100 Years of Insulin: Lifesaver, immune target, and potential remedy for prevention

Abstract

In this review, we bring our personal experiences to showcase insulin from its breakthrough discovery as a life-saving drug 100 years ago to its uncovering as the autoantigen and potential cause of type 1 diabetes and eventually as an opportunity to prevent autoimmune diabetes. The work covers the birth of insulin to treat patients, which is now 100 years ago, the development of human insulin, insulin analogues, devices, and the way into automated insulin delivery, the realization that insulin is the primary autoimmune target of type 1 diabetes in children, novel approaches of immunotherapy using insulin for immune tolerance induction, the possible limitations of insulin immunotherapy, and an outlook how modern vaccines could remove the need for another 100 years of insulin therapy.

eTOC blurb

Ziegler et al. showcase insulin from its breakthrough discovery as a life-saving drug 100 years ago to its uncovering as the autoantigen and potential cause of type 1 diabetes and eventually as a target of future immunotherapies to prevent autoimmune diabetes.

My uncle was born the year surgeon Frederick Banting and student Charles Best began their experiments to extract insulin from the pancreas of dogs, tells Anette Ziegler. At age 21, he developed type 1 diabetes. The diagnosis was made in the military by a medical student. The diagnosis saved him from the rest of the war and the discovery of insulin saved his life. At the age of 55, he died from diabetic complications. He himself studied medicine and founded one of the first diabetes outpatient clinics in Germany. Together with his friend Hellmut Mehnert, my later teacher and mentor, he scripted a play ‘Better than Insulin’ which first introduced me to the groundbreaking advances in insulin treatment.

Insulin saves lives

The birthday of insulin

Studying medicine in his home town Berlin, Germany, the story of insulin always fascinated Thomas Danne. It started in Berlin in 1869, when the medical student Paul Langerhans described in his dissertation the pancreatic islet cells named after him¹. In 1909, de Meyer gave the name “insulin” to the unknown substance formed in Langerhanś islets². In 1921 two Canadians, Frederick Banting, a surgeon with a vision but no formal research training, and Charles Best, a medicine student made it happen³. The birthday of insulin on January 11, 1922, was also the beginning of pediatric diabetology, when 14-year-old Leonard Thompson, diagnosed with diabetes in 1919 was treated at Toronto General Hospital with the extract prepared by Banting and Best. However, the researchers considered it a complete failure, since the glucose decreased only from 440 to 320 mg/dl, ketones remained present and the local outcome was sterile abscess in the buttocks⁴. James Collip refined the extraction progress and his extract on Jan. 23–24, 1922 normalized the glucose and made the ketones disappear. On February 5, 1922, Banting and Best published the article “The internal secretion of the pancreas”⁵. Banting and his boss and head of the laboratory J.R.R. Macleod were rewarded with the Nobel Prize in 1923. Best and Collip were not nominated but the two winners shared the money with them. With one of the greatest discoveries in medical history, the insulin era of diabetes mellitus began. The horror of a life-threatening disease had been taken away. Diabetes could not be cured with the help of insulin substitution, but it could be treated effectively (Figure 1). Collip and Best sold the insulin patent to the University of Toronto for only $1. They wanted everyone who needed their drug to be able to afford it. List prices of insulin in the United States have nearly tripled from 2002 to 2013, according to a report from a working group at the American Diabetes Association⁶. This problem is certainly not limited to the United States and reasons for this increase are not entirely clear but are due in part to the complexity of drug pricing in general and of insulin pricing in particular. Worldwide, some patients are still rationing their insulin because taking too little is better than having none to take at all. Thus, even one hundred years after the discovery, we need to encourage innovation in the development of more effective and affordable insulin preparations.

Figure 1. Insulin: Lifesaver, immune target, and potential remedy for prevention.

Schematic representation of the different facets of the role of insulin for type 1 diabetes including insulin to treat patients, insulin as primary autoimmune target of type 1 diabetes in children and novel approaches of immunotherapy using insulin for immune tolerance induction.

The story of insulin immunity begins

Until about 1980, insulin was obtained exclusively from the pancreases of slaughtered animals, especially cattle and pigs. One porcine pancreas contains approximately 27mg Insulin corresponding to 780 IU of insulin which allowed the extraction of 14 mg i.e. 400 IU of active insulin. A child with a daily insulin need of 40 IU/day would need pancreas extracts from 37 pigs per year. Berson and Yalow were the first to demonstrate that days to weeks after the start of insulin substitution, insulin antibodies are formed, circulate in the blood, and affect the action of the injected insulin⁷. Causes for antigenicity are both the molecular differences between bovine or porcine insulin and human insulin, and the prior priming of the immune system against endogenous insulin. Circulating antibodies can bind insulin and release it again later according to their avidity. At high insulin antibody concentrations, the insulin effect is therefore prolonged, but also reduced⁸.

