OBJECTIVE—We sought to examine the molecular mechanisms underlying permanenent neonatal diabetes mellitus (PNDM) in a patient with a heterozygous de novo L225P mutation in the L0 region of the sulfonylurea receptor (SUR)1, the regulatory subunit of the pancreatic ATP-sensitive K+ channel (KATP channel).
RESEARCH DESIGN AND METHODS—The effects of L225P on the properties of recombinant KATP channels in transfected COS cells were assessed by patch-clamp experiments on excised membrane patches and by macroscopic Rb-flux experiments in intact cells.
RESULTS—L225P-containing KATP channels were significantly more active in the intact cell than in wild-type channels. In excised membrane patches, L225P increased channel sensitivity to stimulatory Mg nucleotides without altering intrinsic gating or channel inhibition by ATP in the absence of Mg2+. The effects of L225P were abolished by SUR1 mutations that prevent nucleotide hydrolysis at the nucleotide binding folds. L225P did not alter channel inhibition by sulfonylurea drugs, and, consistent with this, the patient responded to treatment with oral sulfonylureas.
CONCLUSIONS—L225P underlies KATP channel overactivity and PNDM by specifically increasing Mg-nucleotide stimulation of the channel, consistent with recent reports of mechanistically similar PNDM-causing mutations in SUR1. The mutation does not affect sulfonylurea sensitivity, and the patient is successfully treated with sulfonylureas.
ATP-sensitive K+ channels (KATP channel) consist of four pore-forming Kir6.2 subunits (KCNJ11) and four regulatory sulfonylurea receptor (SUR)1 subunits (ABCC8) (1,2). ATP inhibits the channel by directly interacting with Kir6.2 (3). SUR1, an ATP-binding cassette (ABC) transporter, mediates channel stimulation by Mg-nucleotide interaction with its two nucleotide binding folds (NBF1 and NBF2) (3) and high-affinity sulfonylurea inhibition via interaction with the SUR1 core (4,5). In addition to the transporter core, SUR1 possesses an NH2-terminal transmembrane domain, TMD0, connected to the SUR1 core via the cytoplasmic loop L0 and interacting directly with Kir6.2 (6,7).
Activating mutations in the pancreatic KATP channel lead to β-cell underexcitability and suppression of insulin secretion and underlie permanent and transient forms of neonatal diabetes mellitus (8,9), with the most severe forms associated with extrapancreatic symptoms (Developmental Delay, Epilepsy, and Neonatal Diabetes Syndrome) (10). While the majority of neonatal diabetes mellitus (NDM)-associated KATP mutations are localized to Kir6.2, recent studies have now described such mutations in SUR1 (11,12). Here we describe a patient with permanent neonatal diabetes mellitus (PNDM) with a novel heterozygous SUR1 mutation (L225P) located in L0. L225P causes increased stimulation of recombinant KATP channels by Mg nucleotides and increased activity in the intact cell. The effects of L225P are abolished by mutations in the NBFs of SUR1 that prevent nucleotide hydrolysis, indicating that L225P likely acts on the transduction mechanism that links nucleotide hydrolysis at the NBFs to channel gating.
RESEARCH DESIGN AND METHODS
Genetics and molecular biology.
Genomic DNA was prepared from peripheral blood samples from the patient and both parents, and candidate genes were sequenced. The 39 exons of SUR1 (ABCC8) were directly sequenced (13), and SUR1 cDNA and protein sequences were numbered as previously described (14). Two mutations were found in ABCC8: t674c, predicted to cause an L225P amino acid change, was de novo, and g2638a, predicted to cause a D880N amino acid change, was maternally transmitted. Mutations were engineered into hamster SUR1 cDNA (GenBank accession no. L40623) (15) by two-step PCR and confirmed by direct sequencing. The study protocol was reviewed and approved by the institutional review board of The Children's Hospital of Philadelphia, and written informed consent was obtained from the parents of the patient.
COSm6 cells were transfected with mouse Kir6.2 cDNA (GenBank accession no. D50581) (16) plus SUR1, as described (17). To recapitulate the heterozygous state, cells were transfected with 1:1 wild-type:L225P cDNA. Membrane patches from transfected cells were voltage clamped with an Axopatch 1-D amplifier (Axon Instruments, Union City, CA). Currents were measured at −50 mV membrane potential and digitized with a Digidata 1322A board (Axon Instruments). Bath and pipette solutions (KINT) contained (in mmol/l) 150 KCl, 10 HEPES, and 1 EGTA (pH 7.4). ATP and ADP were added as dipotassium salts. Where indicated, MgCl2 was added to the bathing solution to a calculated Mg2+free of 0.5 mmol/l. Offline analysis was performed using pClamp (version 8.2; Axon Instruments) and Microsoft Excel.
Open probability in the absence of nucleotides (Po,zero) was estimated from stationary fluctuation analysis of macroscopic currents in 0 ATP and in 10 mmol/l ATP, as described (17). The ATP dose response in the absence of Mg2+ was quantified by fitting the data with a Hill equation (17). The tolbutamide dose response was quantified by fitting the data with a double-Hill equation accounting for high- and low-affinity fractions (5).
