Acute Sulfonylurea Therapy at Disease Onset Can Cause Permanent Remission of K_ATP-Induced Diabetes

Mice carrying the GOF Kir6.2[K185Q, ΔN30] mutant transgene were generated previously (21). By crossing these mice with Pdx-Cre (22), we generated pancreatic β-cell–specific DTG mice. Both wild-type and single transgenic mice were used as controls since these were previously shown to have normal blood glucose levels and insulin secretion (21). DTG mice 8 weeks of age received five consecutive daily doses of tamoxifen (50 mg/g body wt; experimental days 0–4). Blood glucose measurements were taken daily using a Glucometer Elite XL. The limit of detection was 600 mg/dL, and glucose at or above this level is recorded as 600 mg/dL but considered to be a lower limit of the true value. Cre-mediated recombination leads to the expression of the Kir6.2[K185Q,ΔN30] protein but also to the expression of green fluorescent protein (GFP) via internal ribosomal entry site; therefore GFP can be used to detect successful recombination and track transgene expression.

All experiments were performed in compliance with the relevant laws and institutional guidelines and were approved by the Washington University Animal Studies Committee. To test the effect of aggressive SU treatment in neonatal diabetes, mice were subjected to two independent experimental groups: 1) by injecting high doses (0.5 g/kg i.p.) of glibenclamide for 6 consecutive days before the onset of diabetes and 2) by adding glibenclamide (Sigma-Aldrich) to the drinking water (160 mg/L, or reduced accordingly) to titrate the dose required in each individual. Glibenclamide was prepared by dissolving 25 mg in 500 μL of DMSO, which was then diluted in 150 mL of water to a final concentration of 167 mg/L (337 μM) in 0.2% DMSO, with no visible precipitation. For the latter experiments, each animal was caged individually; the drug was freshly prepared every day. For SU injection, the drug was added to PBS buffer solution (Ambion). Similar numbers of males and females were used for these experiments.

Tail blood was assayed for glucose content using Glucometer Elite Xl meter (Bayer Corporation, Elkhart, IN). Plasma insulin was measured using a rat insulin enzyme-linked immunoabsorbent assay kit (ELISA kit).

Intraperitoneal glucose tolerance test was performed in 12-h fasted mice by injection of a bolus of glucose (1.5 g/kg body wt) at day 70 after the first dose of tamoxifen for induction of the disease. Blood was taken at different times (as indicated in Fig. 2) and assayed for glucose content as above. Insulin tolerance test was done on mice after a 6-h fast. Animals were injected intraperitoneally with insulin (0.5 units/kg), and blood was isolated from the tail vein at times indicated in Fig. 3 and assayed for glucose as described above.

Weight-matched wild-type and TG mice were anesthetized with isoflurane and catheters (MRE 025; Braintree Scientific Inc., Braintree, MA) implanted into both the right internal jugular vein and left femoral artery; mice were allowed to recover for 5–7 days. After a 5-h fast, catheters were flushed with saline, and heparin (20 units/kg) was administered to maintain catheter patency. Insulin (4 mU/kg/min) in saline containing 0.1% BSA was infused through the venous catheter (2 μL/min) using a Harvard-11 pump. At 5- to 10-min intervals, blood (100 μL) was removed from the arterial catheter for determination of blood glucose levels using Contour TS glucometer. Dead-space blood was then reinfused into the animal. Dextrose (25%) was infused through the venous catheter at a rate sufficient to maintain a plasma glucose level of 100 to 110 mg/dL.

Mice were anesthetized with isoflurane (0.2 mL) and killed by cervical dislocation. The bile duct was cannulated and perfused with Hanks’ balanced salt solution (Sigma-Aldrich) containing collagenase (0.3 mg/mL, Collagenase Type XI; Sigma-Aldrich). Pancreata were removed and digested for 7 min at 37°C, hand shaken, and washed three times in cold Hanks’ solution. Islets were isolated by hand under a dissecting microscope, and pooled islets were maintained overnight in CMRL-1066 (5.6 mmol/L glucose) culture medium (GIBCO) supplemented with FCS (10%), penicillin (100 units/mL), and streptomycin (100 μg/mL).