Tailoring insulin action to individual needs

The industrial production of human insulin has made the insulin market independent of the limited availability of bovine and porcine pancreas^9,10, and clinical trials of human insulins began in 1980¹¹. High levels of insulin antibodies influencing therapy now have become extremely rare. Industry research and development did not stop with human insulin. By hindering or enhancing the association of insulin molecules to di- and hexamers, absorption can be accelerated or slowed, respectively¹². This has led to the development of several generations of short and long-acting insulin analogs (Figure 2). The different absorption profiles and half-lives of these insulin preparations allow insulin therapy to be tailored for the individual patient, becoming a crucial element of modern treatment of patients with type 1 diabetes^13–21. For example, a child who always eats the same breakfast and school snack may want to cover both meals with a single injection of regular insulin. However, when flexibility is needed and a meal with rapidly absorbed carbohydrates is on the menu, an ultrafast insulin analog would be more appropriate for prandial injection (Figure 3). The same goes for basal insulin substitution. An adolescent who has difficulties keeping regular injection times for basal insulin may profit from a very stable basal insulin analog, which results in less variability of insulin concentrations when injected each day at a different time. In contrast, an individual who needs to tailor basal insulin to the circadian rhythm of insulin sensitivity and exercise may choose three daily injections of the “old-fashioned” NPH insulin²², which has the shortest duration of all basal insulins²³.

Figure 2. Development of insulin preparations over the years.

Timeline indicating the development and classification of different insulin preparations.

Figure 3. Action profile of insulin preparations.

Schematic representation of insulin action over time of different insulin preparations.

Diabetes education amplifies the therapeutic efficacy of insulin

Karl Stolte, a German pediatrician had been breaking new ground in the treatment of children and adolescents with diabetes since the end of the 1920s. Instead of strictly calculated diets he developed an insulin substitution method which, from today’s perspective, can justifiably be called the precursor of intensified insulin therapy. He proposed the daily new adaptation of the insulin dose to a freely chosen food intake²⁴. However, it took until the Diabetes Control and Complications Trial (DCCT) in 1993 before such therapy became standard of care²⁵. Another major advance came with the development of rapid-acting and long-acting insulin analogues with stable action profiles that made it possible to undertake individualized treatment without strict rules for diet and other life style factors. However, even with all the progress in insulin therapy until the present day education remains a key for success.

Rediscovering intensified insulin therapy

In contrast to Stolte, the majority of leading diabetologists at that time practiced a therapy which had the declared goal of sparing the patient from frequent insulin injections, i.e., to inject insulin only once or twice a day. Such practice had been facilitated by the introduction of delayed-release insulin preparations in the mid-1930s. However, application of relatively large amounts of delayed-release insulin resulted in permanent hyperinsulinism, which could only be compensated by frequent, precisely calculated meals containing carbohydrates. As a result, the focus of diabetes treatment was adherence to a strictly calculated diet. Conventional insulin therapy was, by its very nature, a thoroughly restrictive method of treatment that strictly regulated the lives of people with diabetes day and night, year in and year out. Moreover, the common misconception that metabolic control in pre-pubertal age does not contribute to late diabetes complications led to most children living in almost constant hyperglycemia. This error was finally remedied and challenged when Thomas Danne and Bruno Weber presented the results of the population-based longitudinal Berlin Retinopathy Study. This study provided ample evidence that hyperglycemia at every age mattered²⁶. Building on these results and following pioneering efforts of Johnny Ludvigsson²⁷ in Sweden and Carina de Beaufort in The Netherlands²⁸, Danne became an advocate for intensified insulin therapy in children worldwide. Resistance was abundant, but intensified insulin therapy eventually became the gold standard also in the pediatric age group^29,30, in particular after the introduction of insulin analogues that reduced hypoglycaemia and were much better suited to varying life situations than the combination of NPH and regular insulin which was the only option for so long.

