Type 1 Diabetes Mellitus

How similar are human and animal endocrine systems?

This is not an exhaustive list of species-specific differences, nor can one be made given their unknown full extent, but rather an example of how these differences impact the face, construct, predictive, and intrinsic validity of animal models.

Not all species-specific differences can be accounted for in animal models, as there are hundreds of them, their relevance to Type 1 diabetes mellitus (T1DM) is unclear, and their interaction with other organ systems in the animal model makes it difficult to predict how they would have behaved within the human system.

Endocrine systems of human and non-human animals differ at every level, from molecular to organism level, contributing to poor translation of animal studies to humans.

​

Species-specific differences in structure and function of pancreatic islets

Inter-species differences in the cytoarchitecture of the pancreatic islets of Langerhans have consequences on the function of islets, susceptibility to diabetes, and disease pathophysiology.

In the islets of Langerhans, alpha, beta, delta, gamma and epsilon endocrine cells secrete hormones insulin, glucagon, somatostatin, pancreatic polypeptide, and ghrelin, that have distinct functions. While the beta cells-secreted insulin lowers blood glucose by promoting cellular uptake, the alpha cells-secreted glucagon raises blood glucose by stimulating hepatic release. The secretion of both hormones is inhibited by delta cells-secreted somatostatin. By regulating, stimulating and inhibiting each other through paracrine and autocrine communication, endocrine cells maintain balanced blood glucose levels and metabolic homeostasis.

In human islets, most beta, alpha, and delta cells are aligned along blood vessels with no particular order, indicating that islet microcirculation likely does not determine the order of paracrine interactions 12. By contrast, in mice the alignment of endocrine cells relative to blood vessels is more organized with beta cells that cluster around intra-islet capillaries.

In both human and murine beta cells, increase in blood glucose levels leads to increased ATP production, closure of K-ATP channels causing membrane depolarisation, influx of extracellular calcium (Ca2+), and insulin granule exocytosis. However, as described in the section 'Species-specific differences in glucose regulation', there are major inter-species differences at every step of this process.

​

Imaging of cytoplasmic free Ca2+ concentration, [Ca2+]i, showed that beta cell oscillatory activity was not coordinated throughout the human islets as it was in the mouse islets, suggesting inter-species differences in beta-cell network dynamics and insulin release patterns.

Mathematical modelling of Ca2+ activity in human and mouse islets showed that Ca2+ activity in human islets was more vulnerable to inhibition of highly connected beta cells, termed hubs, than in mouse islets 13. In contrast to human islets, mouse islets have a more uniform beta cell distribution, making them less dependent on hubs. This observation suggests that humans may be more susceptible than mice to loss of hubs, that can occur through autoimmune attack (T1DM) or toxicity/inflammation (T2DM).

Furthermore, human islets responded with an increase in [Ca2+]i when lowering the glucose concentration, which could be explained by the fact that human islets contain proportionally fewer beta cells and more alpha cells than do mouse islets 14, pointing to inter-species differences in glucose regulation. Indeed, contrarily to beta cells, alpha cells are electrically active at low glucose concentration, maintaining tonic Ca2+ influx that supports glucagon secretion. 

The neurotransmitter acetylcholine has a major role in the function of the insulin-secreting pancreatic beta cells. Cholinergic input to exocrine acini ducts and endocrine islets of Langerhans stimulates secretion of digestive enzymes and hormones.

In contrast to mouse islets, cholinergic innervation of human islets is sparse. Instead, the alpha cells of human islets provide paracrine cholinergic input that primes the beta cell to respond optimally to increases in glucose concentration. Cholinergic signaling pathways within human islets could potentially be a therapeutic target in diabetes 15. However, because of inter-species differences in pancreatic islets structure and function, this therapeutic lead, and probably many other leads that have the potential to benefit patients, are likely to have failed in animal models of diabetes.

​

Species-specific differences in gene expression

As it is the case for the overwhelming majority of species, humans have a single insulin gene. Rodents, however, have two non-allelic insulin genes that, despite their differences in structure, are both equally expressed. It is unclear why rodents have two insulin genes and whether these genes might have different functions, which limits extrapolation to the single insulin gene system in humans 16, 17.

