Altered β-Cell Prohormone Processing and Secretion in Type 1 Diabetes Free

The pancreatic β-cell integrates humoral, metabolic, neural, and paracrine inputs to regulate the production, release, and processing of the hormones insulin and islet amyloid polypeptide (IAPP) (1). In type 1 diabetes, immune-mediated death and dysfunction of pancreatic β-cells lead to reduced circulating levels of insulin and IAPP. The physiological role of insulin as a critical regulator of carbohydrate metabolism is well established. Whereas the biological function of IAPP is not fully understood, studies suggest that it may act centrally to suppress food intake and gastric emptying; and within the islet, IAPP may suppress glucagon secretion (2). Both insulin and IAPP are synthesized as prohormones that undergo sequential processing in the secretory pathway within the β-cell (1,3) (Fig. 1). Multiple studies have identified evidence of increased proinsulin or proIAPP relative to mature insulin or IAPP expression, either in circulation or in the islet, at various stages in the natural history of type 1 diabetes (4–8). For years, conventional dogma suggested that by the time of clinical onset, a large majority of β-cell mass (i.e., 85–90%) has been destroyed, with ultimate progression to complete loss of insulin and IAPP. However, large cohort studies have shown that many individuals continue to secrete low levels of C-peptide for years after type 1 diabetes diagnosis (9). Recent studies have also highlighted persistent secretion of incompletely processed forms of proinsulin and proIAPP in patients with long-duration type 1 diabetes, even in cohorts characterized by the absence of detectable serum C-peptide, suggesting the continued presence of proinsulin-producing cells that do not effectively process prohormones (6,8,10). Analyses of pancreata from organ donors with type 1 diabetes have extended these findings to the tissue level (4,8).

Mechanistic studies have begun to link β-cell stress, occurring either from extrinsic sources or due to activation of intrinsic molecular pathways within the β-cell, with disruptions in normal prohormone processing (1,11) (Fig. 2). Moreover, T cell autoreactivity has been demonstrated against regions of the incompletely processed forms of proinsulin and proIAPP, suggesting a role for altered prohormone processing in immune activation (12,13). In response to these emerging findings, levels of circulating prohormones relative to mature hormones are actively being studied for their potential as noninvasive indicators of type 1 diabetes risk, as well as biomarkers of therapeutic efficacy, in efforts designed for disease prevention or intervention (14).

A substantial body of work suggests that elevations in relative levels of circulating prohormones can provide insights into β-cell stress at different stages during the evolution of type 1 diabetes. Most studies in individuals at risk for or with a diagnosis of type 1 diabetes have shown elevations in ratios of proinsulin or proIAPP relative to mature insulin/C-peptide or IAPP rather than absolute elevations in circulating prohormones in comparison with control samples (6–8,15,16).

Analyses from natural history cohorts of islet autoantibody-positive (Aab⁺) individuals without diabetes who ultimately develop type 1 diabetes have shown that these individuals exhibit elevated proinsulin–to–C-peptide (PI/C) ratios in comparison with Aab⁻ control subjects. These efforts also suggest that fasting PI/C ratios are inversely associated with first-phase insulin responses during intravenous glucose tolerance tests or hyperglycemic clamps (15,16). Among Aab⁺ relatives, elevated PI/C ratios are also associated with increased risk of progression to type 1 diabetes and can complement the use of Aab for risk prediction, with longitudinal ratios inversely related to time to diagnosis (7,15,17,18). PI/C ratios appear to be most elevated in younger at-risk children (i.e., age <10 years) who progress to type 1 diabetes, suggesting that this subgroup of individuals may exhibit more severe β-cell stress during disease progression (7).

