“Our insulin is too slow,” Aaron Kowalski, head of the JDRF’s Artificial Pancreas Project, said when discussing some of today’s biggest roadblocks for better artificial pancreas systems. The cry for faster insulin is familiar to many diabetics, but the benefits of a faster method of delivery are not limited to the individual diabetic’s desire to have blood glucose levels normalize more quickly after bolusing. Faster insulin will also be a huge boon for systems designed to automatically calculate bolus amounts, as the faster response time allows both humans and automated systems to more accurately judge the overall effects of a given bolus, and therefore to more accurately anticipate how much more insulin might be needed given the observed change in glucose levels.
How fast is fast enough, though? In a non-diabetic, the pancreas begins to secrete stored insulin almost immediately after glucose is ingested and blood sugar levels are liable to rise above the normal levels. This initial release of insulin causes a peak in insulin levels within ten minutes of glucose infusion; then, insulin levels in the blood drop off slightly until newly manufactured insulin starts taking effect, with a second peak approximately 40 minutes after glucose infusion.
The fastest insulin available commercially for pumps and injections, however, takes an average of 10 – 30 minutes to begin action, and peaks once, somewhere between 30 minutes and 3 hours after it is taken. Recently, studies have shown that orally-administered insulin, delivered via a spray that allows patients to inhale insulin molecules, can transfer insulin throughout the body more rapidly, but the oral means of delivery makes inhaled insulin a less viable option for artificial pancreas systems and those already on pump systems.
The clear disparity between native, pancreatic insulin and the currently available synthetic insulins leaves a lot to be desired. Insulin that more closely mimicked the body’s natural hormone would be a large step forward in the development of the next era of diabetes medical devices. So, what’s the hold up? Why is it so hard to make faster insulin?
Insulin, as a result of its complex chemical and kinetic properties, is not an easy protein to make or to perfect. A single insulin molecule is what’s called a peptide hormone– a short protein created and managed by the endocrine system. Each insulin molecule is composed of 51 amino acids total, split into an A-chain and a B-chain that are linked together.
The single-molecule form of insulin, called an insulin monomer, is the active form of the protein; when absorbed into the bloodstream and used by the body to increase glucose uptake into cells, insulin must be in its monomer form. However, the active state is also a very unstable one, and when insulin is stored in the body, therefore, it is stored not as a monomer but as a hexamer– that is, in a set of six monomers, linked together by zinc ions. This state proves much more stable when insulin is inactive. When the stored hexamer insulin is needed to respond to glucose, it undergoes a process of self-assembly, in which the hexamer is signaled so that it disassembles into six discrete insulin molecules that can subsequently be absorbed into the bloodstream and used by the body’s cells.
Imitating this self-assembly behavior is one of the biggest challenges of creating an effective synthetic insulin; the insulin molecules must be stable enough to be stored and used over time in needles and pumps, but its self-assembly process must be fast enough to allow the insulin monomers to be absorbed quickly and effectively into the bloodstream. The longer a synthetic form of insulin takes to separate into monomers, the longer its onset and peak effect times will be.
But that’s just one of the main challenges; the other trick for insulin manufacturers is that though the hexamer form is more stable and has a longer shelf-life, the hexamers also have a tendency to form amyloid fibrils. Amyloid fibrils are long protein chains that form when the shorter molecules bond together into tight beta-sheets. Once insulin forms fibrils, it can no longer undergo self-assembly, and therefore can no longer separate into monomers with glucose-lowering abilities.
In the body, insulin has been known to form amyloid fibrils when a patient suffers from a set of diseases like Parkinson’s or injection amyloidosis, but synthetic insulin, depending on how it is designed, might form fibrils as a result of environmental factors like temperature or pH levels.
What is needed, then, is a form of insulin that is stable enough to store and that won’t either degrade rapidly or form amyloid fibrils in sub-optimal conditions, and yet can quickly destabilize and allow the rapid-acting insulin monomers to enter the bloodstream. This is no small task, but recent forays into the chemical manipulation of the protein structure of insulin have had some promising results, two of which were presented at the American Diabetes Association Scientific Sessions by Michael Weiss of Case Western Reserve University.
The first approach to the problem of faster insulin which Weiss presented was that being taken by Biodel’s new insulin candidate, VIAject. VIAject is a form of recombinant human insulin in which the B-chain of each insulin molecule has been chemically modified. Following the lead of recent developments with statin drugs, Biodel’s researchers found that they could alter the the pharmacokinetic properties of insulin by changing just a single atom at a critical point in the chain.
The researchers focused on the B-chain’s central phenylalanine amino acid, and added a single chlorine atom. The addition of chlorine, a member of the group of nonmetal elements known as halogens, had important effects on the polarity and stability of the insulin structures; Biodel’s chloro-insulin molecule, with its chlorine add-on, does not form hexamers in standard concentrations or temperatures. Instead, the insulin molecules retain their separate monomeric form, with the polarity of the chlorine atom on the B-chain pushing apart each monomer and preventing hexamers from forming.
