Novel dominant KATP channel mutations in infants with congenital hyperinsulinism: Validation by in vitro expression studies and in vivo carrier phenotyping

Abstract Inactivating mutations in the genes encoding the two subunits of the pancreatic beta‐cell KATP channel, ABCC8 and KCNJ11, are the most common finding in children with congenital hyperinsulinism (HI). Interpreting novel missense variants in these genes is problematic, because they can be either dominant or recessive mutations, benign polymorphisms, or diabetes mutations. This report describes six novel missense variants in ABCC8 and KCNJ11 that were identified in 11 probands with congenital HI. One of the three ABCC8 mutations (p.Ala1458Thr) and all three KCNJ11 mutations were associated with responsiveness to diazoxide. Sixteen family members carried the ABCC8 or KCNJ11 mutations; only two had hypoglycemia detected at birth and four others reported symptoms of hypoglycemia. Phenotype testing of seven adult mutation carriers revealed abnormal protein‐induced hypoglycemia in all; fasting hypoketotic hypoglycemia was demonstrated in four of the seven. All of six mutations were confirmed to cause dominant pathogenic defects based on in vitro expression studies in COSm6 cells demonstrating normal trafficking, but reduced responses to MgADP and diazoxide. These results indicate a combination of in vitro and in vivo phenotype tests can be used to differentiate dominant from recessive KATP channel HI mutations and personalize management of children with congenital HI.

K ATP channel HI mutations can act in either recessive or dominant fashion, depending on the specific defect. Homozygous or compound heterozygous recessive K ATP channel mutations cause diffuse HI that is unresponsive to medical treatment with diazoxide and, therefore, many of these patients require near-total pancreatectomy to control hypoglycemia. Recessive K ATP channel mutations can also cause focal diazoxide-unresponsive HI when a paternally-transmitted mutation becomes isodisomic due to somatic loss of the maternal 11p imprinted region (11pUPD; de Lonlay et al., 1997). Dominant K ATP channel HI mutations are less common than recessive mutations; however, they account for all K ATP -HI cases with diazoxide-responsive HI and for approximately one-third of cases with diazoxideunresponsive diffuse K ATP -HI (Macmullen et al., 2011;Pinney et al., 2008;Snider et al., 2013). Dominant K ATP channel mutations may be difficult to distinguish from recessive mutations based on family history, because clinical manifestations in carriers of dominant mutations may fail to be recognized (Macmullen et al., 2011;Pinney et al., 2008).
Genetic diagnosis has become an essential tool for clinical management of children with congenital HI, particularly since identification of a monoallelic paternally-transmitted recessive K ATP channel mutation can accurately diagnose the presence of a focal HI lesion (Snider et al., 2013). In these cases, 18 F-DOPA PET/CT scan can be used to localize the lesion and guide curative surgical excision (Laje et al., 2013;Otonkoski et al., 2006). Unfortunately, one-third to onehalf of K ATP channel mutations detected by genetic testing are novel variants that are interpreted as variants of unknown significance. This is particularly a problem with novel missense K ATP channel variants, since these can potentially be benign polymorphisms, recessive inactivating mutations, dominant inactivating mutations, or even dominant diabetes-causing mutations (Babenko et al., 2006;Gloyn et al., 2004;Snider et al., 2013). In silico prediction software is not able to accurately predict whether novel missense K ATP channel variants are pathogenic and cannot identify the mode of inheritance (Adzhubei et al., 2010;Sim et al., 2012). The ability to categorize a novel mutation as dominant or recessive is also important for counseling families about recurrence risk and for identifying other family members at risk of hypoglycemia. Thus, characterization of novel variants and their associated phenotypes is essential for improving the interpretation of genetic tests and for accurate diagnosis of affected children. For this purpose, the present report adds six novel dominant K ATP channel mutations to the 62 mutations currently known. These novel mutations were confirmed to be dominant based both on in vitro expression studies and by in vivo phenotype tests of mutation carrier parents.

| Consent
Written informed consent was provided by all subjects or their parents. This study was approved by the Children's Hospital of Philadelphia Institutional Review Board (IRB 05-004037, IRB 07-005772, IRB 09-007046).