After overnight incubation in CMRL-1066 medium containing 5.6 mmol/L glucose, islets (10/well in 12-well plates) were preincubated in glucose-free CMRL-1066 plus 3 mmol/L glucose for half an hour and then incubated for 60 min at 37°C in CMRL-1066 plus varying glucose, 1 μM glibenclamide, or 30 mmol/L KCl, as indicated. After the incubation period, the medium was removed and assayed for insulin release. Islets were disrupted using ethanol-HCl extraction and sonicated on ice before estimation of insulin content. Experiments were repeated in triplicate. Rat insulin radioimmunoassay according to manufacturer’s procedure (Millipore, St. Charles, MO) was used to determine insulin secretion and content.

Data are presented as mean ± SEM. Differences among groups were tested using ANOVA and post hoc Duncan test. When only two groups were compared, unpaired t tests were used to assess significance. Differences were assumed to be significant in each case if P < 0.05, and nonsignificant (ns) differences are indicated.

To induce transgene expression, 2-month-old Pdx-Cre/Kir6.2[K185Q,ΔN30] DTG and littermate control mice were injected daily for 5 consecutive days with tamoxifen (50 μg/g body wt). Without SU treatment, DTG mice develop severe diabetes within 2 weeks after induction, whereas single transgenics and wild-type littermates are unaffected (21) (Fig. 1A). The clinical observation that SU dose requirement typically declines with time in NDM patients provides a hint that some secondary compensatory changes may be occurring. We demonstrated previously two key relevant features of disease progression in DTG (21). First, chronic SU treatment (implanted slow-release glibenclamide pellets) provides long-term protection against hyperglycemia, as in the human disease. Second, protection against hyperglycemia either by this treatment or by transplantation of exogenous islets preserves insulin content of endogenous islets, but, without protection, insulin content is lost, and subsequent administration of SU is ineffective. This leads us to hypothesize that part of the drug sensitivity variability in human NDM is a result of long-term poor control of the diabetes and that aggressive early therapy is critical for maintaining islet insulin.

To gain further insight into SU sensitivity in the disease, DTG animals were initially treated daily with a high dose intraperitoneal injection of glibenclamide (0.5 g/kg) for 6 consecutive days, during and immediately after tamoxifen induction (Fig. 1B). For the period of glibenclamide injection the rise in blood glucose was similarly suppressed in all treated animals (Fig. 1C). However, the subsequent response of DTG animals fell into two clear groups. In the first, blood glucose was maintained normal during SU treatment, but rose and did not fall once SU treatment was terminated, as expected for persistent expression of the GOF mutation (i.e., PNDM). In this noncured group, 24/34 (∼70%) animals developed severe diabetes; blood glucose was >500 mg/dL by 14 days and stayed high until approximately day 70 when the animals were killed (Fig. 1B, blue), comparable with untreated DTG animals (Fig. 1A). It is noteworthy that, however, ∼30% (10/34) were apparently cured by this early treatment, maintaining blood glucose <200 mg/dL for ∼70 days after treatment was terminated until being killed, without any further intervention (Fig. 1B, red); i.e., this group exhibited a chronic remission, essentially a cured phenotype.

To further examine the implications of this important finding (that early aggressive SU treatment can essentially cure the disease and provide subsequent prolonged normoglycemia without further therapy), an additional group of DTG animals was treated orally with glibenclamide from disease onset, and then the drug dose was titrated daily to attempt to maintain stable blood glucose levels (Fig. 2A). Glibenclamide was initially dissolved in DMSO and then added to the drinking water at the highest concentration used for these experiments (160 mg/L). Glucose was measured daily, and the drinking water dose for each animal was then adjusted according to the following regimen: glibenclamide was initiated at the highest dose (160 mg/L) as soon as the blood glucose was >200 mg/dL and kept at this concentration for 4 days, independently of the blood glucose level. Subsequently, starting on day 5, if blood glucose was >250 mg/dL, the dose was restored to the preceding dose; if blood glucose was 200–250 mg/dL, the dose was reduced by half; if glucose was <200 mg/dL, glibenclamide was removed from the water.

