Genetically engineered bacteria have been producing human insulin at industrial scale since the early 1980s, and the vast majority of insulin used by people with diabetes today traces back to this technology. The basic idea is straightforward: scientists insert a synthetic copy of the human insulin gene into a bacterium, and the bacterium’s own protein-making machinery reads those instructions and churns out insulin. Turning that simple concept into a safe, injectable medicine, though, required decades of refinement in molecular biology, fermentation, and purification.
How Scientists First Got Bacteria to Make Insulin
The story begins in the late 1970s, when a team led by Arthur Riggs, Keiichi Itakura, and Herbert Boyer at Genentech and the City of Hope National Medical Center set out to prove that a common laboratory bacterium, Escherichia coli, could manufacture a human protein. They had already demonstrated the principle with somatostatin, a small hormone, and turned to insulin next. Rather than extracting the insulin gene directly from human cells, the team built it from scratch: they chemically synthesized the DNA sequences encoding insulin’s two peptide chains (called A and B) and cloned each one separately into E. coli using a standard laboratory plasmid. To trick the bacteria into reading those foreign genes efficiently, they fused each synthetic gene to a native E. coli gene so the bacterium’s own machinery would transcribe and translate it. The insulin peptides were then split off from the bacterial protein, purified individually, and chemically combined. When the researchers tested the product, it behaved like authentic human insulin in immunological assays. That work was published in January 1979.1PubMed Central. Expression in Escherichia coli of chemically synthesized genes for human insulin
The leap from a laboratory proof-of-concept to an approved drug took only a few more years. In October 1982, the U.S. Food and Drug Administration approved Humulin, making it the first therapeutic product based on recombinant DNA technology ever cleared for human use.2Endocrine Reviews. Making, Cloning, and the Expression of Human Insulin Genes in Bacteria: The Path to Humulin Before that point, all injectable insulin came from the pancreases of slaughtered pigs and cattle. The supply was limited, extraction was messy, and some patients developed immune reactions to the animal proteins. Recombinant human insulin solved all three problems in one stroke.
What Happens Inside the Bacterium
Modern production methods have evolved considerably since the original two-chain approach, but the core logic is the same. An engineered DNA construct encoding human proinsulin (or a proinsulin-like precursor) is inserted into E. coli on a small circular piece of DNA called a plasmid. The bacterium is grown in a nutrient broth, and at a chosen moment, a chemical or temperature signal triggers the bacteria to begin reading the insulin gene and translating it into protein at high volume.
The bacteria make so much proinsulin that it piles up inside the cell in dense clumps known as inclusion bodies. These clumps are actually useful: they are easy to separate from everything else in the cell by simple centrifugation, giving manufacturers a head start on purification.3Peptide Science. Refolding of therapeutic proteins produced in Escherichia coli as inclusion bodies The trade-off is that the protein inside inclusion bodies is misfolded and biologically inactive. It has to be dissolved using chemical agents, then coaxed into folding correctly through a carefully controlled refolding step. After refolding, the connecting peptide (called the C-peptide) that links the A and B chains in proinsulin is snipped away by enzymes, typically trypsin and carboxypeptidase B, to yield mature two-chain insulin.4PubMed. Effects of citraconylation on enzymatic modification of human proinsulin using trypsin and carboxypeptidase B
Each of these steps requires careful optimization. The enzymatic cleavage, for instance, can accidentally chop insulin at the wrong sites if reaction conditions drift, producing inactive fragments that then have to be removed. One strategy to prevent this involves temporarily shielding vulnerable sites on the protein with a chemical group before cleavage and then removing the shield afterward, which increases the yield of correctly processed insulin.4PubMed. Effects of citraconylation on enzymatic modification of human proinsulin using trypsin and carboxypeptidase B
Squeezing More Insulin Out of Each Batch
Researchers have spent decades fine-tuning every variable in the fermentation process to push yields higher. The temperature at which bacteria grow before and after induction, the concentration of the chemical inducer, the sugar content of the growth medium, how dense the bacterial culture is when you flip the “on” switch, and even the timing of harvest all affect how much proinsulin ends up in those inclusion bodies.5PubMed Central. Optimizing proinsulin production in E. coli BL21 (DE3) using taguchi method and efficient one-step insulin purification by on-column enzymatic cleavage Under optimized high-cell-density fed-batch conditions, published yields have reached into the hundreds of milligrams per liter of culture broth for the final insulin product, and some groups have reported fusion-protein yields in the range of several grams per liter before the insulin portion is cleaved free.6Journal of Biotechnology. Temperature-induced production of recombinant human insulin in high-cell density cultures of recombinant Escherichia coli
One recent study reported recovering over 500 mg/L of human insulin in the soluble fraction of E. coli under optimized induction conditions, a notable figure because it suggests a meaningful amount of the protein can fold correctly even inside the bacterial cell, rather than all of it aggregating into inclusion bodies.7PubMed Central. A novel and more efficient biosynthesis approach for human insulin production in Escherichia coli (E. coli) Getting soluble protein is preferable because it eliminates the harsh dissolving and refolding steps, which are time-consuming and reduce overall yield. Still, most large-scale manufacturing continues to rely on inclusion-body routes because the downstream processing, while more complex, is well-characterized and reliable at scale.