The special characteristics of childhood and adolescence make highly individualized treatment necessary. In contrast to most adults, insulin sensitivity in children varies due to the influence of growth and hormonal changes, different daily routines and frequent infectious diseases. Since the insulin dose must be adapted to the food intake, the unpredictability of physical activity (running around) and the, sometimes irregular, food intake in children, treatment must be flexible. Parents or guardians bear the responsibility for their child’s daily diabetes therapy well into adolescence and must also deal with the variability of their child’s daily needs. Fortunately, the original pedagogic concepts and therapeutic principles developed by the pioneers of insulin therapy such as Joslin and Stolte were rediscovered after the DCCT. These strengthen the concept that different training modules (structure, content, didactic concept) are required for preschool children, elementary school children, adolescents in puberty and adolescents during the transition to adult diabetological care. Modern diabetes education aims to promote the self-management skills of affected children and adolescents and their families. However, transferring sole responsibility to adolescents with diabetes too early has proven to be unfavorable³¹.

Insulin pumps on the rise

Shortly after the principle of continuous insulin infusion (CSII) was introduced into diabetes therapy, insulin pump regimens were studied in children³², allowing intricate programming of basal and prandial insulin³³. Recently, the SWEET registry demonstrated that the management of type 1 diabetes has diversified³⁴. Analysis of more than 25,000 pediatric patients with type 1 diabetes from 101 centers worldwide found an increasing use of insulin pumps and continuous glucose monitoring. Although the most frequently used combination among patients in these specialized pediatric diabetes centers was still injections and blood glucose monitoring, 60% of patients were using at least one technological component such as continuous glucose monitoring (CGM) or pump for diabetes management. Even after adjustment for demographics, region, and gross domestic product–health per capita, a combination of pump and CGM resulted in a greater likelihood of accomplishing treatment targets. Current data shows the superiority of clinical outcomes when starting pump CSII directly at onset as compared to later during the course of the disease³⁵. In pediatrics, insulin pump therapy is becoming established as the standard method of insulin substitution in countries with economic resources to pay for diabetes technology³⁶. In countries or regions with low socioeconomic status, the use of diabetes technology is still modest with the consequence that HbA1c levels are high³⁷. These data raise the concern that youth with type 1 diabetes from lower socioeconomic status will be at a systematic disadvantage to achieve optimal diabetes outcomes as advances such as automated insulin delivery systems are made.

Automated insulin delivery is the future

The ability to continuously trace glucose levels by CGM has changed the everyday (and every night) life of the patients, particularly among children and their parents. CGM also allows the calculation of time in range (TIR) as a measure of glycemic control. TIR goes beyond HbA1C in representing blood glucose levels because it captures variation – the highs, lows, and in-range values that characterize life with diabetes. Treatment targets of 70% with less than 4% in the hypoglycemic range have been proposed³⁸. CGM is also an essential part of the future of insulin dosing, and various stages of automatic insulin regulation are currently under development³⁹. In a first step, the combination of insulin pump and glucose sensor allowed a prospective interruption of insulin delivery in case of imminent hypoglycemia (PLGM-predictive low glucose management)⁴⁰. Patients do not even notice this process if the acoustic alarm function is switched off. In the next step, the first trial of automated insulin delivery (AID) including insulin suspension for lows but also correction boluses for highs outside the hospital was performed in the typical pediatric setting of diabetes camps, and demonstrated a nearly threefold reduction of hypoglycemic episodes for AID under these challenging conditions⁴¹. Ultimately, the knowledge of the technical feasibility of closed-loop systems as well as the increasing exchange of information via social networks has led to the use of so-called “do-it-yourself” systems (DIY)⁴², in which insulin pumps and sensors are combined with algorithms that are freely available on the internet. Legally, these DIY systems are currently considered problematic because they do not meet the standards of a medical device. On the other hand, we as health care professionals have an ethical duty to support people living with diabetes to achieve the best glycemic control they are capable of, at least if they understand the potential risks of their chosen therapy. Thus, as a diabetes community we need to work together to ensure that life changing closed loop systems can be safely made available to those who wish to use them whether this be via a commercial option or DIY.

The next commercially available solution is the so-called hybrid closed loop insulin pump, that can dynamically adapt basal insulin delivery to glucose values but requires manual entry of carbohydrates at mealtime^43–49. Importantly, detailed education remains essential despite the increased degree of automation in insulin delivery and glucose sensing⁵⁰. Bihormonal systems offer the possibility of more aggressive insulin dosing when protection against hypoglycemia is available through counterregulatory glucagon infusion^51,52, In view of the improvements achieved with insulin-only AID⁵³, it is likely that bihormonal systems may be appropriate for a small minority of patients. It also should be considered that, despite the advances, diabetes technology is not for everyone and that treatment options will and should remain a shared decision between the diabetes team and the individual with diabetes tailored according to personal preferences after unbiased information and education.

