Leptin treatment prevents impaired hypoglycemic counterregulation induced by exposure to severe caloric restriction or exposure to recurrent hypoglycemia

Abstract

Hypoglycemia-associated autonomic failure (HAAF) is a maladaptive failure in glucose counterregulation in persons with diabetes (PWD) that is caused by recurrent exposure to hypoglycemia. The adipokine leptin is known to regulate glucose homeostasis, and leptin levels fall following exposure to recurrent hypoglycemia. Yet, little is known regarding how reduced leptin levels influence glucose counterregulation, or if low leptin levels are involved in the development of HAAF. The purpose of this study was to determine the effect of hypoleptinemia on the neuroendocrine responses to hypoglycemia. We utilized two separate experimental paradigms known to induce a hypoleptinemic state: 60% caloric restriction (CR) in mice and three days of recurrent hypoglycemia (3dRH) in rats. A sub-set of animals were also treated with leptin ((0.5–1.0 μg/g) during the CR or 3dRH periods. Neuroendocrine responses to hypoglycemia were assessed 60 minutes following an IP insulin injection on the terminal day of the paradigms. CR mice displayed defects in hypoglycemic counterregulation, indicated by significantly lower glucagon levels relative to controls, 13.5 pmol/L (SD 10.7) versus 64.7 pmol/L (SD 45) (p = 0.002). 3dRH rats displayed reduced epinephrine levels relative to controls, 1900 pg/mL (SD 1052) versus 3670 pg/mL (SD 780) (p=0.030). Remarkably, leptin treatment during either paradigm completely reversed this effect by normalizing glucagon levels in CR mice, 78.0 pmol/L (SD 47.3) (p=0.764), and epinephrine levels in 3dRH rats, 2910 pg/mL (SD 1680) (p=0.522). These findings suggest that hypoleptinemia may be a key signaling event driving the development of HAAF and that leptin treatment may prevent the development of HAAF in PWD.

1. Introduction

Patients with diabetes are particularly vulnerable to hypoglycemia, and exposure to hypoglycemia is associated with an increased risk of all-cause mortality in these patients (Khunti et al. 2014). Patients with longstanding diabetes frequently experience recurrent hypoglycemic episodes due to poorly regulated hyperinsulinemia and inadequate glucagon responses. These recurrent bouts of hypoglycemia often lead to the development of hypoglycemia associated autonomic failure [(HAAF) (Cryer 2013; Rogers et al. 2017; Lontchi-Yimagou et al. 2018)], which is characterized by critically reduced neuroendocrine responses to hypoglycemia as well as hypoglycemia unawareness.

Although HAAF is commonly experienced by persons with diabetes (PWD) on insulin therapy, diabetes is not required for its development. It is well established that HAAF can be induced in metabolically heathy humans and rodents via recurrent exposure to insulin-induced hypoglycemia (Senthilkumaran and Bobrovskaya 2017; Sankar et al. 2020). Thus, HAAF represents a pathophysiological adaptation brought about by repetitive exposure to hypoglycemia irrespective of disease state. Furthermore, the known antecedent of HAAF, insulin-induced hypoglycemia, is a relatively modern phenomenon that has only arisen following the adoption of exogenous insulin therapy as a treatment for diabetes. Thus from an evolutionary perspective, severe hypoglycemia due to hyperinsulinemia would be exceedingly rare and unlikely to drive such distinct and ubiquitous physiological adaptations (Beall et al. 2012). In fact, glucose homeostasis is so well controlled in mammals that hypoglycemia is rarely experienced except during prolonged starvation (Goldstein et al. 2011; Watford 2015). Thus, the physiological adaptations associated with starvation could provide insights into underlying mechanism contributing to HAAF in PWD.

Starvation and its physiological antecedents, such as hypoleptinemia, depletion of hepatic glycogen, and the induction of ketosis, are all associated with impairments to the neuroendocrine responses to hypoglycemia, a defining feature of HAAF. More specifically, the counterregulatory responses to insulin-induced hypoglycemia are significantly blunted directly following a 72-h fast in humans (Adamson et al. 1989). Starvation also induces profound hypoleptinemia in both rodents (Frederich et al. 1995; Perry et al. 2018) and humans (Boden et al. 1996; Chan et al. 2006), and exposure to insulin-induced hypoglycemia, the primary antecedent of HAAF, also causes hypoleptinemia in rats (Reno et al. 2015) and relative hypoleptinemia in human (Sandoval et al. 2003; Davis et al. 2015). Depletion of hepatic glycogen, a known consequence of starvation, also leads to deficits in hypoglycemic counterregulation in dogs (Winnick et al. 2016; Warner et al. 2021). Furthermore, induction of nutritional ketosis via ketogenic diets alters hypoglycemic counterregulation in both mice (Morrison et al. 2020) and humans (Ranjan et al. 2017), and is associated with increased frequency of hypoglycemia in PWD on insulin therapy (Leow et al. 2018). Taken together, these studies implicate a clear relationship between the starvation phenotype and impairments in hypoglycemic counterregulation. Yet, determining the individual contribution of hypoleptinemia, depletion of hepatic glycogen, and ketosis to impairments in counterregulation is challenging because each of these phenomena occur under similar conditions.

The purposes of the current study were to 1) establish a paradigm for determining the effect of this starvation-induced hypoleptinemia on hypoglycemic counterregulation independent of concurrent ketosis and hepatic glycogen depletion, and 2) to test whether the prevention of hypoleptinemia could reverse impairment of hypoglycemic counterregulation following exposure to either starvation or recurrent insulin-induced hypoglycemia. In order to establish our hypoleptinemia protocol, we leveraged the fact that leptin levels are independently influenced by food restriction and changes in fat mass (Morrison 2009). In this way, short-term food restriction drives hypoleptinemia without significant reduction in fat mass (Weigle et al. 1997; Ahren 2000), whereas a prolonged caloric restriction that produces a significant reduction in fat mass leads to a sustained hypoleptinemic state following refeeding in both humans (Chan and Mantzoros 2005; Korbonits et al. 2007) and rodents (Zhan et al. 2009; Zhao et al. 2013).