New host strains also contribute to higher productivity. Researchers have engineered specialized E. coli strains and new expression vectors specifically designed for insulin production, achieving greater efficiency than standard laboratory strains.8PubMed. Expression and purification of recombinant human insulin from E. coli 20 strain
Why Purification Is the Hard Part
Growing bacteria that make insulin is, in a sense, the easy part. What makes or breaks a manufacturing process is everything that comes after: washing the inclusion bodies, dissolving and refolding the protein, cleaving the precursor, and then running the resulting insulin through multiple rounds of chromatographic purification to strip away host-cell proteins, misfolded variants, and other impurities.9PubMed Central. Downstream processing of recombinant human insulin and its analogues production from E. coli inclusion bodies Improvements to each step compound: one research group showed that optimizing just the inclusion-body washing protocol and the conditions for dissolving and chemically treating the precursor protein tripled the recovery of proinsulin before the refolding step even began.10PubMed. New and efficient purification process for recombinant human insulin produced in Escherichia coli
One contaminant deserves special mention: endotoxin. Endotoxins are fragments of the outer membrane of gram-negative bacteria like E. coli, and even tiny amounts can cause fever and dangerous inflammatory reactions if they end up in an injectable drug. Because insulin is made inside E. coli cells, the raw product is inevitably contaminated with endotoxin. Standard filtration methods that work well for purifying water are not effective for removing endotoxin from protein solutions. Instead, manufacturers use specialized adsorption materials that selectively bind endotoxin while letting insulin pass through.11PubMed. Chromatographic removal of endotoxin from protein solutions by polymer particles Advanced column materials can reduce endotoxin levels by more than a hundred-million-fold in a single pass, bringing contamination well below safety thresholds even when starting levels are extremely high.12PubMed. The efficient removal of endotoxins from insulin using quaternized polyethyleneimine-coated porous zirconia
Yeast and Other Alternatives to E. coli
E. coli is not the only microorganism used to make insulin. The two yeast species Saccharomyces cerevisiae (baker’s yeast) and Pichia pastoris are also widely used in commercial insulin production.13PubMed Central. Cell factories for insulin production Yeast cells are eukaryotic, meaning they share more of their cellular machinery with human cells than bacteria do. This matters because yeast can perform certain chemical modifications to proteins after they are made, and it can fold and secrete the insulin precursor into the growth medium rather than trapping it in inclusion bodies. That secretion step simplifies purification significantly.
E. coli’s advantages are speed and low cost: it doubles every 20 to 30 minutes, grows on cheap media, and is arguably the most thoroughly understood organism in molecular biology. Its drawbacks include the inclusion-body problem, the absence of eukaryotic protein-processing machinery, and the endotoxin issue. Yeast grows more slowly but compensates by secreting a more correctly folded product. P. pastoris in particular has become a favored platform for industrial-scale insulin manufacturing because it does not heavily modify secreted proteins with excess sugar chains, a problem that can occur with S. cerevisiae. Overall, the total productivity of bacterial and yeast systems has converged to a similar range, though each has distinct advantages depending on the specific insulin product being made.14BioMed Research International. Human Insulin: History, Recent Advances, and Expression Systems for Mass Production
Researchers have also begun testing other bacterial species. A recent study explored Pseudomonas fluorescens as a production platform using non-standard promoters to drive insulin expression, reaching titers of around 55 mg/L in a small bioreactor and demonstrating that it is possible to produce functional insulin without the usual chemical inducer used in E. coli systems.15Process Biochemistry. Production of recombinant human insulin using constitutive and non-IPTG inducible promoters in Pseudomonas fluorescens and scale-up study These alternatives are still in early stages compared with the established E. coli and yeast platforms, but they expand the toolkit available to biosimilar manufacturers seeking to lower costs or work around existing patents.