The effects of insulin gene duplication in rodents may be further amplified, since insulin transcriptionally regulates the expression of more than 150 genes in various tissues.

Differences in sequences of insulin genes between humans and animals affect the regulation of expression of insulin genes, as well as the structure and function of encoded proteins 18.

Overall homology of the insulin promoter between humans and rodents is only around 45 to 48%. Many features of cis-regulatory elements and trans-acting factors are differentially regulated in human and mouse islets 16.

There are also marked inter-species divergences in DNA binding between mouse and human glucose regulatory transcription factors. As many as 41%-89% of  orthologous promoters bound by a protein in one species were not bound by the same protein in the second species.

Species-specific differences in cis regulatory elements (transcription factor binding motifs, epigenetic marks) and trans regulatory elements (expression levels, post-translational modifications) affect how glucose related-genes respond to insulin.

Consequently, the auto-regulatory mechanisms by which genes involved in glucose metabolism regulate their own expression and expression of other related genes via cis and trans inputs, diverge between humans and other species.

For instance, while the gene expression of gluconeogenic enzyme phosphoenolpyruvate carboxykinase is highly regulated by insulin in rodents, this insulin-mediated repression is less pronounced in humans, suggesting alternative regulatory pathways 21.

​

The inter-species differences in insulin protein sequence and post-translational modifications are likely to manifest in inter-species differences in insulin function and proteolytic processing.

For example, guinea pigs have highly divergent sequences of insulin genes and the biological activities of these insulins differ, acting more as a growth factor than as a metabolic hormone 17.

Similarly, as a consequence of specific changes in sequences of insulin genes, some species of New World monkeys have insulin hormones with lower potency, despite sharing 85.5% identity with the human insulin precursor.

Of particular relevance for T1DM, inter-species differences in insulin protein sequence can influence how the immune system recognizes insulin and targets insulin-producing beta cells. In case of T2DM, they may impact how insulin binds to its receptor and thereby produce inter-species differences in pathophysiology of insulin resistance. The response in preclinical studies of T1DM animal models to gene therapy via introduction of the human insulin gene may yield results that are not translatable to humans.

​

Species-specific differences in glucose regulation

Even though insulin is essential for maintaining optimum blood glucose levels in all vertebrate species, its function has evolved in species-specific manners.

The two most frequently used species to model T1DM/T2DM, mice and rats, differ significantly from humans at gene, protein, pathway, cellular, tissue, organ and organism levels of glucose regulation 18.

​

The entry of glucose into beta cells is enabled by glucose transporters (GLUT) via concentration-dependant facilitated diffusion. Inter-species differences in glucose transport have major repercussions for understanding genetic risk and how beta cells respond to glucose.

The principal glucose transporter present in rodent beta cells is glucose transporter 2 (GLUT2). Greatly reduced GLUT2 expression levels have been shown to correlate with elements of T2DM in various diabetic rodent models. However, human islets predominantly express GLUT1 and GLUT3, and GLUT2 expression levels do not correlate with human T2DM 19, 23. 

In comparison to GLUT2, GLUT1 has higher binding affinity and lower transport capacity for glucose, meaning that human beta cells are more sensitive than rodent beta cells to small changes in glucose, supporting a fine-tuned insulin secretion. The lower binding affinity combined with higher transport capacity of GLUT2-rich rodent beta cells that produce burst-like insulin release, is believed to correlate with higher baseline glucose levels and higher metabolic rates in rodents 20.

​

The specificities of human glucose metabolic pathways are critical for the regulation of human glucose homeostasis under normal and diabetic conditions.

In both mice and humans, upon entry into beta cells, glucose undergoes glycolysis to produce pyruvate, which then enters mitochondria to undergo either the Krebs cycle pathway or the anaplerotic pathway. In the former, pyruvate is converted into acetyl CoA which reduces NAD+ to NADH, that is subsequently fed into mitochondrial electron transport chain, ultimately producing ATP by oxidative phosphorylation. In the latter, pyruvate is converted into oxaloacetate that replenishes the Krebs cycle intermediates.    

However, in human islets, glucose metabolism via the anaplerotic pathway, that employs pyruvate carboxylase (PC) enzymes to convert pyruvate into oxaloacetate, was reported to be lower compared to rodent islets. Consequently, human islets appear to be less dependent on PC for insulin secretion, whereas PC-driven anaplerosis is the major contributor to insulin secretion in rodents 22. 

