Spontaneous restoration of functional β-cell mass in obese SM/J mice

Maintenance of functional β-cell mass is critical to preventing diabetes, but the physiological mechanisms that cause β-cell populations to thrive or fail in the context of obesity are unknown. High fat-fed SM/J mice spontaneously transition from hyperglycemic-obese to normoglycemic-obese with age, providing a unique opportunity to study β-cell adaptation. Here, we characterize insulin homeostasis, islet morphology, and β-cell function during SM/J’s diabetic remission. As they resolve hyperglycemia, obese SM/J mice dramatically increase circulating and pancreatic insulin levels while improving insulin sensitivity. Immunostaining of pancreatic sections reveals that obese SM/J mice selectively increase β-cell mass but not α-cell mass. Obese SM/J mice do not show elevated β-cell mitotic index, but rather elevated α-cell mitotic index. Functional assessment of isolated islets reveals that obese SM/J mice increase glucose stimulated insulin secretion, decrease basal insulin secretion, and increase islet insulin content. These results establish that β-cell mass expansion and improved β-cell function underlie the resolution of hyperglycemia, indicating that obese SM/J mice are a valuable tool for exploring how functional β-cell mass can be recovered in the context of obesity.


88
Like in humans, diabetic risk in obese mice depends on genetic background (44,48,80). Variation in β-cell 89 heterogeneity likely underlies variability in islet stress response, and thus needs to be accounted for when comparing 90 nondiabetic-obese and diabetic-obese populations. Loss of function mutations in leptin (ob/ob) and leptin receptor 91 (db/db) provide insight into β-cell physiology in nondiabetic-obese and diabetic-obese states within individual 92 mouse strains (8,40,46,53), however leptin and its receptor play a critical role in β-cell function independent of 93 obesity, limiting interpretations of these studies (22). No current mouse model is well-suited to examine 94 physiological differences in β-cell health between nondiabetic-obese and diabetic-obese states.

95
The SM/J inbred mouse strain has traditionally been used to study interactions between diet and metabolism, 96 and more recently has uncovered genetic architecture underlying diet-induced obesity and glucose homeostasis(17, 97 [49][50][51][52]63). After 20 weeks on a high fat diet, SM/J mice display characteristics of diabetic-obese mice, including 98 elevated adiposity, hyperglycemia, and glucose intolerance(27). We have previously shown that by 30 weeks of 99 age, high fat-fed SM/J mice enter diabetic remission, characterized by normalized fasting blood glucose, and greatly 100 improved glucose tolerance and insulin sensitivity(11). Importantly, these changes occur in the context of sustained 101 obesity. Given the central role of β-cell health in susceptibility to diabetic-obesity, we hypothesize that obese SM/J 102 mice undergo restoration of functional β-cell mass during the resolution of hyperglycemia. This study focuses on 103 how insulin homeostasis, β-cell morphology, and β-cell function change during this remarkable transition and 104 establishes SM/J mice as a useful model for teasing apart diabetic-obese and nondiabetic-obese states.

107
Animal husbandry and tissue collection. SM/J mice were obtained from The Jackson Laboratory (Bar Harbor, ME).

123
Insulin Tolerance Test. At 19 or 29 weeks of age, mice were fasted for 4 hours prior to procedure. Insulin (humulin) 124 was prepared by mixing 10 ul insulin with 10 ml sterile saline. Mice were injected with 3.75 ul insulin mixture/g 125 bodyweight. Blood glucose levels were assessed from a tail nick at times = 0, 15, 30, 60, and 120 minutes via 126 glucometer (GLUCOCARD).

127
Islet Histology and Analyses. At the time of tissue collection, whole pancreas was placed in 3 mL of neutral buffered 128 formalin. These samples were incubated at 4°C while gently shaking for 24 hours. Immediately afterwards, samples 129 were placed into plastic cages and acclimated to 50% EtOH for 1 hour. Samples were then processed into paraffin 130 blocks using a Leica tissue processor with the following protocol: 70% EtOH for 1 hour x 2, 85% EtOH for 1 hour, 131 95% EtOH for 1 hour x 2, 100% EtOH for 1 hour x 2, Xylenes for 1 hour x 2, paraffin wax. Pancreas blocks were 132 sectioned into 4 µm thick sections. Four samples per individual were randomly selected, at least 100 µm apart.

