If you have an ACS member number, please enter it here so we can link this account to your membership. (optional)

ACS values your privacy. By submitting your information, you are gaining access to C&EN and subscribing to our weekly newsletter. We use the information you provide to make your reading experience better, and we will never sell your data to third party members.



Insulin’s second century

100 years after insulin was first used to treat type 1 diabetes, researchers are still finding ways to improve it

by Celia Henry Arnaud
January 30, 2022 | A version of this story appeared in Volume 100, Issue 4
Illustration of a glass vial and syringe.

Credit: Will Ludwig/C&EN


In brief

Insulin was first used to treat diabetes 100 years ago. The peptide is a lifesaving drug for people with type 1 diabetes, and it can improve the health of people with type 2 diabetes. In the century since it was first used as a therapy, researchers have learned much about how insulin works, developed processes for making large quantities of it, and engineered analogs that work faster or for longer periods. As insulin embarks on its second century, researchers continue to find ways to improve its performance. Read on to learn about innovations such as glucose-responsive insulin, which can keep blood sugar in check and reduce the risk of hypoglycemic episodes.

Before the discovery of insulin, type 1 diabetes was a death sentence. That prognosis started to change on Jan. 11, 1922, when 14-year-old Leonard Thompson became the first person with diabetes to receive an injection of a pancreatic extract from dogs. We now know that extract contained insulin.

That first extract wasn’t very pure, so although Leonard’s blood sugar decreased, the overall effect was disappointing. About 2 weeks later, Leonard received an injection of a purer extract, and the results were more promising. That same month, six other people with diabetes also received injections that reduced their blood sugar levels. Those trials at the University of Toronto marked the beginning of a lifesaving treatment for type 1 diabetes.

In the 100 years since Leonard received his first dose, insulin has become well known. Scientists purified, identified, and characterized the peptide; elucidated its role in glucose metabolism; developed biotechnology processes for producing large quantities of it; and engineered analogs that can work faster or for longer periods.

Yet even with all these advances, academic and industry researchers continue to look for ways to improve insulin and thus improve outcomes for people with diabetes. “There is as much to be done chemically with insulin today as there’s ever been in its history,” says Richard D. DiMarchi, a chemistry professor at Indiana University Bloomington who was previously a vice president at Eli Lilly and Company and Novo Nordisk, two of the three major producers of insulin. “While this is a miraculous substance, it is an imperfect and, one might say, lousy drug by conventional standards.”

Researchers especially want to develop insulin that can turn itself on and off in response to glucose in the blood. That glucose responsiveness can be imparted by formulating the insulin with materials engineered to release it in proportion to the amount of glucose or through modifications to the insulin itself.

Insulin Timeline

Insulin basics

Insulin is a peptide hormone that regulates the metabolism of carbohydrates, fats, and proteins. β Cells in the pancreas secrete insulin in response to glucose in the blood. Insulin, which has an A chain and a B chain held together by disulfide bonds, exists in an equilibrium of monomers, dimers, trimers, hexamers, and dodecamers, says Michael A. Weiss, an insulin expert at the Indiana University School of Medicine. Whereas the main storage form of insulin in β cells and in most drug formulations is a hexamer coordinated with zinc, the active form in the blood is the monomer.

Insulin’s binding to its receptor on the surface of cells initiates a signaling cascade that results in the insertion of glucose transporters into cell membranes so that they import glucose. Recent cryo-electron microscopy structures show that as many as four insulin molecules bind to the insulin receptor dimer (eLife 2019, DOI: 10.7554/eLife.48630). Multiple insulin molecules trigger a large conformational change in the insulin receptor and stabilize the active conformation to initiate downstream signaling, says Eunhee Choi, a cell biologist at Columbia University who was part of the team that acquired the structures.

Diabetes is a metabolic disease in which the body either no longer makes insulin or can’t properly use the insulin that it does make. In type 1 diabetes, the body destroys β cells in the pancreas, so it doesn’t produce insulin. In type 2 diabetes, the pancreas still makes insulin, but cells develop resistance to it. All people with type 1 diabetes must take insulin. Many people with type 2 diabetes also need insulin.

Ribbon structure of an insulin hexamer with each monomer shown in a different color.
Credit: Wikimedia Commons
The storage form of insulin in β cells and pharmaceutical formulations is as a hexamer stabilized with zinc. Each monomer is shown here in a different color. The histidines that bind the zinc are shown as stick structures.