The stimulus-secretion coupling in human beta cell diverges from other species with respect to ion channel composition and function.

The ATP-sensitive potassium (KATP) channel plays a critical role in insulin secretion, by linking glucose metabolism-related ATP production to membrane depolarisation. At low ATP, KATP channels are open, allowing the potassium (K+) ions to flow out. When ATP rises, KATP channels close, producing depolarisation that opens voltage-gated Ca2+ channels. The sharply increased intracellular Ca2+ binds to Ca2+-sensing synaptotagmines on insulin granules, triggering the fusion of insulin-containing vesicles with the plasma membrane. In humans, polymorphisms in the KATP channel subunit genes have been associated with increased risk of T2DM. However, the loss of function in subunits that compose KATP channels does not cause dysregulation of insulin secretion in mice, probably due to unknown species-specific compensatory mechanisms 24, 25.

​

Voltage-gated Na+ channels amplify membrane depolarisation necessary for activation of voltage-gated Ca2+ channels. In contrast to mice, human beta cells carry voltage-gated Na+ channel isoforms Nav1.6 and Nav1.7. This means that Nav1.7 antagonists developed for use as analgesics could have serious implications for human patients who face potential impairment of insulin secretion as a side effect. Such adverse effects of drugs would not be predictable in mice since they lack functional beta cell Nav1.7 27.

The expression of voltage-gated delayed phase ionic current and delayed rectifier potassium channels, that ensure controlled insulin release by modulating depolarisation/repolarisation of the beta cell membrane, also differ between humans and mice.

Additionally, the kinetics of insulin release varies across species owing to inter-species differences in the functional importance of L-type Ca2+ plasma membrane channels 26. In contrast to rodents, L-type Ca2+ channels are less dominant and other Ca2+ channels, such T-type and P/Q-type, appear to play a considerable role in humans. As a result, drugs targeting L-type Ca2+ channels are likely to produce misleading results on safety and efficacy in mouse models of T1DM. 

​

At the tissue level, in humans the skeletal muscle is the primary site of glucose clearance, accounting for 50% to 90% of glucose uptake, making it the primary insulin-sensitive tissue and the primary site of dysregulation in human peripheral insulin resistance. By contrast, liver is the primary site of glucose clearance in rodents, with 5 to 10-fold higher glycogen storage in the liver than in muscle, versus 10-fold more glycogen storage in muscle than in liver in humans. This inter-species difference has functional implications since various aspects of glucose regulation differ between human skeletal muscle and rodent liver 18.

Glucose homeostasis relies on multiple glucose-sensing cells in the body that monitor blood glucose levels and respond to adjust its glycemia 28. Hepatocytes express the enzyme glucokinase that acts as a glucose sensor, and can switch between gluconeogenesis and glycolysis depending on glucose availability.

Every animal species has a signature blood glucose level or glycemic set point, likely reflecting their evolutionary adaptation. Normoglycemia of one species would be life threatening for another. For example, mouse normoglycemia can be considered diabetic for humans 29.

As a consequence of reliance on animal research for insights on glucose regulation, several erroneous concepts were established. For example, in contrast to rodent islets, glucagon input from the alpha cell to the insulin-secreting beta cell is necessary to fine-tune the distinctive human set point.

This and many other inter-species differences have major implications for safety of patients and success of therapies to treat diabetes. For instance, treatments that inhibit glucagon pathways might also eliminate their crucial input to beta cells in humans with diabetes.

​

Species-specific immunological and genetic differences in T1DM

Given the crucial role of autoimmunity in T1DM, species-specific features of innate and adaptive immune systems are likely to play a part in poor predictive validity of animal models of diabetes.

For example, treatment of T1DM by antithymocyte globulin, that targets multiple T cell antigens, had shown promising results in animal models of T1DM. However, after receiving the same treatment in clinical trials, patients with T1DM had suffered cytokine release syndrome accompanied by a depletion of regulatory T (Treg) cells 30.