We utilized a well-established mouse model of starvation, 60% caloric restriction (Zhao et al. 2010; Goldstein, Zhao et al. 2011; Mani et al. 2016), followed by a refeeding period to produce a hypoleptinemic state without concurrent ketosis or glycogen depletion. We then then assessed neuroendocrine responses to hypoglycemia during the refeeding period. We hypothesized that hypoglycemic counterregulation would be impaired during the refeeding period and that this impairment would be prevented with exogenous leptin treatment. We similarly tested whether exogenous leptin treatment could prevent impairment of hypoglycemic counterregulation in rats following exposure to 3 days of recurrent insulin-induced hypoglycemia (3dRH), a common experimental model of impaired hypoglycemic counterregulation (Senthilkumaran et al. 2016; Sankar, Khodai et al. 2020).

2. Material and Methods

2.1. Overview/Summary

Animals were subjected to one of two experimental paradigms designed to induce a hypoleptinemic state: either 60% CR (mice) or 3dRH (rats). For 60% CR, mice were fed at 40% of their average daily food intake for six days before entering a 1–4 day refeeding period. For 3dRH, rats received an intraperitoneal insulin (IP) injection on three consecutive days to induce severe hypoglycemia for approximately 150 minutes. The 3dRH paradigm has been shown to induce hypoleptinemia in rats (Reno, Ding et al. 2015). Ad libitum fed mice and saline treated rats served as controls during the CR and 3dRH paradigms, respectively. A subset of CR and 3dRH exposed animals also received concurrent IP leptin injections (0.5–1.0 μg/g), either daily (rats) or twice daily (mice). Following exposure to either the CR or the 3dRH paradigm, the neuroendocrine response to hypoglycemia was assessed by determination of counterregulatory hormone levels 60 minutes following IP insulin injection. In order to evaluate the efficacy of CR paradigm in producing a sustained hypoleptinemia state, fasting hormones and metabolites, as well as body composition were measured in a subset of CR exposed mice at baseline and during the refeeding period.

2.2. Ethical Approval

All animal experiments were approved by the Institutional Animal Care and Use Committee at Pennington Biomedical Research Center [(PBRC); (Approval numbers 981P and 1058P)]. All animals were reared in an AAALAC (Assessment and Accreditation of Laboratory Animal Care) accredited animal facility in accordance with all conditions specified by the United States Department of Agriculture’s Office of Laboratory Animal Welfare.

2.3. Animals

Male and female C57BL/6J wild-type mice 8–12 weeks of age (PBRC breeding colony) and male Sprague Dawley rats 8–14 weeks of age (Envigo) were used in these studies. The majority of experiments were conducted in males with a cohort of female mice used to confirm the major findings of some experiments. Unless explicitly noted, results are from male only cohorts. All animals were single-housed with a 12-h light and 12-h dark cycle, with the dark cycle beginning at either 18:00 or 19:00. All animals were fed Purina 5001 Rodent Laboratory Chow. Purified tap water was provided ad libitum. For mice, following completion of experiments, animals were euthanized via hypoxia induced by a gradual increase in CO₂, followed by exsanguination (adjunctive method). For hypoxia, animals were maintained in their home cages and placed in a chamber that contained an inflow of CO₂ supplied from a compressed gas cylinder at a displacement rate of 20% of the chamber volume per minute. All rats were euthanized via overdose (1.5 mL/kg) with commercial euthanasia solution (Euthasol, 390 mg/mL sodium pentobarbital + 50 mg/mL sodium phenytoin; Med-Vet, Mettawa, IL) followed by exsanguination (adjunctive method).

2.4. 60% Caloric Restriction and Refeeding Paradigm

Animals were randomly assigned to one of two groups: Mice exposed to caloric restriction for six days (CR mice) or ad libitum fed mice (Ad-lib mice). Littermates were assigned in equal numbers to each group and began the paradigm concurrently. See top border of Figure 1 for graphical representation of the timeline of the paradigm relative to the start of the paradigm. Following a five day acclimation period to wire-bottomed cages (Days 0–4), food intake was monitored daily in each mouse for six days prior to initiation of caloric restriction, and each mouse’s average daily food intake was determined (Days 5–10). CR mice were then fed 40% of their daily average food intake for six days (Days 11–16) while Ad-lib mice continued to have unrestricted access to food. Mice were fed at ≈ 1.5 h prior to lights off each day. Body weight and non-fasting blood glucose were measured daily immediately prior to feeding. All mice were administered 1.0 mL of warmed saline subcutaneously to prevent dehydration during the six days of CR. These injections occured immediately following body weight and blood glucose measurements.

Figure 1. Six days of 60% caloric restriction leads to significant reductions in body weight and blood glucose while inducing hyperphagia and sustained reductions in blood glucose following refeeding.

Following a five day acclimation period and six days of baseline food intake data, wild type B6 male mice were assigned to either a 60% caloric restriction paradigm (CR mice; black circles) or ad libitum access to chow (Ad-lib mice; grey squares) for six days. Following the caloric restriction period, both groups of mice, CR and Ad-lib, were given ad libitum access to chow for four days (see top border for timing of each experimental period). CR mice experienced significant reductions in both body weight and blood glucose during the CR paradigm with both measurements reaching their nadir on the final day of the paradigm. The reduction in body weight was reversed within three days of refeeding, while blood glucose levels remained lower than those of Ad-lib mice only on the first day of refeeding. CR mice also displayed a marked hyperphagia during the refeeding period, which continued despite restoration of pre-restriction body weight. Data are expressed as mean ± SD. All data were analyzed via a two-way ANOVA with Bonferroni's multiple comparisons tests. Days marked with * represents significant differences between CR and Ad-lib groups (p-values < 0.05). n=52 per group days 1–17, n=35 per group days 18–19, and n=22 per group on day 20.