| Probands
Probands were identified during inpatient evaluations in the Congenital Hyperinsulinism Center at The Children's Hospital of Philadelphia between 2006 and 2018. The diagnoses of HI was based on previously described criteria: fasting hypoglycemia accompanied by inadequate suppression of plasma insulin, inappropriate suppression of plasma free fatty acid and β-hydroxybutyrate concentrations, and inappropriate glycemic response to glucagon at the time of hypoglycemia (Finegold, Stanley, & Baker, 1980;Stanley & Baker, 1976

| Family members of probands
Family members of probands were identified by parent of origin testing for genetic mutations found in the probands. Family members completed a thorough review of their hypoglycemic history, awareness of symptoms, and any treatment.

| Mutation screening
Mutation screening was performed in commercial laboratories for all probands and parents included in this study. Additional mutation analysis for other family members was performed on a research basis. Genomic DNA was isolated from peripheral blood (5 PRIME, Gaithersburg, MD) or from saliva (Oragene DNA self-collection kit; DNA Genotek, Kanata, Ontario, Canada). Coding sequences and intron/exon splice junctions were amplified and directly sequenced on an ABI 3730 capillary DNA analyzer (Applied Biosystems, Carlsbad, CA). The nucleotides of ABCC8 and corresponding SUR1 amino acids were numbered according to the sequence reported by Nestorowicz et al. (1996) that includes the alternatively spliced exon 17 sequence (NCBI accession no. L78224). The functional consequences of novel, missense mutations were predicted with bioinformatics software SIFT (Sim et al., 2012) and PolyPhen2 (Adzhubei et al., 2010). Genetic variants were searched against the gnomAD Browser (v2.1; Lek et al., 2016).

| Phenotype studies
Fasting test: Subjects were fasted for 20-24 hr during which time blood samples were taken at regular intervals to measure plasma glucose, β-hydroxybutyrate, and insulin. Fasting tests were terminated after 20-24 hr or when plasma glucose fell to <50 mg/dl (2.8 mmol/l).
Oral protein tolerance test (oPTT): Subjects consumed a protein drink containing 1.5 g/kg of protein powder (maximum 60 g; Resource Instant Beneprotein Powder, Nestlé Health Science) mixed with water and sugar-free flavoring (Sugar-free Crystal Light™, Kraft Foods, Inc.) after at least 3 hr of fasting. Plasma glucose concentrations were measured at baseline and every 30 min for 3 hr. Abnormal responses were interpreted as either a decrease of plasma glucose greater than 10 mg/dl (0.56 mmol/l) or a drop in plasma glucose to below 70 mg/dl (3.9 mmol/l; Fourtner, Stanley, & Kelly, 2006;Hsu et al., 2001).
Oral glucose tolerance test (oGTT): Subjects consumed a glucose drink containing 1.75 g/kg of glucose (maximum 75 g) after at least 3 hr of fasting (NERL™ Trutol™, ThermoFisher Scientific™, East Providence, RI). Blood samples were taken at baseline and 30 min for 3-4 hr. Results were compared to published norms for adults (American Diabetes, 2010). Abnormal response to glucose was interpreted as a peak glucose >150 mg/dl (8.3 mmol/l). Immunoblotting: COSm6 cells grown in 6-well culture plates were transfected with 0.6 μg human SUR1 and 0.4 μg human Kir6.2 cDNA (~1:1 M ratio) per well using FuGENE ® 6 (Roche Applied Science, Penzburg, Germany). Cells were lysed in 20 mM Hepes, pH 7.0/5 mM EDTA/150 mM NaCl/1% Nonidet P-40 with cOmplete™ protease inhibitors (Roche Applied Science, Penzburg, Germany) 48-72 hr post-transfection. Proteins were separated by SDS/PAGE (8%), transferred to nitrocellulose, analyzed by a rabbit anti-SUR1 antibody recognizing the C-terminal 13 amino acids of SUR1  followed by horseradish peroxidase (HRP)-conjugated anti-rabbit secondary antibodies (GE Healthcare, Chicago, Illinois), and visualized by enhanced chemiluminescence (Super Signal West Femto; Pierce, ThermoFisher Scientific, Waltham, MA). 86 Rb + efflux assay: Response of channels to metabolic inhibition was assessed by 86 Rb + efflux assays. COSm6 cells transfected with human SUR1 and Kir6.2 cDNA were incubated overnight in medium containing 86 RbCl (1 μCi/mL). Before the assay, cells were washed with Krebs-Ringer solution and incubated with metabolic inhibitors (1 mM 2-deoxyglucose and 2.5 μg/mL oligomycin) in the presence of 86 RbCl for 30 min. Cells were then washed with Krebs-Ringer to remove 86 Rb + in the medium. 86 Rb + efflux was monitored over a 40-min cumulative time period by adding and collecting fresh Krebs-Ringer containing metabolic inhibitors repeatedly at 2.5, 5, 7.5, 15, 25, and 40 min. At the end of the 40 min, cells were lysed with 1% SDS. 86 Rb + in collected solutions and the cell lysate was counted. The percentage efflux at each time point was calculated as the cumulative counts in the solution(s) divided by the total counts from all solutions plus the cell lysate. To calculate the K ATP channel mediated efflux rate constant, efflux data were fitted to the following equation: efflux = 1-e ((−k1t) + (−k2t)) ; where k1 represents the rate constant calculated from untransfected cells and k2 is the rate constant for the K ATP channel mediated efflux (Cooper, Sala-Rabanal, Lee, & Nichols, 2015).