The lack of understanding of the precise role of individual components of the immune system in T1DM, combined with species-specific characteristics of the immune system, makes it difficult to know to which extent immune-mediated responses in animal models of T1DM match that of T1DM patients.

In T1DM, autoreactive CD4+ and CD8+ T cells destroy beta cells, driven by autoantigens like GAD65 enzymes and insulin that are presented by MHC class II on the surface of antigen-presenting cells.

In humans, the proinflammatory cytokines IFN-α, IFN-β, IFN-γ, and IFN-λ, secreted by T cells, macrophage, NK cells and other immune cells, play a major role in mounting of innate and adaptive immune responses to islet antigens 31.

While in humans these cytokines activate the transcription factor STAT4, that plays a pivotal role in insulitis, beta cells apoptosis, and cytotoxicity, the activation of STAT4 is much more restricted in mice 32, leading to underestimation of its role in T1DM mouse models.

The gold standard non-obese diabetic (NOD) mouse model of T1DM mimics the autoimmune beta cell destruction and STAT4 responses characteristic of human T1DM. Nonetheless, the immune response in NOD mice diverges from that in human T1DM. For example, NOD mice exhibit a stronger Th1 response, leading to direct cytotoxic CD8+ T-cell-mediated beta-cell destruction, whereas in humans, Th17 plays a more significant role, contributing to chronic islet inflammation and beta cell dysfunction, rather than immediate cytotoxicity. The fact that T1DM mouse models is likely underrepresent Th17-mediated mechanisms additionally limits their translational relevance.

In addition, insulitis - the inflammation of the islets of Langerhans - shows both qualitative and quantitative species-specific differences between humans with T1DM and animal models of T1DM. In humans, insulitis is less intense and is found in less than 10% of islets. In contrast, the NOD mouse model of T1DM exhibit a more aggressive form of insulitis that touches over 50% of pancreatic islets. As a result, studies in NOD mice are likely to underestimate the role of Treg cells and environmental modulators that have the potential to slow down disease progression in humans.

As part of innate immunity, complement system enhances inflammation, opsonization, and cell lysis. Human studies showed elevated complement components in pancreatic islets of T1DM patients with insulitis. In humans, the complement system is tightly regulated by inhibitors, such as complement factor H (CFH) and factor-H related proteins (CFHR), and their dysregulation can contribute to T1DM 33. However, as genomic studies have revealed, mice lack certain CFHR genes present in humans, such as CFHR-C 34.

Although the impact of species-specific disparities in CFH is currently unknown, they have the potential to produce inter-species divergencies in complement system regulation, immune response, and severity of insulitis.

The anti-inflammatory cytokine IL-10 suppresses immune responses and promotes Treg function, thereby mitigating beta cell destruction. While in humans, IL-10 is produced both during Th1 and Th2 responses, in mice IL-10 is much less prominent in Th1 than in Th2 35.

This inter-species difference in IL-10 production has significant implications for T1DM mouse models, including for the Th1-biased NOD mouse, in which limited Il-10 production leads to a more aggressive disease phenotype than in humans, under-represented immune regulatory mechanisms compared to humans, and under-estimated human therapeutic responses.

In NOD mice, the binding pocket of the MHC class II allele H2-Ag7 protein, responsible for binding peptides and presenting them on the surface of antigen-presenting cells to T cells, is similar to the binding pocket of the analogous human HLA-DQ. Yet, NOD mice do not express the class II MHC molecule I-E, meaning that they rely solely on I-Ag7 which has a unique peptide binding groove.

The human class II MHC alleles associated with T1DM have a greater degree of polymorphism than the NOD mouse MHC genes, which allows for a more diverse range of antigen presentation 36. Unlike in NOD mice, the HLA complex in humans includes several loci that independently confer risk of T1DM 37.

The efficacy of antigen-specific treatments might differ depending on the MHC alleles involved and the way in which peptides bind in the MHC groove. For example, a single amino acid difference in the B chain of human insulin seems to affect the ability of porcine insulin, but not human insulin, to prevent development of T1DM in NOD mice.

Cluster of differentiation 28 (CD28) is involved in T1DM autoimmune response as a co-stimulatory receptor on T cells that binds to CD on antigen-presenting cells and thereby provides signals for T-cell activation. In T1DM patients, the expression of CD28 infiltrating T cells is elevated.

