N-Hydroxyethyl-1-Deoxynojirimycin (Miglitol) Restores the Counterregulatory Response to Hypoglycemia Following Antecedent Hypoglycemia Free

Hypoglycemia presents a significant barrier in achieving effective glycemic control in patients with type 1 diabetes, particularly long-term patients with impaired sympathoadrenal responses who may be at a 25-fold greater risk for severe hypoglycemia (1). The impaired sympathoadrenal response, known as hypoglycemia-associated autonomic failure (HAAF), derives directly from insulin therapy and recurring bouts of hypoglycemia (2). Attempts to mitigate HAAF have relied primarily upon strategies to avoid hypoglycemia during intensive insulin therapy, but these have only partially curbed the incidence of hypoglycemia among patients with type 1 diabetes (3). This has led to the pursuit of alternative measures, including repurposing existing pharmacological agents (e.g., naloxone [opiod receptor antagonist], fluoxetine [selective serotonin reuptake inhibitor], and dehydroepiandrosterone). While these have all been shown to enhance or restore the counterregulatory response (CRR) following hypoglycemia, they present some limitations in their application, requiring either chronic administration (4,5) or administration during the antecedent bout (6–8).

While the above treatments have targeted the central nervous system, other critical glucose-sensing loci are potential targets for pharmacological intervention. Portal-mesenteric vein (PMV) glucosensing neurons are of particular importance for the detection of “slow-onset” hypoglycemia, which characterizes most clinical events (9,10). Absent PMV glucosensory input, the central nervous system is unable to mount an appropriate CRR to slow-onset hypoglycemia (11,12). Further, a single bout of antecedent hypoglycemia results in a 35–50% suppression in the sympathoadrenal response to a subsequent bout of slow-onset hypoglycemia with concomitant loss of PMV glucosensor responsiveness (13,14). Importantly, normalizing PMV glycemia alone during the antecedent bout of hypoglycemia prevented suppression of the CRR, despite the brain being exposed to antecedent hypoglycemia (14). Delaere et al. (15) have described a PMV glucose-sensing mechanism involving the sodium–glucose cotransporter 3 (SGLT3), sensitive to small changes in blood glucose and mediated by PMV spinal afferents, characteristics consistent with our own observations (12,16,17). SGLT3 presents an interesting target for pharmacological intervention as N-hydroxyethyl-1-deoxynojirimycin (miglitol), approved for the treatment of type 2 diabetes, is a potent agonist for human SGLT3 (18). Thus, in the current study, we sought to determine whether the portal infusion of miglitol might preserve or restore the CRR following antecedent hypoglycemia.

Male Wistar rats (weight 327 ± 4.5 g; n = 47) were studied in the conscious relaxed state with all surgical and experimental procedures preapproved by the University of Southern California Institutional Animal Care and Use Committee. One week prior to the experiments, animals were chronically cannulated under anesthesia in the right carotid artery (PE-50: arterial sampling), left jugular vein (dual silastic: glucose and insulin infusion), and PMV (silastic: miglitol or saline infusion). The tip of the PMV cannula was positioned either in the superior mesenteric vein, upstream from the liver (PORups) to engage the PMV glucosensors, or in the portal vein adjacent to the liver (PORadj), bypassing PMV glucosensors (19). Six days following surgery, animals were randomly assigned to one of six experimental groups: 1) control (CON; hyperinsulinemic-euglycemic clamp day 1 and hyperinsulinemic-hypoglycemic clamp day 2); 2) antecedent hypoglycemia without miglitol (HYPO; hyperinsulinemic-hypoglycemic clamp day 1 and day 2); or 3–6) one of four miglitol infusion groups (PORadj1, PORadj2, PORups1, and PORups2) exposed to hyperinsulinemic-hypoglycemic clamps on days 1 and 2 with PMV infusion on day 1 (PORadj1 and PORups1) or day 2 (PORadj2 and PORups2) (Fig. 1).

On day 1, animals were placed in an infusion chamber, basal samples drawn, and insulin infusion initiated (25 mU/kg/min) with variable exogenous glucose infusion (GINF). A euglycemic clamp was established for 30 min (−30 to 0 min), followed by a gradual induction of hypoglycemia over 75 min (0–75 min), with the nadir (2.44 ± 0.04 mmol/L) sustained through minute 105 (75–105 min). At minute 105, the insulin infusion ceased, and GINF increased to reestablish euglycemia within 10–20 min (105–135 min). CON animals were exposed to an identical protocol on day 1, with the exception that euglycemia was maintained throughout.