After six days of restriction, the CR mice entered a refeeding period and returned to ad libitum access to food for up to four days, Days 17–20 (Figure 1, top x-axis), which encompassed Days 0–4 of the refeeding period (Figure 1, bottom x-axis). During the refeeding period, food intake, non-fasting blood glucose, and body weight continued to be measured daily up to the terminal day of the experiments. In a subset of CR and Ad-lib mice, body composition was measured on Day −6 (baseline), Day 0, Day 1, Day 2, and Day 4 relative to refeeding via nuclear magnetic resonance spectroscopy (Bruker LF110 BCA-Analyzer, Billerica, MA).

2.5. Three Day Recurrent Hypoglycemia Paradigm (3dRH)

Animals were randomly assigned to one of two groups, 3dRH or Control. 3dRH rats received IP insulin injections on three consecutive days at a dose of 10 U/kg, 9U/kg and 8 U/kg on the first, second, and third days, respectively, at ≈1–2 h following lights on, while control rats received IP saline during the same days and times. Food was removed just prior to insulin or saline injection and returned 180 min later. This dosing strategy in known to produce sustained hypoglycemia, blood glucose 30–50 mg/dL for ~120 min, for three consecutive days (e.g. Reno, Ding et al. 2015), and is commonly used to model impaired counterregulation in rats (Senthilkumaran, Zhou et al. 2016). Blood glucose was measured immediately prior to, and at ≈ 40 min intervals for 180 min following insulin or saline injection on each day of the paradigm in order to confirm that the targeted level of hypoglycemia was achieved.

2.6. Quantification of Hormones, Metabolites, and Hepatic Glycogen

Hormones and metabolites were measured in either serum or plasma samples obtained from trunk blood that was collected following CO₂ euthanasia (mice) or via cardiac puncture following administration of euthanasia solution (rats). Fasting measurements in mice were made following a 4–5 h fast at ≈ 13:30, and fasting measurements in rats were made following an overnight fast (12–14 h) at ~09:00. For glucagon, leptin, insulin, and β-hydroxybutyrate (BHB) analysis, blood was collected into a 1.7 mL micro-centrifuge tubes and allowed to clot at room temperature for 15–20 min. Samples were then centrifuged at 2,000 g at 4 °C for 10 min and serum collected. For ghrelin and epinephrine assays, blood was collected into micro-centrifuge tubes containing EDTA only for epinephrine, or EDTA plus and 2 μL of the enzyme inhibitor MAFP (Cayman Chemical Ann Arbor, MI) for ghrelin. Samples were centrifuged at 2,000 g at 4 °C for 15 min and plasma was collected. Serum and plasma samples were then stored at −80°C until analysis.

Serum glucagon concentrations were determined by ELISA (10–1281-01, Mercodia, Winston Salem, NC). Serum corticosterone concentrations were determined by ELISA (K014-H1, Arbor Assays, Ann Arbor, MI). Plasma epinephrine concentrations were determined by ELISA (KA1882, Abnova, Taipei, Taiwan). Plasma ghrelin, serum leptin, and insulin concentrations in mice were determined by ELISA (EZGRA-90K, EZML-82K, and EZRMI-13K, respectively, MilliporeSigma, St. Louis, MO). For mouse leptin, values that read as undetectable were replaced with the lowest detectable limit of the assay, 0.02 ng/mL. Rat leptin levels were determined by ELISA (EZRL-83K, MilliporeSigma, St. Louis, MO). Serum BHB concentrations were determined by colorimetric assay (700190, Cayman Chemical, Ann Arbor, MI). Hepatic glycogen concentration was determined by colorimetric assay (Ab169558, Abcam, Cambridge, MA). Blood glucose measurements were made by hand held glucometer, Contour Next EZ (Bayer), via tail nick.

2.8. Leptin Treatment

For mice, after a 4–5 h fast animals received an IP injection of insulin, (Humulin R, Lilly USA Indianapolis, IN) at a variable dose (1.6–2.3 U/kg) in order to induce hypoglycemia (blood glucose: 40–60 mg/dL) for 30 min. Blood glucose levels were measured at 15, 0, 10, 20, 30, 45, and 60 min relative to insulin administration, in order to confirm that similar level of hypoglycemia was achieved at 30–60 min post insulin in all groups. At the 60 min time point, all animals were euthanized by CO₂ inhalation followed by rapid decapitation and trunk blood was collected for hormone measurements.

For rats, following an overnight fast (≈14 h), animals received 10 U/kg of insulin IP (Humulin R, Lilly USA Indianapolis, IN). Blood glucose levels were measured just prior to, and 60 min following insulin administration. At the 60 min time point, all animals were euthanized with euthanasia solution and blood was collected via cardiac puncture for hormone measurements. This protocol has been used in prior studies of the neuroendocrine responses to hypoglycemia (e.g. Senthilkumaran and Bobrovskaya 2017), where it was demonstrated that hypoglycemia-induced increases in plasma epinephrine are maximal at 60 min post insulin treatment (Senthilkumaran et al. 2016).

Animals received either twice daily (mice) or once daily (rats) IP injections of recombinant mouse or rat leptin (0.5–1.0 μg/g). For mice, injections occurred at ≈1.5 h following lights on and ≈0.5 h before lights off. For rats, injections occurred at ≈1.5 h before lights off. A cohort of control animals received IP injections of equivalent volumes of phosphate buffered saline (PBS) during the same time periods. Leptin was procured from Dr. A. F. Parlow, National Hormones and Peptides Program, Harbor-UCLA Medical Center, Los Angeles, CA and dissolved in sterile PBS.

