The Peptide That Needed Its Own Factory

By the end of 2002, a Roche plant in Boulder, Colorado was running around the clock, every day, on a single molecule, in a process that took more than a hundred separate steps to make one batch of drug substance. The molecule was enfuvirtide, a chain of 36 amino acids, and it was about to become the first of a new class of anti-HIV drug 14. The building existed because there was no way to make enough of the peptide without it. That is why a drug approved in 2003 still shapes how long peptides are manufactured, and why its launch was a lesson in what happens when chemistry, supply and price all become the news at once.

This is a manufacturing and business history. It describes the discovery only as far as it explains the size of the problem, and it says nothing about how the drug is used.

A fragment of a viral protein that jammed the virus

HIV enters a cell by fusing its outer membrane with the cell's. The fusion is driven by a viral protein called gp41, which unfolds and then collapses into a compact bundle of helices that pulls the two membranes together. In the early 1990s, researchers at Duke University, including Dani Bolognesi and Thomas Matthews, noticed that short synthetic peptides copied from one of gp41's helical regions could inhibit infection in cell culture. The proposal was that the peptide competed with the viral protein's own helix and jammed the bundle before it could close 2.

The 1994 paper reporting those results is a laboratory study in cultured cells, and it said nothing about a medicine. But the idea that a viral protein could be blocked by a fragment of itself was attractive to a pharmaceutical industry that, in the 1990s, was looking for any new way to attack the virus. The peptide that emerged as the lead, a stretch of 36 residues, would eventually be called enfuvirtide, after the code T-20.

Trimeris: a university company with one idea

A discovery like this had no obvious home. A big pharmaceutical company would have regarded a 36-residue peptide as a manufacturing nightmare, and a university cannot make clinical material. The answer was a start-up. Trimeris was formed in North Carolina in the 1990s around the Duke discovery, with the peptide as its main asset, and it did what early biotechnology companies did when they met a problem beyond their size: it found a partner. Roche joined the project at the end of the decade and took on the work of making the drug at scale.

The arrangement is a template for many peptide programmes since. A small company holds the idea and the early clinical data, and a large one holds the capital and the plant. In this case the plant was the whole question. Whether the molecule could be made at all in the quantity a global programme would need was not obvious, and the answer was going to determine whether the drug existed.

The arithmetic of a long chain

Chemical synthesis builds peptides one residue at a time, and every coupling is an opportunity to fail. The article on the solid-phase synthesiser sets out the arithmetic: the fraction of chains that come out complete is the per-step success rate multiplied by itself once per coupling. For a chain of 36, the numbers are unforgiving.

At a bench scale, those losses are tolerable, because a purification column can separate what remains from the by-products. At the scale of hundreds of kilograms a year, they are not. Every by-product has to be removed, the mixture of near-relatives is hard to separate from the target, and the solvents and resin consumed to produce material that is discarded are the largest part of the cost 1.

A hundred and six steps

The route developed for enfuvirtide avoided the problem by not building the chain in one run. As described in a 2003 review from the company that developed the process, protected peptide fragments are made separately on solid supports, released with their protecting groups still in place, purified, and then joined in solution to form the full-length chain, which is deprotected and purified at the end 1. Making three or so shorter fragments is much more efficient than making one long one, and each fragment can be checked before it is committed to the next stage.

The price of that strategy is the step count. Roche stated that manufacturing the drug substance took 106 steps, against 8 to 12 for a typical anti-HIV pharmaceutical. One report put the raw material requirement at 45 kilograms of input for every kilogram of product. That is the context for the price that followed. A hundred-step process is not only difficult but long, so that each batch ties up a plant for a considerable time between starting materials and finished product.

Boulder: a factory for one peptide

Roche's answer to the scale problem was a dedicated facility in Boulder, Colorado. By December 2002 it was described as running continuously, day and night, on a molecule that needed more than a hundred production steps. A dedicated plant is an expensive commitment: if the drug failed or the market was smaller than hoped, the building could not easily make anything else. Without the plant, there would have been no drug to sell.

It did not entirely work. Shortly after approval, Trimeris and Roche announced that supply would cover only about 8,000 to 10,000 of the neediest cases, down from the 12,000 to 15,000 first projected 5. The plant that had been built to prove peptide synthesis could be done on an industrial scale was struggling to keep up with demand it could see coming.

2003: approval, price, and who it was for

The US Food and Drug Administration approved enfuvirtide in March 2003, under its accelerated approval pathway, as the first drug that blocked viral entry. The trials that supported approval enrolled people whose virus had stopped responding to the other available drugs, a group with few remaining options 3. In that context the new drug was medically important, and the supply shortage meant that physicians were using personal connections to secure it.

The price was a second source of anger. Press reports at launch put the annual cost at around twenty thousand dollars, making it the most expensive anti-HIV drug on the market; sources differ on the exact figure and on regional prices. Activists, who had campaigned for years for new drugs for patients with no remaining options, objected both to the cost and to the shortage. Roche and Trimeris answered that the manufacturing was extraordinarily complex, which was true, and which did nothing to lower the number.

Commercially the launch disappointed. Treatment Action Group's contemporary account reports 2003 sales of $36 million against an initial estimate of $120 million, with the peak once expected near a billion; the company had predicted 8,000 to 10,000 patients by the end of the year and supplied 4,350 5. Newer classes of anti-HIV drugs arrived in the following years, and the drug settled into the niche for which it had been built.

What the plant taught the industry

Enfuvirtide was the existence proof. Before it, there was a widely held view that chemical synthesis could make milligrams to grams of a long peptide for research, and kilograms of a short one for a drug, but that a peptide of this length could not be made by chemistry in the quantity and consistency a global medicine needs. Boulder showed that it could, at a price, and set out in the open literature how it was done 1.

The lessons it left were practical. Fragment condensation became a standard strategy for long sequences. The cost of solvents and waste became a central concern of process chemistry. And the experience was a warning about the difference between making something and making enough of it, a gap that the peptide medicines of the 2020s, with demand measured in millions of patients, are now living through at a far larger scale.

Where the synthesis and cost articles take over

The building in Boulder is a physical record of a bet: that a peptide of 36 residues could be made at industrial scale and sold, in the face of a process that took more than a hundred steps to complete. On the chemistry the bet paid off; on the economics the outcome was mixed. For the day-to-day mechanics of making a peptide, the article on the solid-phase synthesiser carries the story forward, and for the longer history of how the cost of a gram has fallen, the article on the price of a gram does the same.