On day 2, all animals were exposed to hypoglycemic clamps identical to day 1, with the exception that serial plasma samples were drawn at −30, 0, 60, 75, 90, and 105 min for the measurement of insulin, epinephrine (EPI), and norepinephrine (NE), with additional plasma drawn at minute 105 for glucagon. Following the final plasma sampling at minute 105, rats were anesthetized with tribromoethanol, perfused transcardially with PBS (0.1 mol/L) and 4% paraformaldehyde, and the brains harvested for postfixation, cryoprotection, and sectioning as previously described (20).

Plasma catecholamines were extracted by adsorption to aluminum oxide, eluted off in 0.1 N perchloric acid, and then processed by reversed-phase high-performance liquid chromatography (Ultimate 3000; Dionex) with electrochemical detection (Coulochem III; ESA). Plasma insulin and glucagon concentrations were determined using commercial ELISA kits (Crystal Chem).

Frozen coronal sections (30 μm) were cut at levels 68–73 of the Swanson rat brain atlas (21) and immunocytochemically processed for dopamine β-hydroxylase (DBH) and Fos-expressing neurons in the dorsomedial medulla (i.e., nucleus of the solitary tract [NTS], area postrema [AP], and dorsal motor nucleus of the vagus [DMX]). Sections were incubated with two primary antibodies: mouse monoclonal anti-DBH antibody (MAB308) and rabbit polyclonal anti-Fos antibody (AB-5; Calbiochem). Primary antibody binding was detected with a donkey anti-mouse IgG secondary antibody conjugated to Alexa 488 (Invitrogen) for DBH and donkey anti-rabbit IgG conjugated to CY3 (Jackson ImmunoResearch Laboratories) for Fos. Images were acquired with a Zeiss AxioImager Z1 microscope (Carl Zeiss Microimaging) using epifluorescence optics and Hamamatsu ORCA-ER camera/Volocity acquisition module (v6.0; PerkinElmer). The number of labeled neurons within the DMX, NTS, and AP were counted manually with reference to adjacent Nissl sections.

To determine the efficacy of miglitol infused at lower dosages approaching those currently prescribed for the treatment of type 2 diabetes, rats (n = 6/group) received PMV miglitol infusions of either 0.48 μmol/kg/min (POR-0.48) or 0.96 μmol/kg/min (POR-0.96). Surgeries and the experimental protocol for this experiment were carried out essentially as described above for PORups (i.e., hyperinsulinemic-hypoglycemic clamps on days 1 and 2 with PMV infusion of miglitol on day 2 upstream from the liver). In contrast to experiment 1 above, plasma samples for EPI, NE, and glucagon were taken on day 1, as well as day 2, so animals could serve as their own controls.

To determine the efficacy of miglitol administered orally in correcting HAAF impaired counterregulation, rats (n = 3–4/group) received oral doses of 0, 40, or 60 mg/kg (Oral-0, Oral-40, and Oral-60, respectively), equivalent to the total amount infused during PMV infusions of 0.96 and 1.45 μmol/kg/min (shown to be effective in experiments 1 and 2). Surgeries and the experimental protocol for this experiment were carried out essentially as described above in experiment 2, with the exception that no catheter was implanted in the portal vein and miglitol was administered as an oral bolus. As in experiment 2 above, plasma samples for EPI, NE, and glucagon were taken on day 1, as well as day 2, so animals could serve as their own controls.

Data were analyzed using one-way or two-way ANOVA with repeated measures where appropriate; significance was set at P < 0.05 (Prism 9.2; GraphPad Software). Post hoc comparisons were conducted using either Dunnett’s test or Bonferroni test as dictated by the experimental design. Experimental results are graphically expressed as scatter plots with bars (mean ± SEM) or line graphs (mean ± SEM) for time course data.

The data generated during the current study are available from the corresponding author upon reasonable request.

On day 1, blood glucose was lowered over 75 min to 2.44 ± 0.04 mmol/L with no significant differences between treatment groups, while CON animals were maintained at euglycemia (Fig. 1B). On day 2, blood glucose was again lowered over 75 min to the same hypoglycemic nadir for all groups with no significant differences between groups.