What makes insulin a “lousy drug” is its narrow therapeutic index. The therapeutic index is the difference between an effective dose and a toxic dose—and in the case of insulin, either too much or too little can be dangerous. If a person takes too little insulin, they don’t metabolize glucose effectively. That can lead to long-term complications, such as cardiovascular disease, nerve damage, and kidney damage. But if they take too much insulin, they run the risk of hypoglycemia, an emergency condition that can lead to coma and death if not reversed in time.

“When you talk to a lot of diabetics, they’ll choose to keep their blood glucose higher than would be ideal because of how scary hypo events are,” says Matthew Webber, a biomedical engineer at the University of Notre Dame who’s working on materials to improve insulin delivery. “They’re making a gamble that they’d rather have cardiovascular disease and risk of stroke and other sorts of things in 20 years than have a hypo event tonight.”

Companies have developed insulin analogs that work faster or last longer than natural human insulin. These changes are achieved by modifying the amino acid sequence or by attaching side chains. But none of these analogs is perfect.

“There’s clearly room for improvement because every insulin that’s marketed today is capable of inflicting a life-threatening hypoglycemic event,” DiMarchi says. He was involved in the discovery and development of the first approved insulin analog, lispro, which acts more quickly than native insulin and is marketed by Lilly under the brand name Humalog.

Faster and slower

Insulin is typically injected into the fatty tissue under the skin. Before it can start working, it needs to get from there to the bloodstream. Even rapid-acting insulin takes about 15 min to start working and continues to work for 3–4 hours after administration, says Eric Appel, a materials scientist at Stanford University who’s working on ways to get insulin to act faster and turn off more quickly.

One challenge to speeding up insulin response times is that insulin has to dissociate from hexamers to monomers, its active form.

Although monomeric insulin would be absorbed quickly, it is too unstable for direct use as a drug. “You simply cannot make a drug product of it and ship it around the world,” Appel says. The insulin molecule has a disordered portion that predisposes the peptide to aggregate into amyloid fibrils. In the hexameric form, the floppy bits that initiate aggregation are sequestered. “It has to be formulated as a hexamer, but the dissociation of the hexamer is what makes it so slow” to act, he says.

Glucose import, activate!
Before insulin binds, the insulin receptor dimer (blue and green) is in an autoinhibited state that prevents signaling. The insulin receptor consists of multiple fibronectin III domains (FnIII-1, -2, and -3), leucine-rich domains (L1 and L2), a cysteine-rich domain (CR), a tyrosine kinase domain (TK), and the C terminus of the alpha chain (α-CT). Insulin’s binding (yellow and pink) induces conformational changes in the receptor, activating a signaling cascade that leads to the production of glucose transporters.
Scheme showing the conformation change induced by insulin binding to its receptor, and a cryo-electron microscopy structure of the full-length insulin receptor bound to four insulin molecules.
Credit: Adapted from eLife

Appel wants to eliminate the dissociation step by stabilizing monomeric insulin in the vial. “As much as protein stability sounds boring, in this particular case, it’s an example where it is absolutely the challenge that keeps drugs from working as well as they could,” he says.

He has identified a polymeric additive that prevents aggregation by blocking protein adsorption to interfaces, such as the air, glass, rubber, and plastic in vials, pen injectors, or pump systems. The interfaces become an even bigger problem if the vial or delivery device is shaken, because the interfaces keep breaking up and re-forming. Because the additive doesn’t interact specifically with the protein, it can be used for any protein therapeutic, Appel says.

He has started a company called Surf Bio to commercialize the new additive. The company has started nonclinical toxicology studies. Appel expects it will take 5 years to get through Phase 3 trials and fully commercialize the new ultrafast monomeric insulin drug candidate and the additive that enables it.

Indiana University’s Weiss is also working on ways to stabilize insulin against amyloid formation. To do that, his team is exploring ways to make heat-stable insulins in which the A and B chains are covalently connected by short peptide linkers.

The tether needs to be long enough to allow receptor binding but too short to allow for amyloid formation, which requires a conformational change in the insulin, according to Weiss. “You can make active insulin molecules that are totally refractory to forming amyloid—even if you boil them,” he says.

Weiss founded the company Thermalin more than a decade ago to develop novel insulin molecules, such as the heat-stable ones and others that are optimized to enable the miniaturization of devices for the automated delivery of insulin.

Longer-lasting insulins are also under development. Novo Nordisk and Lilly are working on slow-acting insulins that could be taken once a week. Such products, also called basal insulins, could be used in combination with a mealtime insulin by people with type 1 diabetes or on its own by people with type 2 diabetes.

