Last month, the news started circling: they cured diabetes in dogs! Mice we hear about often– but dogs is new and exciting. So what’s the story there?
The publication in question is “Treatment of Diabetes and Long-term Survival Following Insulin and Glucokinase Gene Therapy,”  published out of Spain in the journal Diabetes. Unlike many of the recent therapies that have made it to clinical trials, this study doesn’t try to reset the diabetic immune system or prevent the development of autoimmunity. Instead, the authors demonstrate successful treatment of diabetes in dogs using a gene therapy approach they had previously used in mice.
What exactly is their gene therapy approach? First, a quick primer on gene therapy (and here I insert my disclaimer: gene therapy is outside of my area of expertise, and so I read and comment on this paper only as an observer, and not as an informed expert. Take it all with a grain of salt!). Gene therapy is the therapeutic application of techniques that have been used in labs for many years to insert particular strings of DNA into cells such that the machinery of the cell takes the DNA, transcribes it into RNA, and then makes the protein that is encoded in the DNA. This extremely powerful technique allows us to force cells to make proteins that they might not otherwise make, and in the last decade it has started to see success in small trials to treat a number of diseases. For example, in 2006, scientists used gene therapy to generate T cells that made a receptor that recognized specific cancer cells, allowing the T cells to kill the cancer cells . Another example is the use of gene therapy in 2010 to make a patient’s bone marrow cells produce the beta-globin gene that they lacked due to a genetic defect .
These and other similar successes are leading many research groups to study gene therapy and the ways we might use it to treat disease. However, in addition to demonstrating the power of gene therapy, these success stories implicitly demonstrate some of the difficulties; the ability to arbitrarily express any protein in any cell would be incredible, so why don’t we just use gene therapy to treat everything? As it turns out, designing and targeting a string of DNA at a cell is not easy. Currently, we use what are called viral vectors to get the DNA inside cells– basically, we make a virus that contains the DNA we want to insert, and infect cells with the virus. The virus enters the cell, and can integrate its DNA segment into the chromosomes of the host cell, or can leave the DNA separate, floating around as if it were a separate chromosome. There are a number of different virus types that are used, with a variety of advantages and disadvantages, but I’m sure you can imagine what makes people nervous about gene therapy– we’re leveraging the existing natural ability of viruses to invade cells to our advantage, but our knowledge of how exactly to direct and control the virus is still developing. Sometimes there are off-target effects, where the viral vector ends up affecting a gene we didn’t intend to affect. And it’s not easy to hit every type of cell– what do you do if the target cell is not sitting somewhere easy to reach? Further, any time you introduce a foreign gene into the body, you risk the immune system deciding to attack. And even if it doesn’t, the cells that do take up the DNA and make the target protein might only do so for a little while before dividing, dying, or otherwise changing. In other words, gene therapy is incredibly powerful and full of promise– but it’s still nascent, and there’s lots of work to be done, one study and one trial at a time.
And the paper published in Diabetes was one such study– a step forward in using gene therapy. The authors in this paper discuss an approach to curing hyperglycemia using two viral vectors that force the expression of two genes in skeletal muscle (that is, the muscles you use to move around that are connected to your bones). Skeletal muscle is not a standard target in treating type 1 diabetes, where we more often hear about T cells and beta cells. However, skeletal muscle plays an important role in glucose control because it is the end destination of much of the body’s circulating glucose. Normally, in cells in the muscle, special proteins called glucose transporters take glucose into the cell and allow it to be converted into energy for the cell’s use. Insulin binding to the outside of the cell is what turns the glucose transport system on, so, under normal circumstances, a person eats food, which gets processed in the gut, releasing glucose into the blood stream. At the same time, a delicate concert of signals causes the pancreas to release insulin, which travels through the body and makes cells present glucose transporters on their outsides such that they can take glucose in from the blood, and blood sugar levels remain steady. In a diabetic, with no beta cells to make and release insulin, the sugar stays in the blood unless insulin is injected directly.
Scientists look at this glucose-insulin signaling system and see several places to intervene: what if we could make other cells produce insulin? Then we would have insulin in the diabetic. The trick here, though, is that too much insulin is bad, too. The usual approach, therefore, is to try to make new beta cells or to make cells that can specifically respond to glucose. The authors of this paper, though, take a less traveled path: instead of trying to manufacture cells that respond to glucose and produce insulin, they say, let’s just force the cells that need the glucose, like those in the skeletal muscle, to take up all the extra sugar that’s in the blood, even if there’s not enough insulin.
How do they do this? That’s where gene therapy comes in. The researchers thought to use viral vectors to force skeletal muscle cells to make the insulin protein. Not enough that the host animal becomes hypoglycemic, but just a little bit, so that cells throughout the body are stimulated to put glucose transporters on the surface of the cell, waiting for glucose. At the same time, a second viral vector forces the expression of a protein called glucokinase (Gck). Gck is normally only expressed in the liver, and it functions downstream of the glucose transporter to help bring glucose in and ready it for processing. Skeletal muscle cells express a similar kinase, but it is less effective at processing the glucose. More importantly, the researchers in previous studies had found that if they forced skeletal muscle to express Gck, it only started working when glucose levels were high, helping to pick up the slack left by the version of the protein naturally in skeletal muscle cells.