| Mutation expression studies
Because there was a time-dependent divergence of efflux from the monoexponential fit (due to the effect of metabolic inhibition on background efflux), only data from the first three time points (2.5, 5, 7.5 min) were used for fitting.
Patch-clamp recordings: Functional properties of channels were studied using the inside-out patch-clamp recording technique (Taschenberger et al., 2002). Briefly, COSm6 cells transfected with 3 | RESULTS

| Illustrative case (Family 1)
The proband in Family 1 ( Figure 1 and Table 1, 1-III-a) was a female born at 41 weeks gestation (birth weight 3.77 kg, Z-score + 0.57) following an uncomplicated pregnancy, labor, and delivery. She appeared to be growing and developing normally until 9 months of age when her parents first noted an episode of weakness, drooling, stupor, and pallor. This was interpreted as possible apnea and she was taken to an emergency department where an EKG and EEG were found to be normal. One month later at day care, she had two brief episodes of cyanotic apnea that lasted less than a minute; a plasma glucose test was not done. Subsequently, her parents witnessed an episode of shaking during sleep, after which she awakened in a stupor. She was again taken to an emergency department where the plasma glucose was 29 mg/dl (1.6 mmol/l). Hypoglycemia was confirmed by a 6-hr fasting test, which was followed by an inappropriately large glycemic response (delta 73 mg/dl) to a pharmacologic dose of glucagon. A diagnosis of HI was made and she was transferred to The Children's Hospital of Philadelphia. Treatment with diazoxide at a high dose of 20 mg/kg/day failed to control the hypoglycemia: the plasma glucose fell to 50 mg/dl (2.8 mmol/l) after only 5 hr of fasting with suppressed plasma beta-hydroxybutyrate concentration (0.82 mmol/l) and an inappropriately large glycemic response to glucagon (delta 40 mg/dl).
Diazoxide treatment was then discontinued. After an 18 F-DOPA PET/CT scan failed to detect a focal lesion, the child was begun on treatment with octreotide, 11 mcg/kg/day, together with overnight continuous intragastric infusion of dextrose at 5 mg/kg/min. On this regimen, she was able to fast for 3-6 hr and discharged home. At 22 months of age, lanreotide, a long-acting somatostatin analog, was begun and overnight tube feedings were discontinued. At 4 years of age, she maintains good glycemic control on monthly injections of lanreotide.
Mutation analysis of genes associated with congenital HI in a commercial laboratory found a maternally-transmitted novel missense variant in ABCC8 (p.Val715Met; Table 2) that was interpreted as a "variant of unknown significance." The mother denied having symptoms of hypoglycemia; however, phenotype testing demonstrated that she was affected (shown in Table 3): a 24-hr fasting test revealed hypoketotic hypoglycemia; an oPTT provoked an abnormal drop in plasma glucose to 65 mg/dl (3.6 mmol/l; delta 27 mg/dl or 1.5 mmol/l). The mother's oGTT showed mild glucose intolerance; the mother also had gestational diabetes during the pregnancy, consistent with previous reports of gestational diabetes and impaired glucose-stimulated insulin secretion in patients with K ATP channel inactivating mutations (Grimberg et al., 2001;. These clinical tests indicated that ABCC8 (p. Val715Met) was a dominant, disease-causing mutation which was subsequently confirmed by in vitro expression studies (see Figure 2). F I G U R E 1 Pedigrees of 11 families with dominant K ATP channel mutations. The pedigrees are listed in ascending order of SUR1 codons followed by Kir6.2 codons. Arrows indicate probands. Black shapes, hypoglycemia diagnosed; gray shapes, symptoms of hypoglycemia reported without clinical confirmation; white and black diagonal lines, hypoglycemia identified on phenotype test without prior knowledge of symptoms; gray and black diagonal lines, hypoglycemia identified on phenotype test and prior knowledge of symptoms. n/M, mutation positive; n/n, mutation negative Based on Genome Aggregation Database (gnomAD) v2.1, which includes data from 141,456 individuals.
Review of his son's birth history revealed that 1-III-d had low plasma glucose detected at birth requiring treatment with continuous intravenous dextrose infusions; however, his plasma glucose was thought to normalize and he was discharged home at 1 week of age.
After 1-III-d was found to be heterozygous for the ABCC8 p.
Val715Met mutation, his parents tested plasma glucose levels at home and he was found to have abnormal low overnight fasting values of 50-60 mg/dl (2.8-3.3 mmol/l). Evaluation at The Children's Hospital of Philadelphia confirmed that 1-III-d had congenital HI, but appeared to be only mildly affected and did not require medical therapy (fasting tolerance of 12-15 hr).