In mice, CD28 is expressed on all CD4+ and CD8+ T cells, whereas in humans it is expressed on all CD4+ but only about 50% of CD8+ T cells 38.

Because of species-specific differences in types of CD and surface expression of CD types, the effects on human patients of drugs that target CD are likely to be underestimated or overestimated.

For example, the CD28 antagonist Abatacept showed efficacy in preclinical studies, but not in T1DM patients 39.

In the same manner, the CD28 agonist Theralizumab, also known as TGN1412, caused catastrophic cytokine storm with systemic organ failures in its first human trial, despite being approved as safe in mice and in primates 40.

Administration of low-dose IL-2, alone or combined with rapamycin, prevented hyperglycemia in NOD mice. Additionally, low-dose IL-2 was reported to cure recent-onset T1DM in NOD mice 41. However in humans, the same treatment had significant detrimental effects on beta cell function and survival 42. 

Natural killer (NK) cells of NOD mice exhibit a defect in functional maturation and have impaired cytotoxic functions, which may render these mice resistant to the toxic effects of IL-2 treatment, leading to failure to predict toxicity to humans.

While a single dose of expanded antigen-specific Tregs  could reverse the T1DM phenotype in NOD mice, the attempts to reproduce the same results in T1DM patients have repeatedly failed 43.

As the above examples show, the faith in the ability of animal models of T1DM to recapitulate human insulitis pathology has deleterious repercussions for understanding underlying disease mechanisms and for developing safe and effective therapies for T1DM. Decades of massive failure in clinical trials of hundreds of immune interventions approved in animal models of T1DM have prompted numerous calls for a human-centred approach to research 44, 45.

It was suggested that using animal models with a humanized immune system might improve translatability to humans, however, such an approach would face persistent, insurmountable challenges: the role of the human immune system in T1DM is complex and not fully understood, the equivalence of humanized animals to the human immune system was never demonstrated by objective measures 46, and the cross-talk between the human immune system and the rest of human organ systems cannot be recapitulated in animals.

Additionally, another aspect that cannot be recapitulated in humanized animal models, is the development and adaptation of the immune system in response to the environment, including the microbiome, diet and housing. This missing aspect in animal models is crucial, since as twin studies show, the human immune system is more shaped by the environment than by genes 47.

Species-specific differences in tropism for T1DM-associated viruses

Several viruses were associated with the development of T1DM, particularly through their potential to trigger autoimmune responses that damage insulin-producing beta cells 48, 49, 50.

Species-specific genetic variations in the host's cell surface receptors and the immune system have important implications for species’ and individuals' responses to viral infections.

For example, humans possess human-specific single nucleotide polymorphisms (SNPs) in IFIH1 gene, that are absent in non-human Ifih1 gene, and that confer human-specific vulnerability (gain-of-function variants) or protection (loss-of-function variants) to T1DM.

The IFIH1 gene encodes the melanoma differentiation-associated protein 5 (MDA5), a cytosolic pattern recognition receptor expressed in immune cells and in pancreatic beta cells, that can detect dsRNA from viral infections and trigger an innate immune response.

Human-based genetic and population studies have shown that increased activity of IFIH1/MDA5 in genetically susceptible individuals can promote the development of T1DM, by increasing type I interferon production, beta cells apoptosis, and innate-adaptive immune system crosstalk that subsequently amplifies autoimmune responses 51, 52.

None of the animal species used in T1DM animal research, including NOD/STZ/BB mice and non-human primates (NHP), possess the human-specific SNPs in IFIH1 gene, hindering the study of MDA5-mediated autoimmunity and development of MDA5-targeting therapies 53.

This also raises the question of whether NHP, that do not naturally develop T1DM, possess other protective SNPs in orthologues of human genes involved in T1DM and that could jeopardize safe and effective translation of preclinical studies to T1DM patients.

Moreover, in different animal species employed to model T1DM, different viruses may show varying degrees of tropism for pancreatic cells, leading to differences in disease outcomes that hamper translatability to humans 54.

Species-specific differences in gut microbiome

The immune system homeostasis is modulated by the gut microbiome which varies significantly across species, making findings from microbiological studies in animals not directly translatable to humans.