2.9. Data Analysis

All data were analyzed with GraphPad Prism 8 (GraphPad Software, San Diego, CA). Data are expressed as mean (SD). For all statistic tests, a p-value ≤ 0.05 was selected a priori as the threshold for statistical significance. For experiments comparing CR to ad libitum fed mice, body weight, non-fasting glucose, food intake, and body composition data were analyzed using repeated measures two-way ANOVAs (or with a mixed-effects model when missing values were present) with two levels of diet, CR and ad libitum, and multiple levels of time depending on the analyses. For terminal measurements of hormones, metabolites, and hepatic glycogen, normal two-way ANOVAs were used with two levels of diet, CR and ad libitum. When two-way ANOVAs were used, differences between the CR and Ad-Lib groups at each time point were determined via Bonferroni’s multiple comparison tests. In situations where data were not collected at multiple time points, unpaired t-tests were used.

For Experiments involving leptin treatment, data were analyzed via repeated measures two-way ANOVA (blood glucose) or one-way ANOVA (hormones) with two experimental groups (CR+PBS and CR+leptin mice, or 3dRH+PBS and 3dRH+leptin rats) and one control group (either Ad-lib mice or Control rats). For blood glucose data, differences between groups were determined via Bonferroni’s multiple comparison tests comparing all three groups at each time point. For hormone data, differences between groups were determined via Bonferroni’s multiple comparison tests comparing the two experiments groups to their respective control group. Correlational analyses were performed using a simple linear regression model.

Study data were collected and managed using REDCap electronic data capture tools hosted at PBRC (Harris et al. 2009; Harris et al. 2019). REDCap (Research Electronic Data Capture) is a secure, web-based software platform designed to support data capture for research studies, providing 1) an intuitive interface for validated data capture; 2) audit trails for tracking data manipulation and export procedures; 3) automated export procedures for seamless data downloads to common statistical packages; and 4) procedures for data integration and interoperability with external sources.

3. Results

3.1. 60% Caloric Restriction and Refeeding Paradigm

3.1.1. Body weight, non-fasting glucose, and food intake

Within one day of exposure to severe caloric restriction, CR mice had significantly lower mean body weight compared to Ad-lib mice. The mean body weights of CR mice continued to diverge from Ad-lib mice as the CR paradigm continued, reaching a maximum difference on the final day of restriction [(p < 0.001); (Figure 1, top panel)]. Mean body weight of CR mice remained significantly lower than Ad-lib mice during the first two days of the refeeding period (p < 0.001 and p = 0.048, respectively), before normalizing on the third and fourth day of refeeding (Figure 1, top panel). Blood glucose values showed a similar pattern to body weight during the CR paradigm (Figure 1, middle panel). The restoration of body weight and blood glucose during the refeeding period was accompanied by significant increases in mean daily food intake in male CR mice relative to male Ad-lib mice on each of the four days of the refeeding period [(p < 0.001 for all 4 comparisons); (Figure 1, lower panel)]. Body weights and blood glucose of female CR and Ad-lib mice showed a similar pattern during the exposure to the caloric restriction paradigm (Figure S1A).

3.1.2. Body composition

CR and Ad-lib mice had nearly identical body composition at baseline, and exposure to caloric restriction cause significant changes in both % fat mass and % lean mass that persisted through the first two days of the refeeding period (Figure 2A and 2B). The CR-induced effect on % fat mass was particularly robust with CR mice experiencing an average reduction in fat mass of 1.6 g (SD 0.4) which represented an approximately 50% reduction in % fat mass from baseline levels (Figure 2A and 2C). CR-induced changes in both fat mass and lean mass persisted through the first two days of the refeeding period before normalizing on the fourth day of refeeding (Figure 2C and 2D).

Figure 2. Temporal changes in body composition following refeeding after 60% caloric restriction.

Six days of 60% caloric restriction in male mice led to significant reductions in both fat mass and lean mass when quantified as either % of body weight (Day 0, panels A and B,) or absolute change in mass (Day 0, panels C and D). The effect on fat mass persisted during the first two days of the refeeding period (Day 1 and 2, panels A and C). In terms of overall body composition, the reduction in fat mass represented a 50% reduction in % fat mass relative to ad libitum fed (Ad-lib) controls (Day 0, panel A). Data are expressed as mean ± SD. Data were analyzed via two-way ANOVA. Group differences on each Day were determined via Bonferroni’s multiple comparison test with significance levels as follows: ***- p ≤ 0.001. n=52 per group baseline, Day 0 and Day 1, n=35 per group Day 2, and n=21 per group on Day 4.

3.1.3. Fasting glucose, hormones, and metabolites.

Fasting glucose, leptin levels, and hepatic glycogen content were all reduced in CR mice relative to Ad-lib mice on the final day of the caloric restriction [(p < 0.001); (Figure 3A–C, Day 0)]. In contrast to hepatic glycogen content, which was rapidly replenished during refeeding, reductions in fasting glucose and leptin levels persisted through the 1^st day of refeeding in CR mice relative to controls [(p ≤ 0.001); (Figure 3A–C, Day 1). Fasting levels of ghrelin and BHB displayed a pattern similar to hepatic glycogen content, where levels were significantly higher in CR mice relative to Ad-lib mice on the final day of the caloric restriction (p < 0.001 and p=0.017, respectively), while there was no difference in these levels in CR mice relative to controls during any portion of the refeeding period (Figure 3D and 3E). Fasting glucagon levels were not significantly different in CR relative to Ad-lib mice at any time point (Figure 3F). Thus, our CR paradigm successfully produced a hypoleptinemic state on the 1^st day of refeed without concurrent ketosis or depletion of hepatic glycogen.

Figure 3. Temporal changes in hormones and metabolites following refeeding after 60% caloric restriction in male mice.