| Clinical features of novel dominant K ATP channel HI mutations
We analyzed six novel K ATP channel variants as possible pathogenic dominant HI mutations by clinical phenotyping of carrier parents and by in vitro expression studies. Five of these mutations were novel and one had previously been reported both as a dominant and a recessive defect (ABCC8: p.Ala1458Thr; Macmullen et al., 2011;Ovsyannikova et al., 2016;Reimann et al., 2003). As shown in Table 1, three of the novel mutations were in ABCC8 and three in KCNJ11. All were single amino acid variants: five were missense substitutions while the sixth was a deletion of a single amino acid. All five missense substitutions were predicted to be damaging by SIFT and/or PolyPhen. All six mutations were very rare since none were listed in population frequency databases (gnomAD). As shown in Figure 1, these six novel mutations were identified in 13 children with clinically-diagnosed HI from 11 families. Six of these children inherited their mutation from their mother and five from their father; mutations in the remaining two cases arose de novo.  Table 2, seven of the 11 probands had birth weights that were large for gestational age (a common feature of congenital HI due to the growth-promoting effects of insulin in utero) and four had birth weights that were appropriate for gestational age. Twelve of the 13 affected children presented with hypoglycemia on the first day of life. However, in three of these, the diagnosis of HI was not made until much later (6, 15, and 16 months, respectively). One proband did not present with hypoglycemia symptoms (seizure) until 9 months of age. Seven of the 11 probands were responsive to medical therapy with diazoxide, four were not responsive. Proband 3-III-a was not fully responsive to diazoxide. Biopsies of her pancreas showed diffuse disease; her parents declined further surgery and she was ultimately managed on diazoxide. The remaining three children who were diazoxide unresponsive were managed with octreotide or lanreotide plus tube feedings and did not have pancreatectomies.
Among the 11 families shown in Figure 1, 16 family members were found to carry the same novel K ATP channel mutation as the proband.
As described in the illustrative case, two of these family members had In Family 7, the proband's mother reported undiagnosed symptoms of hypoglycemia in herself and her brother; both were found to carry the mutation identified in the proband (ABCC8: p.Glu1517Gly).
Clinical features of these carriers are shown in Table 3. Four were female and three were male. Ages ranged from 25 to 39 years of age and BMI ranged from 22.5 to 37.5 kg/m 2 . Six of the seven tested carriers denied having symptoms of hypoglycemia; 7-II-b reported symptoms of hypoglycemia that had not been clinically confirmed.
Four of the six carrier adults who underwent a 24-hr fast ended with a plasma glucose below 70 mg/dl and suppressed ketones, consistent with HI. All of the seven adult carriers who underwent an oPTT had a fall in plasma glucose levels of greater than 15 mg/dl; four of the seven also dropped their plasma glucose levels below 70 mg/dl (3.9 mmol/l). Based on previous studies in control subjects and patients with K ATP -HI and HI due to GLUD1 activating mutations, all of the carriers were classified as protein sensitive (Fourtner et al., 2006;Hsu et al., 2001). Huopio and Laakso first described evidence of mild glucose intolerance in K ATP -HI, based on responses to oral glucose in families with the Finnish ABCC8 founder mutation for dominantly inherited congenital HI (Huopio et al., 2000). However, only mild abnormalities were seen in adult carriers in our series, and, therefore, not all were investigated; of the three carriers who underwent an oGTT, two had mild to moderate glucose intolerance (glucose peak >150 mg/dl) and all three had a glucose nadir <60 mg/dl (3.3 mmol/l).  3.5 | In vitro expression studies of novel dominant K ATP channel HI mutations 3.5.1 | Immunoblotting