Smart insulin or smart devices

Glucose responsive insulin, also known as “smart insulin”, is chemically activated in response to changes in blood glucose levels. Intelligent insulin remains inactive until the blood glucose level rises above normal. At that time, the chemical component activates the insulin. Once blood glucose returns to normal, the insulin action ceases, thus preventing low blood glucose levels. To be a practical remedy, smart insulin should act long enough to avoid the need for multiple daily injections. Not only smart insulins, but also smart pens have been designed. Indeed, an automated dosing advisor was found to be non-inferior compared to expert diabetologist advice in a multinational multicenter randomized trial⁵⁴. The current challenge for diabetes teams around the world is fighting for affordable diabetes technology and counseling the individual person with diabetes regarding available systems that matches his or her needs⁵⁵.

From Savior to Immune Target

Insulin becomes an Autoantigen

Ezio Bonifacio remembers the day when Jerry Palmer and colleagues had published the presence of autoantibodies to insulin in patients prior to having their first insulin injection⁵⁶. It was a pivotal moment for diabetes researchers. However, the acceptance of insulin autoantibodies (IAA) was not instant. Part of the problem was that IAA were not visible in the immunofluorescence islet cell antibody (ICA) test. The other reason was the emergence of new and different assays that attempted to measure IAA. These included several solid phase ELISAs that lacked both sensitivity and specificity. It wasn’t until a series of workshops were carried out that the scientific community at large started to believe and recognize their importance in the disease process⁵⁷.

Anette Ziegler had gone to the Joslin Diabetes Center in Boston for a post-doc period in the late 80s when Stuart Soeldner developed what turned out to be the gold standard radio-binding assay for IAA measurement⁵⁸. Using this test, she measured IAA in a collection of samples from first-degree relatives of patients with type 1 diabetes from Germany as well as the Joslin Family Study collection of samples⁵⁹. This was where and when she realized that IAA were predictive of future type 1 diabetes, and that the combination of both, IAA and ICA were associated with accelerated disease progression⁶⁰. The antibodies were present in many of the children who later developed type 1 diabetes, they were present in younger children and patients and were associated with the genetic marker of type 1 diabetes, HLA DR4-DQ8^61,62 (Figure 4 bottom). The savior of type 1 diabetes was now also an important autoimmune target of the disease.

Figure 4. Insulin as critical autoantigen.

Top: influence of the proinsulin gene on the development of autoantibodies, insulitis and diabetes in NOD mice. Bottom: influence of genetic risk on childhood type 1 diabetes.

Insulin as the primary autoimmune target in animal models

George Eisenbarth was Anette Ziegler’s mentor during her time in Boston and also understood how important insulin was as an antigen in the pathogenesis of type 1 diabetes. He and others showed that antibodies were present in spontaneous animal models of autoimmune diabetes and that their presence was also predictive of progression to disease in these models⁶³. He helped develop the notion that autoimmunity to insulin had an essential role in the pathogenesis (Figure 1). Now placed in Denver, he together with Dale Wegmann had identified that the dominant peptide recognized by islet T-cells from animal models was the insulin B-chain peptide B9–23⁶⁴. Maki Nakayama from his group performed some elegant experiments where she engineered non-obese diabetic (NOD) mice so that they had no native insulin but instead a functional insulin that had been mutated at what had been proposed to be a critical epitope for autoimmune recognition⁶⁵ (Figure 4). Remarkably, these mice failed to develop autoantibodies against insulin and neither developed insulitis nor diabetes. Effectively, her studies placed insulin as a primary cause in the what is the quintessential animal model for type 1 diabetes. This was icing on the many studies by many investigators that showed how antigen specific immunotherapy with insulin was able to prevent or delay diabetes in the NOD mouse^66–70. Thus, the notion that insulin could again be the savior had already been foreseen.