Hypoglycemia led to a 66-fold increase in EPI for CON on day 2, which was suppressed by 64% in HYPO animals previously exposed to hypoglycemia on day 1 (Fig. 2A) (P < 0.001). Administering miglitol on day 1 had no significant impact on epinephrine responses to antecedent hypoglycemia, with values for PORadj1 and PORups1 remaining significantly below CON (P ≤ 0.001). Administering miglitol on day 2 (PORadj2 and PORups2) resulted in EPI responses not significantly different from CON (P > 0.49). Similar results were observed for NE, in which HYPO demonstrated a 42% suppression in peak NE responses compared with CON (P < 0.001), which was not significantly affected by administering miglitol on day 1, but effectively restored with day 2 miglitol infusion. HYPO animals also demonstrated a 47% suppression in glucagon values for day 2 (Fig. 2B) (P < 0.01), which remained significantly below CON for PORups1 (P < 0.05), though partially restored for PORadj1 (P = 0.16). However, administering miglitol on day 2 resulted in glucagon values for PORadj2 and PORups2 comparable to those for CON (P = 0.50).

Consistent with hormonal responses, mean GINF rates for HYPO, PORadj1, and PORups1 during the sustained hypoglycemic nadir (GINF_75–105) were all elevated relative to CON (Fig. 2C), though these did not achieve conventional standards of statistical significance (P = 0.06, 0.08, and 0.07, respectively). Administering miglitol on day 2 resulted in GINF_75–105 values for PORadj2 and PORups2 not significantly different from those for CON (P > 0.43).

As observed previously (17), CON demonstrated significant Fos expression in the NTS, DMX, and AP in response to hypoglycemia (Fig. 3). Animals exposed to HYPO demonstrated a significant suppression in FOS expression for the DMX (↓73%; P < 0.0001), NTS (↓27%; P < 0.05), and AP (↓34%; P < 0.05). Mean Fos expression in the DMX remained suppressed with day 1 miglitol administration for PORups1 and PORadj1 (P < 0.0005) but was not significantly different from CON following day 2 miglitol administration (P > 0.49). For the NTS, miglitol administration, day 1 or day 2, resulted in FOS values not significantly different from CON (P > 0.39). The AP demonstrated evidence of FOS suppression in PORadj1 (P = 0.09), but this failed to meet the standard criteria for statistical significance, and all other groups were not significantly different from CON (P > 0.33).

As in experiment 1, blood glucose was lowered over 75 min to a nadir of 2.48 ± 0.02 mmol/L, thereafter sustained to minute 105 on both day 1 and day 2 with no significant differences between treatment groups (Fig. 4A). Hypoglycemia on day 1 resulted in elevations above basal of >50-fold for EPI and 8-fold for NE in both POR-0.48 and POR-0.96, with no significant differences between groups (P ≥ 0.95). POR-0.48 failed to restore EPI, which remained suppressed by 50% compared with day 1 (Fig. 4B) (P = 0.0018), while POR-0.96 day 2 EPI values were not significantly different from day 1 (P = 0.83) and significantly elevated above day 2 POR-0.48 (↑82%; P = 0.011). Neither POR-0.48 or POR-0.96 demonstrated significant differences between day 1 and day 2 NE values (P > 0.80), nor were values significantly different between groups (P ≥ 0.77). Day 1 glucagon values were not significantly different between groups (P = 0.029). Day 2 glucagon values were significantly suppressed for POR-0.48 (↓50%; P < 0.0001), while not significantly different from day 1 for POR-0.96 (P = 0.07), which were significantly elevated compared with day 2 POR-0.48 (↑71%; P = 0.0004). Consistent with hormonal responses, the mean GINF rate for POR-0.48 was significantly elevated during the sustained hypoglycemic nadir (GINF_75–105; P < 0.015) on day 2 versus day 1, while POR-0.96 demonstrated no such difference (Fig. 4F).