Not surprisingly, comparisons with Aab⁻ nonrelative control subjects show elevations of PI/C in individuals with new or recent-onset type 1 diabetes, especially younger children, consistent with high levels of β-cell stress at diagnosis (19,20). However, other efforts have shown differing results regarding whether improvements versus maintained elevations in PI/C occur after exogenous insulin therapy is started, during the clinical remission often referred to as the honeymoon period (19–21). This could be related to varied participant age at the time of diagnosis (youth vs. adult), as well as the degree and/or duration of improved β-cell function over this period (19–21). A recent study testing treatment of children and young adults with recent-onset type 1 diabetes with an IgG1-κ monoclonal antibody specific for human TNF-α showed stable PI/C ratios in drug-treated participants compared with progressive increases over time in placebo (22). Intriguingly, higher PI/C ratio at the time of type 1 diabetes diagnosis was noted in studies decades ago showing an improved response to cyclosporin treatment, suggesting that high PI/C ratios, as a proxy or readout of β-cell stress, could be leveraged to identify individuals who might benefit from a disease-modifying intervention (19). Such findings are consistent with data from other immunomodulatory interventions identifying individuals with active disease and β-cell dysfunction as robust treatment responders (23). Ideally, such biomarkers could also be leveraged to help in selection and optimization of specific interventions.

Very elevated PI/C ratios, suggestive of severe β-cell stress, are consistently present in individuals with long-standing type 1 diabetes and residual detectable C-peptide decades after diagnosis (8,10). Ratios tend to be higher in those diagnosed at very young ages (i.e., <7 years), again suggesting a more severe disease phenotype in younger children (8). Even in those individuals with long-standing type 1 diabetes and undetectable stimulated C-peptide, circulating proinsulin can be identified. The reported percentage of individuals with undetectable C-peptide but persistent proinsulin secretion has varied from 16% to up to 90%, depending on the characteristics of the assays used and cohorts studied (8,10,24,25). In addition, many individuals with undetectable or minimal residual C-peptide, but detectable proinsulin, do not exhibit the expected increase in stimulated circulating proinsulin that is observed in control subjects or those with type 1 diabetes exhibiting higher residual C-peptide (10). While measurements of the levels of the IAPP precursor, proIAPP, relative to mature IAPP have not been performed across the entire natural history of type 1 diabetes, children with long-standing type 1 diabetes and islet transplant recipients with type 1 diabetes exhibit elevated ratios of a proIAPP processing intermediate (proIAPP_1–48) relative to total IAPP (6). Interestingly, unlike PI/C, ratios of proIAPP_1–48 to mature IAPP do not appear to be elevated in individuals with type 2 diabetes (6,26).

Several older studies in relatives of individuals with type 1 diabetes testing negative for islet autoimmunity have described elevations in circulating absolute and normalized proinsulin in comparison with unrelated control subjects, suggesting that some individuals may inherit a predisposition to abnormal prohormone processing that predates detectable autoimmunity. Islet cell cytoplasmic antibody (ICA)⁺ and ICA⁻ individuals without diabetes with an identical twin affected by type 1 diabetes showed an absolute increase in fasting proinsulin in comparison with unrelated control subjects (27). Despite normal insulin and glucose values, testing in groups of Swedish and Swiss first-degree relatives of patients with type 1 diabetes who were ICA⁺ and ICA⁻ also showed HLA-independent increases in fasting absolute proinsulin values in comparison with control subjects, although ICA⁻ relatives showed less pronounced elevations (28,29). An important caveat to these results is the use of older, less sensitive assays for assessment of islet autoimmunity. In contrast, more recent testing in first-degree relatives from the Belgian Diabetes Registry demonstrated similar random PI/C ratio values among Aab⁺ relatives that did not progress to diabetes during follow-up, Aab⁻ relatives, and a small group of nonrelative Aab⁻ control subjects (17).

Increased PI/C ratios in the circulation months to years before clinical diagnosis (7,15,17,18) point to a potential functional β-cell defect appearing very early in the type 1 diabetes disease process. However, until recently, whether this relative increase in circulating proinsulin could be linked to abnormal proinsulin expression in the pancreas remained largely unknown. Recent access to human pancreas samples through the Network for Pancreatic Organ donors with Diabetes (nPOD) and the Exeter Archival Diabetes Biobank (EADB) coupled with improvements in image analysis methodologies has provided a unique opportunity to study these phenomena in samples from control subjects without diabetes (Fig. 3) as well as across multiple phases of type 1 diabetes (4,5,8) (Figs. 4 and 5). Here, analysis has shown that islets from Aab⁺ individuals exhibit an increase in absolute pancreatic proinsulin area and in the proinsulin-to-insulin ratio without a significant reduction in insulin area or β-cell mass (5) (Fig. 4A). Moreover, islets from Aab⁺ donors show changes in the subcellular localization of proinsulin, which was predominantly localized in secretory vesicles in Aab⁺ individuals with multiple Aab, as opposed to the normal juxtanuclear localization. Taken together, these findings suggest that proinsulin maturation might already be defective in the prediabetic pancreas, independently of the loss of β-cell mass. The same analysis reported elevations in the proinsulin-to-insulin ratio in pancreatic samples from living individuals with recent-onset type 1 diabetes collected through the Diabetes Virus Detection Study (DiViD) (5). Continued elevation of ratios despite exogenous insulin therapy suggests that β-cells remain dysfunctional at the time of diagnosis, even when their workload is diminished.