145
Background was subtracted from DAPI, insulin, glucagon, and phospho-histone H3 images using ImageJ. DAPI 146 channel was used to identify total nuclei in CellProfiler. Insulin and glucagon channels were combined and overlaid 147 on the DAPI image to identify islet nuclei. Insulin (INS + ) staining overlaid with DAPI identified β-cell cells, glucagon (GCG + ) staining overlaid with DAPI identified α-cells. Phosphohistone H3 (PHH3 + ) staining identified 149 mitotic nuclei. Total nuclei, islet cells, β-cells, α-cells, and mitotic nuclei were summed across 4 slides for each 150 individual. Islet, β-cell, and α-cell mass is reported as fraction of total nuclei. Mitotic islet index is reported as 151 proportion of β-cells and α-cells positive for phosphohistone H3. Islets with diameter < 50 µm were discarded.

152
Islet isolation. Pancreas was removed and placed in 8mL HBSS buffer on ice. Pancreas was then thoroughly minced.

153
Collagenase P (Roche) was added to a final concentration of 0.75 mg/ml. Mixture was then shaken in a 37°C water 154 bath for 10-14 minutes. Mixture was spun at 1500 rpm for 2 minutes. The pellet was washed twice with HBSS. The 155 pellet was re-suspended in HBSS and transferred a petri dish. Hand-selected islets were placed in sterile-filtered 156 RPMI with L-glutamine (Gibco) containing 11mM glucose, supplemented with 5% pen/strep and 10% Fetal Bovine 157 Serum (Gibco). Islets were rested overnight in a cell culture incubator set to 37°C with 5% CO2.

171
Statistical analyses. Phenotypes were assessed for normality by a Shapiro-Wilk test, and outliers removed. A 172 student's t-test was used to assess significance between two cohorts, while a one-way ANOVA with Tukey's Post Hoc test was used to assess significance among multiple cohorts. Pearson's correlation was used to determine 174 strength of correlation among variables. P-values < 0.05 were considered significant.

176
Obese SM/J mice increase insulin levels and improve insulin sensitivity. The resiliency of β-cells distinguishes 177 nondiabetic-obese and diabetic-obese individuals (8,46,47,63,66,72,74,81). While both groups develop 178 hyperinsulinemia, diabetic-obese individuals become insulin resistant, leading to β-cell dysfunction, 179 hypoinsulinemia, and hyperglycemia. Our previous work shows that obese SM/J mice spontaneously transition 180 from hyperglycemic to normoglycemic with age(11). Principle to this is a 40 mg/dl decrease in fasting glucose 181 levels in high fat-fed SM/J mice between 20 and 30-weeks (Fig. 1A). We first sought to characterize how insulin 182 homeostasis changes during this transition. Interestingly, 20-week high fat-fed SM/J mice have comparable levels 183 of plasma and pancreatic insulin levels compared to age-matched low fat-fed mice ( Fig. 1B-C). By 30 weeks, high 184 fat-fed SM/J mice increase circulating insulin levels 5.3-fold and pancreatic insulin levels 1.9-fold, in line with 185 other models of hyperinsulinimic nondiabetic-obesity (33,36,55). We sought to test for peripheral insulin resistance 186 via an insulin tolerance test (ITT), as insulin resistance is a known mechanism for increasing circulating and 187 pancreatic insulin levels. Surprisingly, 20-week high fat-fed SM/J mice display insulin resistance compared to low 188 fat-fed mice, however, insulin sensitivity is restored by 30 weeks (Fig. 1D-E). The simultaneous increase in insulin 189 production and improved insulin sensitivity is unprecedented and suggests a novel mechanism beyond insulin 190 resistance for enhancing β-cell insulin secretion.

199
Obese SM/J mice increase islet mass during resolution of hyperglycemia. In humans and mice, obesity initially 200 increases islet mass, and maintenance of that mass in part differentiates nondiabetic-obese individuals from diabetic-201 obese individuals (2,9,25,59,76,85). To understand the source of increased insulin production in obese SM/J 202 mice, we examined islet morphology during the resolution of hyperglycemia. To quantify islet mass and number, 203 β-cell mass, α-cell mass, and mitotic index, we randomly selected 4 sections per fixed pancreas and stained with 204 antibodies against insulin, glucagon, and phospho-histone H3. Representative images of immuno-stained pancreatic 205 sections for 30-week high fat-fed mice and 30-week low fat-fed mice are shown in Figure 2A

