fails any limit is held, retested or discarded.

The factory also runs a system called HACCP, hazard analysis and critical control points, which maps every step of the process, names the hazards at each step, and sets a control for each one. A critical control point might be the steam sterilisation of the tank, the filtration that removes cells, or the drying temperature that kills microbes. Each control point has a limit, a check, and a written action for what happens if the limit is missed. The system is checked by inspectors, and the records must be ready to show at any time.

Traceability runs through the whole system. Each batch of ingredient carries a number that links it to the cell bank vial, the tank, the feed lots, the date and the people who ran it. Each batch of finished food that uses the ingredient carries its own number, linked back to the ingredient batch. If a problem appears, the maker can find every food that used the faulty batch, recall it, and prove that the next batch is sound. That chain, from vial to tank to food to shop, is what makes a new ingredient trustworthy.

Trust and the shopper

A new food technology succeeds or fails on trust, and trust is earned by openness. A maker who publishes the safety file, or a summary of it, who names the host microbe, who explains the purification, and who labels the allergen plainly, gives the shopper the facts to decide. A maker who hides the method, who uses a vague name, or who prints a footprint claim without numbers, asks the shopper to trust a slogan, and slogans do not survive a hard question.

The history of chymosin is a useful guide. It has been made by fermentation since the 1990s, it is used in most cheese making, and shoppers eat it every day without a second thought, because it works, it is safe, and its use is open. The lesson is that a technology becomes ordinary when it is proven, labelled and used well over time. Precision fermentation is on the same path: each ingredient that passes its tests, is labelled plainly, and tastes right in the food, makes the next one easier to accept.

The shopper's part in this is simple and powerful: read the label, ask what the ingredient is, check the allergen, and judge the claim against the numbers. A shopper who does that for every new food is doing the same work as the authority, on a smaller scale, and a food system where shoppers read labels is a food system where makers must tell the truth. This unit has given you the tools for that reading, and the next unit shows what the numbers say about cost and about the planet.

Tracing a batch: a worked example

Traceability is proved by tracing a batch, and the best way to learn it is to trace one by hand. A shopper complains about a protein powder, and the maker must find which batch it came from, what went into it, and whether other packs share the problem. The batch number on the pack is the start: it links to the packing record, which names the drying run, which names the purification run, which names the fermentation run, which names the tank, the cell bank vial, the feed lot and the date. Each link is a record, and the chain is only as strong as its weakest record.

A worked example shows the chain. Batch P-104 was packed on 12 May from drying run D-88, which dried purification run C-71, which purified fermentation run F-52. Run F-52 grew in tank 3 from cell bank vial W-19, fed with sugar lot S-203 and minerals lot M-44, run by two operators whose names are on the record. The maker checks the other batches from run F-52: batches P-103 and P-105 share the same fermentation run, so they are held and tested too, while batch P-201, from a different run, is not affected. The trace takes an hour if the records are in order, and a week if they are not.

The exercise that proves the system is a mock recall: the maker picks a batch at random, traces it back to its vial and its feed, and then traces forward to every pack that shares any part of it, without warning the team in advance. A factory that passes its mock recall can recall a real fault quickly, and can prove to a customer that the next batch is sound. A factory that fails its mock recall learns the lesson cheaply, in a drill, rather than dearly, in a real recall. The batch number on a pack is not paperwork. It is the thread that pulls the whole chain back.

Questions to think about

  • Why must a safety file cover both the finished ingredient and the process that made it?
  • A fermentation made whey protein is the same molecule as milk whey. Why can no test change the fact that it must be labelled as a milk protein?
  • Why do different countries take different amounts of time to approve the same ingredient, even though they ask the same core questions?
Think about the safety file question before you read on

The finished ingredient is what the shopper eats, so it must be tested for identity, purity, toxins and allergy. The process is how it was made, so it must be checked for the host microbe, the gene, the purification, and what remains of the host. A pure ingredient from a dirty process is not safe, and a clean process that makes the wrong molecule is not the ingredient claimed. Both checks are needed.

How a new ingredient is approved

Build the file: identity, process, composition
↓
Test allergy risk against known allergens
↓
Run feeding studies in laboratory animals
↓
Estimate how much people will eat each day
↓
Send the file to the food safety authority
↓
Answer questions, run more tests if asked, wait for acceptance

The diagram shows the approval chain in order. The authority sees the file before the food reaches the shop, and it can ask for more tests or refuse, so the file must be complete and honest.

Practical note: In a practical session, take the label of a food that uses a fermentation made ingredient, if one can be found, or of a cheese made with fermentation chymosin. Write down the exact words used for the ingredient, the allergens named, and any claim on the front of the pack. Then check each claim against the numbers on the back: does a high in protein claim meet the usual threshold, and does the ingredient list support the front of pack words? Present your findings as a short table and note any gap between the front and the back.

Quick self-check

Question: What is HACCP, and what does a critical control point do?

HACCP is a system that maps every step of a process, names the hazards at each step, and sets a control for each one. A critical control point is a step where control is essential, such as steam sterilisation of a tank, with a limit, a check, and a written action if the limit is missed.

Question: Why must the label name a fermentation made whey protein as a milk protein?

Because the protein is the same molecule as the milk protein, so a person allergic to milk can react to it. The allergy is to the molecule, not to the animal, so the label must warn allergic shoppers plainly.

Question: Why does a mock recall trace a batch both backwards and forwards?

Backwards to find what went into it, the vial, the tank, the feed and the date, and forwards to find every pack that shares any part of it, so all affected packs are held and tested, not only the one that was complained about.

Try it yourself

Go to a shop and pick up five foods that make a claim on the front of the pack: high in protein, lower sugar, natural, or a planet claim. For each one, write down the claim, then turn the pack over and write down the numbers that would support it: protein per 100 g, sugar per 100 g, or the ingredient list. Judge each claim in one sentence: is it backed by the numbers on the back, or does the back tell a different story? Keep your notes and compare them with a classmate who shopped in a different store.

Key ideas to carry forward

  • A safety file covers identity, process, composition, allergy, feeding studies and expected intake, and the authority reads it before sale.
  • Allergy is checked by comparing the protein sequence with known allergens and by testing blood from allergic people.
  • Countries share the same core questions but differ in procedure and timing, and no single country's rule is the world's rule.
  • Labels must name the ingredient, name the allergen, and not mislead the shopper about what the food is.
  • A claim on a pack must meet a legal definition and be backed by measurement, including planet claims.
  • HACCP maps hazards at each step and sets a control point with a limit, a check and a written action.
  • Trust is earned by openness: naming the host, explaining the process, and labelling the allergen plainly.

  • Traceability links a pack to its vial, tank, feed and date, and a mock recall proves the chain works before a real fault appears.

Glossary

Dossier

The set of documents sent to a food safety authority before a new ingredient is sold. Food tech example: the dossier for a fermentation made protein holds identity, process, composition and feeding studies.

HACCP

Hazard analysis and critical control points, a system that maps hazards and sets controls. Food tech example: steam sterilisation of a tank is a critical control point in a HACCP plan.

Critical control point

A step where control is essential for safety. Food tech example: filtration that removes cells is a critical control point in making a fermentation protein.

Allergen

A food that can trigger an allergic reaction in some people. Food tech example: milk protein is an allergen that must be named on a label even when made by fermentation.

Feeding study

A study where laboratory animals are fed a new food to look for harmful effects. Food tech example: a new protein is fed to animals for weeks before it is accepted for sale.

Intake estimate

A calculation of how much of an ingredient people will eat each day. Food tech example: the intake estimate for a sweet protein counts every food and drink that uses it.

Specification sheet

A document that lists what a delivered ingredient must meet. Food tech example: each batch of fermentation made protein is checked against its specification sheet.

Traceability

The ability to follow a batch back to its cells, tank, feed and date. Food tech example: a batch number traces a protein powder to the tank it grew in.

Animal free

A label term for a food made without an animal. Food tech example: an animal free ice cream uses fermentation made whey instead of dairy whey.

New food

A food or ingredient that has not been eaten widely before and needs approval. Food tech example: a fermentation made sweet protein is treated as a new food in many countries.

Food safety authority

The body in a country that judges whether a new food is safe. Food tech example: the maker sends the safety file to the food safety authority before sale.

Serum

The liquid part of blood used in allergy tests. Food tech example: serum from people allergic to milk is tested against a fermentation made whey protein.

Life cycle assessment

A method that adds up the land, water, energy and gas releases of a product across its whole life. Food tech example: a lower carbon claim for a food must be backed by a life cycle assessment.

Mock recall

A practice recall that traces a batch without warning, to test the system. Food tech example: a factory that passes its mock recall can recall a real fault quickly.

Batch number

A number that links a pack to its production records. Food tech example: a batch number traces a protein powder to the tank it grew in.

References

  • Stanbury, P. F., Whitaker, A. and Hall, S. J., Principles of Fermentation Technology, 3rd edition, 2017.
  • Bamforth, C. W., Food, Fermentation and Micro-organisms, 2005.
  • Waites, M. J., Morgan, N. L., Rockey, J. S. and Higton, G., Industrial Microbiology: An Introduction, 2001.
  • Demain, A. L. and Davies, J. E., Manual of Industrial Microbiology and Biotechnology, 2nd edition, 1999.
  • Damodaran, S., Parkin, K. L. and Fennema, O. R., Fennema's Food Chemistry, 4th edition, 2007.
  • Fellows, P. J., Food Processing Technology, 4th edition, 2017.

Unit 7: Economics and Sustainability

Note

Unit 7: Economics and Sustainability

A question to start with

A fermentation tank is a factory, and a factory must pay its way. The sugar must be bought, the tank must be built, the power must be paid for, and the finished protein must sell for more than it cost to make. At the same time, the tank must earn its place on the planet: it must use less land, less water and fewer gas releases than the food it replaces, or its promise is empty. This unit puts numbers to both sides: what a fermentation made ingredient costs, what drives that cost down, and how its footprint is counted, honestly, from the field to the factory gate.

What a fermentation made ingredient costs

The cost of a fermentation made ingredient has four parts. The first is the feed, mainly sugar, which is often the largest single cost. The second is the capital cost, the price of building the tanks, the filters, the dryers and the plant, spread across the tonnes made over the life of the plant. The third is the running cost: power for stirring, air and cooling, water, cleaning, labour and maintenance. The fourth is the cost of purification and drying, which rises with the purity demanded.

A simple way to see the balance is to split a kilogram of protein into its costs. If the feed costs the most, then a cheaper sugar or a strain that makes more protein per kilogram of sugar will cut the cost fastest. If the capital cost dominates, then a larger plant that spreads the building cost across more tonnes will cut the cost fastest. If purification dominates, then a host that releases the protein outside the cell, so that it is easy to collect, will cut the cost fastest. A maker who knows which part dominates knows where to work.

A worked example shows the feed arithmetic. A yeast makes 50 g of protein from each kilogram of sugar it eats, a yield of 5 percent. To make 1 kg of protein, the tank needs 20 kg of sugar. If sugar costs 0.50 per kilogram, the feed alone costs 10 per kilogram of protein. If a better strain makes 100 g of protein per kilogram of sugar, the feed needed halves to 10 kg, and the feed cost halves to 5 per kilogram. Strain improvement is therefore the fastest route to a cheaper protein, and it is where much of the research money goes.

How cost falls as scale rises

Cost falls with scale in three ways. First, a larger tank makes more protein for each unit of steel, power and labour, because a tank twice as large does not need twice the staff or twice the control system. Second, a larger plant buys sugar in bulk, at a lower price per tonne, and uses its power and water more efficiently. Third, a plant that runs all year, with short stops for cleaning, spreads its capital cost across more tonnes than a plant that runs half the year.