| Locations of novel dominant mutation on the K ATP channel structure
To assess the molecular effects of these six novel amino acid mutations on the K ATP channel, we first determined the processing efficiency of mutant channels by examining the glycosylation patterns of SUR1 by western blotting (Figure 2c). Complex-glycosylation of SUR1, which requires successful assembly and trafficking of the channel beyond the medial Golgi and correlates well with channel expression in the membrane, was used to monitor channel maturation and surface expression as described in our previous studies (Macmullen et al., 2011;Pinney et al., 2008;Yan et al., 2007).

| 86 Rb + efflux assay
To assess how the mutations impact the ability of the K ATP channel to open in response to metabolic inhibition, we measured 86 Rb + efflux in cells expressing WT or mutant channels and pretreated with metabolic inhibitors, as described in the Methods. As shown in Figure 2, among the six mutations, p.Ala1458Thr and p.Glu1517Gly in SUR1 significantly reduced the efflux rate, and p.Val715Met in SUR1 as well as p.Leu164Phe, p.Arg206His, and p.Thr298del in Kir6.2 reduced the efflux rate to nearly 0 after subtracting background efflux observed in non-transfected cells (see Section 2), confirming a causal role of these mutations in the disease (Figure 2a). Because all six mutations are heterozygous, we also simulated heterozygous expression of the mutants and assessed how the resulting channels responded to metabolic inhibition in 86 Rb + efflux assays. Figure 2b shows that all mutations caused a significant reduction in the efflux rate, even in simulated heterozygous expression conditions.  (Nichols, 2006). Physiological changes in ATP and ADP concentrations as blood glucose levels fluctuate will tip the balance toward ATP inhibition or MgATP/MgADP stimulation to favor channel closure or opening, respectively. Reduced sensitivity to MgATP/MgADP stimulation is one of the most common defects seen in HI-causing K ATP channel mutations (Nichols et al., 1996;Shyng et al., 1998). We therefore deter-   (Cartier, Conti, Vandenberg, & Shyng, 2001;Pinney et al., 2008). The subunits of the K ATP channels, SUR1 and Kir6.2, are assembled into hetero-octameric channels in the endoplasmic reticulum and then must traffic via the Golgi apparatus where the SUR1 acquires mature N-linked glycosylation before trafficking to the plasma membrane. Recessive mutations of either subunit characterized to date, ranging from nonsense mutations to missense mutations, result in inability of the mutant channels to traffic to the plasma membrane (Cartier et al., 2001;Crane & Aguilar-Bryan, 2004;Lin et al., 2008;Nestorowicz et al., 1997;Yan et al., 2007). Thus, patients with diffuse hyperinsulinism due to homozygous or compound heterozygous recessive mutations have a lack of K ATP channels and severe hyperinsulinism that is not responsive to treatment with the channel opener, diazoxide.
Heterozygous carriers of recessive K ATP mutations appear to be asymptomatic (Grimberg et al., 2001), presumably because the mutant protein is unable to incorporate into the channel complex and is degraded in the ER while the normal allele is able to generate sufficient numbers of functional channels. In contrast, dominant mutations of K ATP subunits characterized to date are missense mutations that produce subunits which can be incorporated into mature octameric channel complexes capable of trafficking to the plasma membrane, but resulting in disruption of normal channel gating activity. Some of these dominant mutations result in a severe impairment of responsiveness to diazoxide, whereas mutations associated with milder impairment of channel gating activity may be diazoxide-responsive. Recessive K ATP channel mutations can cause focal hyperinsulinism lesions through isodisomy for a paternally derived recessive K ATP channel mutation as a consequence of loss of the maternal 11p region. Focal lesions due to a dominant mutation have not yet been described, but in such a case, resection would not be curative due to the ongoing effect of the mutation in the remaining pancreas. In a child with diazoxide-unresponsive HI who has a novel monoallelic missense mutation in ABCC8, determining whether the mutation is dominant or recessive is of great clinical importance, because if the monoallelic mutation is clearly recessive there is >95% likelihood of a focal lesion that can be cured by F-DOPA PET scan localization and targeted resection (Snider et al., 2013). In contrast, a monoallelic dominant mutation is diagnostic of diffuse disease. Differentiating between dominant and recessive mutations is also essential for counseling about recurrence risk as well as for identifying other family members at risk for hypoglycemia (as illustrated by the Case Summary of Family 1).