Further studies demonstrated that residual β-cells containing insulin and proinsulin could be found in individuals with short and long duration of type 1 diabetes (4). Islets from donors with type 1 diabetes exhibit overall reductions in insulin, proinsulin, and C-peptide in pancreatic protein extracts in comparison with nondiabetic control subjects. However, the levels of proinsulin were higher than expected and comparable with those of Aab⁺ individuals. Both the proinsulin-to-insulin and PI/C ratios were significantly elevated in individuals with type 1 diabetes. Furthermore, insulin mRNA was low but detectable in almost all subjects with type 1 diabetes (4). In agreement with these studies, populations of proinsulin-enriched, insulin-depleted cells have been described in a subset of individuals with long-standing type 1 diabetes (30) (Fig. 5A). This depletion of insulin-positive granules might suggest β-cell exhaustion (degranulation) with (or perhaps even without) a dramatic reduction in β-cell mass.

Recent work has focused on the influence of age at diagnosis on the expression and distribution of insulin and proinsulin in the pancreas (8). Analysis of samples from the EADB and the nPOD collections showed that in children diagnosed early in life (before 7 years), most β-cells exhibited a high degree of insulin and proinsulin colocalization (8) (example in Fig. 4B). This population also displayed a hyperimmune profile characterized by abundant islet-infiltrating CD20⁺ and CD8⁺ cells. The authors proposed that these features represent one particular “endotype” coined as type 1 diabetes endotype 1 (T1DE1) (6). By contrast, in samples from individuals diagnosed at or after the age of 13 years, >70% of islets displayed little colocalization and had a lower proportion of CD20⁺ cells in inflamed islets. It was also noted that, irrespective of age at diagnosis, for the islets of people in whom β-cells persisted for >5 years after diagnosis there were low rates of proinsulin-to-insulin colocalization. Such islets were most prevalent in subjects diagnosed at age ≥13 years. Taken together with work described above, these findings suggest that there is a strong interplay among abnormal proinsulin processing, β-cell function, and immune activity in children diagnosed very early in life. Continued study of pancreata from children and adults at multiple stages of disease should help to confirm whether disease endotypes exist (rather than a continuously variable set of pancreatic phenotypes) and may suggest routes to personalized therapeutic intervention.

The studies described above demonstrate that proinsulin is elevated in islets before and after onset of type 1 diabetes and that relative circulating levels of proinsulin remain increased in individuals with long-standing disease. From these findings, it seems very likely that the efficiency of prohormone processing is unable to keep up with prohormone production during the course of type 1 diabetes. Potential etiologies of these findings are further discussed in Table 1. A likely contributing etiology is increased insulin secretory demand or rapid insulin release in association with reduced β-cell mass (31), resulting in mature secretory granule depletion and resultant depletion of processing enzymes. It is also possible that an intrinsic prohormone processing defect may arise during the course of the disease. Expression patterns of proinsulin and proIAPP processing enzymes (prohormone convertase 1/3 [PC1/3], prohormone convertase 2 [PC2], and carboxypeptidase E [CPE]) have been investigated in islets from individuals with short and long disease duration. One study found significant reductions in PC1/3 mRNA but not in PC2 or CPE (4). Similarly, in another study, with use of laser-capture microdissection and mass spectrometry (MS), significant reductions were reported in PC1/3 and a tendency to lower CPE but no difference in PC2 (30).