The learning curve adds a fourth fall. Each run teaches the operators something: a better feed rate, a better stirring speed, a better harvest time. Each improvement is written into the procedure, and the next run is a little cheaper. Over years, those small gains add up to a large fall in cost, which is why a technology that looks dear in its first year can look affordable in its fifth. The history of chymosin shows this plainly: it was dear when it was new, and it is now the standard, cheap way to make cheese.

There is a limit to scale, and it is worth naming. A tank cannot grow without limit, because mixing and cooling get harder as the tank grows, and a single very large tank is a single point of failure: if it is contaminated, a large amount of output is lost. Most plants therefore use several large tanks rather than one giant one, so that one failure does not stop the plant. The balance between large tanks and many tanks is an engineering choice, and it is set by the cost of failure as well as by the cost of steel.

The feed question: where the sugar comes from

Every fermentation needs a feed, and the feed is usually sugar from a crop: maize, wheat, sugar cane or sugar beet. That sugar has its own footprint: the land it grew on, the water it used, the fertiliser that fed it, and the energy used to turn it into syrup. A fermentation made protein is therefore not free of farming: it moves the farming from the animal to the sugar crop, and the footprint of the sugar must be counted in the footprint of the protein.

The choice of sugar matters. Sugar from a crop grown on land that was forest a year ago carries a large footprint, because clearing the forest released CO₂. Sugar from a crop grown on existing farmland, with modest fertiliser, carries a smaller one. Sugar from a by-product, such as the whey left from cheese making or the molasses left from sugar refining, carries a smaller footprint still, because the crop was grown for another product and the by-product would otherwise be waste. A maker who wants a low footprint chooses the feed with care, and says which one was used.

The future of feed is wider than sugar. Researchers are testing feeds made from food waste, from straw, from wood, and from CO₂ itself, using microbes that eat gas instead of sugar. Each new feed cuts the link to farmland a little further, and each one brings its own cost and its own engineering problems. For now, sugar from crops is the feed that works, and the honest footprint counts it, rather than pretending the tank runs on nothing.

Counting the footprint: life cycle assessment

The footprint of a fermentation made ingredient is counted by a life cycle assessment, which adds up the land, water, energy and gas releases across the whole chain: growing the sugar crop, turning it into syrup, running the tank, purifying and drying the protein, packing it, and taking it to the food maker. The gases counted are CO₂, CH₄ and N₂O, and they are added together as CO₂ equivalent, a single number that lets different gases be compared.

A fair comparison counts the same chain for the food being replaced. A fermentation made whey protein is compared with dairy whey, counted from the farm that grew the feed, through the cow, the milk, the cheese plant and the whey dryer. A fermentation made fat is compared with the fat it replaces, counted from its own farm or plantation. A comparison that counts the tank but not the sugar crop, or that counts the cow but not the feed crop, is not a comparison: it is a trick, and a careful reader should ask for the full chain.

The results so far are encouraging but not uniform. Fermentation made proteins usually need far less land than the animal food they replace, because a tank uses a small floor area and the sugar crop uses far less land than the feed crop and pasture of a cow. Water use is usually lower too, though the tank needs clean water and the sugar crop needs irrigation in some regions. Gas releases are usually lower, but they depend on the power source: a plant run on coal power carries a larger footprint than the same plant run on wind or solar power, so the energy source is part of the answer, not a footnote.

Water, energy and waste

Water is used in a fermentation plant in three ways: as part of the broth, as cooling water, and as cleaning water. The broth water leaves the plant as part of the product or as waste water, which must be treated before it is released, because it holds sugars, salts and microbes. The cooling water is usually reused in a loop, and the cleaning water is the largest single use, because a food plant must be cleaned thoroughly between runs. A plant that reuses its cooling water and treats its waste water well uses far less fresh water than one that does not.

Energy is used for stirring, for pumping air, for cooling, for heating during sterilisation, and for drying the finished powder. Drying is often the largest single energy use, because removing water takes heat, and spray drying a tonne of protein liquid uses a large amount of gas or electricity. A plant that dries efficiently, that recovers heat from its dryers, and that buys power from a low carbon source, has a smaller footprint than one that does not, and the difference can be larger than the difference between two strains.

Waste from a fermentation plant is modest but real. The spent cells, after the protein is taken out, can be dried and sold as feed or as fertiliser, turning a waste into a product. The waste water, after treatment, can be reused for cleaning or released safely. The packaging, usually sealed bags and drums, must be recycled or disposed of properly. A plant that sells its spent cells and treats its water is not only cleaner: it earns money from a stream it would otherwise pay to dispose of.

Who buys, and what happens next

The buyers of fermentation made ingredients are food makers: the companies that make ice cream, protein drinks, cheese, bakery foods and meat analogues. They buy because the ingredient lets them make a food that is animal free, or lactose free, or consistent in supply, or better in function than the alternative. The first buyers are usually makers of high value foods, where the shopper will pay more for an animal free label, and the market widens as the cost falls.

The future of the field depends on three things happening together. First, strains must keep improving, making more protein per litre and per kilogram of sugar. Second, plants must keep growing larger and running better, spreading costs across more tonnes. Third, the rules must stay clear and the labels must stay honest, so that shoppers trust the foods and keep buying them. If any one of the three fails, the field stalls: a dear protein with no buyers, a cheap protein with no approval, or an approved protein that shoppers do not trust are all dead ends.

The wider lesson of this unit is that a new food technology is judged twice: once on its cost, and once on its footprint, and both judgements need numbers, not slogans. A fermentation made protein that is dearer than the food it replaces will stay a niche product, whatever its footprint. A protein that is cheap but carries a hidden footprint, from a sugar crop grown on cleared forest, will fail the second judgement when the numbers are counted. The makers who succeed will be the ones who publish both sets of numbers, cost and footprint, and who improve both, year after year.

Break-even: a worked example

A fermentation product is judged by its break-even point: the amount that must be sold before the maker starts to earn. The derivation is simple. The maker pays fixed costs, which do not change with output, such as the rent, the tank and the salaries, and variable costs, which grow with output, such as sugar, power and packaging. Each kilogram sold brings in a price, and the contribution of each kilogram is the price minus the variable cost per kilogram. The break-even amount is the fixed costs divided by that contribution, and it is the number of kilograms that must be sold to cover the fixed costs.

A worked example shows it. A plant pays 1,000,000 in fixed costs a year. It sells its protein at 50 per kilogram, and the variable cost is 30 per kilogram, so each kilogram contributes 20. The break-even amount is 1,000,000 divided by 20, which is 50,000 kg a year. Below that, the plant loses money. Above that, each extra kilogram earns 20. If the price falls to 40, the contribution falls to 10, and the break-even doubles to 100,000 kg, which shows how sensitive the business is to price.

The maker has three levers to lower the break-even: cut the fixed costs, cut the variable costs, or raise the price. Cutting the variable cost is the most powerful, because it raises the contribution of every kilogram sold, and it is done by lifting the yield, by buying sugar cheaper, or by using less power. Raising the price is the riskiest, because a dearer product sells less. The break-even is the first number an investor asks for, and a maker who can work it out by hand, and can show which lever moves it most, understands the business as well as the biology.

Questions to think about

  • A yeast makes 50 g of protein per kilogram of sugar. How much sugar is needed for 1 kg of protein, and what happens to the feed cost if the yield doubles?
  • Why must a life cycle assessment count the sugar crop as well as the tank when judging a fermentation made protein?
  • Why do most plants use several large tanks rather than one giant tank, even though a giant tank looks cheaper?
Think about the feed arithmetic before you read on

At 50 g per kilogram, 1 kg of protein needs 20 kg of sugar, because 1,000 divided by 50 is 20. If the yield doubles to 100 g per kilogram, the sugar needed halves to 10 kg, and the feed cost halves too. That is why strain improvement, making more protein from the same sugar, is the fastest route to a cheaper ingredient.

Where the cost of a fermentation protein goes

Feed: sugar, nitrogen, minerals and vitamins
↓
Capital: tanks, filters, dryers and the plant itself
↓
Running: power, water, cleaning, labour and maintenance
↓
Purification and drying: filtration, chromatography and spray drying
↓
Total cost per kilogram falls as yield rises and the plant grows

The diagram shows the four parts of the cost. Knowing which part dominates tells the maker where to work: on the strain, on the plant size, or on the purification.

Practical note: In a practical session, work out the feed cost of a protein you eat: buy a small pack of dried yeast and a bag of sugar. Grow a small yeast culture as in Unit 2, weigh the sugar you used and the yeast you harvested, and work out the grams of yeast per gram of sugar. Then look up the price of sugar per kilogram in your shop and work out the feed cost per kilogram of yeast. Compare your number with a classmate who used a different sugar or a different temperature, and answer: which change cut the feed cost most?

Quick self-check

Question: Why is the feed often the largest cost of a fermentation made protein?

Because the tank needs a large amount of sugar to make each kilogram of protein, and that sugar must be bought, stored, sterilised and pumped. A strain that makes more protein per kilogram of sugar cuts that cost directly.

Question: Why does the power source change the footprint of the same protein?

Because the tank uses power for stirring, air, cooling and drying, and the gas releases of that power depend on how it was made. A plant run on coal power carries a larger footprint than the same plant run on wind or solar power.

Question: A plant pays 1,000,000 in fixed costs, sells at 50 per kilogram and spends 30 per kilogram on variable costs. What is the break-even amount?

Each kilogram contributes 50 minus 30, which is 20, so the break-even is 1,000,000 divided by 20, which is 50,000 kg a year.

Try it yourself

Find a product in a shop that claims to be animal free, plant based, or lower carbon, and read its pack. Write down the claim, the ingredient list, and any numbers given for carbon or for protein. Then search for the maker's own report or web page about that product and write down whether it gives a life cycle assessment number, and what chain that number covers: does it count the farm, the factory and the packaging, or only the factory? Answer in writing: is the claim backed by a full chain number, or only by a part of the chain?

Key ideas to carry forward

  • The cost of a fermentation protein splits into feed, capital, running cost, and purification and drying.
  • Feed is often the largest cost, and a strain that makes more protein per kilogram of sugar cuts it fastest.
  • Cost falls as plants grow larger, run all year, and learn from each run, but one giant tank is a single point of failure.
  • The sugar feed has its own footprint from the crop, and that footprint must be counted in the footprint of the protein.
  • A life cycle assessment counts land, water, energy and gases across the whole chain, and a fair comparison counts the same chain for both foods.
  • Drying is often the largest energy use, and the power source can change the footprint more than the strain does.
  • A new technology is judged twice, on cost and on footprint, and both judgements need published numbers, not slogans.

  • Break-even amount equals fixed costs divided by price minus variable cost per kilogram, and it shows how much must be sold before the maker earns.

Glossary

Capital cost

The cost of building a plant, spread across the tonnes made over its life. Food tech example: the price of tanks and dryers is a capital cost spread across years of output.

CO₂ equivalent

A single number that adds together CO₂, CH₄ and N₂O by their warming effect. Food tech example: a life cycle assessment reports gas releases as CO₂ equivalent per kilogram of protein.

Feedstock

The raw material fed to a fermentation, usually sugar. Food tech example: maize syrup is a feedstock for making whey protein by fermentation.

Learning curve

The fall in cost as operators learn and improve a process over time. Food tech example: each run of a fermentation plant teaches a better feed rate, cutting cost a little.

Life cycle assessment

A method that adds up the land, water, energy and gas releases of a product across its whole life. Food tech example: a life cycle assessment of a fermentation protein counts the sugar crop, the tank and the drying.

Molasses

A thick syrup left after sugar refining, used as a fermentation feed. Food tech example: molasses is a cheap feedstock for growing yeast.

Running cost

The day to day cost of operating a plant: power, water, labour and cleaning. Food tech example: power for stirring and cooling is a running cost of a fermentation plant.

Spent cells

The cells left after the product is taken out of the broth. Food tech example: spent yeast cells are dried and sold as animal feed.

Yield

The amount of product made per amount of feed. Food tech example: a yeast yield of 100 g of protein per kilogram of sugar is twice as good as 50 g per kilogram.