Novo Nordisk’s weekly insulin, called icodec, is in Phase 3 clinical trials, according to Peter Kurtzhals, the firm’s chief scientific adviser. Icodec has three substitutions in insulin’s amino acid sequence and a 20-carbon fatty diacid side chain. The backbone substitutions reduce the rate that cells take up insulin. The fatty diacid gives icodec a strong affinity for albumin, which increases the time the insulin spends in circulation. Icodec has a circulating half-life of nearly a week, Kurtzhals says.

Lilly’s once-weekly insulin is a single-chain variant—meaning the A and B chains are connected by a covalent tether—linked to part of a human antibody, says Ruth Gimeno, vice president for diabetes research and clinical investigation at Lilly. It has a circulating half-life of 17 days, and the concentration in serum varies by only 10%, which gives people added flexibility in when they take each dose. It’s intended for people with type 2 diabetes.

An insulin that can be taken once weekly will encourage people with type 2 diabetes to adopt insulin therapy earlier and to stick to it, leading to better outcomes, Gimeno said at Lilly’s investor community meeting in December.


One of the best ways to avoid hypoglycemic events would be to develop insulins that can respond to glucose. Researchers are taking multiple approaches to create such designer insulins.

One approach is to encapsulate insulin in glucose-responsive materials. These insulin storage depots have the advantage of lasting for a long time, says Danny Hung-Chieh Chou of Stanford University, who is working on various forms of glucose-responsive insulin.

You have to do some advanced chemistry to equip the insulin molecule with those chemical features.
Peter Kurtzhals, chief scientific adviser, Novo Nordisk

Another advantage of depots is that they can incorporate any already-approved insulin analog, which could ease the approval process, Webber says. But the insulin is also only a fraction of the material in the formulation. “That raises costs and causes all kinds of other potential pitfalls in terms of manufacturing,” he says. Plus, once the insulin is released from the depot, there’s no way to get it back, so a person can still become hypoglycemic.

Zhen Gu, a materials scientist at Zhejiang University, and his coworkers are developing a glucose-responsive microneedle patch that can deliver insulin across the skin. Insulin and phenylboronic acid are embedded in the polymeric microneedles. When glucose forms a complex with the phenylboronic acid, the microneedles swell and release the insulin. Gu and his colleagues have used such patches to control blood glucose levels in diabetic minipigs (Nat. Biomed. Eng. 2020, DOI: 10.1038/s41551-019-0508-y). A patch about the size of a coin can control the blood glucose of a 25 kg minipig for about a day, Gu says.

Gu believes that the patch could ease some of the timing constraints around insulin injections. People with diabetes must inject themselves before meals—and then eat soon enough to prevent hypoglycemia. People would still apply Gu’s patch before meals, but they would have more flexibility in the timing of the meals. He and his colleagues have started a company called Zenomics to develop and commercialize the patches.

A glucose-responsive microneedle patch for delivering insulin. In the background is a micrograph of the microneedle array.
Credit: Courtesy of Zhen Gu
A patch with an array of insulin-embedded microneedles releases the insulin into the skin when glucose binds to another substance in the needles. In the background is a micrograph of the microneedle array.

The second general approach to creating glucose-responsive insulins is to add a glucose-sensing moiety directly to insulin. This has proved to be challenging, but researchers are optimistic that it will eventually work.

Chou is trying to impart glucose responsiveness by inducing conformational changes in insulin through a combination of amino acid changes and chemical modifications. Cryo-electron microscopy structures reported in the past few years show that insulin changes conformation when it binds to the insulin receptor, Chou says. Glucose responsiveness could be achieved by switching between conformations that bind more strongly or weakly to the insulin receptor. Initially, Chou says, researchers would focus on creating linear responses to changes in glucose levels. But to better mimic the pancreas’s activity, insulin would respond nonlinearly, Chou says.

Merck & Co. had a glucose-responsive insulin reach Phase 1 clinical trials. Merck acquired this program when it bought SmartCells, which explored the concept of adding saccharide units that bind to the mannose receptor. Merck developed the compound that reached the clinic. The molecule can be removed from the blood via a mechanism that doesn’t involve activating the insulin receptor and driving glucose levels too low. When glucose levels are low, insulin that might otherwise induce hypoglycemia is cleared through the mannose receptor. And the lower that glucose levels get, the faster that the mannose pathway works.


“We really thought that the big unmet need was to open up the therapeutic index, which is not as readily addressable by modulating release from a depot,” says David E. Kelley, a retired leader of diabetes and obesity drug discovery at Merck. “It is addressable by modulating clearance in a manner that accelerates as you approach hypoglycemia.”

But the addition of the carbohydrates reduced the potency of the insulin, so people had to inject higher doses to get the desired effect, Kelley says.