Recombinant Insulin Versus Animal-Derived Insulin in the Clinic
When recombinant human insulin first entered clinical trials in the early 1980s, a critical question was whether a protein made inside bacteria would be as safe and effective as the animal-derived insulins that had kept people with diabetes alive for decades. The answer, confirmed across multiple trials, was yes. A Cochrane systematic review found no meaningful differences in blood-sugar control or rates of low blood sugar between human recombinant insulin and animal insulins.16PubMed Central. ‘Human’ insulin versus animal insulin in people with diabetes mellitus
Recombinant insulin did, however, show an advantage in how the immune system responded to it. In a 12-month trial comparing bacterially produced human insulin with purified pork insulin in patients who had never used insulin before, those on the recombinant product developed less antibody binding over time. More than half of patients treated with recombinant human insulin remained antibody-free at one year, compared with about 40% of those on pork insulin.17PubMed. Immunogenicity of recombinant DNA human insulin While antibody formation to insulin is usually clinically mild, lower immunogenicity reduces the risk of allergic reactions and unpredictable changes in how long a dose lasts.
Engineered Insulin Analogs
Once scientists mastered making exact copies of human insulin in bacteria, the next step was to tweak the molecule itself. Insulin analogs are engineered variants in which one or a few amino acids have been swapped, added, or removed to change how quickly the insulin is absorbed or how long it lasts in the body. Fast-acting analogs like lispro, aspart, and glulisine are absorbed more rapidly after injection than regular human insulin, which helps control blood sugar spikes after meals. Long-acting analogs like glargine and detemir are designed to provide a slow, steady baseline level of insulin over 12 to 24 hours.
These analogs are produced using essentially the same bacterial or yeast platforms. The precursor gene is simply altered to encode the desired amino-acid substitutions, and the rest of the manufacturing process follows a similar path. Some newer ultra-long-acting analogs are being designed with chemical modifications attached after the protein is produced. One research group recently used P. pastoris to produce the main chain of an insulin analog and then chemically attached a fatty-acid side chain to extend its duration, achieving purity above 95%.18Nature (Scientific Reports). Design of a novel long-acting insulin analogs by acetylation modification and compared with insulin Icodec This type of post-production chemical modification is becoming increasingly common as manufacturers pursue once-weekly insulin formulations.
What Recombinant Manufacturing Means for Cost and Access
Despite the efficiency of bacterial manufacturing, insulin remains expensive in many parts of the world, a fact that frustrates both patients and public-health advocates. A detailed cost analysis estimated that the raw manufacturing cost for regular human insulin biosimilars could translate to a price of roughly $48 to $71 per patient per year, and even analog insulins could be produced at prices between $78 and $133 per patient per year for most types.19PubMed Central. Production costs and potential prices for biosimilars of human insulin and insulin analogues Those figures stand in stark contrast to the hundreds or thousands of dollars per year that patients in some countries actually pay, a gap driven largely by patents on analog formulations, limited market competition, and supply-chain markups rather than by the biology of making the protein itself.
The manufacturing processes for regular human insulin and its analogs are remarkably similar at the molecular level. Both start with a microbial host expressing an insulin precursor, accumulating it as inclusion bodies, and purifying it through enzymatic cleavage and chromatography. The active pharmaceutical ingredient for regular human insulin costs an estimated $25,000 per kilogram, while glargine, one of the most widely prescribed analogs, costs closer to $69,000 per kilogram.19PubMed Central. Production costs and potential prices for biosimilars of human insulin and insulin analogues The price difference between regular and analog insulin, in other words, reflects the complexity of downstream processing and intellectual property rather than any fundamentally different production biology. Biosimilar manufacturers entering the market in recent years have begun to narrow these price gaps, particularly in countries with regulatory pathways that encourage competition.
Beyond the Fermentation Tank
Some of the most striking recent work moves away from industrial fermenters altogether. One emerging concept uses engineered bacteria not as factory organisms that produce insulin in a tank, but as living therapeutic agents that colonize the gut and synthesize insulin or other peptide drugs directly inside the body. Research into these “living materials” has shown that engineered bacteria can survive in the gastrointestinal tract, release therapeutic peptides on site, and potentially overcome the long-standing barrier that has prevented insulin from being taken as a pill: the digestive system destroys most proteins before they can be absorbed.20PubMed. Engineered Bacteria as Living Materials for Oral Delivery of Peptide Drugs This work is still in early experimental stages and faces significant safety and regulatory hurdles, but it represents a conceptual extension of the same principle that launched the insulin industry: if you give bacteria the right genetic instructions, they will follow them.
Plants are another frontier. Researchers have engineered tobacco and lettuce chloroplasts to produce proinsulin at high levels, and the resulting protein remained stable in dried leaves, raising the possibility of a shelf-stable, low-cost insulin source that could be grown in a greenhouse rather than manufactured in a sterile bioprocessing facility.21PubMed Central. Low-cost production of proinsulin in tobacco and lettuce chloroplasts for injectable or oral delivery of functional insulin and C-peptide The appeal for global health is obvious: plant-based production could sidestep the expensive infrastructure of stainless-steel bioreactors and cold-chain storage that currently limits insulin access in low-resource settings. Whether plant-produced insulin can meet the purity and consistency standards required for an injectable drug at commercial scale remains an open question, but the biology works.