Retaining the insulin monomers is key to VIAject’s speed and promise: as a result of its default monomeric form, VIAject does not have to self-assemble after injection, and can be absorbed into the bloodstream much more quickly. In studies comparing VIAject to the current set of rapid-acting insulins on the market like lispro, VIAject began working within minutes, and reached half-maximal capacity between 16 and 50 minutes, a full 17 minutes faster on average than lispro. Additionally, given that the insulin molecules are loathe to form hexamers, they are also resistant to forming amyloid fibrils, and therefore less likely to degrade in suboptimal conditions.
A second approach toward similar speeds is being attempted by Weiss himself at Case Western University in conjunction with the biotech startup Thermalin Diabetes, LLC. Instead of focusing on the two-chained insulin molecule, Weiss has turned his eye to proinsulin, a single chain of amino acids that serves as a precursor to the two-chained insulin molecule in the beta-cell’s process of creating new insulin. In proinsulin, the A-chain and B-chain of the eventual insulin molecule are linked together in a single chain by a short peptide. Using the body’s natural model, Weiss has created over a hundred insulin analogs based on variations of the proinsulin chain, with an A-chain-like segment, B-chain-like segment, and a custom configuration of peptides holding both together.
There are several complexities to making an effective insulin analog with the single-chain method. Firstly, each chain must be short enough so that it doesn’t form amyloid fibrils. Secondly, the configuration of the A-chain, B-chain, and intervening peptide must be carefully designed and controlled such that the single chain of insulin binds effectively to insulin receptors in the body’s cells. The natural form of proinsulin, for example, is more than fifty times less effective than the two-chain insulin monomer, and so clearly any single-chain molecule would have to be designed to mimic the two-chained monomer’s binding capabilities. And finally, the proinsulin precursor and some of its single-chain imitators have a tendency to be mitogenic, which means they trigger cell-division within the body. This would be a dangerous and undesirable side effect of insulin action, and therefore must be avoided in any viable single-chain insulin.
Despite these complexities, Weiss and his team have come up with several promising configurations of single-chain insulin. These single-chain molecules prove to be ultra-stable and resistant to forming amyloid fibrils, even in high temperatures. This stability alone is promising for insulin production and storage, but the additional benefit of some of the specially designed single-chain insulins is that, like VIAject’s monomers, they do not need to undergo self-assembly, and therefore can be absorbed into the bloodstream extremely rapidly.
So where are these two promising formulations of faster, better, more stable insulin? Rapid as the insulin may be, the drug development process is subject to the pace of the FDA; currently, Biodel is in the process of submitting a New Drug Application under section 505(b)(2). This means that the FDA needs only approve VIAject as a modification of an already existing and tested drug, so the FDA approval process, like the insulin itself, should hopefully be faster than average. Thermalin, meanwhile, in its partnering with Weiss and Case Western, has collected over $250,000 in federal grants, and last year began soliciting for angel funding.
Hopefully, for the sake of blood glucose control for individual diabetics and for artificial pancreas solutions being developed, chemically modified insulins like VIAject and Thermalin will indeed be available soon.
Karmel Allison was diagnosed with type 1 diabetes when she was nine. She lives in San Diego with her husband. For more from Karmel, visit her blog Where Is My Robot Pancreas?
* Insulin Hexamer image created by Isaac Yonemoto
* long protein chains image taken from “The protofilament structure of insulin amyloid fibrils“
* Weiss’s single-chain insulin molecule image taken from “Design of an Active Ultrastable Single-chain Insulin Analog Synthesis, Structure and Therapeutic Implications”
Karmel Allison is science editor of ASweetLife. She writes the blog Where is My Robot Pancreas?.
Having had type 1 diabetes for 39 years, I couldn’t carry on a normal life without an insulin pump,being somewhat brittle. How close are we to finding a structure which reins in insulin’s action and then releases it only if and when the glucose warrants it? Even if such an insulin were available-there would need to be a way to monitor the actual amount of insulin released along with real time blood glucose monitoring in a catheter, providing there were no leaks or malfunctions in the method of delivery to prevent disasterous drops in glucose or sudden, unexpected highs. How… Read more »
The slowness comes from the need to get insulin from its subcutaneous depot (where you inject it) into the blood stream. Synthetic “fast” insulins like insulin glargine have a mutation that blocks its ability to form hexamers. It works somewhat faster than regular insulin since regular insulin needs to dissociate from its hexamer form and then can move towards the bloodstream. A structure which holds insulin in (or close to) the blood stream and releases upon sensing glucose is needed. People have experimented with catheter-like connections directly to the blood stream in animal models but leaks in the system tend… Read more »
This is fascinating. I have long wondered why the most rapid-acting insulins on the market are such poor comparisons to the real thing — and am grateful to have such a coherent and interesting answer. Thank you!
Thank you Karmel… “fastest insulin available…takes an average of 10 – 30 minutes to begin action, and peaks once, somewhere between 30 minutes and 3 hours after it is taken.” This is what makes synthetic insulin a poor model to endogenous. Faster isn’t the only issue – but this notion of stability you write of is one part of the conundrum. If a patient is peaking it can take anywhere from half a day to one hour for blood sugars to come down. Another important issue – hormones that are vital in glucose homeostasis and the dynamic interplay of hormones… Read more »