Mechanistic studies testing human islet prohormone processing in the context of type 1 diabetes are fairly limited in number. Cytokine treatment for 24 h of isolated islets from donors without diabetes was associated with a significant reduction in mRNA expression levels of PC1/3, CPE, and PC2 (30). Longer (48–72 h) in vitro exposure of human islets to cytokine combinations resulted in reduced protein concentrations of PC1/3 and PC2 in association with an increase in medium proinsulin-to-insulin ratio, supporting the idea that islet inflammatory stress might be associated with intrinsic reductions in prohormone processing (32). Accumulation of misprocessed proteins can have dramatic effects on β-cell survival and mRNA translation through activation of cell-intrinsic stress pathways. Indeed, β-cells exposed to proinflammatory cytokines are prone to endoplasmic reticulum (ER) stress and apoptosis (33), while islets from individuals with type 1 diabetes have increased levels of the ER stress markers BIP and CHOP, the latter of which has been associated with β-cell death (34). Defects in protein processing could theoretically contribute to β-cell death in type 1 diabetes, perhaps through exacerbation of increased secretory demands or in ways that are analogous to some monogenic forms of diabetes that occur due to proinsulin misfolding and ER stress (35,36). Misfolded proinsulin has not been identified in secretory granules; however, β-cells under conditions of ER stress independent of cytokines could also potentially contribute unprocessed proinsulin and/or proIAPP to the circulation via “leakage” or cell death. Aberrant processing of proIAPP to mature IAPP has been hypothesized to contribute to IAPP aggregation and amyloid formation (37), which has recently been detected in pancreata of some donors with long-standing type 1 diabetes (38) (Fig. 5B). As IAPP aggregates are a trigger for inflammation and β-cell stress (6,37), it is plausible that impaired handling of proIAPP or IAPP could contribute to β-cell dysfunction and loss in type 1 diabetes, as is suggested in the case of type 2 diabetes (38).

In addition, accumulating evidence indicates a link between prohormone processing and T cell–mediated autoreactivity (12). T cell repertoire studies have identified autoreactivity against multiple epitopes of preproinsulin, proinsulin, and proIAPP, while analysis of the β-cell HLA-I peptidome has confirmed that many of these epitopes can be presented by the β-cell (12,39,40). Specifically, in the case of insulin, subcellular targeting and protein modifications impact processing pathways, as well as efficiency and regulation of antigen presentation (41). Errors in processing can lead to posttranslational modifications of the native protein; posttranslationally modified insulin-derived epitopes have improved HLA-I binding capacity and induce T cell activation (42). Moreover, the physiologic release of granule contents directly into the bloodstream may allow for an opportunity to interact with distant antigen-presenting cells outside of the pancreas (12). Importantly, islet autoantibodies are absent at diagnosis of inherited forms of diabetes related to protein misfolding and ER stress, pointing away from widespread activation of islet autoimmunity under conditions of β-cell stress (35,36). However, it is possible that β-cell stress is linked to increased antigen presentation and T cell–mediated autoreactivity in individuals with a genetic predisposition to autoimmunity.

The existing body of work quantifying circulating and tissue levels of proinsulin and proIAPP has provided convincing data that during the natural history of type 1 diabetes, prohormone processing cannot keep pace with prohormone production. PI/C ratios are increased in many at-risk individuals before, at, and after diagnosis and are associated with disease progression, particularly in young children. Even many years after diagnosis, persistent secretion of proinsulin and proIAPP has been documented, reflecting the existence of severely dysfunctional β-cells in long-standing disease. However, several lingering questions regarding the natural history and physiology of circulating prohormones remain. These will be important to answer moving forward in order to implement use of circulating prohormones as biomarkers to guide type 1 diabetes prediction, progression, and treatment. Similarly, despite the increasing number of studies in islet tissue on absolute or relative prohormone accumulation and processing, there are still conceptual gaps in our understanding of prohormone conversion dynamics and storage, both in “normal” β-cells of humans without diabetes and in cells influenced by different types of stressors.

Studies often show significant elevations in PI/C ratios in comparisons with control subjects, but in many cases, values overlap with those of control subjects. This might be related to several factors, including heterogenous phenotypes or endotypes of type 1 diabetes, which do not uniformly involve high levels of intrinsic β-cell stress (43). Overlap with values of control subjects could also reflect disease heterogeneity due to physiologic differences in “normal” PI/C ratios during different life stages, especially for pediatric age-groups advancing through puberty, or groups of adults compared with children. Both adults and pubertal children can exhibit increases in insulin resistance relative to prepubertal children, and knowing the expected “normal” range for PI/C ratios and proIAPP-to-IAPP ratios more definitively during growth and aging will be important for understanding what constitutes an “abnormal” value (44).

