The insulin in that pen was made by microbes that were handed a human gene

For sixty years, treating diabetes depended on abattoirs. Insulin was extracted from the pancreases of pigs and cattle collected at slaughterhouses — something like a tonne of glands for a few dozen grams of usable hormone — which meant the supply of a life-sustaining drug was tied to the meat trade, and the product was never quite right. Pig insulin differs from the human protein at one amino acid, cattle insulin at three, and for some patients that was enough to provoke an immune response against their own treatment.

In 1982 that ended. A modified strain of E. coli carrying a synthetic version of the human insulin gene became the source, and insulin became the first medicine ever approved that was made by recombinant DNA. What is worth understanding is not the headline — a microbe was made to produce a human protein — but the part nobody puts on the poster: growing the cells is the straightforward step, and almost everything difficult and expensive happens afterwards.

The gene has to be written, not copied

The obvious plan would be to cut the insulin gene out of human DNA and paste it into a bacterium. That does not work, and the reason is a genuine difference in how the two kinds of organism read their genomes.

Human genes are interrupted. The coding sequence is broken up by stretches called introns, which are transcribed and then spliced out before the message is translated into protein. Bacteria have no splicing machinery. Hand E. coli a raw human gene and it will faithfully translate the introns too, producing nonsense.

So the gene is not copied — it is written. The amino acid sequence of insulin is known, so a DNA sequence encoding it is designed and chemically synthesised, with intron-free coding and codons chosen to suit the host's preferences. That synthetic gene is inserted into a plasmid — a small circular DNA carrier — behind a promoter that lets the operator switch production on at a chosen moment, along with a selection marker so that only cells which took up the plasmid survive. The plasmid goes into the host, and a single successful cell is grown out into the master cell bank that every future batch will start from.

The fermenter, and the mess inside the cells

Production begins from a frozen vial of that bank, scaled through progressively larger flasks and seed vessels into a fermenter of several thousand litres, with oxygen, temperature, pH and feed rate held under tight control. When the cell density is right, the promoter is induced and the culture switches from growing to manufacturing.

What happens next depends on the host, and the two common choices behave very differently.

Bacteria make protein fast and cheap, but a foreign protein produced in bulk usually crashes out inside the cell as inclusion bodies — dense, insoluble aggregates of misfolded chains. This looks like failure and is actually convenient: inclusion bodies are easy to isolate. But the protein in them is inert. It must be dissolved with a chaotropic agent, then refolded under carefully controlled oxidising conditions so that the correct disulfide bridges form and not the many possible wrong ones. Refolding yield is one of the central economics of the whole process.

Yeast is slower and more expensive to run, but it folds proteins properly and can secrete the product into the broth, so the first purification step becomes simply separating cells from liquid. Both routes are used commercially for insulin.

There is one more problem specific to this molecule. Insulin is not a single chain — it is two chains held together by disulfide bonds, and a cell asked to make the two separately will not reliably join them. The solution mirrors human biology: express proinsulin, a single continuous chain that folds correctly and forms its own disulfide bridges, then enzymatically cut out the connecting C-peptide afterwards. Trypsin and carboxypeptidase B do in a tank what the pancreas does in a secretory granule.

Cells will make almost any protein you ask them to. Getting it out intact, correctly folded, and free of everything else the cell made is the actual discipline, and it is where most of the cost and nearly all of the failures live.

Downstream is the expensive half

After harvest comes a sequence of chromatography steps, each separating on a different property — charge, size, hydrophobicity — until the product is pure enough for injection. Purity here is not a rounding exercise: the specification concerns individual related impurities, misfolded variants and residual host-cell proteins, each measured and limited.

Bacterial hosts add a specific burden. Gram-negative bacteria carry endotoxin in their outer membrane, which causes fever and shock in even trace amounts, so its removal and measurement is a whole workstream of its own. Then comes crystallisation or formulation with zinc and stabilisers, sterile fill-finish into vials or pen cartridges, and a cold chain that must hold from the plant to the patient's refrigerator.

