Industrial biotechnology

The oldest commercial use

Recombinant insulin was the first genetically engineered therapeutic to be approved, produced by putting the human insulin sequence into bacteria rather than extracting it from animal pancreases. It set the template for everything that followed: identify a protein worth making, choose a host organism that grows fast and cheap, express the gene, and spend most of the engineering effort on purification rather than on the genetics.

Cheese made the same transition quietly. Chymosin, the enzyme that curdles milk, was traditionally taken from calf stomachs. Fermentation-produced chymosin, expressed in engineered microbes, now accounts for a large share of hard cheese production in many countries, and it is usually the reason a cheese can be labelled vegetarian. The genetic engineering happens in a fermenter; nothing engineered ends up in the cheese.

Enzymes: the volume business

Industrial enzymes are proteins sold by the tonne to replace harsher chemistry or lower a process temperature. Proteases and lipases in laundry detergent let cold washes remove protein and fat stains, which is a direct energy saving in the home. Amylases do the same job in baking and in starch processing. Cellulases soften and finish textiles. Phytase added to pig and poultry feed releases phosphorus that the animal would otherwise excrete, reducing both the phosphate added to feed and the phosphorus in the manure.

Almost all of these are made by engineered production strains — filamentous fungi such as Aspergillus and Trichoderma, or Bacillus species — and almost all have been improved by directed evolution, where variants are generated and screened for stability at a target pH and temperature. Companies such as Novozymes and DSM built large businesses on exactly this, and the competitive question is rarely whether an enzyme exists but whether it survives the conditions of the customer's process.

Precision fermentation

The newer label covers the same mechanism aimed at food ingredients. Soy leghemoglobin, the heme protein that gives Impossible Foods' products their meat-like character, is produced in an engineered yeast rather than extracted from soy roots. Dairy proteins made by fermentation rather than by cows follow the same route. Vitamin B2 and several amino acids have been made this way for years, and the argument for it is straightforward: a fermenter decouples production from land, weather and animal husbandry.

The argument against is cost. A fermentation plant is capital-heavy, the feedstock is usually sugar, and the finished ingredient competes with commodities produced at enormous scale. Where the target molecule is high value and hard to source, fermentation wins comfortably. Where it is a bulk commodity, the margin is thin and the plant's utilisation rate matters more than the strain does.

Fuels, chemicals and materials

Fuel ethanol from starch is mature and mostly a matter of yeast performance and process efficiency. Cellulosic ethanol, made from crop residues and woody material, has been technically demonstrated repeatedly and has struggled commercially: the pretreatment needed to free the sugars is expensive, and the engineered yeasts that ferment five-carbon sugars such as xylose add complexity. It is a fair reminder that a working strain is not the same as a working business.

Chemicals have done better where the biological route is genuinely shorter. Bio-based 1,3-propanediol, produced from corn sugar by engineered microbes and used in fibres and personal care, is the standard example of a fermentation product displacing a petrochemical one on process merit. Farnesene, made by engineered yeast, feeds into fragrances, lubricants and cosmetics. In materials, recombinant structural proteins — spider-silk-like polymers among them — are being spun into textiles, still at specialty rather than commodity volumes.

The three things a plant is judged on

Fermentation has its own vocabulary, and it is worth knowing because it explains most decisions. Titre is how much product ends up in a litre of broth. Rate is how quickly it gets there. Yield is how much of the feedstock carbon ends up in the product rather than in biomass or carbon dioxide. A strain that looks excellent on one of the three can be useless on the other two, and improving one usually costs you another.

Downstream processing — separating and purifying the product — frequently costs more than the fermentation itself, which is why secreted products are preferred over ones locked inside cells. Contamination, foaming, oxygen transfer and heat removal all get worse as vessels get larger, so performance at laboratory scale predicts industrial performance only loosely.

How it is regulated

Contained use of engineered microorganisms is governed separately from deliberate release, with facility-level containment requirements rather than environmental risk assessment. Where the output is a food ingredient, it meets food law: a GRAS determination or notification in the United States, a novel food authorisation through EFSA in the European Union, or the equivalent national process elsewhere. Where it is an enzyme used as a processing aid, labelling rules differ by country, which is why the same cheese may be described differently in two markets.