A total of 68 monoallelic mutations in ABCC8 and KCNJ11 associated with dominant HI have been reported, including the six described here (see Figure S1 and Table S1a,b in the Supporting Information).
These dominant mutations have primarily been single amino acid substitutions, but single amino acid deletions and an insertion of two amino acids have also been reported (Kapoor et al., 2011;Shemer et al., 2012;Thornton et al., 2003). Dominant mutations in ABCC8 have been associated with both diazoxide-responsive and diazoxideunresponsive disease; in contrast, all of the dominant missense HI mutations in KCNJ11 reported to date have been diazoxideresponsive (Shyng, Bushman, Pratt, & Zhou, 2012).
Predicting whether novel missense mutations in ABCC8 and KCNJ11 are recessive or dominant is not possible based on mutation locations since mutations that impair channel trafficking have been identified throughout both proteins (Martin, Chen, Devaraneni, & Shyng, 2013;Shyng et al., 2012). (See Figure S1 in the Supporting  (Macmullen et al., 2011) and p.Ala1458Val, (Saint-Martin et al., 2015 are dominant mutations, while clinical evidence indicates that the third, p.Ala1458Pro, is a recessive (non-trafficking) mutation since this patient was homozygous and had diffuse HI (Bellanne-Chantelot et al., 2010).
Finally, it is important to note that the number of HI mutations which have been identified in the K ATP channel subunits has not yet reached saturation, since approximately 70% of pathogenic missense HI mutations identified in ABCC8 and KCNJ11 continue to be novel variants not previously reported (Snider et al., 2013). Therefore, uncertainty in interpretation of monoallelic novel missense variants of the K ATP subunits continues to be a handicap in applying genetic mutation testing to clinical decision-making in congenital HI and especially hinders the identification of children with potentially curable focal lesions.
Software programs, such as SIFT (Sim et al., 2012) and PolyPhen2 (Adzhubei et al., 2010), are commonly used to predict whether a novel mutation will be damaging. In the present series of patients, both SIFT and PolyPhen2 correctly predicted that four of the five amino acid substitutions would be damaging (4/5, 80%) but gave conflicting predictions for the fifth (1/5, 20%); neither program could be used for the single amino acid deletion mutation, KCNJ11:p.Thr298del. This is similar to the accuracy of these software programs for previously reported dominant and recessive K ATP channel mutations (79% correctly predicted as disease-associated, 17% had conflicting predictions, and 4% incorrectly predicted to be benign, (Snider et al., 2013). Despite the fact that the accuracy of software programs is high for predicting disease causation for K ATP channel mutations, these programs are not capable of predicting whether a mutation will act recessively or dominantly.
In vitro expression studies of the six novel K ATP channel HI mutations reported here demonstrated features consistent with them acting in a dominant fashion. All were associated with normal trafficking of channels from the Golgi apparatus to the plasma membrane surface in COSm6 cells, whereas recessive mutations severely interfere with normal trafficking (Cartier et al., 2001 (Yang, Fang, Fromondi, & Chan, 2005). It is also well documented that the most prevalent cystic fibrosis transmembrane conductance regulator (CFTR), p.delPhe508, traffics efficiently to the plasma membrane in Xenopus oocytes but not in mammalian cells (Denning et al., 1992;Drumm et al., 1991) (Christesen et al., 2002;Dullaart, Hoogenberg, Rouwe, & Stulp, 2004;Glaser et al., 1998;Gloyn et al., 2003;Wabitsch et al., 2007) and glutamate dehydrogenase (Stanley et al., 1998) and in a family with HI associated with the hexokinase-1 locus (Pinney et al., 2013). Several factors may contribute to the under-recognition of patients carrying dominant HI mutations, including indications that the hypoglycemia in K ATP -HI may become less severe in adolescents and adults than in young infants (Glaser et al., 1999), failure to diagnose hypoglycemia in children having seizures or other symptoms of hypoglycemia, and possible hypoglycemia unawareness induced by recurrent hypoglycemia (hypoglycemia-associated autonomic failure [HAAF]; Dagogo-Jack, Craft, & Cryer, 1993;Segel, Paramore, & Cryer, 2002).
The present studies suggest that evidence of HI can be demonstrated in carriers of dominant K ATP channel mutations and that an oral protein tolerance test might be the simplest method for this purpose. Our results also suggest that the penetrance of dominant K ATP channel mutations is near 100%, but that in many carriers hypoglycemia symptoms are not diagnosed.