By-product

A useful material left over from another process. Food tech example: whey from cheese making is a by-product that can feed a fermentation.

Carbon dioxide

A gas released when fuels burn, written CO₂. Food tech example: the CO₂ from growing a sugar crop is counted in the footprint of a fermentation protein.

Methane

A gas released from farming and waste, written CH₄. Food tech example: methane is counted in the footprint of dairy whey in a life cycle assessment.

Nitrous oxide

A gas released from fertilised soils, written N₂O. Food tech example: fertiliser on a sugar crop adds N₂O to the footprint of a fermentation protein.

Spray drying

Drying by blowing a liquid as a fine mist into a hot chamber. Food tech example: spray drying is often the largest energy use in making a protein powder.

Fixed costs

Costs that do not change with output, such as rent and salaries. Food tech example: a fermentation plant pays its fixed costs whether it makes 1 kg or 1,000 kg.

Variable costs

Costs that grow with output, such as sugar, power and packaging. Food tech example: sugar is a variable cost that rises with every kilogram made.

References

  • Stanbury, P. F., Whitaker, A. and Hall, S. J., Principles of Fermentation Technology, 3rd edition, 2017.
  • Bamforth, C. W., Food, Fermentation and Micro-organisms, 2005.
  • Waites, M. J., Morgan, N. L., Rockey, J. S. and Higton, G., Industrial Microbiology: An Introduction, 2001.
  • Demain, A. L. and Davies, J. E., Manual of Industrial Microbiology and Biotechnology, 2nd edition, 1999.
  • Damodaran, S., Parkin, K. L. and Fennema, O. R., Fennema's Food Chemistry, 4th edition, 2007.
  • Fellows, P. J., Food Processing Technology, 4th edition, 2017.

Unit 8: Strain Engineering and Synthetic Biology: How Microbes Are Designed

Note

Unit 8: Strain Engineering and Synthetic Biology: How Microbes Are Designed

A question to start with

A yeast in a bakery makes bread rise. The same kind of yeast, given a new instruction, can make a whey protein, a sweet protein or a colour. The instruction is written in DNA, and the work of writing it is called strain engineering. This unit explains how a microbe is designed: how scientists read the code of life, find the gene for a food protein, put that gene into a microbe, tune the microbe so it makes enough of the protein, and keep the strain safe and stable for the factory.

Reading the code: genes and proteins

Every protein is built from a recipe written in DNA, and the recipe is called a gene. A gene is a stretch of DNA that spells out the order of amino acids in one protein, using a four letter code. The cell reads the gene, copies it into a working message called RNA, and then builds the protein amino acid by amino acid, following the message. That chain, DNA to RNA to protein, is the central idea of this unit, and every step of strain engineering is a way of changing one link in that chain.

To make a food protein by fermentation, the scientist first finds the gene for that protein. For a whey protein, the gene comes from the animal that makes it, read from its DNA and copied in the laboratory. The gene is a string of letters, and the string can be checked letter by letter against the known protein, so the scientist knows the copy is right before it is ever put into a microbe. The gene alone is not enough: the microbe must also be told when to read it, how often to read it, and where to send the protein once it is made.

The microbe that receives the gene is called the host, and the finished microbe is called a production strain. The host is chosen for its strengths: a yeast that secretes protein well, a fungus that pours out enzymes, a bacterium that grows fast. The host is not chosen for the protein it already makes, but for the factory it already is. A good host is safe, well understood, easy to grow in a tank, and easy to clean out of the final product. The gene is the new job. The host is the worker who will do it.

Finding the right gene and preparing it

Finding a gene starts with knowing the protein. The scientist knows the amino acid order of the target protein, from the food it comes from, and works backwards to the DNA letters that spell it. Often the gene is copied from the original organism, but it is usually rewritten first, in a process called codon optimisation. The DNA code has more than one spelling for most amino acids, and each microbe prefers its own spellings. Rewriting the gene in the host's preferred spellings helps the host read it faster and make more protein, without changing a single amino acid in the finished protein.

The rewritten gene is joined to two more pieces of DNA before it goes into the host. The first is a promoter, a switch that tells the host when to read the gene. Some promoters are always on, so the protein is made all the time. Others switch on only when a certain sugar is added, or when the temperature changes, so the maker can grow the cells first and switch the protein on later, when the tank is full. The second piece is a signal that tells the cell where to send the protein: either to keep it inside the cell, or to secrete it outside, into the broth, where it is far easier to harvest.

The three pieces, promoter, gene and signal, are joined into a circle of DNA called a plasmid, or built into a piece that will slot into the host's own DNA. The plasmid is a delivery vehicle: it carries the new instructions into the cell and holds them there. Building it is careful bench work, done with enzymes that cut and join DNA at exact places, and each join is checked by reading the DNA letters again, so a mistake is caught before the strain is ever grown.

Putting the gene into the microbe

Putting new DNA into a microbe is called transformation, and the method depends on the host. A yeast can be made to take up DNA by treating its wall so it opens briefly, or by a short electric pulse that opens tiny pores in the membrane. A bacterium can be treated the same way, or given the DNA by a virus that carries it in. A fungus can be transformed by mixing its spores with the DNA and a chemical that helps it in. In each case, only some cells take up the DNA, so the scientist must find the few that did.

Finding them uses a marker, a second gene on the same DNA that gives the transformed cell a visible advantage, such as the ability to grow without a nutrient that other cells need. The cells are spread on a plate that lacks that nutrient, and only the transformed cells grow into colonies. Each colony is a candidate strain, and each one is tested for how much protein it makes. The best colonies are kept, the rest are discarded. The marker itself is chosen with care for a food strain: a marker that needs an antibiotic is avoided in food production, because no antibiotic should be anywhere near a food process, so food strains use markers based on nutrients instead.

A worked example shows the scale of the search. A transformation may give 100 colonies on a plate, and the scientist tests each one for protein output. If the outputs range from 0.1 g per litre to 1 g per litre, the best colony makes ten times more than the worst, even though all of them carry the same gene. That spread is why screening matters: the gene is only the start, and the strain that makes the most protein is found by testing many candidates, not by building one and hoping. The best colony becomes the parent of the production strain.

Tuning the factory: more protein, less waste

Once a strain makes the protein, the work turns to tuning: making more of it, with less waste. The first tune is the copy number, how many copies of the gene the cell holds. More copies usually mean more protein, but each copy costs the cell energy, so there is a sweet spot, and the scientist tests strains with one, two, three or more copies to find it. The second tune is the promoter: a stronger promoter reads the gene more often, and a promoter that switches on late lets the cells grow first and make protein later, when the tank is full of healthy cells.

The third tune is the cell's own machinery. A protein that must be secreted needs a path out of the cell, and the cell's secretion path can be widened by adding more of the helper proteins that carry cargo out. A protein that folds badly can be helped by adding chaperones, proteins that help other proteins fold into the right shape. A protein that the cell chews up can be protected by removing the cell's own protein cutting enzymes, called proteases, that would otherwise destroy it. Each tune is a small change, tested one at a time, so the scientist knows which change helped and which did not.

A worked example shows how the tunes add up. A first strain makes 0.5 g of protein per litre. Doubling the copy number lifts it to 0.9 g per litre. Switching to a stronger promoter lifts it to 1.5 g per litre. Improving secretion lifts it to 2.2 g per litre. Each step is less than a doubling, but together they lift the output more than four times. That is the real work of strain engineering: not one clever trick, but a series of small, measured improvements, each tested, each kept only if it helps, until the strain is ready for the pilot tank.

Testing, improving and keeping the strain stable

A production strain must be stable: it must make the same protein, in the same amount, every time it is grown, for months and years. Stability is tested by growing the strain again and again, through many generations, and checking the output at each step. A strain that loses its gene, or that makes less protein after ten generations, is not ready for the factory. The gene is therefore usually built into the host's own DNA, rather than left on a plasmid that can be lost, because a gene in the chromosome is copied every time the cell divides, and it stays.

Improvement continues after the strain is stable, by a method called directed evolution. The scientist makes many small random changes to the gene or to the strain, tests each one, keeps the best, and repeats. It is slow, deliberate breeding at the level of DNA, and it can lift the output further, or make the protein more heat stable, or easier to purify. Each round is a small step, and the steps add up over months. The strain that reaches the factory is the product of many rounds, not of one design.

The finished strain is stored in a cell bank: a set of frozen vials, each holding the same strain, kept at very low temperature. One vial is the master bank, opened only to make more vials. The others are the working bank, used to start each production run. Every run starts from the same bank, so every batch starts from the same cells, and the product stays the same. The cell bank is the strain's memory, and it is guarded as carefully as the recipe, because without it the factory has no factory.

Safety and the questions a designer must answer

Designing a strain for food carries duties that a laboratory strain does not. The host must be safe, with a history of safe use in food, or a full safety case of its own. The gene must be known, and the protein it makes must be the protein intended, with no extra proteins made by accident. The marker must not be an antibiotic resistance gene, and no DNA from the design work should remain in the finished food unless the safety file has judged it. The finished ingredient is tested to prove that the host is gone, that the DNA is gone or judged safe, and that the protein is the right molecule.

The designer must also answer what the strain cannot do. It cannot escape and grow in the open, because it is built to live in a tank on a rich feed, and it dies without it. It cannot pass its gene to other microbes in the factory, because the gene is built into its chromosome and the factory is cleaned and sterilised between runs. It cannot make a toxin, because the gene it carries makes only the intended protein, and the strain is tested for toxins before it is used. Each of those answers is written into the safety file, with evidence, not with a promise.

The last duty is honesty about what was changed and what was not. A precision fermentation ingredient is not the animal it imitates, and it is not the microbe that made it. It is one molecule, made by a designed microbe, purified out of the broth. The label says what the molecule is, and the safety file says how it was made. The reader who understands strain engineering can read both, and can judge the food on its evidence, not on its novelty.

Questions to think about

  • Why is a gene rewritten in the host's preferred spellings before it is put into the microbe, and does that change the finished protein?
  • A transformation gives 100 colonies that all carry the same gene, but their protein output ranges from 0.1 to 1 g per litre. Why, and what should the scientist do?
  • Why must a food strain avoid an antibiotic resistance marker, and what kind of marker is used instead?
Think about the 100 colonies before you read on

The colonies differ because the gene can land in different places, in different copy numbers, and the cells can read it more or less well, even though the gene itself is the same. The scientist tests each colony for protein output and keeps the best, because the gene is only the start and the best strain is found by screening, not by building one and hoping.

From a gene to a production strain

Find the gene for the target protein and rewrite it in the host's spellings
↓
Join the gene to a promoter switch and a secretion signal
↓
Put the DNA into the host by transformation
↓
Find the transformed cells with a nutrient marker and test each colony
↓
Tune the strain: copy number, promoter, secretion and folding
↓
Test for stability over many generations and store the strain in a cell bank

The diagram shows strain engineering in order: from a gene on paper to a stable strain in a frozen bank, ready to start a production run.

Practical note: In a practical session, take a familiar recipe, such as a bread recipe, and rewrite it as a set of instructions for a stranger: list the ingredients, the order, the temperatures and the times. Then give the rewritten recipe to a classmate and ask them to follow it exactly, and note where they hesitate or go wrong. Record the unclear steps and rewrite them, and answer: how is this like codon optimisation, where the same recipe is rewritten so a new reader can follow it better?

Quick self-check

Question: Question: What is a promoter, and why does it matter?

A promoter is a piece of DNA that switches a gene on, telling the cell when to read it. It matters because a strong promoter makes more protein, and a promoter that switches on late lets the cells grow first and make protein later, when the tank is full.

Question: Question: Why is the production gene usually built into the host's own DNA rather than left on a plasmid?

Because a gene in the chromosome is copied every time the cell divides and stays, while a plasmid can be lost as the cells grow. A strain that loses its gene stops making the protein, so stability needs the gene in the chromosome.