249
Obese SM/J mice increase islet insulin secretion and insulin content. In conjunction with changing β-cell 250 morphology, diabetic-obesity is associated with altered β-cell function, including diminished first phase insulin 251 secretion, increased basal insulin secretion, and decreased β-cell insulin production (16,23,57,66). We sought to 252 examine if changes in β-cell insulin secretion and content corresponded with the resolution of hyperglycemia and 253 expanded β-cell mass we observe. To test this, we isolated islets from high and low fat-fed 20-and 30-week SM/J 254 mice. After allowing islets to rest overnight, we performed a glucose-stimulated insulin secretion assay by 255 subjecting islets to low (2.8 mM) or high (11 mM) glucose conditions. We find that high fat-fed SM/J mice week, high and low fat-fed mice (C). Islet insulin content normalized by total protein measured via protein BCA 272 (D). Correlation between insulin secretion ratio and islet insulin content (E). Open circles -20-week high fat-fed, 273 closed circles -30-week high fat-fed. *p<0.05, **p<0.01, ***p<0.001, N.S. Not Significant.

277
The ability to maintain appropriate insulin production and secretion, termed functional β-cell mass, 278 is a central determinant of diabetic risk. In this study, we describe insulin homeostasis, islet morphology, 279 and β-cell function in obese SM/J mice as they transition from hyperglycemic to normoglycemic. We 280 determine that increased insulin production and insulin sensitivity accompany improved glycemic control, 281 driven by expanded β-cell mass and improved glucose-stimulated insulin secretion. Our results show 282 obese SM/J mice undergo restoration of functional β-cell mass, providing an opportunity to explore how 283 compensatory insulin production can be achieved in the context of obesity. 284 Susceptibility to high fat diet-induced diabetes in mice depends on genetic background. Several 285 strains and sub-strains develop diabetic-obesity, including hyperglycemia, glucose intolerance, and insulin 286 resistance, consistent with the diabetic phenotypes observed in obese SM/J mice at 20 weeks (3,44,83). 287 Remarkably, by 30 weeks, obese SM/J mice have characteristics of diabetic-resistant obese strains, 288 retaining glycemic control by dramatically increasing insulin production and improving insulin sensitivity 289 (3,79,83). To our knowledge, this is the first report of transient hyperglycemia in an inbred strain, 290 although similar phenomena have been reported in mice with the leptin receptor (db/db) mutation. 291 C57bl/6J (db/db) and 129/J (db/db) mice are obese and initially develop mild hyperglycemia at 8-10 weeks of 292 age, but this resolves by 20-30 weeks, concurrent with increased insulin production and β-cell mass(40, 293 54). Unfortunately, leptin and its receptor play an important role in β-cell growth and function independent 294 of obesity, which confounds understanding of how genetic background mediates diabetic risk in 295 obesity(22). 296 High fat diet-induced obesity in mice can result in increased islet mass, no change, or decreased 297 mass (3,39,66,79). Across these studies, inability to expand islet mass is associated with hyperglycemia. 298 In humans, islet mass correlates with BMI in nondiabetic obese-individuals, while diabetic-obese 299 individuals have low islet mass compared to nondiabetic individuals (26,47,54). High fat-fed SM/J mice 300 are unique because they have expanded islet mass at 20 weeks, yet normal insulin levels and insulin 301 resistance. By 30-weeks, islet mass continues to expand, driven by increased islet area and increased islet 302 number, corresponding with increased insulin production and improved insulin sensitivity. Islet 303 neogenesis may contribute to the increased islet number, and fission of large islets has been reported 304 during development, suggesting islets have mechanisms to maintain an appropriate size (41,80). 305 β-cell expansion is the primary driver of islet expansion in mouse models of obesity (8,46). Some 306 nondiabetic obese mice experience increased β-cell mass, but do not show evidence for elevated β-cell replication in immunostaining of pancreatic sections (38,83). This has been attributed to islets in the tail 308 of the pancreas being substantially more proliferative in response to high fat diet than the body and head 309 regions(28), thus technical artifacts in sampling could result in inflated variances which mask biological 310 differences. This is could be the case here, given that high fat-fed SM/J's β-cell mass is far above low fat-311 fed controls, that their β-cell mass expands 2-fold during the resolution of hyperglycemia, yet we find no 312 evidence for increased β-cell replication. However, α-cell mass also expands in obesity (29,37,61). While 313 α-cell mass is elevated in high fat-fed SM/J mice compared to low fat-fed controls, we find it does not 314 change between 20 and 30 weeks, despite substantial elevation of α-cell mitotic index. 