Six days of 60% caloric restriction in mice led to significant reductions in fasting glucose, leptin, and hepatic glycogen levels (Day 0, Panels A, B, and C), while significantly increasing fasting ghrelin and β-hydroxybutyrate levels (Day 0, Panels D and E). Following a single day of refeeding, levels of hepatic glycogen, ghrelin, and β-hydroxybutyrate normalized and were not significantly different from ad libitum fed control mice for the remainder of the refeeding period, with the exception of Day 4 hepatic glycogen (Day 1, 2, and 4, Panels C, D, and E). In contrast, fasting glucose and leptin remained significantly lower during the first day of refeeding in mice exposed to 60% caloric restriction before returning to levels similar to Ad-lib mice on Days 2 and 4 of the refeeding period (Panels A and B). Fasting glucagon levels were not significantly different between the two groups on the final day of caloric restriction (Day 0, Panel E), nor were there any differences between the groups during the refeeding period (Days 1, 2, and 4, Panel E). Data are expressed as mean ± SD. Data were analyzed via two-way ANOVA. Group differences on each day were determined via Bonferroni’s multiple comparison test with significance levels as follows: *- p ≤ 0.05; **- ≤ <0.01; ***- p ≤ 0.001. n= 7–8/group (A and B); n= 5–8/group (D, E, and F); n= 4–6/group (C).

3.1.4. Neuroendocrine responses to insulin-induced hypoglycemia

Hypoglycemic counterregulation was measured on the first, second, and fourth day of the refeeding period in separate groups of CR and Ad-lib mice. Due to differences in baseline blood glucose levels and insulin sensitivity between CR and Ad-lib mice, a variable dose of insulin was administered to achieve equivalent levels of hypoglycemia (Figure 4A and 4B). Mean blood glucose at 20, 30, 45, or 60 min post insulin injection was not significantly different between any of the six groups of male mice (Figure 4A). In addition, when comparisons were limited to only CR mice versus their respective controls on each day of the refeeding period, e.g. CR vs. Ad-lib mice on the first day of refeeding, there was no different in glucose levels at the 10, 20, 30, 45, or 60 min time points (p>0.999). Thus, an equivalent exposure to hypoglycemia was achieved in all experimental groups.

Figure 4. Hypoglycemic counterregulation is impaired following exposure to six days of 60% caloric restriction and is associated with hypoleptinemia.

One, two, and four days following refeeding, neuroendocrine responses to insulin-induced hypoglycemia were assessed (A) in male mice previously exposed to a six-day 60% caloric restriction paradigm (CR, black) and ad libitum fed controls (Ad-lib, grey). In order to overcome differences in insulin sensitivity and fasting blood glucose levels between groups, a variable dose of insulin (IP) was used to produce equivalent exposure to hypoglycemia (A,B). Mice exposed to 60% caloric restriction displayed a significant reduction in hypoglycemia-stimulated glucagon release on Day 1 and a significant increase in glucagon release on Day 4 relative to control mice, while no between-group differences occurred on Day 2 (C). Hypoglycemia-induced corticosterone levels were also reduced one day following refeeding (D, p = 0.034). The impairment in hypoglycemia-induced glucagon release on Day 1 was significantly corredlated with the degree of hypoleptinemia experienced by CR mice, while there was no relationship between leptin levels and glucagon levels in Ad-lib mice (E). Data are expressed as mean ± SD, and were analyzed via two-way ANOVA or t-test (corticosterone data). Group differences on each day of refeeding were determined via Bonferroni’s multiple comparison test with significance levels as follows: *- p ≤ 0.05; **- p <0.01; n= 8–10/group (A and C), n= 5–7/group (B), and 14–15/group (D). P-values and r-squared values, along with curve fits (black and grey lines), were determined by simple linear regression analysis (E), and the values of these parameters for each group can be found in the figure legend.

On the first day of the refeeding period CR mice had significantly reduced hypoglycemia-induced glucagon levels compared to Ad-lib mice [(p = 0.002);(Figure 4C)]. There were no differences in glucagon levels in on the second day, but CR mice had significantly increased hypoglycemia-induced glucagon levels on the fourth day of refeeding compared to Ad-lib mice [(p = 0.023);(Figure 4C)]. Female mice showed a similar pattern, with female CR mice having significantly reduced glucagon levels compared to female Ad-lib mice [(p = 0.028); (Figure S1D)]. In order to further examine starvation-induced effects on hypoglycemic counterregulation, hypoglycemia-induced changes in corticosterone levels were assessed in mice on the first day of the refeeding period. CR mice had significantly reduced corticosterone levels on the first day of refeeding relative to Ad-lib fed mice [(p=0.034);(Figure 4D)].

Serum leptin levels were also significantly reduced at 60 min following insulin treatment in CR mice relative to Ad-lib mice on the first and second day of the refeeding period. These reduced leptin levels were significantly correlated with reduced hypoglycemia-induced glucagon release in individual male CR mice [(p = 0.031, r² = 0.212); (Figure 4E, CR group)]. Serum insulin levels 60 min following insulin injection are shown in Figure S2A.

3.2. The effect of leptin treatment on hypoglycemic counterregulation following exposure to caloric restriction

In order to further evaluate the relationship between hypoleptinemia and starvation-induced loss of glucose counterregulation, we tested whether leptin supplementation during CR would prevent impairment of the neuroendocrine response to hypoglycemia. Blood glucose and hypoglycemic counterregulation were assessed in three addition groups of mice: CR+PBS, CR+Leptin, and Ad-lib+PBS. Non-fasting blood glucose levels during the period of caloric restriction were similar in CR+PBS and CR+Leptin mice, but reduced in both groups relative to Ad-lib mice on the last five days of CR (Figure 5A). Neuroendocrine responses to hypoglycemia were then assessed on the first day of the refeeding period.

Figure 5. Leptin treatment during caloric restriction restores normal hypoglycemia-induced glucagon release without reversing hypoleptinemia.