Try it yourself

Choose a food protein you know, such as the whey in milk or the main protein of egg white, and write a one page plan for making it by precision fermentation: which host would you choose and why, would you secrete the protein or keep it inside the cell, and what would you test first? Then list the three things that could go wrong, and what you would change for each one.

Key ideas to carry forward

  • A gene is the recipe for one protein, and the cell reads DNA to RNA to protein.
  • Codon optimisation rewrites the gene in the host's preferred spellings without changing the finished protein.
  • A promoter switches the gene on, and a signal sends the protein inside the cell or out into the broth.
  • Transformation puts the DNA in, a nutrient marker finds the cells that took it up, and screening finds the best colony.
  • Tuning lifts output step by step: copy number, promoter, secretion, folding and protection from proteases.
  • Stability is tested over many generations, and the finished strain is stored in a frozen cell bank.
  • Food strains must avoid antibiotic markers, and the safety file must show the protein is the intended molecule and the host is gone.

Glossary

Gene

A stretch of DNA that spells out the order of amino acids in one protein. Food tech example: the gene for a whey protein is copied and put into a yeast.

Host

The microbe that receives a new gene and makes the protein. Food tech example: a yeast host is chosen because it secretes protein well.

Production strain

A microbe designed and tuned to make one chosen ingredient. Food tech example: a production strain of yeast makes whey protein in a tank.

Promoter

A piece of DNA that switches a gene on. Food tech example: a promoter that switches on late lets cells grow first and make protein later.

Codon optimisation

Rewriting a gene in the host's preferred DNA spellings, without changing the protein. Food tech example: codon optimisation helps a yeast read a whey gene faster.

Plasmid

A circle of DNA that carries new instructions into a cell. Food tech example: a plasmid carries a gene, a promoter and a marker into a yeast.

Transformation

Putting new DNA into a microbe. Food tech example: transformation gives a yeast the gene for a sweet protein.

Marker

A gene that lets the scientist find the cells that took up new DNA. Food tech example: a nutrient marker lets only transformed cells grow on a plate.

Copy number

How many copies of a gene a cell holds. Food tech example: raising the copy number from one to two can lift protein output.

Secretion

Sending a protein out of the cell into the broth. Food tech example: secreting a whey protein makes it far easier to harvest.

Chaperone

A protein that helps another protein fold into the right shape. Food tech example: adding chaperones helps a new protein fold correctly.

Protease

An enzyme that cuts proteins apart. Food tech example: removing a protease stops the cell from destroying the new protein.

Cell bank

A set of frozen vials holding the same strain, used to start each run. Food tech example: every production run starts from a vial of the working cell bank.

Directed evolution

Making small random changes, testing each one, and keeping the best, repeated many times. Food tech example: directed evolution lifts a strain's output over months.

Chromosome

The main DNA of a cell, copied every time the cell divides. Food tech example: building the gene into the chromosome keeps it stable.

References

  • Stanbury, P. F., Whitaker, A. and Hall, S. J., Principles of Fermentation Technology, 3rd edition, 2017.
  • Bamforth, C. W., Food, Fermentation and Micro-organisms, 2005.
  • Waites, M. J., Morgan, N. L., Rockey, J. S. and Higton, G., Industrial Microbiology: An Introduction, 2001.
  • Demain, A. L. and Davies, J. E., Manual of Industrial Microbiology and Biotechnology, 2nd edition, 1999.
  • Damodaran, S., Parkin, K. L. and Fennema, O. R., Fennema's Food Chemistry, 4th edition, 2007.
  • Fellows, P. J., Food Processing Technology, 4th edition, 2017.

Unit 9: Downstream Processing and Purification in Depth

Note

Unit 9: Downstream Processing and Purification in Depth

A question to start with

The tank has done its work and the broth is full of protein, but the broth is not a food ingredient. It also holds cells, leftover sugar, salts, other proteins and water, and the protein a shopper will eat must be taken out clean. That work is called downstream processing, and it decides the purity, the cost and the yield of every fermentation made ingredient. This unit follows the product out of the tank, step by step, through breaking cells open, filtering, purifying and drying, and it shows how each step trades purity against yield.

Where the product sits, and why it matters

The first question in downstream processing is where the product sits. Some proteins are secreted, sent out of the cell into the broth by the microbe itself. Others stay inside the cell, held in the cytoplasm until the cell is broken open. The difference decides the whole process: a secreted protein can be harvested by removing the cells and keeping the liquid, while an inside protein needs a breaking step first, and that step adds cost, releases everything else the cell holds, and makes purification harder.

Strain designers therefore prefer secretion whenever the protein allows it, and Unit 8 explains how the signal for secretion is added. Secretion is not always possible: some proteins fold badly outside the cell, some are too large to pass through the cell wall, and some are damaged by the broth. When secretion fails, the process must break the cells, and the breaking step must be gentle enough to leave the protein whole while opening the cell.

A worked example shows the difference in yield. A tank holds 100 kg of protein in total. If the protein is secreted, the harvest may recover 95 kg in the liquid, because the cells are simply spun away. If the protein is inside the cells, breaking them may release 90 kg, and the debris may trap 10 kg more, leaving 80 kg in the liquid. The same tank, the same strain, a different product place, and the downstream yield differs by 15 kg before purification has even begun. Where the product sits is a design decision, not a detail.

Breaking cells open: release without damage

Cells are broken open by force or by chemistry. The common force method is homogenisation: the broth is forced through a narrow gap at high pressure, and the shear tears the cell walls apart. It is fast and it works at factory scale, but it is violent: it also breaks the protein if the pressure is too high, and it heats the broth, so the liquid is cooled as it passes through. The second force method is bead milling, where the broth is stirred with tiny glass beads that grind the cells open. It is gentler, but slower, and it suits smaller runs.

The chemical method uses enzymes that dissolve the cell wall, or a change in salt and pH that weakens it. Enzymatic breaking is gentle and it leaves the protein whole, but the enzyme itself is a protein that must later be removed, and it costs money. The choice is a balance: force is cheap and fast but harsh, enzymes are gentle but dear and they add a new impurity. Some processes use both, a mild enzyme step to weaken the wall, then a gentle homogeniser pass to finish the job, so neither step needs to be harsh.

Whatever the method, the broken broth is a soup of everything the cell held: the target protein, other proteins, DNA, fats, sugars and cell wall fragments. The next steps must sort that soup, and the sorting is easier if the breaking was gentle, because a gentle break leaves the debris in large pieces that are easy to spin out, while a violent break shreds the debris into fine pieces that clog filters. The breaking step is where downstream processing is won or lost, and the operator who breaks too hard pays for it in every step that follows.

Separating solids: spinning and filtering

The first separation removes the solids: whole cells for a secreted protein, and cell debris for an inside protein. The workhorse is the centrifuge, a machine that spins the broth fast so the heavy solids fly to the outside and the clear liquid stays in the middle. A large factory centrifuge spins tonnes of broth an hour, and it runs continuously, with solids leaving one way and liquid leaving another. The liquid that comes out, called the clarified liquid, still holds the protein, the sugars and the salts, but the solids are gone.

Filtration is the second tool. A microfilter holds back cells and large debris while the liquid passes through, and it is used where a centrifuge would damage the product, or where the liquid must be sterile. A depth filter, a thick pad of fibres, catches fine particles that a centrifuge misses. Filters clog, and a clogged filter slows the whole line, so the broth is usually clarified in a centrifuge first and filtered second, with the filter as a polish, not as the main worker.

A practical rule runs through this stage: never let a step do a job that an earlier, cheaper step can do. The centrifuge removes the bulk of the solids cheaply. The filter removes the fine remainder. A process that sends unclarified broth straight to a fine filter will clog it in minutes, stop the line, and cost a day. The order of the steps is part of the design, and it is tested at pilot scale before the production plant is built.

Membranes: sorting molecules by size

Once the solids are gone, the liquid is sorted by size with membranes, in steps called ultrafiltration and diafiltration. An ultrafiltration membrane holds back large molecules, such as proteins, and lets water, salts and small sugars pass through. The protein liquid is pushed across the membrane, the small molecules leave, and the protein stays behind, more concentrated than before. The step is repeated, with fresh water added, in a washing process called diafiltration, which flushes out more of the small molecules each time.

Membranes are chosen by their cut off, the size above which a molecule is held back. A membrane with a cut off smaller than the target protein holds the protein and lets the rest go. A membrane with a cut off larger than the protein would lose it, so the cut off is set with care, and it is tested with the real broth, not only with clean water, because the broth fouls the membrane: proteins and fats stick to it and slow the flow. The operator watches the flow rate, and when it falls, the membrane is cleaned with a wash cycle, and if cleaning fails, the membrane is replaced.

A worked example shows the concentration. A clarified liquid holds 10 g of protein per litre in 10,000 litres, which is 100 kg of protein in total. Ultrafiltration removes half the water, leaving 5,000 litres at 20 g per litre, still 100 kg of protein, but half the volume to handle. Diafiltration then washes out the salts and sugars, so the liquid that goes to the next step is smaller, cleaner and cheaper to dry. The membrane step does not purify the protein from other proteins, because other proteins of similar size stay with it. That job belongs to the next step, chromatography.

Chromatography: the fine sort

Chromatography sorts molecules by how they stick to a material, not only by their size. The liquid is passed through a column packed with beads, and the target protein sticks to the beads while most other molecules flow through. The column is then washed, and the protein is released with a change in salt or pH, coming out in a clean stream. The method is powerful: one chromatography step can lift a protein from 50 percent pure to 95 percent pure, and it is the step that makes a food grade ingredient out of a crude broth.

Chromatography has two costs: the beads are dear, and the step loses protein. Some protein never sticks, some never comes off, and each run loses a little, so the operator balances the gain in purity against the loss in yield. A food ingredient rarely needs the extreme purity of a medicine, so the chromatography step for food is usually a single, gentle pass, not the three passes a medicine might need. The choice of beads matters: some bind by charge, some by size, some by a specific tag on the protein, and the maker tests each type with the real broth before choosing.

The column is cleaned and reused many times, and its life is counted in cycles. A column that loses its binding power is reconditioned or replaced, and the cost of the beads is spread over the cycles they serve. The operator keeps a record of each cycle: how much protein went in, how much came out, and how pure it was. That record is the proof that the step works the same every time, and it is part of the batch record that follows the ingredient to the customer.

Drying, polishing and the balance of yield and purity

The last steps are polishing and drying. Polishing removes the last traces: a final filtration through a sterile filter that catches any microbe, and a test for endotoxins, the fragments of bacterial walls that can cause fever. The liquid is then dried, usually by spray drying, where it is blown as a fine mist into a hot chamber and the water evaporates, leaving a powder. Freeze drying is gentler and dearer, and it is used where the protein is fragile and would be damaged by heat.

The balance of yield and purity runs through every step, and it can be worked out step by step. Suppose each of four steps recovers 90 percent of the protein that enters it. The overall yield is 0.9 multiplied by 0.9 multiplied by 0.9 multiplied by 0.9, which is about 0.66, so 66 kg comes out of every 100 kg that entered. If one step is tightened to recover only 80 percent, the overall yield falls to 0.9 times 0.9 times 0.9 times 0.8, which is about 0.58. That is the derivation a downstream engineer does by hand: overall yield equals the product of the step yields, so a small loss at any step multiplies through the whole chain.

The lesson is to spend the losses where they buy the most purity, and to keep the other steps gentle. A step that loses 20 percent of the protein to gain 1 percent of purity is a bad trade, unless that 1 percent is what the safety file needs. A step that loses 5 percent to gain 20 percent of purity is a good trade. The engineer tests each step, measures its yield and its purity gain, and designs the chain so the total yield stays high while the final purity meets the specification. The powder that leaves the plant is the result of that arithmetic, not of luck.