315 Retention of β-cell function separates diabetic-obesity and nondiabetic obesity (5,35,45). 20-316 week high fat-fed SM/J mice have an insulin secretion profile similar to diabetic-obese mice and humans, 317 including blunted glucose-stimulate insulin release, elevated basal insulin secretion, and low islet insulin 318 content, which resolves by 30 weeks. Underscoring this transition is the positive correlation between 319 glucose-stimulated insulin release and islet insulin content. Care was taken to select normal sized islets 320 across all cohorts for functional assessment (~100µm in diameter) indicating this robust improvement in 321 β-cell functional mass is due to changes to β-cell physiology. 322 Three current, non-mutually exclusive components of β-cell stress response may shed light on the 323 perplexing improvement in glycemic control seen in SM/J mice: β-cell dedifferentiation, nascent β-cell 324 maturation, and changes in β-cell subtype proportions. While early studies concluded overworked β-cells 325 undergo apoptosis (10, 56, 67, 73), recent studies have suggested β-cells dedifferentiate into a 326 dysfunctional, progenitor-like state, potentially as a defense mechanisms against prolonged glycemic 327 stress (18,43,58,84). These dedifferentiated β-cells have low insulin content and poor glucose-stimulated 328 insulin secretion. Further, the dedifferentiated state is reversible in cultured conditions, revealing potential 329 for therapeutic intervention(24). It is feasible that obese SM/J mice have β-cells in the dedifferentiated 330 state at 20-weeks, which would explain the low insulin content and poor functionality despite the elevated 331 β-cell mass. Improvement in insulin sensitivity could ease glycemic stress, allowing dedifferentiated β-332 cells to redifferentiate by 30 weeks, reestablishing insulin production and secretion. 333 Work from several groups suggests β-cells can be divided into two broad categories: functionally 334 immature and functionally mature cells. Immature β-cells have greater proliferative potential and are 335 resistant to stress, at the expense of functional maturity (4, 7, 69, 89). These immature β-cells have low 336 insulin content, high basal insulin secretion, and a lack of glucose stimulated insulin secretion. The large 337 β-cell expansion seen in obese SM/J mice, suggests nascent β-cells must undergo maturation at some 338 point. We have no evidence of enhanced β-cell replication at 20-weeks, but it is possible these β-cell are 339 still functionally immature and reach maturity by 30-weeks. This could explain why islets from these mice 340 lack glucose stimulated insulin release, show elevated basal insulin secretion, and have low insulin 341 content, despite elevated mass. 342 Recent advances in single cell technology has allowed for identification of β-cell subtypes, based 343 on functional characteristics and gene expression. These include β-cells that specializes in basal insulin 344 secretion, characterized by low mature insulin content, and enriched in db/db diabetic islets(32). While 345 these cells are not equipped to respond to elevated glucose, they are enriched for maturity markers 346 including Ucn3 and Glut2, distinguishing them from immature β-cells. Pancreatic multipotent progenitors 347 (PMPs) are rare insulin positive cells capable of generating endocrine cells in vivo including functionally 348 mature β-cells(73, 85). These cells resemble immature β-cells, with low insulin content and Glut2 349 expression, whose proliferation is stimulated by glycemic stress in STZ-treated and NOD mouse models. 350 Lastly, β-cell hub cells coordinate calcium signaling and insulin release of surrounding β-cells(42). These 351 cells have markers for both mature and immature β-cells, including expression of Gck and Pdx1, but low 352 insulin content, and are especially sensitive to glycemic and inflammatory stress. Ablation of these cells 353 results in loss of coordinated insulin release, suggesting they are necessary for mature islet function. Given 354 the elevated β-cell mass, poor insulin secretion, and low insulin content in 20-week high fat-fed SM/J 355 mice, it is possible islets are enriched for basal insulin secretors and PMP's, while devoid of hub cells. At 356 30 weeks, as glycemic stress diminishes, basal insulin secretors and PMP populations decline, while hub 357 cells rise, reestablishing β-cell functionality. 358 Clearly, the interplay between β-cell dedifferentiation, nascent β-cell maturation, and β-cell 359 subtype identity in diabetic-obesity needs to be clarified. SM/J mice are a useful model because they allow 360 for appropriate comparisons across diabetic-obese, nondiabetic-obese, and nondiabetic-lean populations. 361 Future studies interrogating these differences will provide insight into the physiological mechanisms that 362 allow β-cell functionality to be maintained and improved in the context of obesity.  The authors declare no conflicts of interest.