This is why biologics do not behave like ordinary chemistry. For a small-molecule drug you can characterise the finished product completely and prove it is what you say it is. For a protein, the molecule is large, its folding and any sugar decoration depend on the cell line and the exact conditions it grew in, and no analytical panel fully captures it. In this industry the saying is that the product is the process — change the cell line, the feed, the refolding step, and you may have changed the medicine.

Rewriting the molecule to change the clock

Once you can manufacture a protein to order, you can also redesign it, and modern insulins are protein engineering rather than pharmacology.

Natural insulin clusters into six-unit assemblies in the vial, and those must break apart before absorption, which delays onset. Rapid-acting analogues work by swapping a couple of residues so the molecules no longer stack as readily — the same biological activity, absorbed far faster, so an injection can be taken with a meal rather than half an hour ahead of it.

Long-acting analogues go the other way, and one of them is a small masterpiece. Add two arginines to the end of the B chain and the protein's isoelectric point shifts to around neutral, which means it is soluble in the acidic solution in the pen and precipitates gently under the skin, dissolving back over many hours. Others attach a fatty acid so the molecule binds to albumin in the blood and is released slowly. In each case the change is a handful of atoms, deliberately chosen to alter timing rather than action.

Why a biosimilar is not a generic

A generic tablet is chemically identical to the original, and proving it needs little more than analysis and a bioequivalence study. A protein made in a different cell line by a different process is never quite identical, so copies are approved as biosimilars through their own regulatory pathway: extensive analytical comparison, plus clinical work to show there is no meaningful difference in effect or immune response.

That pathway is where a great deal of Indian biotechnology sits. India is among the world's larger producers of recombinant insulin and one of the most active biosimilar developers, and it built a domestic regulatory framework for these products years before several richer markets did. It matters commercially, and it matters practically: insulin has been in use for over a century, and the reason it remains costly in many places is not the science but manufacturing scale, quality systems and supply chains — precisely the things a competitive biosimilar industry pushes on.

Why it matters for students and researchers

Insulin is the reference case for an entire industry. Monoclonal antibodies, vaccines, clotting factors, enzyme replacements and engineered cell therapies all run on the same chain of decisions: choose a host, design an expression construct, control a fermentation, recover and refold, purify, formulate, and prove batch-to-batch consistency to a regulator.

The teaching point is where the difficulty actually sits. Molecular biology gets you a cell that makes the protein — that part is now routine enough for an undergraduate laboratory. Turning that into thousands of consistent doses is bioprocess engineering: refolding chemistry, chromatography design, endotoxin control, analytics, validation. India has the fermentation capacity and the market; the constraint is people who understand downstream processing and quality systems, not people who can clone a gene. For anyone in biotechnology or bioengineering, that is where the unglamorous, genuinely scarce expertise lies.

Frequently asked questions

How is insulin made today?

A synthetic gene for human insulin is inserted into bacteria or yeast, which are grown in large fermenters and induced to produce the protein. It is then extracted, refolded if necessary, cut to its final form by enzymes, purified through several chromatography steps and formulated for injection.

Why can't the human insulin gene simply be copied into bacteria?

Because human genes contain introns that must be spliced out before translation, and bacteria cannot do that. An intron-free version of the coding sequence is designed and chemically synthesised instead, often with codons optimised for the host organism.

Why is insulin made as proinsulin first?

Insulin consists of two chains joined by disulfide bonds, which do not reliably assemble if the chains are made separately. Expressing the single-chain precursor lets it fold and bond correctly, after which enzymes remove the connecting C-peptide to yield mature insulin.

What is the difference between insulin and an insulin analogue?

An analogue is human insulin with a few deliberate changes to its amino acid sequence or an attached fatty acid, made to alter how quickly it is absorbed. Rapid-acting versions resist clustering, while long-acting versions form a depot under the skin or bind to blood albumin.

Why are biosimilars not called generics?

Because a protein produced in a different cell line and process cannot be made exactly identical to the original — folding and modifications depend on the manufacturing route. Biosimilars therefore require detailed analytical comparison and clinical evidence rather than the simple equivalence testing used for generic tablets.