Insulin as a potential primary immune target in human disease

While it was established that IAA appeared prior to disease onset and were predictive of disease onset, little was known regarding the timing of the development of these antibodies. The infamous concept that type 1 diabetes is a chronic disease proposed by George Eisenbarth had focused the community on the presence of an autoimmune pre-symptomatic disease phase. It was the BABYDIAB study, however, that first provided concrete data on how early the autoimmunity starts in life^71–73 (Figure 4). The BABYDIAB study was the first to prospectively follow children who had a genetic risk of type 1 diabetes from birth at regular intervals throughout childhood and adolescence. Auto-antibodies were already present at the age of 9 months in some children⁷¹. It later became evident that this 9 months to 3 years age-period was the peak period for the development of type 1 diabetes-associated autoimmunity^74,75. These findings indicated that children who develop type 1 diabetes usually start the process of autoimmunity sometime in the first 3 years of their life. Importantly, it was IAA that was present already in in the age 9 months samples. Ezio Bonifacio, who had come from the Bottazzo school, which believed that ICA was the most important of the antibodies⁷⁶ was skeptical and measured ICA in these children. However, the data soon convinced him that insulin was indeed an important target, as case after case showed that IAA were always present at that first positive sample (Figure 1). More antigens were discovered during the 90s^77,78, including those that contributed to the ICA⁷⁹, but even with very sensitive assays to measure these antibodies, the message remained the same: IAA were there first.

Ezio Bonifacio, Anette Ziegler, and others sought to find out more about these antibodies. IAA were predominantly of IgG1 subclass⁸⁰, and very quickly acquired high affinity in children who subsequently develop diabetes, but often remained low affinity IAA in children who did not progress to type 1 diabetes^81,82 (Figure 4). High affinity, disease-related IAA also bound proinsulin and it remains unknown whether the IAA target of origin is insulin or proinsulin. In addition to their association with HLA DR4-DQ8^62,83, IAA were also associated with the type 1 diabetes susceptible genotype of the INS gene⁸⁴, which itself was associated with a higher methylation status of multiple CpG sites on the insulin gene as compared to the non-susceptiple genotype⁸⁵, and with decreased insulin protein expression in the thymus⁸⁶. Therefore, like in the NOD mouse, the evidence in humans suggested that insulin could be the primary autoantigen of type 1 diabetes and that a strong genetic determination of insulin autoimmunity was a primary cause of type 1 diabetes in children.

As more studies have been conducted, in particular the TEDDY multi-center study, the role of insulin autoimmunity in relation to autoantibodies against other islet proteins was recognized^87,88. While the first early peak of autoimmunity is dominated by the presence of IAA, children who develop islet autoimmunity later in childhood, often have auto-antibodies against, for example, GAD without the initial presence of IAA. Notably, IAA is rare in patients who develop type 1 diabetes in adulthood⁶¹. It is now considered that there may be age-related multiple forms of type 1 diabetes autoimmunity with different etiologies⁸⁹. In this regard, the TEDDY study has suggested that different environmental and genetic factors predispose for the development of insulin and GAD autoimmunity⁹⁰. Maternal type 1 diabetes, for example, decreases the risk of early IAA but is less protective against a GADA first autoantibody phenotype. The same was true for early probiotic use in TEDDY study participants⁹¹. Endotypes with different etiologies may require different prevention strategies.

Insulin as protector against disease

Insulin antigen-specific immunotherapy

Targets of aggression deserve protection. In this case, the aggressor is the immune system, which could be suppressed or re-educated. Re-education when applied to autoimmunity has substantial advantages. It is likely to be very specific to that particular target, and, therefore, have fewer side effects. In this way it could be applied rather broadly. It has been proposed and considered that the targeting of insulin by the immune system is a crucial step in the development of childhood type 1 diabetes. This posits that the aberrant recognition of insulin as foreign by the immune system in children who have HLA DR4 and INS susceptibility genotypes is necessary for pathogenesis. This is also supported by the finding of insulin-specific T cells in pancreatic islets of patients^92,93. If this holds true, insulin could be a source to correctly educate the immune system in recognizing insulin as a safe protein (Figure 1). The notion that insulin could immunologically save a child from developing type 1 diabetes was proposed already in the 80s and 90s. One question was how? Early studies in allergy had shown that repeated exposure to an allergen in certain doses could lead to some form of desensitization and this was associated with a shift from IgE response to an IgG response. Animal models of autoimmune diabetes also showed that disease could be prevented by insulin immunization. Subcutaneous injection of insulin with incomplete Freund’s adjuvant, repeated exposure to insulin orally, or exposure to proinsulin peptides intranasally or intradermally, all showed a positive effect with a reduction in the development of the disease in rodent models^{66–70,94–97}.