| Limitations
There are several limitations in differentiating between dominant and recessive K ATP channel missense mutations. Pedigrees of patients with novel mutations are often too small to be able to clearly distinguish dominant vs recessive modes of transmission. in vitro mutation expression studies can provide insight into the functional consequences of the amino acid change on channel trafficking and channel activity characteristics of recessive or dominant defects. However, the turnaround time involved in expression studies is generally too long to support clinical decision-making in a novel case (e.g., defining the possibility of a resectable focal lesion). In addition, there appear to be differences in the reliability of different methods used for expression studies, depending on the type of cell used and the species origin of the constructs for channel subunits; the method employed in the present studies appears to have advantages over studies based on amphibian cells.
The results of phenotype testing of mutation carriers can be obtained more quickly than in vitro expression studies to distinguish whether a K ATP channel missense HI mutation is dominant or recessive; as in the present studies, phenotype testing of adult carriers may require only bedside glucose monitoring. The oral protein tolerance test appeared to be more reliable in confirming that carriers were affected than a 24-hr fasting test or an oral glucose tolerance test. However, since these tests have been carried out only in a limited number of carriers and controls, their accuracy in identifying whether a missense mutation is dominant vs recessive remains uncertain.

| CONCLUSION
In summary, these studies identify six novel K ATP channel inactivating mutations associated with single amino acid changes that cause a dominant form of HI; three in ABCC8 and three in KCNJ11. In silico prediction tools and mutation location were not adequate to define whether these mutations act in dominant vs recessive fashion. However, determination of the functional consequences of the mutations by in vitro expression in a COSm6 mammalian cell line and in vivo phenotype testing of adult carriers confirmed that all six mutations acted in dominant fashion. In vitro expression studies, such as 86 Rb + efflux, appear to provide the most definitive assessment of pathogenicity; however, results are difficult to obtain quickly enough to inform clinical management.
Phenotype studies of mutation carriers, such as the patient's parent, and, in particular, the glucose response to oral protein challenge, may provide a timely method for determining whether a novel K ATP channel missense variant is a dominant HI mutation.

CONFLICT OF INTEREST
None.

DATA AVAILABILITY STATEMENT
The data that support the findings of this study are available from the corresponding author upon reasonable request.