Two groups of male mice were exposed to a six-day 60% caloric restriction paradigm and received either twice-daily leptin treatment (CR+Leptin, black semicircles) or PBS injections (CR+PBS, black circles) during the CR paradigm. A third group consisted of ad libitum fed control animals (Ad-lib, grey squares). Non-fasted blood glucose levels during the period of caloric restriction were similar in CR+PBS and CR+Leptin mice, but reduced in both groups relative to Ad-lib mice on the last five days of restriction (A). One day following refeeding, counterregulatory hormone levels were assessed following insulin-induced hypoglycemia in all three groups (B). In order to overcome differences in insulin sensitivity and fasting blood glucose levels between the control and CR groups, a variable dose of insulin (IP) was used to produce equivalent exposure to hypoglycemia (C). Hypoglycemia-induced glucagon levels were significantly reduced in the PBS treated CR mice relative to ad-lib mice, while leptin-treated CR mice showed no deficit (D). There were no differences in corticosterone levels between the three groups 60 minutes following insulin treatment (E), but leptin levels were significantly reduced in both CR+PBS and CR+Leptin mice at this time point relative to ad-lib mice (F). Data are expressed as mean ± SD. Data were analyzed via two-way ANOVA (A and B) or one-way ANOVA (C-F) and group differences were determined via Bonferroni’s multiple comparison test relative to control mice. *- p ≤ 0.05.

Similar to the previous experiments, there was no difference in glucose levels in the three groups at 20, 30, 60 min post insulin treatment (Figure 5B), but the dose of insulin administered was lower in CR+PBS and CR+Leptin mice relative to Ad-lib+PBS mice (Figure 5C). In contrast, hypoglycemia-induced glucagon levels were similar in Ad-lib+PBS and CR+Leptin mice (p>0.999), but significantly reduced in CR+PBS mice relative to Ad-lib-PBS controls [(p=0.022);(Figure 5D)], thus demonstrating a reversal of impaired glucagon secretion in leptin treated mice. Hypoglycemia-induced corticosterone levels in both CR groups were not significantly different from controls (Figure 5E), while serum leptin levels 60 min following insulin treatment were lower in both CR+PBS and CR+Leptin mice relative to Ad-lib+PBS mice [(p=0.004 and p<0.001, respectively); (Figure 5F)].

3.3. The effect of leptin treatment on hypoglycemic counterregulation following exposure to recurrent hypoglycemia

Given that preventing starvation-induced hypoleptinemia prevented impairment of hypoglycemic counterregulation during refeeding, we sought to determine if a similar effect could be achieved by leptin treatment during recurrent exposure to hypoglycemia in rats. Recurrent hypoglycemia is a known antecedent of both hypoleptinemia (Reno, Ding et al. 2015) and impairment of hypoglycemic counterregulation in rats (Senthilkumaran, Zhou et al. 2016; Sankar, Khodai et al. 2020). Analogous to our leptin experiments in CR mice, three groups were used: 3dRH+PBS, 3dRH+Leptin, and Control rats. Both 3dRH groups experienced nearly identical exposure to hypoglycemia during the 3dRH paradigm, and there was no difference in mean glucose during the assessment of the neuroendocrine response to hypoglycemia between the three groups (Figure 6A). Hypoglycemia-induced epinephrine levels were similar in Control and 3dRH+Leptin rats (p=0.522), but significantly reduced in 3dRH+PBS rats relative to controls [(p=0.030);(Figure 6B)], thus demonstrating prevention of 3dRH-induced impairment of epinephrine secretion with leptin treatment. Hypoglycemia-induced corticosterone levels in both 3dRH groups were not significantly different from controls (Figure 6C). Similarly, hypoglycemia-induced glucagon levels in 3dRH+PBS and 3dRH+Leptin were also not significantly different form controls. Serum leptin levels 60 min following insulin treatment were similar in Control and 3dRH+PBS rats, but significantly increased in 3dRH+Leptin rats relative to controls [(p=0.004); (Figure 6D)].

Figure 6. Leptin treatment during exposure to three days of recurrent hypoglycemia restores normal hypoglycemia-induced epinephrine levels in male rats.

Counterregulatory hormone and leptin levels were assessed following insulin-induced hypoglycemia in three groups of rats. Two groups were exposed to three days of recurrent hypoglycemia (3dRH) via IP insulin and received either daily leptin (3dRH-Leptin, black semicircles) or PBS injections (3dRH-PBS, black circles) during the 3dRH paradigm. One group received neither leptin nor insulin (Control, grey squares). Glucose levels of the three groups during both the 3dRH paradigm and final bout of hypoglycemia are shown in Panel A. Following exposure to 3dRH, hypoglycemia-induced epinephrine levels were significantly reduced in the 3dRH-PBS rats relative to controls, while 3dRH-Leptin rats showed no deficit (B). There were no differences in corticosterone levels between the three groups 60 minutes following insulin treatment (C). Leptin levels at the 60-minute time point in 3dRH-PBS rats were not different from control rats, while leptin levels of 3dRH-Leptin rats were significantly increased (D). Data were analyzed via two-way ANOVA (glucose) or one-way ANOVA (hormones), and group differences were determined via Bonferroni’s multiple comparison test relative to control rats. *- p ≤ 0.05.

4. Discussion

4.1. Major findings

There is increasing evidence linking both exposure to starvation and leptin signaling with alterations in glucose homeostasis (Adamson, Lins et al. 1989; Reno, Ding et al. 2015; Mani, Osborne-Lawrence et al. 2016; Mani and Zigman 2017; Perry, Wang et al. 2018). In this study we extended these results to demonstrate that both male and female mice exposed to starvation, via six days of 60% CR, exhibit deficits in their counterregulatory response to insulin-induced hypoglycemia following refeeding, and that this effect is associated with hypoleptinemia. More importantly, we established that normal hypoglycemic counterregulation was rescued by antecedent leptin supplementation in both our CR paradigm and rats exposed to recurrent hypoglycemia. Our data also adds to the body of knowledge regarding the physiological response to refeeding following exposure to starvation. These findings reinforce our original hypothesis and demonstrate that 60% CR is a useful model to explore the mechanisms underlying the development of HAAF. Furthermore, we identify hypoleptinemia as a potential necessary signaling event driving the development of HAAF.