Pragmatically, another important hurdle in the implementation of prohormone measurements as biomarkers for use in prediction and intervention efforts will be development of a better definition of the expected natural history of increasing prohormones/mature hormones as type 1 diabetes develops. Longitudinal data are needed in adequately powered studies of diverse at-risk populations that allow for improved characterization of patterns of change over time. While proIAPP has been measured in established type 1 diabetes, it is not clear how patterns of secretion change during earlier disease stages or whether there is concordance between proIAPP and proinsulin secretion at different disease stages. Such analyses will allow for a better understanding of whether certain groups of individuals are at risk for abnormal values, either at baseline or in association with environmental stressors. Along these lines, older data in ICA⁻ first-degree relatives are intriguing but must be followed up using newer biochemical Aab assays. Additionally, ratios may increase progressively over time or could change in a relapsing-remitting pattern, in association with periods of active autoimmune disease or intermittent environmental stressors. Such factors have important implications for the timing of screening and prediction approaches—but will also be informative in tracking of the disease course in individuals. In addition, samples from successful intervention studies should be tested to identify whether circulating prohormones can be used as an early marker of successful treatment response. Such analyses would also serve to identify whether circulating prohormones could be used to identify reversible β-cell stress and predict treatment “responders” to immunomodulation.

Several important unanswered questions revolve around the pathophysiologic relevance of increases in circulating prohormones in type 1 diabetes (outlined in Table 1). First, given that multiple pathways could theoretically increase the relative proportion of β-cell prohormone expression and secretion, what are the specific pathologic etiologies leading to these outcomes in type 1 diabetes? Exploration of this question will lead to a better understanding of whether defects in prohormone processing can serve as a readout of particular immune phenotypes and/or disease endotypes (8).

Work from several investigators suggests that the immune system could react differently to β-cells under situations in which abnormal prohormone expression occurs (reviewed in 12). Alternatively, increased susceptibility to β-cell stress could exacerbate β-cell failure and death once autoimmunity is established. In contrast, recent preclinical work suggests a model wherein partially dedifferentiated β-cells may be selected for their ability to evade autoimmunity. Along these lines, loss of processing enzyme expression and the reduced ability to complete prohormone processing could be viewed as part of a dedifferentiated β-cell phenotype (45,46). How these potentially conflicting paradigms contribute to type 1 pathophysiology is not clear and should be a focus of future studies.

Although both intact/unprocessed proinsulin and partially processed des-31,32 proinsulin forms are elevated in type 2 diabetes (47), the relative proportion of intermediate and intact forms of proinsulin and proIAPP is still unclear in type 1 diabetes. Separate quantification of both intact and partially processed prohormone forms in the circulation and pancreas from individuals with type 1 and type 2 diabetes, and of relationships between increased relative proportions of proinsulin and proIAPP, may clarify the pathophysiologic etiologies of altered ratios.

Studies testing both pancreas tissue and serum from the same individual would be optimal to address these questions, but such samples are challenging to obtain from living individuals. However, recent advances have provided the opportunity to study live pancreatic tissue from organ donors and therefore correlate anatomical and physiological data. Slices of whole pancreas tissue with preserved endocrine, exocrine, and immune tissue compartments can be obtained from donors with or without type 1 diabetes (48). With use of these, it may be feasible to measure basal and stimulated prohormone secretion as well as to evaluate enzymatic prohormone processing in both nondiabetic and diabetic tissue environments. A preliminary study in pancreas slices from a donor with type 1 diabetes has confirmed that, in some cases, significant β-cell mass is preserved at disease onset and that insulin deficiency can result from β-cell dysfunction (5,48). These findings corroborate data generated from islets, isolated from living donors with recent-onset type 1 diabetes in DiViD, that were dysfunctional initially but recovered glucose responsiveness during culture. Expanding these studies will be key to defining the role of β-cell dysfunction in type 1 diabetes.