Essentially the outcome in the orally treated animals was the same as observed in animals injected with glibenclamide; ∼60% (6/9) developed severe diabetes by day 14 (noncured), with blood glucose levels of >500 mg/dL, even though glibenclamide dose was not stopped or lowered. In the orally treated group (cured; 3/9), glibenclamide dose requirement declined, and in each case the drug was eventually removed permanently (between days 20 and 40) (Fig. 2A). Most strikingly, after glibenclamide removal, these cured animals all maintained blood glucose <200 mg/dL for more than 60 days (Fig. 2A). All mice drank ∼10 mL daily, implying a maximum oral dose of ∼1.67 mg/25 g/day. This is considerably lower than the daily dose received by injection in Fig. 1 and may have been insufficient to stop the development of hyperglycemia in the noncured subset, even during treatment.

These two drug trials reveal that in a subset of animals (cured) the initial pathology that follows onset of K_ATP overactivity can actually remit, since affected islets must subsequently release sufficient insulin to maintain normal blood glucose levels. Why only a subset of animals remits is unclear. Retrospective analysis reveals no difference in weight (20.4 ± 0.82 g, n = 10; vs. 20.1 ± 0.41 g, n = 24; cured vs. noncured) or blood glucose (129 ± 4.68 mg/dL vs. 127 ± 1.14 mg/dL, cured vs. noncured) at the time of induction. As shown in Fig. 2B, the level of transgene expression (assessed by GFP fluorescence) in islets from cured DTG mice is at least as high as in islets from noncured DTG mice, indicating that glibenclamide did not reduce the efficiency of Cre-recombination and transgene expression, nor survival/replication of nonrecombinant β-cells.

Consistent with severe diabetes, plasma insulin is dramatically reduced in untreated DTG mice, but, strikingly, cured DTG mice treated for only a short term with SU maintained nearly normal glycemia reflecting a sufficient, albeit reduced, circulating level of insulin to halt the development of diabetes (Fig. 1D). All DTG mice showed lower plasma insulin after SU treatment was terminated, but this was more dramatic in the noncured mice (Fig. 1D).

Mice underwent glucose tolerance testing at days 57 and 60 after tamoxifen induction, ∼6 weeks after SU treatment was terminated. Mice were intraperitoneally injected with glucose after a 12-h fast. Noncured DTG mice show a dramatic impairment in glucose tolerance, with fasting blood glucose of >500 mg/dL. It is noteworthy that, however, cured mice at days 57 and 60 show fasting blood glucose of ∼220 mg/dL and are only mildly glucose-intolerant compared with control littermates (Fig. 2C).

Isolated islets from noncured DTG mice that develop severe diabetes show negligible insulin secretion not only in response to glucose, but also to glibenclamide, or high [K⁺] (Fig. 2D), consistent with the drastic reduction of insulin content (Fig. 2E). However, even though they still show no significant glucose-dependent secretion (predicted given the constitutive expression of overactive β-cell K_ATP channels), islets from cured DTG animals show enhanced insulin secretion in response to glibenclamide, and to high K⁺ depolarization, reflecting preservation of insulin content long after SU treatment is terminated (Fig. 2D and E).

Because there is still very little glucose stimulation of insulin secretion in cured mice, we examined the possibility that normoglycemia results from greater insulin responsiveness. We determined the glucose-lowering effect of intraperitoneal insulin (0.5 units/kg) on 6-h fasted mice. Noncured mice do not show any obvious lowering of blood glucose, even though absolute levels of blood glucose are much higher than in control mice. Conversely, cured mice show a prolonged drop in blood glucose levels, indicating a relative increase in insulin sensitivity with respect to control littermates (Fig. 3A).

To more directly determine peripheral insulin sensitivity in cured DTG mice, hyperinsulinemic-euglycemic clamps (33) were performed in control and in cured DTG mice. The lack of glucose-stimulated insulin secretion is reflected in the higher overshoot of blood glucose in response to onset of infusion during the establishment of the clamp (Fig. 3B), but once comparable, stable glucose levels are attained (∼60 min), the rate of glucose disposal is markedly higher (by ∼30%) in cured DTG mice compared with control littermates.