Macroscopic 86Rb+ efflux assays.
Under basal conditions or metabolic inhibition (1 mmol/l 2-deoxy-d-glucose and 2.5 μg/ml oligomycin), 86Rb+ efflux from transfected COS cells was measured as previously described (17). The rate constant for KATP-specific 86Rb+ efflux (k2) was obtained by fitting the data with a double-exponential equation (17). The glibenclamide dose response was quantified by fitting with a Hill equation. Results are presented as means ± SE. Statistical tests and P values are noted in figure legends.
RESULTS AND DISCUSSION
A neonatal diabetes patient with a L225P mutation in SUR1.
A 6-week-old male presented with diabetic ketoacidosis. He was born at term, weighing 2,100 g (<10th percentile for gestational age) and measuring 47.5 cm. Despite a voracious appetite, he had not regained birth weight. Physical exam revealed an alert but malnourished-appearing and dehydrated baby with sunken anterior fontanelle. Serum glucose was 950 mg/dl, and bicarbonate was 11 mEq/l; ketonuria was evident. He developed transient evidence of cerebral edema, but uneventfully recovered after mannitol treatment. There was no history of diabetes in either unrelated parent. Although abdominal ultrasound suggested an abnormally small pancreas, there was no evidence of exocrine pancreatic insufficiency. Insulin treatment was continuously required from the time of initial illness at full replacement doses of ∼1 unit · kg body wt−1 · day−1. Subsequent growth was normal, without diabetic vascular, peripheral neural, or renal complications. The patient is now 8 years of age.
Direct sequencing of the patient's DNA revealed no mutations in the coding regions of PDX-1, glucokinase, or KCNJ11 (Kir6.2). However, two heterozygous point mutations were found in ABCC8 (SUR1): L225P, which was de novo (Fig. 1A), and D880N, which was maternally transmitted.
L225P channels are more stimulated by Mg nucleotides.
L225P and D880N were engineered into SUR1 to examine their functional consequences. Coexpression of wild-type Kir6.2 and SUR1 in COS cells leads to expression of KATP channels that are inhibited by ATP and stimulated by Mg nucleotides in excised membrane patch-clamp experiments (Fig. 1C and E). Channel activity in the on-cell configuration (before excision) was significantly increased by L225P but not altered by D880N (Fig. 1B). Since KATP channel overactivity may result from alterations of intrinsic gating, we measured open probability in the absence of ligands (Po,zero). Compared with wild-type channels, L225P did not alter (Po,zero) (0.55 ± 0.02 vs. 0.58 ± 0.03) nor did it alter sensitivity to inhibitory ATP in the absence of Mg2+ (half-maximal inhibitory [ATP], K1/2, was 8.2 ± 1.2 vs. 8.9 ± 1.5 μmol/l, Fig. 1D). However, L225P-containing channels were significantly more active in MgATP compared with wild type (K1/2 = 19.0 ± 2.6 vs. 67.6 ± 15.8 μmol/l, P < 0.05 by Student's t test) and were significantly more stimulated by MgADP than wild type (Fig. 1E and F), while D880N-containing channels were stimulated comparably with wild type (data not shown). These results indicate that L225P causes KATP channel overactivity by specifically increasing channel stimulation by Mg nucleotides and is therefore the likely cause of PNDM in the patient.
L225P channels are more active in the intact cell.
To examine channel activity in the intact cell in more detail, we measured macroscopic 86Rb+ fluxes from transfected cells. Consistent with expectations, L225P flux was significantly higher than wild-type flux under both basal conditions and metabolic inhibition (Fig. 2A and B). The fractional activity of L225P channels under basal conditions (relative to metabolic inhibition) was significantly increased compared with WT (Fig. 2C), indicating that L225P channels are already close to maximal activity under basal conditions. Since L225P did not significantly alter current density in excised membrane patches compared with wild type (1.3 ± 0.2 vs. 1.2 ± 0.1 pA/patch), the increased Rb fluxes measured for L225P channels must reflect differences in activity rather than in expression.
Since the patient is heterozygous for L225P, we measured Rb fluxes from cells transfected with a 1:1 mixture of wild-type and L225P cDNA. The resulting mixed WT + L225P channels exhibited fluxes that were intermediate between wild-type and homomeric L225P channels (Fig. 2A and B). The fractional activity of mixed wild-type + L225P channels was also significantly increased compared with wild type (Fig. 2C), consistent with KATP channel overactivity leading to suppression of insulin secretion and diabetes in the heterozygous patient.
L225P does not alter sulfonylurea sensitivity of the channel.
Sulfonylurea sensitivity of recombinant KATP channels was unaffected by L225P, both in excised membrane patches and in intact cells (Fig. 3B-D). Consistent with this, the patient responded to an initial trial with the oral sulfonylurea glyburide, and insulin therapy has been completely discontinued. On treatment with 15 mg/day glyburide and no insulin, premeal blood glucose levels remained nearly normal (mean 116 mg/dl [range 70–192]). However, on this dose, he displayed glucose intolerance when given large loads of carbohydrate, such as sugared drinks. No other beneficial or adverse effects of glyburide were noted.