Strategies and challenges in optimizing prohormone measurement are outlined in Table 2. Of note, the use of highly specific and reliable assays for multiple forms of prohormone (e.g., intact, des-64,65, and des-31,32 proinsulins) will be critical to gaining a more detailed understanding of prohormone processing in type 1 diabetes. Immunoassays have been developed for detecting intact proinsulin (49), total proinsulin (which includes intact + partially processed proinsulin forms) (50,51), C-peptide, insulin, and proIAPP_1–48 (6). Unfortunately, the performance characteristics of prohormone immunoassays, which are limited by the inherent exclusive dependence on reproducible antibody reagents, are often not well documented (52). For proinsulin, obtaining optimal specificity can be particularly challenging due to the structural similarities among proinsulin, its partially processed forms, insulin, and C-peptide and the low concentration of proinsulin in comparison with insulin/C-peptide (52). Specifically, proinsulin and mature insulin or C-peptide immunoassays often cross-react with intact proinsulin as well as a combination of proinsulin split products, resulting in a degree of inaccuracy in quantification of proinsulin or insulin levels from clinical data. Although to date there are no widely accepted guidelines or standardized approaches, careful independent validation is needed in confirming consistency of findings (53,54).

A promising alternative means to develop highly specific assays for prohormone products is the use of targeted MS. These techniques (55,56) provide direct measurements of surrogate peptides from any given protein with well-characterizable specificity and multiplexing capacity but without the need for affinity reagents. In principle, it is highly feasible to develop targeted MS assays for specific forms of prohormone processing products in serum, tissue, and secretome (10); however, the current limitations of targeted MS assays lie in the achievable sensitivity and analytical throughput. Further developments are needed to achieve the sensitivity required for such detection. The rigor and reproducibility of immunoassays could be substantially improved if the specificities and performance are validated and standardized with use of antibody-free targeted MS-based assays along with improved calibration standards.

In summary, analysis of pancreatic tissue and data from clinical cohorts indicate that prohormone processing insufficiency is an early and persistent component of type 1 diabetes pathogenesis. The body of work outlined herein has identified a number of exploratory questions for the type 1 diabetes research community and suggests several key next steps that should be taken to clarify the role of altered prohormone expression and secretion in type 1 diabetes (Fig. 6). These questions largely focus around continued and longitudinal testing of relevant clinical populations, studies that probe the genetics of impaired prohormone processing in type 1 diabetes, combined functional and histologic testing of human pancreas tissues, and continued interrogation of how prohormone processing intersects with immune activation and autoreactivity. A key tenet of this work is the need for reliable and well-validated assays that can detect and distinguish intact prohormones and processing intermediates. Successful resolution of these questions may offer the potential to use altered prohormone processing as an anchor that can subsequently be used to cluster different endotypes of disease enveloping a large set of variables (e.g., age, genetic background, severity of insulitis and its composition, and rate of progression). Ultimately, such an approach could inform clinical therapeutic strategies aimed at intervening in the natural history of type 1 diabetes.

Acknowledgments. Schematic figures were created with BioRender (BioRender.com). The authors apologize to investigators whose work they were unable to include in this article due to space limitations. The authors also acknowledge the support of the nPOD Slice Prohormone Processing Interest Group, which includes the coauthors as well as Clive Wasserfall, Irina Petrache, Mingder Yang, Harry Nick, Gladys Teitelman, Adam Swenson, and Sirlene Cechin.

Funding. This effort was supported by nPOD (Research Resource Identifier: SCR_014641), a collaborative type 1 diabetes research project sponsored by JDRF (grant 5-SRA-2018-557-Q-R) and The Leona M. & Harry B. Helmsley Charitable Trust (2018PG-T1D053). T.R.-C. receives funding from JDRF (5-CDA-2020-949-A-N), E.K.S. receives funding from the National Institute of Diabetes and Digestive and Kidney Diseases (NIDDK) (R03 DK117253 01A1S1 and R01DK121929) and JDRF (grant 2-SRA-2017-498-M-B), P.A. receives funding from NIDDK (R01DK48280), and C.B.V. receives funding from JDRF (1-INO-2019-794-S-B) and Canadian Institutes of Health Research (PJT-153156).

Duality of Interest. No potential conflicts of interest relevant to this article were reported.