In diabetes, inflammatory responses can be triggered by an imbalance between beneficial and pathogenic bacteria (dysbiosis), by an increased intestinal permeability to bacterial endotoxins, by bacterial stimulation of pro-inflammatory cytokines, or by elevated levels in the blood of gram-negative bacteria cell walls-derived lipopolysaccharides. 

In T1DM patients, it has been reported that the proinflammatory environment in the gut is associated with the combination of an increased relative abundance of Bacteroidetes and a decreased level of Firmicutes, regardless of geographical location 55.

Studies from human cohorts suggest that dysbiosis contributes to T1DM primarily by promoting Th17/Th1 autoimmunity, gut barrier permeability, and beta cells destruction 56, 57.

Dysbiosis of the gut microbiota is suggested to occur early in life, aggravating gut inflammation and influencing the immune system, before the onset of T1DM. The relationship between T1DM autoimmunity and gut microbiome dysbiosis is likely to be bidirectional and needs to be further investigated using human-based metagenomics and metaproteomics 58.

The study of the microbiome has historically relied on mouse studies, based on coarse comparisons at higher taxonomy hierarchy level with a poorly characterized mouse gastrointestinal (GI) microbiota. However, advances in metagenomics that have enabled a much more granular view of mouse microbiota at lower taxonomy level, have shed light on previously unknown profound species-specific differences in GI microbiome composition, quantity, and function.

A comparative survey of the phylogenetic composition of 16 human subjects and 3 commonly used mouse lines indicated that their microbiota show similarity at the genus level but is quantitatively very different 59.

Comparison of comprehensive mouse microbiota genome to microbiota from the human GI genomes shows an overlap of 62% at the genus but only 10% at the species level, demonstrating that human and mouse gut microbiota are largely distinct 60.

What is more, mouse and human microbial strains of the same species can have divergent gene content and function, as is exemplified by the species Limosilactobacillus reuteri which has mouse-adapted and human-adapted strains, however, with very different functions.

Previously, DNA sequencing of fecal samples of mouse strains of diverse genetic, provider, housing and diet backgrounds and comparison to the human microbial GI genome demonstrated that only 4% of mouse GI microbial genes were shared with human GI microbial genes 61.

Rats tend to have a higher abundance of the phylum Bacteroidetes compared to mice. Mice, on the other hand, have a higher abundance of the family Muribaculaceae within the phylum Bacteroidetes 62.

Even within closely related species, like striped hamsters and Djungarian hamsters, there can be differences in bacterial families and genera that reflect their distinct lifestyles and dietary habits 63, 64.

Even though the gut microbiota in humans was closer to that in NHP, significant differences remain between the two species. For example, African green monkeys show opposite microbial taxa and genes responses to a typical high-protein, high-fat, low-fiber Western diet compared to humans, suggesting that NHP are not an appropriate model organism for studying the effect of the human gut microbiota on host metabolism and human diseases 65.

Efforts have been made to humanize mice with human microbiota and to inter-cross inbred strains. However, it was found that the GI microbiome in mice varied immensely depending on the housing conditions, even more so than on the genetic background, and that the inter-individual variation in GI microbiome decreased after 10 generations 59.

The so called human microbiota-associated (HMA) mouse models, that are designed to study the contribution of a dysbiotic microbiome to development of human diseases by comparing disease phenotypes of germ-free mice colonized with the fecal microbiota of patients to those of mice colonized with the microbiota of healthy controls, do not replicate the patient microbiome 66. Indeed, engrafted HMA mice showed only a partial resemblance to the donor microbiota. It is believed that only the microbiome phylotypes that are adapted to the mouse intestinal environment, and that survive an ecological shift after engraftment, will remain in the mouse gut.

Human-based research has shown that the gut microbiome plays a crucial role in shaping both the innate and adaptive immune system from infancy through adulthood through human host-specific interactions between human host-specific immune system and human host-specific microbiota.

Therefore it is not surprising that HMA mice have a low immune maturation with reduced numbers of adaptive and innate intestinal immune cells when compared with mice that harbor a murine microbiota.

This indicates that the engrafted human microbiota does not accurately mirror the microbial composition found in patients, nor does it capture the dysbiosis and immune responses associated with human diseases.