As in experiments 1 and 2, blood glucose was lowered over 75 min to a nadir of 2.49 ± 0.03 mmol/L, thereafter sustained to minute 105, with no significant differences between treatment groups (Fig. 5A). Hypoglycemia on day 1 resulted in elevations above basal of >42-fold for EPI and 8-fold for NE for all three groups, with no significant differences between groups (P > 0.66) (Fig. 5B). As expected, Oral-0 demonstrated a significant suppression on day 2 when compared with day 1 for both EPI (↓87%; P < 0.0001) and NE (↓40%; P = 0.0036). Oral-40 and Oral-60 also demonstrated a significant suppression in EPI for day 2 when compared with day 1 (P < 0.02) but were significantly elevated on day 2 when compared with Oral-0 (P < 0.0001) (Fig. 5C). Neither Oral-40 or Oral-60 demonstrated any significant difference in NE between day 1 and day 2 (P > 0.44). A significant suppression in glucagon was observed between day 1 and day 2 for both Oral-0 (↓45%; P = 0.0025) and Oral-40 (↓40%; P = 0.016), but not Oral-60 (Fig. 5D). However, both Oral-40 and Oral-60 demonstrated significantly greater day 2 values when compared with Oral-0 (↑44%; P < 0.01). Mean GINF rates for all groups were not significantly different during the sustained hypoglycemic nadir (GINF_75–105) on day 1. However, consistent with the hormonal response, the day 2 GINF values were significantly elevated for Oral-0 (P ≤ 0.0007), but not for either of the groups receiving oral miglitol (Fig. 5E).

We have previously demonstrated that HAAF, induced by a single prior bout of hypoglycemia, can largely be ameliorated by selectively normalizing portal vein glycemia during the antecedent bout of hypoglycemia (14). These findings suggest that PMV glucosensors may be an early casualty in the development of HAAF, one that might be addressed by targeting the putative portal vein glucose receptor SGLT3. Toward that end, we examined the impact of infusing miglitol, a potent SGLT3 agonist, via the PMV on the CRR to hypoglycemia following an antecedent bout of hypoglycemia. As expected, exposing animals to HYPO significantly suppressed the EPI (↓68%), NE (↓61%), and glucagon (↓47%) responses to day 2 hypoglycemia. As shown in Fig. 2, PMV infusion of miglitol during the antecedent bout of hypoglycemia (day 1) had no significant effect on any of the counterregulatory hormonal responses to a subsequent bout of hypoglycemia on day 2, which remained suppressed (Fig. 2). In contrast, PMV infusion of miglitol during the second hypoglycemic episode (POR2) fully restored all three counterregulatory hormone responses to control levels on day 2. Consistent with the restoration of CRRs, GINF initially elevated by antecedent hypoglycemia was restored to control values with day 2 administration of miglitol, indicative of restored endogenous glucose production. We subsequently examined the impact of lowering the miglitol dose to 0.96 and 0.48 μmol/kg/min, values approaching those currently prescribed for type 2 diabetes, adjusting for species differences (i.e., rats vs. humans). While infusion of 0.96 μmol/kg/min proved effective in restoring hormonal responses following antecedent hypoglycemia, hormonal responses remained suppressed for POR-0.48, suggesting the minimal effective dose lies somewhere between these two infusion rates. Perhaps more promising, oral administration of miglitol (i.e., the standard method of administration) proved effective in restoring hormone responses on day 2 at concentrations comparable to portal infusions in rats. These data suggest that miglitol may prove an effective intervention in treating HAAF when administered following an initial bout of hypoglycemia.

While our data demonstrated the efficacy of PMV miglitol administration in restoring the CRR, they do not support the PMV SGLT3 as the critical site of action. Unlike glucose infusion, which is only effective when infused upstream from the liver, thereby perfusing the PMV, the infusion of miglitol appeared equally effective whether administered upstream or adjacent to the liver. Further, miglitol was not effective when administered during the initial hypoglycemic episode (day 1), as was the case previously for glucose, suggesting that the mechanism underlying miglitol’s effect in restoring CRRs must be different. That miglitol is not acting through the SGLT3 is further supported by the observation that Fos expression in the hindbrain in response to hypoglycemia was restored by miglitol, an SGLT3 agonist. This would be consistent with the work of Aljure and Diez-Sampedro (22) and Barcelona et al. (23), who have shown that, unlike human SGLT3, the murine form of SGLT3 fails to induce membrane currents at pH 7.4 in the presence of glucose or 1-deoxynojirimycin (DNJ), from which miglitol is derived. While they did not specifically examine the sensitivity of rat SGLT3 to DNJ, they did demonstrate similarities in glucose responsiveness for the murine form of SGLT3a and rat SGLT3a, the latter being the dominant form of SGLT3 in the PMV (15).