Ultimately, Merck discontinued the program. “It was a business decision about where Merck wanted to invest,” Kelley says. He hopes that others will be able to build on the science. He still finds the tactic of accelerating insulin clearance attractive because it would increase the range of doses that insulin can be safely given.

Lilly and Novo Nordisk continue to work on glucose-responsive insulins. Neither company’s work is at a stage where the firm is ready to reveal molecular details about its lead molecules. But representatives are willing to talk about why they see the need for such designer insulins.

Novo Nordisk acquired the start-up Ziylo in 2018 for its glucose-responsive insulin platform. Now the company has parallel projects focused on internally developed glucose-responsive technology and the Ziylo technology. “We have not settled on a single technology,” Kurtzhals says.

The goal is to equip insulin with a glucose-binding element that would modulate insulin’s binding affinity for its receptor. Ideally, the effect would be reversible, so that insulin would adopt its active conformation when blood glucose levels are high and would be deactivated when blood glucose levels are low, Kurtzhals says. It also needs to be specific to glucose so that the insulin can’t be activated by other molecules in the blood.

“You have to do some advanced chemistry to equip the insulin molecule with those chemical features,” Kurtzhals says. And that chemistry needs to be cost effective. “You can’t have chemistry that is so complex that it will never be useful on an industrial scale,” Kurtzhals says.

Building glucose responsiveness into insulin is technically challenging, Lilly’s Gimeno agrees. “You need to go from zero activity to 100% activity over a roughly threefold change in glucose levels. And you need to somehow build it into a molecule,” she says. “That’s actually something that hasn’t been done with any other therapeutic.”

Like Novo Nordisk, Lilly bought a start-up—Protomer Technologies—for its glucose-responsive platform. Gimeno won’t go into detail about the technology. “Right now we’re just focusing on getting a molecule into the clinic as quickly as possible,” she says.

Other prospects

Whether insulin acts faster, slower, or in response to glucose, researchers would like to deliver insulin orally instead of via injection. But bioavailability—the amount of drug that gets to the bloodstream—has proved to be a major hurdle.

Novo Nordisk reported a clinical proof-of-concept trial (Lancet Diabetes Endocrinol. 2019, DOI: 10.1016/S2213-8587(18)30372-3). But the bioavailability was only about 2%. That’s too low to be commercially viable, Kurtzhals says.

We are celebrating the 100th anniversary of insulin with another wave of insulin innovation.
Ruth Gimeno, vice president for diabetes research and clinical investigation, Eli Lilly and Company

Because less than 10% of insulin makes it into circulation when delivered orally, Lilly isn’t pursuing oral insulin, Gimeno says.

But there’s still hope for an oral formulation. Robert S. Langer and coworkers at the Massachusetts Institute of Technology are developing ingestible, self-orienting devices with microneedles to inject insulin directly into the stomach lining. In a study with pigs, the bioavailability of insulin delivered that way was more than 50% (Nat. Biotechnol. 2021, DOI: 10.1038/s41587-021-01024-0). Langer is collaborating with researchers at Novo Nordisk to develop the devices.

Future advances in insulin therapy could also involve combining it with other hormones. One example is amylin, which also plays a role in glucose regulation. β Cells produce amylin in addition to insulin, so people with type 1 diabetes lack that hormone too. Ideally the two hormones would be given together.

Amylin Pharmaceuticals, now part of AstraZeneca, developed an amylin analog called pramlintide, but it “was so unstable at regular insulin formulation conditions that it couldn’t be mixed with insulin,” Stanford’s Appel says. “The two hormones had to be dosed separately.” Because the pancreas releases the two hormones at the same time and at a defined ratio, dosing them separately makes it difficult to mimic what happens with a functional pancreas. The new protein-stabilizing additive makes it possible to administer insulin and pramlintide together, and the rapid response time of monomeric insulin may enable the dual-hormone drug to properly mimic normal function of the pancreas.

And other pancreatic cells release the hormone glucagon, which raises the concentration of glucose in the blood. Glucagon is often used as a rescue drug to counteract hypoglycemia. Some researchers, such as Gu and Webber, are working on ways to deliver insulin and glucagon together in a glucose-dependent manner.

“Coformulations are an important next frontier in recapitulating this complex metabolic control that our bodies have,” Appel says.

The ongoing work shows that there’s still plenty of room for improvement in insulin. “We are celebrating the 100th anniversary of insulin with another wave of insulin innovation,” Gimeno says.


This article has been sent to the following recipient:

Chemistry matters. Join us to get the news you need.