4.2. Relevance to prior studies

The neuroendocrine responses to hypoglycemia that we observed following our CR and 3dRH paradigms are largely consistent with prior work in the field. The hypoglycemic-induced glucagon levels in both ad-lib and leptin treated CR mice are nearly identical to levels measured by Berglund et al. (2008) during a hypoglycemic clamp in B6 mice. Our observed reductions in glucagon levels in CR mice without leptin treatment are also consistent with that observed by Jacobson and colleagues (2006) in B6 mice following antecedent exposure to hypoglycemia (Figure S4A). In contrast, our corticosterone levels in CR and Ad-lib mice were significantly higher than these two prior studies, which is most likely due to our method of euthanasia. Yet it is very noteworthy that the magnitude of the reduction in hypoglycemia-induced corticosterone in CR mice relative to Ad-lib mice in the current study is entirely consistent with the magnitude of the reduction seen by Jacobson and colleagues (2006) following antecedent exposure to hypoglycemia in mice (Figure S4B). Furthermore, our hypoglycemia-induced epinephrine levels in both control and 3dRH rats are highly consistent with prior works employing either hypoglycemic clamps (de Vries et al. 2004; Chan et al. 2006; Herzog et al. 2008; Reno, Ding et al. 2015) or terminal cardiac puncture (Senthilkumaran and Bobrovskaya 2017) in the 3dRH model (Figure S4C - S4D). It is also worth noting that the epinephrine levels we report in 3dRH rats treated with leptin (green circle in Figure S4C) are entirely consistent with those reported in control rats, in these same studies. Thus lending further support to our assertion that leptin treatment effectively reverses the effects of prior exposure to recurrent hypoglycemia in rats.

Our data also adds to the body of work related to the physiological response to starvation. Severe caloric restriction paradigms have been routinely used to investigate the physiological response to starvation in both mice (Sun et al. 2008; Zhao, Liang et al. 2010; Li et al. 2012; Mani, Osborne-Lawrence et al. 2016) and rats (Bois-Joyeux et al. 1990; Perry, Wang et al. 2018) via 50–60% CR or a 48-h fast, respectively. The metabolic parameters measured on the final day of our CR paradigm are generally consistent with these prior studies. Yet, little data has been reported on the restoration of pre-starvation physiology during refeeding in rodent models of severe CR, thus our study makes significant contributions to this body of knowledge. Several of the physiological responses to starvation were reversed to control levels within 24 h. These included elevations in ghrelin and BHB, as well as the depletion of hepatic glycogen content. These results are consistent with the known biological regulation of circulating ghrelin (Sun, Butte et al. 2008; Mani and Zigman 2017) and ketone levels (Owen 2005; Watford 2015; Rojas-Morales et al. 2016) and previous measurements of hepatic glycogen following refeeding in 48 h fasted rats (Bois-Joyeux, Chanez et al. 1990).

In contrast, leptin levels remained lower than controls on the first day of refeeding and this coincided with alterations in glucose homeostasis including reductions in non-fasting glucose, fasting glucose, as well as a profound reduction in hypoglycemia-induced glucagon secretion, with milder reductions in hypoglycemia-induced corticosterone levels. In addition, individual CR mice that experienced more pronounced hypoleptinemia during the refeeding period displayed more extreme reductions in hypoglycemia-induced glucagon levels during a hypoglycemic challenge, and leptin treatment during caloric restriction restored the deficit in glucagon release. Taken together, these observations support a role for hypoleptinemia as a signaling event driving impaired hypoglycemic counterregulation following starvation in mice.

Leptin is a known regulator of energy homeostasis and neuroendocrine function (Morrison 2009; Kelesidis et al. 2010) and more recently has gained recognition as a potent regulator of glucose homeostasis (Meek and Morton 2016). Although much of this work has focused on the glucose lowering effect of leptin (Chinookoswong et al. 1999; Yu et al. 2008; Perry et al. 2014), acute leptin treatment has been shown to influence counterregulation in response to both fasting- and insulin-induced hypoglycemia in rats (Reno, Ding et al. 2015; Perry, Wang et al. 2018). Perry et al. (2018) found that the physiological processes that maintain glucose levels during starvation (48-h fast) in rats, were elicited by hypoleptinemia along with insulinopenia, within 16 h of fasting. These studies also demonstrated that acute leptin infusion during the starvation state had significant effects on blood glucose, with physiological doses reducing blood glucose and supraphysiological doses increasing blood glucose. In contrast, our data suggests that leptin administration during starvation in mice has less profound effects on chronic blood glucose levels, as non-fasting glucose levels in CR+PBS and CR+Leptin mice were nearly identical during the caloric restriction period. This discrepancy could be due to either species-specific differences or methodological differences between the experimental paradigms. In is also worth noting that our leptin treatment in rats had no effect on fasting glucose levels, nor glucose levels following insulin treatment. Our data in rats could be explained by the longer interval, 12–16 h, between leptin treatment and any of our blood glucose measurements relative to the work of Perry et al. (2018).