So to put all the pieces together: in order to allow glucose to get into cells in the absence of beta cells, the scientists force skeletal muscle cells to make a little bit of insulin, and also to make a protein in the glucose processing pipeline that works very quickly when extra glucose is present. Thus, skeletal muscle cells are able to take up lots of extra glucose, and hyperglycemia is avoided, even without beta cells.
Does this actually work? The authors previously showed that this dual-gene treatment worked with mice , but in this study, they moved up to large animals– namely, dogs whose beta cells had been killed with a selective toxin, streptozotocin. The treated dogs thus lost all beta cell function and became hyperglycemic, making them a useful model for type 1 diabetes. After becoming diabetic, the dogs were treated with on of:an injected insulin regimen; the Gsk gene alone; the insulin gene alone; or both the insulin and the Gsk genes. Not surprisingly, without treatment, the dogs were hyperglycemic, and with regular injected insulin, the dogs were better but not great. (For those of you who are diabetics or parents of diabetics, it should come as no surprise that multiple daily injections doesn’t really cut it!)
And the gene therapy treatments? Expression of Gsk alone did not ameliorate hyperglycemia; without any insulin, the skeletal muscle presumably did not have any glucose transporters on the surface of cells to even begin to let glucose in to reach the Gsk protein. Expression of low levels of insulin alone on the other hand was an improvement over no treatment at all– the dogs regained some of the weight lost upon induction of diabetes, and fasting glucose levels were near normal. However, the dogs were unable to properly process glucose after meals, and, like the untreated dogs, spiked up to 500 mg/dL and remained elevated for hours.
To test both the insulin and Gsk vectors together, the researchers injected five dogs with one or two doses of the viral vectors. The dogs quickly regained normoglycemia, and even the higher doses were tolerated well by the animals, not causing any toxicity or side effects. Unlike the dogs given insulin alone, the dogs given both genes together responded much better to an oral glucose tolerance test, spiking slightly up to between 200 and 300 mg/dL, but stayed well below the levels seen with the diabetic dogs and returned to normal within three hours. (Healthy, non-diabetic controls peaked around 150 mg/dL.) And the really impressive part: the dogs have remained normoglycemic, with continuing expression of insulin and Gsk in the skeletal muscle, for four years and counting.
So what does this all mean? Have the researchers cured diabetes, and if so, how quickly can we move this sort of treatment into humans?
First, my personal soapbox: the researchers here have cured hyperglycemia in the dogs, not diabetes; type 1 diabetes is a complex autoimmune disorder, and the immune aspect was not addressed in this paper. That’s just semantics, though, and as a diabetic, I would be more than happy with a cure for hyperglycemia!
But, that brings me to a problem I see; the researchers have essentially done two things: (1) They enforce low basal levels of insulin, and then (2) they make skeletal muscle cells very insulin sensitive by making them take up glucose faster than normal.
In this dog model, those two things are sufficient to keep the dogs in range. However, if I extrapolate to humans, this seems to be similar to (1) taking a long-acting basal insulin, and (2) taking an insulin sensitizer such as Metformin to make cells particularly responsive to the low levels of insulin. Now, at this point, if you’re a type 1 diabetic, you’re saying, “Yeah, right, like that would work.” I, as a type 1 diabetic, think there may be a lot of value in incorporating Metformin into treatment, and I know there’s a lot of value in long-acting basal insulins. But the two together do not a cure make– there’s bound to still be too much up and down!
So why would this work in dogs? I see two primary possibilities: (A) I am incorrect in my equating the gene therapy to low basal insulin plus Metformin, and really there is something more intricate and responsive going on, or (B) the metabolism and lifestyle of humans is sufficiently different from even large mammals like dogs that what works in the latter is not necessarily sufficient for the former.
I won’t choose between those here, because I am not an expert and do not understand the details of the paper well enough to rule out (A). Plus, I hope (A) is true, and the treatment does work in humans. That would be amazing. Incredible. If (B) is the case, though, this study perhaps serves to reinforce the value of a combined Metformin and insulin treatment for type 1 diabetics, but does not offer a be-all-end-all cure or even a practically preferable path to insulin sensitization given the relative cost and risk of Metformin.
That said, regardless of what they find with this particular application of gene therapy, I do commend the researchers for a unique approach to treating diabetes (we need those!), and also for taking one more step forward with gene therapy. Gene therapy as a field is very exciting, and I look forward to seeing more applications reach the clinic. (The child in me says, “Gene therapy! The future is now!”)
Anyone out there with more knowledge than me? Is the answer happily (A) or unfortunately (B)?