GOF mutations in the pore-forming Kir6.2 and the regulatory SUR1 subunits of the K_ATP channel are major causes of both PNDM and remitting-relapsing TNDM in humans (4,8,13,15,25,34,35). SU directly inhibit the K_ATP channel, thereby promoting insulin secretion independently of the metabolic state of the cell. The identification of K_ATP mutations as causal in NDM led straight to the realization that SU could be an effective treatment. Now many PNDM and TNDM patients have been successfully transferred from insulin injections to oral SU therapy, frequently demonstrating improved glycemic control (26,32,36–38). However, successful transfer has not proved to be possible in all PNDM cases; in general, transferability correlates negatively with the age of the patient and the severity of the disease (12,31,39). There are several examples of successful transfer of an infant to relatively low SU doses, with the parent afflicted with the same mutation being less responsive (19,26,38). It is also clear that either PNDM or TNDM can result from the same K_ATP mutation (9). These disparities imply that additional genetic or nongenetic factors must contribute to the determination of disease severity and treatability.

We previously demonstrated development of severe diabetes in inducible mouse NDM, in which we express GOF K_ATP channels specifically in pancreatic β-cells (21). Similarly, Girard et al. (40) demonstrated development of diabetes in mice expressing the NDM-associated GOF Kir6.2 V59M mutation using the same approach. It is noteworthy that we identified unpredicted and previously unrecognized secondary consequences of systemic diabetes in these mice: deterioration of pancreatic architecture, with drastic loss of islet insulin content and β-cell mass (21,40). These consequences underlie the lack of insulin secretion in response to glibenclamide, or to high K⁺ depolarization from these islets, at late stages of the disease (21). It is equally noteworthy that maintenance of normoglycemia, achieved either by syngeneic islet transplantation or by chronic SU therapy (glibenclamide pellet implantation, drug present for 90 days), not only completely avoided hyperglycemia, but also prevented the development of the secondary consequences (21). Isolated islets from chronically SU-treated animals retained normal insulin secretion in response to high K⁺, and near-normal response to glibenclamide, yet glucose-dependent secretion above basal was essentially absent, consistent with the expected consequences of GOF K_ATP channels (21). It is noteworthy that, however, if SU therapy was initiated after the disease had developed (blood glucose >300 mg/dL), this treatment was ineffective, a result of the systemic diabetes leading to loss of β-cell mass and insulin content (21). We speculate that exposure to episodic hyperglycemia in human NDM may lead to similar secondary consequences, reflected in the demonstration that ability to transfer to SU in members of a single family carrying the same identified K_ATP mutation declines with duration of the disease, and in some cases this has proved to be completely ineffective (14,26). Thus, our studies in mice provide an immediate potential explanation for why older NDM patients typically show reduced SU responsivity (26,38).

The variable disease outcome in our experiments and dependence of efficacy of SU treatment on the time of therapeutic intervention in human NDM open the possibility that remission of NDM might actually be possible with early aggressive SU treatment. We demonstrate the striking finding that a short period (6 days) of intraperitoneal glibenclamide treatment at disease onset can indeed provide sustained remission from what is otherwise permanent worsening diabetes in mice. Plasma insulin levels remain below normal after glibenclamide therapy is terminated, but they are sufficient to maintain near normoglycemia and to preserve islet insulin content. Maintained islet insulin secretion in response to SU and KCl indicates that the secretory machinery itself is not affected in cured DTG mice. Oral glibenclamide treatment, with the drug added to the drinking water, also prevented diabetes in a responding subset of animals, which were weaned from glibenclamide in ∼4 weeks and subsequently maintained near normoglycemia for an indefinite period without further treatment. Thus aggressive control with glibenclamide only at the onset of disease can lead to subsequently sustained, drug-free recovery from hyperglycemia in these NDM mice.