The work of Reno et al. (2015), which evaluated the effect of acute leptin treatment on counterregulatory responses to concurrent insulin-induced hypoglycemia, is also highly relevant to our work. These experiments demonstrated that acute intracerebroventricular (ICV) leptin infusion had mixed effects on the counterregulatory responses in rats exposed to a single bout of hypoglycemia, with glucagon levels being decreased by 50% and epinephrine levels increased by 19%. In addition, acute ICV leptin infusion concurrent with a 90 min hypoglycemic challenge in 3dRH-exposed rats increased both glucagon and epinephrine levels, 45% and 25% respectively, during the first 40 min of hypoglycemia relative to 3dRH exposed rats that were not infused with leptin. Yet, this increase was nominal relative to the levels reported in control rats (not exposed to 3dRH). Epinephrine levels in leptin treated 3dRH rats remained 60% lower than control rats, thus acute leptin infusion was not sufficient to correct the deficit in epinephrine release caused by exposure to 3dRH. In contrast, our leptin treatment paradigm, which occurred concurrent with the exposure to the 3dRH paradigm, as opposed to during the hypoglycemic challenge itself, increased epinephrine levels by 50% relative to untreated 3dRH rats, while also normalizing epinephrine release relative to rats exposed to a single bout of hypoglycemia. Hypoglycemia-induced glucagon levels were not affected by exposure to 3dRH or leptin. Although our glucagon results are not consistent with Reno et al. (2015), this is likely explained by the known inconsistences of the 3dRH paradigm in reliably producing reductions in glucagon levels (see Senthilkumaran, Zhou et al. 2016 for discussion of this phenomenon).

Reno et al. (2015) also reported that rats exposed to 3dRH had nearly undetectable fasting leptin levels one day following the completion of the 3dRH paradigm, thus demonstrating an association between hypoleptinemia and impaired counterregulation. A further association between impaired leptin signaling and hypoglycemia is supported by experiments in Zucker (fa/fa) rats, which lack functional leptin receptors. Fa/fa rats are known to display impaired counterregulatory hormone release in response to both 2-deoxyglucose-induced glucoprivia (Allars and York, 1986) as well as insulin-induced hypoglycemia (Morrison et al., 2021; Nishikawa and Ikeda, 1981). The results of our CR experiments in mice strongly support these observed associations between impaired leptin signaling and hypoglycemic counterregulation. Yet, we cannot confirm this finding in our rat model because our leptin measurements occurred 60 min following insulin treatment, thus not in a typical fasted state. Insulin administration at the dose used in our experiments has been shown to raise leptin levels in rats (Hardie et al. 1996), which could explain why our leptin levels in control rats were 2-fold higher (~2 ng/mL) than that commonly reported in fasted Sprague Dawley rats (~1 ng/mL); (e.g. Bagnasco et al. 2002; Perry, Wang et al. 2018).

Altogether, our findings in the CR and 3dRH models contributes to the understanding of the restoration of normal physiology following starvation and builds on the body of work of Perry et al. (2018) and Reno et al. (2015) linking leptin and hypoglycemic counterregulation. Our work also significantly moves the field forward by demonstrating a possible causative link between antecedent exposure to hypoleptinemia and the development of impaired hypoglycemic counterregulation, especially in the context of the development of HAAF. This could have profound implications for the clinical management of patients at risk for HAAF.

4.3. Clinical implications

HAAF is highly prevalent in PWD, dangerous, and difficult to manage clinically. Patients with longstanding diabetes are particularly vulnerable to exposure to recurrent, treatment-induced hypoglycemic episodes, which leads to the development of HAAF and further risk of hypoglycemic complications. It is estimated that 40% of PWD are at high risk for hypoglycemic complications (Hopkins et al. 2012), and that 20% have HAAF (Geddes et al. 2008). Various mechanisms underlying the development of HAAF have been proposed, including altered CNS glucose utilization, modified CNS glycogen storage, and alterations in neurotransmission and/or neuromodulation, yet the data supporting each of these mechanisms are largely inconsistent (Beall, Ashford et al. 2012; Cryer 2013; Lontchi-Yimagou, You et al. 2018). The results of the present study along with Reno et al. (2015) support the hypothesis that HAAF is caused by antecedent conditions which may also be present during starvation, most notably hypoleptinemia. Individual with type 1 diabetes on insulin therapy also have been shown to exhibit a relative hypoleptinemia (Roden et al. 2000; Yazici et al. 2012). These results add further credence to the concept, first suggested by Reno et al. (2015), that leptin may have therapeutic value for the treatment or prevention of HAAF in patients with diabetes. Human recombinant leptin, metreleptin, is an FDA-approved therapy for the treatment of generalized dyslipidemia (Chou and Perry 2013; Oral et al. 2019), and therefore could readily be repurposed for the treatment or prevention of HAAF.

4.4. Study Limitations

Some limitations to our experimental design are worth noting. First, our use of a variable dose of insulin to induce equivalent levels of hypoglycemia during our assay of the neuroendocrine response to hypoglycemia in mice is atypical. This was required due to both the increase in insulin sensitivity and decrease in fasting glucose levels experienced by CR mice, which are both well-known consequences of exposure to severe caloric restriction. Although some changes in the counterregulatory responses to with varying insulin doses have been shown, these effects are driven by extreme differences in insulin dosing and have minimal effects on glucagon levels (Davis et al. 1992; Davis et al. 1993). Given our comparatively modest 1.3-fold increase in insulin dose (yielding a 2-fold increase in serum insulin) in Ad-lib mice relative to CR mice on the first day of refeeding, it seems unlikely that this small increase in insulin dose could account for the large differences in hypoglycemia-induced glucagon levels observed in our study.

A second limitation of our study is that we did not measure hypoglycemia-induced levels of epinephrine in our mouse experiments. Catecholamine release, especially epinephrine, is a major component of the acute phase of the counterregulatory response, and is particularly important in individuals with longstanding diabetes, who lack normal hypoglycemia-induced glucagon responses (Cryer 2013). Yet, while human exposed to recurrent hypoglycemia routinely show an effect on both glucagon and epinephrine (e.g. Heller and Cryer 1991; Oz et al. 2017), mice exposed to recurrent hypoglycemia generally show an effect on glucagon and corticosterone and no effect on epinephrine (e.g. Jacobson, Ansari et al. 2006; Poplawski et al. 2011). Thus, we chose to focus on glucagon and corticosterone as opposed to epinephrine in our mouse experiments. This limitation is further offset by our inclusion of epinephrine in our rat experiments.