Questions to think about

  • Why is a secreted protein easier and cheaper to harvest than a protein kept inside the cell?
  • Four steps each recover 90 percent of the protein. What is the overall yield, and what happens if one step is tightened so it recovers only 80 percent?
  • Why is chromatography used for food ingredients as a single gentle pass, rather than the several passes a medicine might need?
Think about the four steps before you read on

Overall yield equals the product of the step yields, so 0.9 times 0.9 times 0.9 times 0.9 is about 0.66, or 66 kg out of every 100 kg. If one step falls to 80 percent, the product becomes 0.9 times 0.9 times 0.9 times 0.8, which is about 0.58, so a small loss at one step multiplies through the whole chain.

From broth to clean powder

Harvest the broth: keep the liquid for a secreted protein, break the cells for an inside protein
↓
Spin out the solids in a centrifuge and filter the liquid clear
↓
Concentrate and wash with ultrafiltration and diafiltration
↓
Purify with chromatography: stick, wash and release the protein
↓
Polish: sterile filter and test for microbes and endotoxins
↓
Dry into a powder, test the batch and pack it

The diagram shows downstream processing in order. Each step removes more of what is not wanted, and each step loses a little of the product, so purity is balanced against yield.

Practical note: In a practical session, filter a cloudy liquid, such as a thin soup, through three stages: a sieve, a cloth, and a coffee filter. Weigh the liquid before and after each stage, and note how much is lost at each stage and how much clearer it becomes. Record the loss and the clarity at each stage, and answer: which stage gave the most clarity for the least loss, and what does that tell you about ordering the steps in downstream processing?

Quick self-check

Question: Question: What is the difference between ultrafiltration and chromatography?

Ultrafiltration sorts molecules by size through a membrane, holding back proteins and letting water and small molecules pass. Chromatography sorts by how molecules stick to beads in a column, and it can lift a protein from about 50 percent pure to about 95 percent pure in one step.

Question: Question: Why is the breaking step the place where downstream processing is won or lost?

Because a gentle break leaves debris in large pieces that are easy to spin out, while a violent break shreds debris into fine pieces that clog filters. A hard break also damages the protein itself, and the damage is paid for in every step that follows.

Try it yourself

Take a food label that lists a fermentation made ingredient, such as an enzyme or a protein, and write down the steps you think it passed through after the tank: breaking, spinning, filtering, membrane, chromatography and drying. For each step, write one sentence on what it removed and what it might have lost. Then answer: which step do you think cost the maker most, and why?

Key ideas to carry forward

  • A secreted protein is harvested by removing the cells, while an inside protein must be broken out first.
  • Homogenisation breaks cells by force, enzymes break them gently, and a gentle break leaves debris that is easy to remove.
  • Centrifuges remove the bulk of solids cheaply, and filters polish the liquid that is left.
  • Ultrafiltration concentrates the protein and diafiltration washes out salts and sugars.
  • Chromatography sorts by sticking, and one gentle pass usually gives a food ingredient its purity.
  • Overall yield equals the product of the step yields, so a small loss at any step multiplies through the chain.
  • Drying turns the clean liquid into a powder, and every batch is tested before it is packed.

Glossary

Downstream processing

All the steps after the tank that turn broth into a clean ingredient. Food tech example: downstream processing filters, purifies and dries a fermentation made protein.

Secreted

Sent out of the cell into the broth by the microbe itself. Food tech example: a secreted whey protein is harvested by removing the cells.

Homogeniser

A machine that forces cells through a narrow gap at high pressure to break them open. Food tech example: a homogeniser breaks yeast cells to release a protein kept inside.

Clarified liquid

Broth with the solids removed, holding the protein in solution. Food tech example: clarified liquid goes to ultrafiltration after the centrifuge.

Ultrafiltration

Filtering through a membrane that keeps large molecules and lets water pass. Food tech example: ultrafiltration concentrates a protein liquid before drying.

Diafiltration

Washing a liquid by adding fresh water during ultrafiltration to flush out small molecules. Food tech example: diafiltration washes salts and sugars out of a protein liquid.

Chromatography

A separation method that sorts molecules by how they stick to a material in a column. Food tech example: chromatography purifies a fermentation made protein in a column.

Endotoxin

A fragment of a bacterial wall that can cause fever if it reaches the body. Food tech example: a finished ingredient is tested for endotoxins before sale.

Spray drying

Drying by blowing a liquid as a fine mist into a hot chamber. Food tech example: spray drying turns a purified protein liquid into a powder.

Freeze drying

Drying by freezing a liquid and removing water as vapour. Food tech example: freeze drying is gentler than spray drying but costs more.

Cut off

The size above which a membrane holds a molecule back. Food tech example: a membrane cut off is set below the size of the target protein.

Fouling

The sticking of proteins and fats to a membrane, which slows the flow. Food tech example: a fouled membrane is cleaned with a wash cycle.

Yield

The share of the product that survives a step or a whole process. Food tech example: four steps at 90 percent each give an overall yield of about 66 percent.

Polishing

The last cleaning steps before drying, such as sterile filtration. Food tech example: polishing removes the last traces of microbes from a protein liquid.

Bead milling

Breaking cells open by stirring broth with tiny glass beads. Food tech example: bead milling is gentler than homogenisation but slower.

References

  • Stanbury, P. F., Whitaker, A. and Hall, S. J., Principles of Fermentation Technology, 3rd edition, 2017.
  • Bamforth, C. W., Food, Fermentation and Micro-organisms, 2005.
  • Waites, M. J., Morgan, N. L., Rockey, J. S. and Higton, G., Industrial Microbiology: An Introduction, 2001.
  • Demain, A. L. and Davies, J. E., Manual of Industrial Microbiology and Biotechnology, 2nd edition, 1999.
  • Damodaran, S., Parkin, K. L. and Fennema, O. R., Fennema's Food Chemistry, 4th edition, 2007.
  • Fellows, P. J., Food Processing Technology, 4th edition, 2017.

Unit 10: Fats and Oils by Fermentation in Depth

Note

Unit 10: Fats and Oils by Fermentation in Depth

A question to start with

A chocolate bar snaps, a croissant flakes, a spread melts on warm toast: each of those moments is the work of a fat, and fats are among the hardest ingredients to replace. Most food fats come from plants and animals, and the most wanted ones, cocoa butter, palm oil and butterfat, come with land, climate and supply questions. Fermentation offers another route: microbes that store oil inside their cells, grown in tanks, harvested and pressed like a crop that never sees a field. This unit explains those oil making microbes, how they are grown, how their oil is taken out and shaped, and where fermentation fats fit in real foods.

Why fats matter, and why they are hard to replace

A fat does more in a food than add calories. It carries flavour, because many flavour molecules dissolve in fat and not in water. It gives texture: the snap of chocolate, the flake of pastry, the creaminess of ice cream are all set by which fats are present and how they crystallise. It gives structure: a fat that is solid at room temperature holds a biscuit together, while a fat that melts at body temperature disappears on the tongue. A fat that melts too early makes a chocolate bar soft in the hand. A fat that melts too late makes it waxy in the mouth.

Fats are judged by their fatty acids, the building blocks of a fat molecule. A fat rich in saturated fatty acids is usually solid at room temperature, while a fat rich in unsaturated fatty acids is usually liquid. Cocoa butter is prized because its fatty acids are arranged so it is solid in the hand and melts sharply at body temperature. Palm oil is prized because it is solid, cheap and stable. Butterfat is prized because its mix of fatty acids gives a flavour no other fat matches. A replacement fat must match not only the melting point, but the crystal form, the flavour and the stability, and that is a harder test than matching a protein.

Fermentation fats are made to match those tests. The microbe is chosen or tuned so its oil holds the right fatty acids in the right proportions, and the oil is then shaped by the same tools used for plant oils: fractionation to separate the solid part from the liquid part, and interesterification to rearrange the fatty acids on the fat molecule. The goal is not a new fat. The goal is a familiar fat, made by a new route.

Oleaginous microbes: cells that store oil

Some microbes store oil the way a seed stores it: as droplets inside the cell, saved for later. They are called oleaginous, which means oil bearing, and they include yeasts, fungi and microalgae. A yeast grown for oil can hold 40 to 70 percent of its dry weight as oil, far more than a seed, and it can do it in a day, while a crop takes months. The oil sits in droplets, and the droplets grow when the microbe has plenty of carbon, usually sugar, and runs short of another nutrient, usually nitrogen.

The nitrogen trick is the heart of oil production. When nitrogen runs out, the microbe can no longer make protein, so it stops dividing and turns the sugar it keeps eating into oil instead. The operator therefore feeds the tank in two stages: first a nitrogen rich feed, so the cells grow and multiply, then a nitrogen poor feed with plenty of sugar, so the cells fill with oil. The switch is watched by measuring the nitrogen in the broth: when it falls to near zero, the oil stage has begun, and the sugar feed is kept going until the cells are full.

Microalgae make oil by a different trick: they use sunlight and carbon dioxide, and they store oil when they are stressed by too much light or too little nitrogen. Algal oil is rich in long chain unsaturated fatty acids, which suit nutrition products, while yeast and fungal oils are richer in the saturated and monounsaturated fatty acids that suit chocolate and spreads. Each microbe is a trade off between speed, oil content, fatty acid pattern and cost, and the choice is made for the food the oil will go into, not for the microbe alone.

From sugar to oil: the process steps

The process runs in four steps. First, the microbe is grown in a bioreactor on a sugar feed, with air, temperature and pH controlled, until the tank is full of cells. Second, the nitrogen is allowed to run out, and the sugar feed continues, so the cells fill with oil droplets. Third, the broth is harvested, and the cells are separated in a centrifuge and dried, because a wet cell is hard to press. Fourth, the oil is taken out of the dried cells, by pressing, by a solvent, or by breaking the cells open and spinning the oil out.

A worked example shows the arithmetic. A tank holds 100,000 litres of broth, and the cells grow to 50 g of dry cells per litre, which is 5,000 kg of dry cells. If the cells hold 50 percent oil, the tank holds 2,500 kg of oil. If the extraction recovers 90 percent of that oil, the tank gives 2,250 kg of oil per run. The sugar needed to make that oil is far larger than the oil itself, because the cells burn much of the sugar for energy and for growth, so the yield of oil per kilogram of sugar is the number the maker watches most, and a strain that lifts that yield by a tenth cuts the cost of every kilogram of oil.

The sugar source matters as much as the strain. Sugar from a crop carries a farm footprint, and the oil is only as sustainable as the sugar that fed it. Side streams, such as the sugars left from other food processes, can feed the tank more cheaply and with a smaller footprint, but they vary in quality, and a variable feed makes a variable oil. The maker tests each feed, measures the oil yield and the fatty acid pattern, and chooses the feed that gives a steady oil at a price the product can carry.

Shaping the oil: fractionation and interesterification

The oil that comes out of the cells is rarely the oil that goes into the food. It is a mix of fatty acids, and it must be shaped. The first shaping tool is fractionation: the oil is cooled slowly, so the fats that solidify first form crystals, and the solid crystals are filtered away from the liquid that stays behind. The solid fraction, called the stearin, is hard and suits chocolate and spreads. The liquid fraction, called the olein, is soft and suits frying and dressings. One oil, split by temperature, gives two ingredients with different jobs.

The second shaping tool is interesterification, which rearranges the fatty acids on the fat molecule. A fat molecule holds three fatty acids on a backbone, and moving them between molecules changes how the fat crystallises and melts, without changing which fatty acids are present. Interesterification can turn a liquid oil into a solid fat that melts at body temperature, or soften a hard fat so it spreads. It is done with an enzyme or a chemical catalyst, and the enzyme route is preferred for food, because it is gentler and it leaves no chemical residue.

The third shaping tool is blending: mixing the fermentation oil with a plant oil to tune the melting point, the flavour and the cost. A fermentation oil that matches cocoa butter closely can be blended with a small amount of cocoa butter to give the exact snap a chocolate maker wants, while most of the fat comes from the tank. Blending is honest when the label says so, and it is practical, because a blend is cheaper than a pure fermentation fat and more reliable than a pure plant fat when the crop is short.