Studies in IAA positive individuals

Pilot studies administering insulin subcutaneously and/or intravenously (IV) to IAA positive humans were indeed promising^98,99. They were associated with a change in the immune response, in this case more antibodies, confirming that exposure in this manner could reach the immune system. They also suggested that there was a delay in the progression. However, when larger multi-center trials were performed, the overwhelming evidence suggested that IV and subcutaneous insulin injection did not delay the progression to the disease¹⁰⁰. Were we wrong to consider insulin as a primary auto-antigen or did we choose the wrong re-education method? Regular oral exposure had been shown to work in NOD mice by Howard Weiner in the early 90s⁶⁷. When applied to humans, there was relatively little knowledge with respect to how much or how frequent it should be given in order to reach the immune system. After a pilot study, a phase II trial orally administering 7.5 mg of insulin daily was performed in IAA positive first-degree relatives of patients¹⁰¹. The trial had difficulties in recruitment and therefore lowered the IAA positivity threshold required to enter the trial. The trial showed no delay in the oral insulin treated group as compared to the placebo group. However, post-hoc analysis showed a significant delay in those who had fulfilled the original higher threshold IAA entry criteria. A second trial was conducted by the TrialNet consortium¹⁰². This failed to meet its primary outcome, but in a pre-specified analysis and again in post hoc analyses did show a delay in the progression in the subset of those who participated¹⁰³. We still don’t know if 7.5 mg of insulin was sufficient to reach the immune system or if it can delay progression to clinical diabetes.

Preventing insulin autoimmunity

One concern that was raised for insulin immunotherapy is that it may be indeed too late to give the therapy after the development of IAA. There was evidence that insulin autoimmunity may be a key primary step in the disease process but that as the pre-symptomatic phase of the disease continues, insulin autoimmunity may be less relevant and autoimmunity against other antigens become the drivers in the development of clinical disease. Therefore, some of us together with George Eisenbarth proposed that protection using insulin may best be achieved by antigen specific therapy in primary prevention trials. Such primary prevention trials would require active exposure to insulin in genetically susceptible young children before they develop autoantibodies and essentially before the peak period of developing autoimmunity. The first task was to identify a safe dose of oral insulin that may reach the immune system. PrePOInT was a small dose-escalating trial conducted in children aged 2 to 7 years who had extreme genetic risk for type 1 diabetes¹⁰⁴ (Figure 5). It included a range of insulin doses from 2.5 mg to 67.5 mg. The highest dose showed some evidence of an immune response and, importantly, was safe and did not induce hypoglycemia. It was particularly important to demonstrate such safety also in children as young as age 6 months since this was the age the antigen specific immunotherapy had to be applied. PrePOInT Early gave the safety green light in this age group¹⁰⁵. Although it did not meet its primary immune efficacy outcome, it showed overall safety and some evidence of reaching the immune system in children who had both HLA DR4 and susceptible INS genotypes.

Figure 5. Development of insulin immunotherapy.

Approaches for antigen-specific immunotherapies based on insulin. Left: therapies which are currently under investigation in man. Right: therapies which are currently in pre-clinical stages.

Embarking on a large primary prevention trial with oral insulin was against the trend some 5 to 10 years ago. There had been substantial skepticism regarding how much antigen would be required to reach immune tolerance including the notion that it was not possible to give a sufficiently high antigen dose orally or sublingually to achieve tolerance for conditions such as peanut allergy. Fortunately, such discussion did not stop all primary prevention investigators, and especially those of the LEAP study for performing clinical trials. Investigators in the UK LEAP study showed that oral antigen specific therapy with peanut antigen was incredibly effective in preventing peanut allergy¹⁰⁶. Although allergy and autoimmunity have very different etiologies and immune responses, the peanut allergy experience showed that oral exposure to an antigen could prevent sensitization in genetically susceptible children. With the courage of a funding agency to substantially invest resources into primary prevention of type 1 diabetes, the phase II randomized controlled POInT trial through the GPPAD consortium was initiated^107,108. With the daunting task of screening hundreds of thousands of newborns for their type 1 diabetes genetic risk¹⁰⁹ and recruiting over 1000 babies into a placebo randomized trial among 5 European countries, the GPPAD consortium is proud to have reached the first milestone which is completion of recruitment ahead of schedule in March 2021. If the trial is allowed to continue its follow-up until the predetermined completion, we should know if there has been a delay in the development of pre-symptomatic type 1 diabetes sometime in 2025.