L225P effects are dependent on intact nucleotide hydrolysis at the NBFs.
L225 is located in L0, the cytoplasmic loop linking the Kir6.2-interacting domain TMD0 to the SUR1 core (Fig. 4A). TMD0-L0 is sufficient to confer SUR1 ligand-independent channel gating to Kir6.2 (18) and is proposed to transduce Mg-nucleotide stimulation and high-affinity SU-inhibition from the SUR1 core to the Kir6.2 pore (6,7). Consistent with this, the effects of L225P on channel activity in the intact cell were abolished by combination with either of two NBF mutations (K719M or D1506A) that abolish Mg-nucleotide stimulation of the channel (19,20) (Fig. 2D). Single-particle electron microscopy of the KATP channel complex suggests that TMD0 is located in close apposition to Kir6.2 but in between two adjacent SUR1 cores (21). Thus, it is unclear whether TMD0-L0 transduces Mg-nucleotide stimulation from the SUR1 core of the same subunit or from the SUR1 core of the adjacent subunit. To address this question, we measured Rb fluxes for all permutations of wild-type, L225P, D1506A, and (L225P, D1506A) combinations (at a 1:1 cDNA transfection ratio) (Fig. 4C). We made the simple assumption that channel activity is directly proportional to the number of actively transduced SUR1 signals (see 22). Consistent with this, the activity of wild type:D1506A corresponded with (50%) of the activity observed for wild type, since D1506A activity is negligible (Fig. 4C). Similarly, the activity of L225P:D1506A corresponded with (50%) L225P activity, the activity of wild type:L225P yielded an arithmetic sum of wild-type and L225P activities ([50%] wild type + [50%] L225P), and the activity of L225P:[L225P, D1506A] yielded (50%) L225P activity (Fig. 4C).
When L225P is mixed with D1506A, however, the resulting activity should then be (50%) L225P, if stimulation occurs through the TMD0-L0 of the same subunit, or (50%) wild type if it occurs through the TMD0-L0 of the adjacent subunit (Fig. 4B). The actual result was (50%) L225P activity (Fig. 4C), consistent with the “same subunit” model. Likewise, when wild type and [L225P, D1506A] are mixed, the activity should be (50%) wild type if stimulation occurs through the same subunit or (50%) L225P if it occurs through the adjacent subunit (Fig. 4B). The result was (50%) wild-type stimulation (Fig. 4C), again consistent with TMD0-L0 transducing stimulation from the SUR1 core of the same subunit.
The lack of an effect of L225P on sulfonylurea sensitivity indicates that the mutation does not affect transduction of high-affinity sulfonylurea block to the Kir6.2 pore, suggesting that there may be structurally segregated transduction pathways in TMD0-L0 for each of these regulatory inputs from the SUR1 core.
Comparison with other NDM-causing mutations in SUR1.
Two reports have identified three SUR1 mutations associated with PNDM (F132L, L213R, and I1424V) and five associated with transient neonatal diabetes (11,12). Common to F132L and I1424V is an increased sensitivity of the channel to Mg nucleotides, such that channel overactivity results at physiological nucleotide concentrations. Our results are consistent with these previous studies, as L225P increases channel sensitivity to Mg nucleotides without altering intrinsic gating or inhibition by ATP. The effects of L225P are similar to those of the PNDM-associated I1424V mutation (12) but not as severe as those of the DEND (Developmental Delay, Epilepsy, and Neonatal Diabetes Syndrome)-associated F132L mutation (11). Although the number of NDM-associated SUR1 mutations is still relatively small, thus far it appears that disease severity correlates with severity of effects of the mutation on channel properties, as is the case with Kir6.2 mutations (10). In contrast with L225P, D880N did not affect Mg-nucleotide stimulation or on-cell channel activity. Since the patient's mother is a carrier of this mutation and does not have diabetes, nor is there a history of diabetes in her family, we can effectively rule out D880N as a cause of the disease.
Consistent with lack of effect of L225P on sulfonylurea sensitivity of recombinant channels, the patient was successfully treated with oral sulfonylureas. The carrier of the I1424V mutation (which is mechanistically similar to L225P) has also been successfully treated with sulfonyureas (12). This bolsters the likelihood that sulfonylurea therapy will be effective in NDM patients with SUR1 mutations, as is the case in most patients with Kir6.2 mutations (23).
NOTE ADDED IN PROOF
We have now sequenced SUR1 exon 5 from 50 healthy control subjects and have not found the L225P mutation in 98 alleles.
Published ahead of print at http://diabetes.diabetesjournals.org on 2 February 2007. DOI: 10.2337/db06-1746.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
This work was supported by National Institutes of Health Grants DK069445 (to C.G.N.), DK56268, RR00240, and DK019525 (to C.A.S.). We are grateful for molecular biology reagents from National Institutes of Health Diabetes Research and Training Grant DK-20579, and we thank Dr. Joseph Koster for editorial improvements.