Nutrition, stability and the fat that keeps

A fermentation fat is judged like any fat: by its fatty acids, its stability and its flavour. The fatty acid pattern decides the nutrition: a fat rich in unsaturated fatty acids is usually judged better for the heart than a fat rich in saturated fatty acids, but a fat that must be solid, such as a chocolate fat, needs saturated fatty acids to stay solid, so the maker balances the nutrition against the function. The pattern also decides the stability: unsaturated fats oxidise faster, going rancid in storage, while saturated fats keep longer. An oil rich in unsaturated fatty acids needs protection, such as storage away from light and air, and sometimes a natural antioxidant.

Rancidity is the main enemy of a stored oil. Oxygen attacks the unsaturated fatty acids, making off flavours that smell stale and painty. The defence is to keep the oil cool, dark and sealed, to remove oxygen from the pack, and to test the oil for oxidation with a simple laboratory test that measures the peroxides formed. A fat that fails the test is not used, whatever its price, because a rancid fat ruins the whole food it goes into, and the shopper tastes it at once.

The nutrition question that matters most for fermentation fats is whether the oil carries the long chain fatty acids that the body needs, such as the omega-3 fats found in fish oil. Microalgae make those fats naturally, and algal oil is already used in foods and supplements for that reason. Yeast and fungal oils are poorer in those fats, and they are used where structure matters more than nutrition, such as chocolate and pastry. The right oil for a food is the one whose fatty acids fit the job, not the one with the longest list of nutrients.

Using fermentation fats in real foods

Fermentation fats are used where a plant or animal fat is hard to get, hard to afford, or hard to defend. In chocolate, a fermentation fat that matches cocoa butter lets a maker keep the snap and the melt when cocoa is dear or short. In spreads, a fermentation fat gives a solid spread that melts on toast, without the palm oil that some shoppers avoid. In bakery, a fermentation fat gives a flaky pastry and a tender crumb, because it holds its shape during mixing and melts during baking. In dairy alternatives, a fermentation fat gives the creaminess that a plant milk lacks.

The maker tests each use in the real food, not only in the laboratory. A fat that melts perfectly in a test tube may bloom in a chocolate bar, forming white streaks on the surface as the crystals change form in storage. A fat that spreads well at room temperature may run in a warm kitchen. A fat that tastes clean on its own may carry a faint yeasty note into a delicate cream. The tests are the same as for any fat: melt it, cool it, store it, taste it, and check the crystals under a microscope if the bar blooms.

The future of fermentation fats depends on the sugar question and the scale question. If the sugar is cheap and clean, the fat is cheap. If the tank is large and the strain is good, the fat is plentiful. If either fails, the fat stays a niche ingredient for premium foods. The technology is ready in the tank. The work now is in the feedstock, the yield and the price, and in proving to a chocolate maker that a fat from a tank snaps like a fat from a bean.

Questions to think about

  • Why does the operator let nitrogen run out in the second stage of oil production, and what would happen if nitrogen stayed plentiful?
  • A tank holds 100,000 litres at 50 g of dry cells per litre, the cells hold 50 percent oil, and extraction recovers 90 percent. How much oil does one run give?
  • Why must a fermentation fat that matches cocoa butter still be tested in a real chocolate bar, not only in the laboratory?
Think about the oil arithmetic before you read on

The tank holds 100,000 litres times 50 g per litre, which is 5,000 kg of dry cells. At 50 percent oil, the cells hold 2,500 kg of oil, and at 90 percent recovery the run gives 2,250 kg of oil. The sugar needed is far larger than the oil itself, because the cells burn much of the sugar for energy and growth, so the oil yield per kilogram of sugar is the number the maker watches most.

From sugar to shaped fat

Grow the oleaginous yeast on a nitrogen rich feed until the tank is full
↓
Let nitrogen run out and keep feeding sugar so the cells fill with oil
↓
Harvest, spin out the cells and dry them
↓
Take the oil out by pressing or by breaking the cells open
↓
Shape the oil: fractionate into stearin and olein, or interesterify it
↓
Test the fat: melting, crystals, stability and flavour, then pack it

The diagram shows fermentation fat making in order: from growing cells to a shaped, tested fat. The nitrogen switch is the step that turns growth into oil storage.

Practical note: In a practical session, compare three fats: butter, a hard vegetable fat and a liquid oil. Put a small piece of each on a plate at room temperature and note how long each takes to soften. Melt each gently, pour it into a small dish, cool it in a fridge, and note the texture when it sets. Record the melting behaviour of each, and answer: which fat would you choose for a chocolate bar, which for a spread, and which for frying, and what does the melting point tell you about each choice?

Quick self-check

Question: Question: What is an oleaginous microbe?

A microbe that stores oil inside its cells, as droplets saved for later. A yeast grown for oil can hold 40 to 70 percent of its dry weight as oil, and it fills with oil when it has plenty of sugar and runs short of nitrogen.

Question: Question: What does interesterification do, and why is the enzyme route preferred for food?

It rearranges the fatty acids on the fat molecule, changing how the fat crystallises and melts without changing which fatty acids are present. The enzyme route is preferred because it is gentler and leaves no chemical residue.

Try it yourself

Read the fat list on a chocolate bar, a spread and a pastry in a shop. Write down each fat named, and note whether it is solid or liquid at room temperature. Then answer: which of those fats could a fermentation fat replace, and what would the replacement need to match: the melting point, the crystal form, the flavour, or all three?

Key ideas to carry forward

  • A fat carries flavour, texture and structure, and a replacement must match its melting, its crystals and its flavour.
  • Oleaginous microbes store oil as droplets, and they fill with oil when sugar is plentiful and nitrogen runs out.
  • The process grows cells first, then switches them to oil storage, then harvests, dries and extracts the oil.
  • Oil yield per kilogram of sugar is the number the maker watches most.
  • Fractionation splits an oil into a solid stearin and a liquid olein by cooling.
  • Interesterification rearranges fatty acids on the fat molecule to change melting and crystallising.
  • Fermentation fats suit chocolate, spreads, bakery and dairy alternatives, and each use must be tested in the real food.

Glossary

Oleaginous

Oil bearing: a microbe that stores oil inside its cells. Food tech example: an oleaginous yeast can hold half its dry weight as oil.

Fatty acid

A building block of a fat molecule, which decides whether a fat is solid or liquid. Food tech example: cocoa butter is prized for its fatty acid pattern.

Stearin

The solid fraction of an oil, separated by cooling. Food tech example: stearin from a fermentation oil suits chocolate and spreads.

Olein

The liquid fraction of an oil, separated by cooling. Food tech example: olein from a fermentation oil suits frying and dressings.

Fractionation

Splitting an oil into solid and liquid parts by cooling and filtering. Food tech example: fractionation gives a hard fat and a soft oil from one fermentation oil.

Interesterification

Rearranging the fatty acids on a fat molecule to change how it melts. Food tech example: interesterification turns a liquid oil into a solid fat.

Rancidity

The stale, painty flavour that forms when unsaturated fats oxidise. Food tech example: an oil stored in light and air goes rancid.

Peroxide

A substance formed when a fat oxidises, measured to test rancidity. Food tech example: a peroxide test shows whether a stored oil is still fresh.

Omega-3 fat

A long chain unsaturated fat the body needs, found in fish oil and algal oil. Food tech example: microalgae make omega-3 fats used in foods and supplements.

Nitrogen limitation

Running short of nitrogen so a microbe stops dividing and stores oil instead. Food tech example: nitrogen limitation switches a yeast from growth to oil storage.

Cocoa butter equivalent

A fat that behaves like cocoa butter in chocolate. Food tech example: a fermentation fat is tested as a cocoa butter equivalent.

Bloom

White streaks on chocolate formed when fat crystals change form in storage. Food tech example: a fat that blooms in a chocolate bar fails the storage test.

Saturated fatty acid

A fatty acid that makes a fat solid at room temperature. Food tech example: a spread needs saturated fatty acids to stay solid.

Unsaturated fatty acid

A fatty acid that makes a fat liquid at room temperature. Food tech example: unsaturated fats oxidise faster, so they need protection in storage.

Antioxidant

A substance that slows oxidation and protects a fat from going rancid. Food tech example: a natural antioxidant is added to an oil rich in unsaturated fats.

References

  • Stanbury, P. F., Whitaker, A. and Hall, S. J., Principles of Fermentation Technology, 3rd edition, 2017.
  • Bamforth, C. W., Food, Fermentation and Micro-organisms, 2005.
  • Waites, M. J., Morgan, N. L., Rockey, J. S. and Higton, G., Industrial Microbiology: An Introduction, 2001.
  • Demain, A. L. and Davies, J. E., Manual of Industrial Microbiology and Biotechnology, 2nd edition, 1999.
  • Damodaran, S., Parkin, K. L. and Fennema, O. R., Fennema's Food Chemistry, 4th edition, 2007.
  • Fellows, P. J., Food Processing Technology, 4th edition, 2017.

Unit 11: Precision Fermentation Compared with Traditional Production

Note

Unit 11: Precision Fermentation Compared with Traditional Production

A question to start with

A whey protein made by a yeast and a whey protein made by a cow are the same molecule, but they arrive by very different routes. One comes from a tank fed on sugar, the other from a field, a cow and a dairy. Which route is better? The answer is not a slogan. It depends on what is compared, how it is counted, and what the eater values. This unit compares the two routes fairly, on the same measures, and shows how to read a comparison claim without being misled.

What is being compared: the same molecule, a different route

A fair comparison starts by naming what is the same and what is different. What is the same is the molecule: a fermentation made whey protein and a dairy whey protein hold the same amino acids in the same order, and they behave the same in a foam, a gel or a drink. What is different is the route: the fermentation route grows a microbe in a tank on sugar, while the traditional route grows a crop, feeds an animal, milks it, and separates the whey from the milk. The molecule is the same. The chain behind it is not.

The comparison must therefore compare chains, not molecules. The fermentation chain is: sugar crop, sugar refining, fermentation tank, purification, drying. The dairy chain is: feed crop, animal rearing, milking, cheese making, whey separation, drying. Each chain has inputs, land, water, energy, feed, and outputs, food and gases. A study that compares only the tank with the cow is not comparing the routes at all. It is comparing a step with a system.

The functional unit, explained in the alternative proteins subject, is what makes the comparison fair: one kilogram of protein, or one kilogram of digestible protein, from each route. Without it, a kilogram of milk, which holds little protein, is compared with a kilogram of powder, which holds much, and the answer is meaningless. The reader should look for the functional unit first, before reading any number.

The traditional route: field, animal and dairy

The traditional route for a dairy protein starts in a field. A crop is grown for feed, using land, water, fertiliser and tractor fuel. The feed is eaten by a cow, which turns part of it into milk and part into body heat, movement and waste. The milk is collected, cooled and carried to a dairy, where it is separated: the curd becomes cheese, and the liquid becomes whey. The whey is filtered, concentrated and dried into a powder. Each step is well known, the technology is mature, and the chain feeds millions of people every day.

The strengths of the traditional route are scale, familiarity and the whole animal. A dairy farm makes milk, meat, leather and manure, and the milk carries not only protein but fat, lactose, calcium and vitamins. The chain is spread across many farms, so a failure in one place does not stop the supply. The shopper knows the food, trusts it, and eats it without a new label to learn. The weaknesses are the footprint of the animal, the land the feed needs, the methane the animal releases, and the slow pace of change: a cow cannot be redesigned in a year, and a farm cannot move.

The traditional route for other ingredients is different in detail but the same in shape. An enzyme from a plant is grown, harvested and extracted. A colour from a crop is grown and pressed out. A sweetener from a plant is grown and refined. Each traditional route depends on a crop and a season, and each one carries the footprint of the field. The comparison with fermentation is therefore a comparison of a field and a tank, and the tank wins only where the field is the problem.