Pancreatectomy in rats leads to development of uncontrolled diabetes with a similar time course to that we observe after induction of GOF K_ATP expression (41). With partial (90%) pancreatectomy, rats demonstrate a marked dichotomy of response; although glucose levels are quite variable in early stages after pancreatectomy, by 10 weeks all rats either show nearly normal glycemia or are overtly diabetic (41). This progression is strikingly similar to the bifurcated progression that we observe in response to early glibenclamide treatment. In both cases, the divergent outcomes indicate an apparent threshold effect at glucose levels of ∼300 mg/dL; if glucose rises above this level, β-cell deterioration and loss of insulin content ensues, but if not, compensation and remission can occur (21,41).

In human NDM, mutant K_ATP channels will also be expressed in muscle, brain, and other tissues, which may complicate the outcome, but it is conceivable that early aggressive SU treatment might also lead to stable remittance in human PNDM. We have recently reported a family case of NDM as a result of Kir6.2[R201H] mutation in which the carrier patient was treated from the time of diagnosis (day 6) with the SU glyburide (0.2 mg/kg/day) (38). By day 270, the drug dose had been reduced to 0.017 mg/kg/day with HbA_1c in the normal range. Strikingly, the proband is currently receiving a 75-fold lower dose of SU than his affected mother and sister, who had been treated with insulin for 26 and 6 years, respectively (38). It is noteworthy that the proband dose is lower than that reported for any of 44 NDM patients who successfully switched from insulin to SU therapy (26) and considerably lower than the standard dose used to treat patients with type 2 diabetes. Potentially the dose is actually subtherapeutic at this juncture, and the disease has essentially entered remission as a result of the early aggressive treatment.

What is the compensatory mechanism behind an SU dose requirement and disease outcome? In TNDM patients, the mechanism of remittance is unknown but has been proposed to be a result of either a reduced insulin requirement at the time of remission or to some compensation at the level of the β-cell, pancreas, or whole body that can overcome the secretory defect (13). It may reflect an increase in insulin sensitivity, a compensatory increase in β-cell function, or both. Here, we demonstrate that cured DTG mice do have higher insulin sensitivity relative to control mice. An unanswered question is why do we only see ∼30% of the animals being cured? It is possible that individual-specific variation in other genes is involved in the compensation, but this will require extensive in-breeding experiments to test.

In humans, the Kir6.2 polymorphism E23K is highly associated with type 2 diabetes (42,43). Like NDM mutations, the E23K mutation induces a weak K_ATP GOF phenotype and is therefore predicted to reduce insulin secretion in vivo. Nondiabetic human subjects homozygous for E23K demonstrate ∼40% lower insulin secretion in response to oral and intravenous glucose (44). However, glucose tolerance is normal, and hyperinsulinemic-euglycemic clamps reveal that these individuals are more insulin sensitive than control subjects. Thus, greater insulin sensitivity in individuals with E23K or other K_ATP GOF mutations may provide a general compensatory mechanism for reduced insulin secretion. Conceivably, this may underlie the remission that characterizes K_ATP-dependent TNDM, which, in the face of environmental or dietary challenges, becomes insufficient. Reduced insulin secretion may then drive development of type 2 diabetes in E23K carriers (44) or subsequent relapse of TNDM patients.

This work was supported by National Institutes of Health grants (DK-69445) to C.G.N. and (DK-064572) to P.W.H. and a Diabetes Research Training Center Grant (DRTC, 5P60 DK020579) to M.S.R.; reagent support was provided by the Washington University DRTC NIH P60 (DK-020579). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

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

M.S.R. designed the study, carried out the experiments, and wrote the manuscript. S.E.A. and A.K.V. carried out the experiments. P.W.H. designed the study. C.G.N. designed the study and wrote the manuscript.

The authors thank Theresa M. Harter (Department of Cell Biology and Physiology, Washington University School of Medicine) for assistance with mouse breeding, maintenance, and genotyping and Maria Payne (Department of Pediatrics, Washington University School of Medicine) for assistance in performing hyperinsulinemic-euglycemic clamps. The authors also thank Maureen Gannon (Department of Molecular Physiology and Biophysics, Vanderbilt University, Nashville, TN) for providing the Pdx-Cre mice.