Although tolerance against peanut antigens has boosted the field, one should remain grounded, cautious and careful with respect to the prospects of achieving tolerance through oral antigen. In our case, it is possible that preproinsulin (and not insulin) is the key antigen and that even if tolerance against insulin is achieved, we are missing tolerance to important epitopes in order to prevent disease. It is also possible that in an autoimmune disease such as type 1 diabetes tolerance against insulin that reaches the oral mucosa and gut will have very little effect on an autoimmune process that is directed to antigens in and at the pancreas where the process may also include altered forms of antigens¹¹⁰. Moreover, re-educating the immune system against insulin may only be part of what is required and that any form of administration of insulin or proinsulin will be ineffective without adjuvant therapy that strengthens this re-education. For these reasons, novel and other forms of antigen specific immunotherapy that have been under investigation are very much welcomed.

Novel approaches of antigen immunotherapy

After Mark Peakman’s group showed that autoreactive T cells from patients with type 1 diabetes predominantly reacted with peptides outside the insulin A and B chain of the preproinsulin molecule, the notion of using these peptides for antigen-specific immunotherapy gathered speed (Figure 5). Delivery strategies have included the use of nucleic acid encoded antigen, immunodominant proinsulin peptides, or coupled systems designed to enhance tolerogenic presentation and action. All related studies are at an early stage of clinical development in humans with promising initial results. Intramuscular DNA vaccination with a proinsulin-encoding plasmid was successfully tested in a phase 1 clinical trial in patients with established diabetes¹¹¹. Novo Nordisk is continuing its development and together with TRIALNET, has initiated a similar study in patients diagnosed with T1D (https://www.trialnet.org/events-news/blog/trialnet-launches-new-study-exploring-plasmid-therapy). An interesting concept of orally delivering proinsulin produced together with the immunomodulatory cytokine IL-10 in a genetically modified Lactococcus lactis is also in trial as a combination therapy¹¹² with low-dose systemic anti-CD3 therapy (ClinicalTrials.gov Identifier: NCT03751007).

Mark Peakman designed single and multi-proinsulin peptide pools that could be used for immune tolerance induction, and performed a phase 1 clinical trial using intradermal injections of an HLA-DR4-restricted immunodominant proinsulin peptide in patients with newly diagnosed diabetes¹¹³ with some promising indication of immune efficacy. The same peptide was used by Bart Roep in a trial testing the safety and feasibility of intradermal injections with peptide pulsed tolerogenic dendritic cells¹¹⁴. Importantly, all these approaches using (pre-) proinsulin and proinsulin peptide therapy appeared to be safe with no signs of immune suppression, allergic reactions, or worsening of the disease course. However, they are still exploratory in nature, highlighting the need to validate these findings in further and larger clinical trials. We also await human studies for promising preclinical results using nanoparticle delivery of insulin¹¹⁵, and glycosylated insulin targeted to hepatic antigen-presenting cells¹¹⁶.

Insulin, the protector as a model, but perhaps not for everyone and not at every disease stage

A quintessential paradigm for all antigen-specific therapies is the induction or expansion of the FOXP3⁺ regulatory T (Treg) cells^117,118. As a Post-Doc in the laboratory of Harald von Boehmer in Boston, Carolin Daniel investigated mechanisms of Treg cell induction in murine models. She and others demonstrated that conversion of naive T cells into ‘induced’ (i)Treg cells in vivo can be achieved through subimmunogenic antigen stimulation with strong-agonistic T cell receptor (TCR) ligands^118–120. In the context of islet autoimmunity and diabetes in NOD mice, she showed that the unfavored way of antigen presentation by MHCII and recognition by autoreactive insulin-specific T cells resulted in a failure of efficient Treg cell induction in the periphery¹²¹. The work evolved to identify properties of the insulin epitopes that would favor Treg cell induction. Building on molecular work from John Kappler’s group¹²² and following insights from structural analyses of a human HLA-DQ8 insulin complex¹²³, two human insulin mimetopes were optimized for Treg cell induction in humanized NSG-HLA-DQ8 transgenic mice^124,125. Under steady state conditions, this system efficiently induced human insulin-specific FOXP3⁺ Treg cell that were stable, had increased abundance of Treg cell signature genes and robust suppressive capacity¹²⁴.