The fermentation route: sugar, strain and tank

The fermentation route starts with a sugar crop, because the microbe must eat. The sugar is refined, sterilised and fed to the tank, where the strain makes the protein. The broth is harvested, the protein is purified, and the powder is dried. The chain is short and it runs all year, in any weather, on a small floor area. The tank does not need a field for the protein itself, only for the sugar that feeds it, and the sugar can come from a side stream, such as the sugars left from another food process, which cuts the farm share further.

The strengths of the fermentation route are precision, consistency and independence from seasons. The protein is one molecule, made the same way every time, with no variation between batches beyond the tests. The tank runs day and night, all year, and it can be built near the factory that uses the protein, cutting transport. A new protein can be designed and scaled in years, not in the decades a breeding programme takes. The weaknesses are the cost of the sugar, the cost of the tank and the purification, and the need for a safety file and an approval before the first sale.

The fermentation route is also honest about its dependence: it still needs a crop for sugar, and a sugar crop still needs land, water and fertiliser. A fermentation protein made from sugar grown on cleared forest is not a low footprint protein, whatever the tank does. The footprint of the sugar is the largest share of the fermentation footprint, and a study that hides it is hiding the answer. The tank is clean. The sugar decides whether the chain is clean.

Comparing on five measures: land, water, energy, cost and consistency

Land is the measure where fermentation usually wins. A tank stacks production vertically and runs all year, so it needs far less land per kilogram of protein than a field and an animal. The land it does need is the land for the sugar crop, and if the sugar comes from a side stream, the land share falls further. Water is closer: fermentation uses water in the tank and in purification, and the sugar crop uses water in the field, so the total can be similar to the traditional route, and it depends on where the water is taken and how much is cleaned and reused.

Energy is the measure where the answer is least clear. A tank needs power for stirring, air, cooling and drying, and drying is often the largest energy step. A traditional dairy needs energy for the farm, the milking, the cooling and the drying too, and the two totals can be close. The source of the power decides the climate answer: a tank on renewable power carries a small climate footprint, while a tank on coal power does not, and the same is true of a dairy. The energy number without the power source is a number without a meaning.

Cost is the measure that decides the market, and it moves with scale. At small scale, fermentation is dear, because the tank, the strain and the purification are paid for by a small output. At large scale, the cost falls, because the same tank makes more, and the sugar is bought in bulk. The traditional route is already at large scale, so its cost is steady and low. A fermentation protein that costs twice as much as the traditional one will sell only where the buyer values something the traditional one cannot give: a protein without an animal, a protein that is purer, or a protein that the traditional route cannot make at all. Consistency is the measure where fermentation wins clearly: the same molecule, the same purity, every batch, while a crop varies with the season and an animal varies with the feed.

When traditional wins, and when fermentation wins

Traditional production wins where the crop is cheap, the extraction is simple, and the ingredient is plentiful. Sugar from cane, starch from maize, oil from a seed: these are cheap because the crop is grown at scale and the extraction is a press or a wash. Fermentation cannot beat a cheap crop at making a cheap molecule, and it should not try. A maker who ferments a protein that a field makes for pennies is spending money to solve a problem that does not exist.

Fermentation wins where the traditional route is rare, dear, slow or tied to an animal. A whey protein without a cow, an egg white protein without a hen, a sweet protein that a plant makes only in tiny amounts, a colour that a crop makes only in one season: these are the wins. Fermentation also wins where purity matters, because the tank makes one molecule and the purification takes it out clean, while a crop extract carries a mix that must be separated. It wins where the shopper wants a food without an animal, and where the supply of the traditional ingredient is short or unreliable.

The honest answer to which route is better is therefore: it depends on the molecule. A food technologist does not choose a route for all foods. The technologist chooses a route for one ingredient, for one product, for one eater, and the choice is revisited as prices, power and scale change. A fermentation protein that is dear today may be cheap in five years, when the tank is larger and the sugar is cheaper. A traditional ingredient that is cheap today may grow dear when the crop fails. The comparison is a snapshot, not a verdict for all time.

How to read a comparison claim without being misled

A comparison claim can be checked with six questions. First, what is the functional unit: per kilogram of product, per kilogram of protein, or per kilogram of digestible protein? Second, what is the boundary: does it include the sugar crop, the farm, the factory, the shop and the waste? Third, is the same thing compared: the same molecule, the same amount of protein, the same use? Fourth, which gases are counted, and are methane and nitrous oxide in the number? Fifth, what was left out, and does the study say so? Sixth, who paid for the study, and was it checked by an independent reviewer?

A claim that answers those questions can be argued about, and that is a good sign. A claim that cannot answer them is an advertisement. The reader should also watch for three tricks. The first is comparing a tank with a cow, a step with a system. The second is comparing a kilogram of powder with a kilogram of milk, a different amount of protein. The third is counting the sugar as free, as if the microbe ate nothing. Each trick makes fermentation look better than it is, and each one is caught by the six questions.

The last check is the most practical: what would change the answer? If the sugar came from a side stream instead of a crop, how far would the footprint fall? If the power came from the sun instead of coal, how far would the climate number fall? If the tank doubled in size, how far would the cost fall? A maker who can answer those questions knows the process deeply. A reader who can ask them cannot be misled by a single number, because the reader is looking at the system, not the slogan.

Questions to think about

  • Why must a comparison use one kilogram of protein as the functional unit, rather than one kilogram of product?
  • A fermentation protein is made from sugar grown on cleared forest land. Why might its footprint still look small if the study hides the sugar crop?
  • When does traditional production win over fermentation, and when does fermentation win? Give one example of each.
Think about the cleared forest before you read on

The sugar crop is the largest share of the fermentation footprint, because the microbe must eat sugar and the sugar comes from a field. A study that draws its boundary at the tank gate leaves the sugar crop out, so the footprint looks small while the forest clearing, the fertiliser and the water of the crop are never counted.

Comparing two routes to the same protein

Name the molecule and the functional unit: one kilogram of protein
↓
Draw the fermentation chain: sugar crop, tank, purification, drying
↓
Draw the traditional chain: feed crop, animal, dairy, separation, drying
↓
Count land, water, energy and gases for each chain, with the same boundary
↓
Compare cost and consistency at the same scale
↓
Ask what would change the answer: sugar source, power source, tank size

The diagram shows a fair comparison in order: same molecule, same unit, same boundary, then the five measures, then the check on what would change the answer.

Practical note: In a practical session, take a food you eat, such as a yoghurt, and sketch its two chains on paper: the traditional chain from field to pot, and a fermentation chain from sugar to tank to pot. Mark the step in each chain you think is the hotspot, and write one sentence for why. Then swap sketches with a classmate and compare: did you draw the same boundary, and did you count the sugar crop in the fermentation chain?

Quick self-check

Question: Question: Why is comparing a tank with a cow a mistake?

Because a tank is a step and a cow is part of a system. A fair comparison compares whole chains: the sugar crop, the tank, purification and drying against the feed crop, the animal, the dairy, separation and drying, with the same boundary and the same functional unit.

Question: Question: Where does fermentation win clearly over traditional production?

On consistency: the tank makes the same molecule at the same purity every batch, while a crop varies with the season and an animal varies with the feed. It also wins where the traditional ingredient is rare, dear, or tied to an animal.

Try it yourself

Find a claim that a fermentation made food is better for the planet, on a pack or a website, and test it with the six questions in the last section. Write down which questions it answers and which it leaves blank, and answer: is the claim a finding you can check, or a slogan? If you cannot find such a claim, test a claim for a plant based food the same way.

Key ideas to carry forward

  • A fair comparison compares chains, not molecules, with the same functional unit and the same boundary.
  • The fermentation chain is sugar crop, tank, purification and drying, and the sugar crop is its largest footprint share.
  • The traditional chain is feed crop, animal, dairy, separation and drying, and it is mature, familiar and spread across many farms.
  • Fermentation usually wins on land and consistency, and the energy answer depends on the power source.
  • Traditional wins where a crop is cheap and extraction is simple; fermentation wins where the ingredient is rare, dear or tied to an animal.
  • A comparison claim must answer six questions about unit, boundary, sameness, gases, omissions and review.
  • The comparison is a snapshot: prices, power and scale change the answer over time.

Glossary

Functional unit

The amount of function being compared, such as one kilogram of protein. Food tech example: comparing two foods per kilogram of protein is fairer than per kilogram of product.

Boundary

The line that says where a life cycle assessment starts and stops counting. Food tech example: a study boundary that stops at the tank gate leaves out the sugar crop.

Chain

The series of steps that makes a food, from raw material to finished product. Food tech example: the dairy chain runs from feed crop to cow to dairy to powder.

Side stream

A stream of material that leaves a process beside the main product, such as starch or fibre. Food tech example: sugar from a side stream feeds a fermentation tank more cheaply.

Methane

A gas that warms the planet far more per kilogram than carbon dioxide over a short period. Food tech example: methane from animals is counted in the climate footprint of meat.

Nitrous oxide

A gas released from fertiliser and soil that warms far more per kilogram than carbon dioxide. Food tech example: nitrous oxide from fertilised fields is counted in a crop footprint.

Carbon dioxide equivalent

A unit that converts the warming effect of different gases into the same scale. Food tech example: methane is converted to carbon dioxide equivalent so it can be added to the total.

Renewable power

Energy from sources that renew, such as sun, wind and water. Food tech example: a fermentation plant that switches to renewable power cuts its footprint.

Consistency

Making the same product the same way every time. Food tech example: a fermentation tank makes the same protein at the same purity every batch.

Scale

The size of production, which decides cost per kilogram. Food tech example: a fermentation protein is dear at small scale and cheaper at large scale.

Hotspot

The step in a chain where most of the footprint sits. Food tech example: the sugar crop is often the hotspot for a fermentation protein.

Independent review

Checking of a study by experts who did not carry it out. Food tech example: a comparison checked by independent review is easier to trust.

Traditional production

Making a food by farming, rearing and extraction, as it has long been done. Food tech example: traditional whey comes from milk, curd and separation.

Digestible protein

Protein the body can absorb and use. Food tech example: a footprint per kilogram of digestible protein adjusts for how much protein the body can use.

References

  • Stanbury, P. F., Whitaker, A. and Hall, S. J., Principles of Fermentation Technology, 3rd edition, 2017.
  • Bamforth, C. W., Food, Fermentation and Micro-organisms, 2005.
  • Waites, M. J., Morgan, N. L., Rockey, J. S. and Higton, G., Industrial Microbiology: An Introduction, 2001.
  • Demain, A. L. and Davies, J. E., Manual of Industrial Microbiology and Biotechnology, 2nd edition, 1999.
  • Damodaran, S., Parkin, K. L. and Fennema, O. R., Fennema's Food Chemistry, 4th edition, 2007.
  • Fellows, P. J., Food Processing Technology, 4th edition, 2017.

Unit 12: The Future Product Pipeline

Note

Unit 12: The Future Product Pipeline

A question to start with

The precision fermentation shelf today holds a handful of ingredients: a whey protein, an egg white protein, a few enzymes, a sweet protein, a colour. Behind that shelf sits a pipeline, a queue of products at different stages, from an idea in a laboratory to a tank at pilot scale to a safety file waiting for a decision. This unit looks down that pipeline, explains the stages a product passes through, names the bottlenecks that slow it, and shows what to watch for as the next products arrive.

What is in the pipeline now

The pipeline holds products in groups, and each group is at a different stage. The dairy group is furthest along: whey proteins are already sold, casein proteins are being built one by one, and the next step is a casein that melts and stretches like dairy cheese, and a milk fat made by fermentation to give creaminess without a cow. The egg group follows: egg white protein is already sold for foaming and baking, and the next step is a whole egg replacer that scrambles and sets like an egg, and a yolk protein that gives colour and richness.