The hindbrain Fos expression data in the current study (Fig. 3) are consistent with the rescue of PMV glucose sensing, as we have previously shown that Fos expression in the hindbrain during slow-onset hypoglycemia is dependent on PMV glucose sensory input (17) and that glucose sensing in the PMV is impaired by antecedent hypoglycemia (13). As would be predicted, antecedent hypoglycemia led to a suppression in Fos expression in the three regions of the hindbrain (DMX, NTS, and AP) dependent on PMV input for hypoglycemic detection (Fig. 3), while miglitol administration on day 2 led to full restoration of Fos expression in all three regions during the subsequent bout of hypoglycemia. While the mechanism by which miglitol rescues the CRR and hindbrain Fos expression following antecedent hypoglycemia remains elusive, miglitol does demonstrate a number of other properties beyond that of a human SGLT3 agonist. Miglitol is a potent α-glucosidase inhibitor, which is currently prescribed to diminish the rate of postprandial glucose absorption via inhibition of intestinal membrane-bound α-glycoside hydrolase enzymes. More recently, miglitol has been reported to reduce myocardial infarct size and inhibit myocardial glycogenolysis (24), enhance β3-adrenergic signaling, and upregulate UCP1 in brown adipose tissue (25), as well as reduce IL-6 levels in obese subjects (26). The related compound miglustat is a potent inhibitor of glucosylceramide synthase used to suppress glycosphingolipid synthesis in the treatment of both Gaucher disease and Niemann-Pick disease type C. Whether any of these properties are related to miglitol’s action in rescuing the CRR is unclear, but miglitol and related DNJ compounds are clearly complex molecules for which functional properties remain to be fully elucidated.

Some caution is warranted in the interpretation of the current findings given they are restricted to rodents, and translation to humans remains to be demonstrated. The rat has proven a valuable model in studies elucidating mechanisms of impaired counterregulation following bouts of antecedent hypoglycemia (27,28). In response to antecedent hypoglycemia, rats demonstrate suppressed counterregulatory hormonal responses and endogenous glucose production, similar to humans (14,29,30). This has been noted both in normal rodents as well as various diabetic models (31,32). That said, there are some significant species differences in the response to hypoglycemia, notably the magnitude of the EPI response, which is 2–10 times greater for rats versus humans (14,17,30,32–34). Further, the magnitude of the suppression shown in the current study, 45–65% for the CRRs, is greater than the 35–50% reported in human studies (33,34). It should also be noted that the vast majority of rodent studies involve multiple (recurrent) bouts of hypoglycemia, with most human studies using two bouts of hypoglycemia the day before testing. While such protocols may mimic conditions experienced by patients with diabetes or even be required to demonstrate certain HAAF-related impairments, our work and that of others suggest the impaired counterregulation that characterizes HAAF can be demonstrated with a single or limited antecedent bouts of hypoglycemia, presumably an early event in the etiology of HAAF (14,33–35). Extending the current studies to diabetic animal models will be an important next step, though it is clear that no current animal model fully reflects the pathology of human type 1 diabetes (36,37).

In summary, miglitol, when administered i.v. or orally during a second bout of hypoglycemia 24 h following the initial hypoglycemic event, effectively restored both the counterregulatory hormonal responses to hypoglycemia and the GINF required to maintain hypoglycemia in rats. As such, miglitol is unique among pharmacological agents shown to be effective in preventing or correcting HAAF, with all others requiring either administration during the antecedent event (6–8) or chronic administration (4,5). While the application of miglitol would require knowledge of the initial hypoglycemic event, such information should become more readily available with the increased use of continuous glucose monitoring. However, as a proof-of-concept, these findings suggest miglitol might be administered as a “day-after” pill, restoring the CRRs, thereby minimizing the chances of a second severe hypoglycemic event. In theory, interrupting subsequent bouts of hypoglycemia would not only diminish the direct impact of hypoglycemia, but also potentially reduce more serious morbidities associated with recurrent hypoglycemia (e.g., impaired brain function, hypoglycemic unawareness, and enhanced risk for severe hypoglycemia) (38).

Funding. This study was funded by JDRF grant 3-SRA-2017-486-S-B.

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

Author Contributions. A.J.J. conducted experiments, analyzed samples, compiled the data, and assisted in writing the manuscript. M.W. conducted experiments, analyzed samples, and compiled data. C.M.D. designed the experiments, compiled data, conducted statistical analyses, and wrote the manuscript. C.M.D. is the guarantor of this work and, as such, had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Prior Presentation. This study was presented in abstract form at the 79th Scientific Sessions of the American Diabetes Association, San Francisco, CA, 7–11 June 2019.