In the process of finding Treg inducing peptide therapeutics, Carolin Daniels’s team had to design reagents to identify insulin-specific Treg cells. They achieved this with the same insulin peptide mimetopes incorporated into HLA DQ8 tetramers¹²⁴. By combining tetramer staining with intracellular FOXP3 staining, an assay that permitted the direct ex vivo analysis of the relevant human insulin-specific FOXP3⁺ Treg cell population was developed and used to address how these cells develop in humans. Anette Ziegler was always intrigued by the strong link of insulin autoantibodies to young age⁵⁸ as well as the decline in the titer and eventual loss of IAA in a number of children who progressed slowly or not at all to clinical diabetes⁸⁸. Ezio Bonifacio had the insight that the insulin became less relevant as a driving antigen as the disease progressed from the model of islet transplantation. He observed that in transplanted adult patients who had an immediate restimulation of their islet autoimmunity, there was no evidence of increased insulin autoimmunity, but a marked increase in GAD autoimmunity¹²⁶. It was possible, therefore, that the immune system in patients naturally took care of the insulin autoimmunity with time and age¹²⁷. Carolin Daniel took her quantitative insulin-specific Treg cell assay and tested samples from IAA positive children with rapid and slow progression. Supporting the notion that insulin autoreactive T cell phenotype changes with age and/or time, increased frequencies of insulin-specific FOXP3+ Treg cells accompanied by reduced frequencies of insulin-specific T follicular helper precursor cells were found in children with ongoing islet-autoimmunity but without progression to overt clinical type 1 diabetes for more than a decade^124,128. Of note, frequencies of insulin-specific Treg cells were found to be severely reduced during human and murine islet autoimmunity onset and in children with a fast progression to clinical type 1 diabetes^124,129.

The findings suggested that insulin use as an antigen-specific therapy to prevent disease would best be achieved in the very young. It appeared that this setting was ideal for inducing insulin-specific Treg cells that may delay or even prevent progression to overt diabetes¹²⁴. Again, we must be cautious, however. Carolin Daniel’s team also identified impairments in in vitro Treg cell induction during the onset of islet autoimmunity ie early. Defects in Treg induction during the onset of autoimmunity were likewise demonstrated in non-autoantigen-specific and in polyclonal Treg cell induction assays, in all cases using naïve CD4⁺T cells as the starting population^129,130. This impaired Treg cell induction was accompanied by increased T cell proliferation. Hence, it is possible that there is a finite optimal period for Treg inducing and/or expanding insulin-based antigen specific immunotherapy and that this may be dependent on an intrinsic ability or defect in generating Treg cells. The latter argues for combining insulin antigen-specific immunotherapy with immunomodulatory agents.

Genetic optimization of insulin-specific immunotherapy

Apart from genetic engineering technologies that may create Treg cells specific for insulin or islet antigens^131,132, there has been recent progress in delivering RNA to improve antigen-specific immunotherapy. One example is the use of microRNAs (miRNAs) as critical regulators of cellular states in T cells¹³³. Carolin Daniel showed that a miR181a-mediated increase in signal strength of stimulation and co-stimulation links Nuclear factor of activated T cells 5 (NFAT5) with impaired Treg cell induction and autoimmune activation. These findings are in line with the concept that low activity of the miR181a-NFAT5 signaling axis links improved Foxp3 inducibility with limited activation of PI3K/Akt/mTOR signaling¹²⁹. In addition to the observed impairments in Treg cell induction, we provided evidence for impaired Treg cell stability contributing to the multiple layers of immune tolerance impairments during ongoing islet autoimmunity¹³⁰. Therefore, targeting or delivery of T cell specific miRNAs could contribute to the development of innovative antigen-specific immunotherapy¹³⁴ and, thereby, broaden the window of application for insulin immunotherapy closer to clinical disease.

Finally, the current pandemic and the successful development of vaccines has led to pronounced and renewed interest in RNA biology and the use of mRNA as a vaccine. In particular, the BioNtech investigators who developed the first mRNA-based vaccine against SARS-CoV-2 also demonstrated the versatility of mRNA immunotherapy by inducing immune tolerance in a multiple sclerosis model¹³⁵. The basis of this work was the use of an mRNA encoding myelin in which the nucleotide uridine was replaced with 1-methylpseudouridine. This modification allows production of the antigen within cells, but removes the ability to trigger TLR6 and thereby restraining its pro-inflammatory characteristics. The vaccine fostered de novo Treg induction and expansion that suppressed Th1 and Th17 autoreactive T cells and prevented experimentally induced multiple sclerosis in the mice. We are hopeful that a similar approach with tolerogenic mRNA encoding preproinsulin and/or other autoantigens may be also be effective to prevent type 1 diabetes in humans.

Highlights.

• Insulin therapy is an art, innovations and new developments are not standing still

• Insulin is the primary autoantigen in childhood diabetes

• Insulin immunotherapy is explored for prevention of type 1 diabetes