The sweet group holds sweet proteins that taste sweet without sugar, made in tiny amounts because a little goes a long way, and the next step is a sweet protein cheap enough for a drink, not only for a premium food. The colour group holds reds, yellows and blues made by fermentation, which replace colours from insects or from crops that vary with the season, and the next step is a blue that stays bright in a drink and a red that survives baking. The fat group, covered in depth in Unit 10, holds cocoa butter and palm oil alternatives, and the next step is a fat that is cheap enough for everyday foods.

The newest groups are the most ambitious. One holds human milk proteins and the sugars of human milk, made for infant foods, where the safety file is the longest and the testing is the strictest. Another holds collagen, the protein of skin and joints, made for foods and drinks that promise firmness and health, where the claim must be proved, not only printed. A third holds flavours, the small molecules that give a food its taste, made by fermentation instead of by extraction from a rare plant. Each group is a different bet, and each bet is judged on the same three things: does it work, does it pay, and is it safe?

From idea to shelf: the stages in order

A product moves through six stages, and each stage has its own test. The first stage is discovery: a scientist finds a protein or a molecule that does a job a food needs, and shows it works in a test tube. The second stage is strain building: the gene is put into a host, the strain is tuned, and the protein is made in a flask, as Unit 8 explains. The third stage is pilot scale: the strain is grown in a 100 litre tank, then a 1,000 litre tank, and the process is tuned at each size, as Unit 5 explains.

The fourth stage is the safety file: the product is tested for identity, purity, microbes, toxins and allergy, and the file is written with the evidence, as Unit 6 explains. The fifth stage is approval: the file is sent to the food safety authority, questions are answered, and a decision is made. The sixth stage is launch: the product is made at production scale, sold to a food maker, and put into a food that reaches a shopper. Each stage takes months or years, and a product can fail at any stage, so the pipeline holds many products at the early stages and few at the late ones.

A worked example shows the time. A strain that makes 1 g per litre in a flask may need two years of tuning to reach 5 g per litre in a pilot tank, a year of safety testing, and a year of approval, so five years can pass between the flask and the shelf. That time is not wasted: it is the time the safety file, the scale up and the cost are proved. A product that is rushed through the stages usually fails at the last one, when the cost or the safety file is found wanting, and the maker must go back to the start.

Bottlenecks: what slows the pipeline

The first bottleneck is scale. There are few large fermentation tanks built for food, and a product that needs a 100,000 litre tank must book time in a plant that may be busy with another product, or build its own plant, which costs far more than the strain. The second bottleneck is cost: a product that works in a flask may cost ten times too much at pilot scale, and the maker must tune the strain, the feed and the purification until the cost falls, or the product will never launch.

The third bottleneck is the safety file and the approval. A new protein needs feeding studies, allergy tests and a full characterisation, and each test takes months. The authority may ask questions that need new tests, and each round of questions adds months. The fourth bottleneck is the shopper: a product that is safe and cheap still fails if the eater does not buy it twice, and the maker must prove the taste and the price before the launch, not after.

The fifth bottleneck is the sugar supply. Every product in the pipeline eats sugar, and a pipeline of many products needs a large, steady, cheap supply of sugar, or of a side stream that can feed them all. A product that depends on a sugar that is dear or short will wait, while a product that can eat a side stream moves ahead. The bottlenecks are not secrets: they are the same five problems every fermentation product faces, and the makers who solve them first are the ones whose products reach the shelf.

What to watch: signals that a product is near

A reader who wants to know whether a product is near can watch for signals, and each signal marks a stage. The first signal is a pilot plant: a maker that builds or books a pilot tank is serious, because a pilot tank costs real money. The second signal is a partnership: a small fermentation maker that signs a deal with a large food maker has a customer waiting, and the food maker has tested the ingredient in a real food. The third signal is a safety filing: a maker that sends a file to a food safety authority has finished its testing and believes its evidence is ready.

The fourth signal is a launch in a small market: a product sold in one food, in one shop, in one region, is a test, and the maker is watching whether shoppers buy it twice. The fifth signal is a second product from the same maker: a maker that launches a second ingredient has learned how to run the pipeline, and the second product will move faster than the first. The sixth signal is a price fall: a product that was dear and becomes cheaper is scaling up, and a cheaper product reaches more foods.

The reader should also watch for the opposite signals. A product that is announced every year and never launches is stuck at a bottleneck, usually cost or scale. A product that launches and then disappears has failed the shopper test, usually taste or price. A product whose maker stops talking about it has often been dropped quietly. The pipeline is public in its announcements and private in its failures, so the reader must watch what happens, not only what is promised.

What the pipeline means for a food technologist

For a food technologist, the pipeline is a map of the skills that will be needed next. The first skill is strain thinking: understanding how a microbe is designed, so the technologist can talk to the strain builder and judge whether a protein can be made at all. The second skill is process thinking: understanding scale up and downstream processing, so the technologist can judge whether a protein can be made cheaply enough. The third skill is food thinking: understanding how a new ingredient behaves in a real food, in a foam, a gel, a melt or a bake, so the technologist can use it well.

The fourth skill is evidence thinking: reading a safety file, a footprint study and a comparison claim, and asking the questions this subject teaches, so the technologist is not misled by a slogan. The fifth skill is eater thinking: running a tasting, listening to a shopper, and knowing that taste and price decide whether a product lives or dies. A technologist who holds all five skills can take a pipeline product and turn it into a food that sells, and that is the job the pipeline is creating.

The pipeline also changes what a food technologist makes. A technologist who once chose between milk and soy may soon choose between milk, soy, fermentation made whey and fermentation made casein, and the choice will be made on function, cost, footprint and label, not on habit. The foods of the next decade will be blends of routes, as the alternative proteins subject explains, and the technologist who understands all the routes will design the blends. The pipeline is not a distant future. It is the next set of ingredients, arriving one product at a time.

A closing check: how to judge a pipeline claim

A pipeline claim can be judged with five questions. First, what stage is the product at: a flask, a pilot tank, a safety file, or a launch? Second, what is the evidence for that stage: a paper, a pilot run, a filing, or a product on a shelf? Third, what is the bottleneck, and has the maker named it honestly? Fourth, who is the customer: a food maker with a real recipe, or only an investor with a hope? Fifth, what would prove the claim wrong: a failed pilot, a safety question, or a shopper who does not buy twice?

A claim that answers those questions is a claim worth following. A claim that cannot answer them is a story, and stories do not feed people. The reader who asks the five questions will follow the pipeline with clear eyes, celebrating the products that pass each stage and letting go of the ones that do not, without disappointment, because the pipeline is wide and the next product is already behind it. The future of precision fermentation is not one product. It is a queue of products, each judged on its evidence, each arriving when it is ready.

Questions to think about

  • Why does a product spend years between the flask and the shelf, and why is that time not wasted?
  • A maker announces a new protein every year but never launches it. What bottleneck is it most likely stuck at, and what signal would show it is moving?
  • Why is a partnership with a large food maker a strong signal that a fermentation product is near launch?
Think about the years between flask and shelf before you read on

The years are spent tuning the strain to make enough protein, scaling up through pilot tanks, testing safety, writing the safety file and waiting for approval. Each stage has its own test, and a product that is rushed usually fails at the last stage, when the cost or the safety file is found wanting, so the time is the proof, not the delay.

From idea to shelf

Discovery: find a protein that does a job a food needs
↓
Strain building: put the gene in a host and tune it in a flask
↓
Pilot scale: grow through 100 litre and 1,000 litre tanks
↓
Safety file: test identity, purity, microbes, toxins and allergy
↓
Approval: send the file, answer questions, wait for a decision
↓
Launch: make at production scale and put it into a food that sells

The diagram shows the six stages of the pipeline in order. A product can fail at any stage, so the pipeline holds many products early and few late.

Practical note: In a practical session, pick a food ingredient you use, such as an enzyme, a colour or a sweetener, and imagine it is being made by precision fermentation. Write a one page pipeline plan: what stage is it at today, what test must it pass next, what is its bottleneck, and what signal would show it is moving? Then ask a classmate to challenge your plan with the five questions in the last section, and note which question your plan answers worst.

Quick self-check

Question: Question: What are the six stages from idea to shelf?

Discovery, strain building, pilot scale, safety file, approval and launch. Each stage has its own test, and a product can fail at any stage, so the pipeline holds many products early and few late.

Question: Question: What are the five bottlenecks that slow the pipeline?

Scale, cost, the safety file and approval, the shopper, and the sugar supply. Each one can stop a product, and the makers who solve them first are the ones whose products reach the shelf.

Try it yourself

Choose a fermentation made ingredient sold near you, or one you have read about, and trace it back through the six stages: when was it discovered, when was the strain built, when did it reach pilot scale, when was its safety file sent, and when did it launch? Write down what you can find for each stage, and answer: which stage took longest, and what does that tell you about where the next product will wait?

Key ideas to carry forward

  • The pipeline holds dairy, egg, sweet, colour, fat, human milk, collagen and flavour products, each at a different stage.
  • A product passes six stages: discovery, strain building, pilot scale, safety file, approval and launch.
  • Five bottlenecks slow the pipeline: scale, cost, safety and approval, the shopper, and the sugar supply.
  • Signals that a product is near include a pilot plant, a partnership, a safety filing, a small launch, a second product and a price fall.
  • A food technologist needs five skills for the pipeline: strain, process, food, evidence and eater thinking.
  • The foods of the next decade will blend routes, and the technologist who understands all of them will design the blends.
  • A pipeline claim is judged by its stage, its evidence, its bottleneck, its customer and what would prove it wrong.

Glossary

Pipeline

The queue of products at different stages, from idea to shelf. Food tech example: the precision fermentation pipeline holds dairy, egg and sweet proteins.

Discovery

The first stage, finding a protein or molecule that does a job a food needs. Food tech example: discovery shows a whey protein works in a test tube.

Pilot scale

Making a product in medium sized tanks to test the process before production. Food tech example: pilot scale grows a strain in a 100 litre tank.

Safety file

The set of evidence that shows a new food is safe to eat. Food tech example: a safety file holds tests for identity, purity, microbes and allergy.

Approval

The decision by a food safety authority that a new food may be sold. Food tech example: approval follows the safety file and the authority's questions.

Launch

Putting a product on the market for the first time. Food tech example: a launch puts a fermentation made protein into a food in a shop.

Bottleneck

The step that slows a whole process down. Food tech example: a lack of large tanks is a bottleneck for the pipeline.

Partnership

A deal between a small maker and a large food company. Food tech example: a partnership shows a food maker has tested the ingredient in a real food.

Human milk protein

A protein found in human milk, made by fermentation for infant foods. Food tech example: human milk proteins face the longest safety file.

Collagen

The protein of skin, joints and connective tissue. Food tech example: fermentation made collagen is used in foods and drinks.

Strain building

Putting a gene into a host and tuning it to make a protein. Food tech example: strain building makes a yeast that produces whey protein.

Scale

The size of production, which decides cost per kilogram. Food tech example: a product is dear at pilot scale and cheaper at production scale.

Signal

A sign that a product is moving through the pipeline. Food tech example: a safety filing is a signal that testing is finished.

Food technologist

A person who designs and makes foods, choosing ingredients and processes. Food tech example: a food technologist turns a pipeline ingredient into a food that sells.

References

  • Stanbury, P. F., Whitaker, A. and Hall, S. J., Principles of Fermentation Technology, 3rd edition, 2017.
  • Bamforth, C. W., Food, Fermentation and Micro-organisms, 2005.
  • Waites, M. J., Morgan, N. L., Rockey, J. S. and Higton, G., Industrial Microbiology: An Introduction, 2001.
  • Demain, A. L. and Davies, J. E., Manual of Industrial Microbiology and Biotechnology, 2nd edition, 1999.
  • Damodaran, S., Parkin, K. L. and Fennema, O. R., Fennema's Food Chemistry, 4th edition, 2007.
  • Fellows, P. J., Food Processing Technology, 4th edition, 2017.