This study was approved by the Ethics Committee of the Central Hospital of Wuhan (2016–2). Informed consent was obtained from the patients and their families. Researchers had access to clinical information and medical records of the participants. Patients with a clinical diagnosis of MODY were included in the study and the criteria were age at diagnosis of ≤25 years, negative islet autoantibodies, and a family history of diabetes.

Genomic DNA was extracted from the participants’ peripheral blood samples (TIANGEN DNA extraction kit, Beijing, China). We used the Agilent SureSelect Human All Exon V6 kit for capture, followed by sequencing on the Illumina hiSeq2500 system, reaching an average depth of approximately 50×. The reference genome for aligning WES was human GRCh37/hg19. SNPs and indels were called by SAMtools, correcting for overestimated mapping quality from Burrows-Wheeler Alignment tool (14).

Variants were filtered to retain protein-altering variants (including non-synonymous, frame shift and splicing, and excluding non-exonic and synonymous variants), and to exclude those with minor allele frequency (<0.001) according to certain public databases (1000genomes, ExAc, dbSNP, ESP and gnomAD). The effects of the identified variants were evaluated using silico prediction tools, namely SIFT, PolyPhen2_HDIV, PolyPhen2_HVAR, LRT, Mutation Taster, Mutation Assessor and FATHMM (15). We selected genes previously implicated in monogenic diabetes, genes with important roles in the beta cell, and genes associated with type 2 diabetes or fasting glucose (16). Variants of uncertain significance (VUS), defined as functional significance, are not recorded in the public database and there are no relevant reports in the literature, such as novel missense variants, variants with conflicting evidence, or variants with insufficient evidence (17).

Candidate variants in the proband and his family were confirmed by direct Sanger sequencing of the PCR products. Afterwards, the candidate variants that cosegregated with diabetes were selected for further analysis. The ABCC8 gene was amplified by PCR using forward primer (proband A) 5′-TCAAGGCCTCCTGCTTCTGT-3^′ and reverse primer 5′-TGTCCAGGCCTCAGCTTCTT-3^′, forward primer (proband B) 5′- AGCAGGAAGGCTTGGTGT-3^′andreverse primer 5′- TTGGAGGAGGTGGGGATT-3^′, forward primer (proband C) 5′- ACACCCAGAACCCCAAACCT-3^′and reverse primer 5′- GTGTCTGTCTGCCCCTCCCT-3^′. The PCR conditions were as follows: 95 C for 5 min; 35 cycles of 95 °C for 1 min, 60 °C for 30 sec, and 72 °Cfor 1 min; and then 72 °C for 10 min. The PCR products were directly sequenced (Beijing Genomics Institute, Wuhan, China) (18).

The patients’ treatment plans were adjusted based on genetic test results and previous literature reports. For example, patients with ABCC8 gene mutations are more sensitive to sulfonylurea drugs. Therefore, according to the results of the gene test, patients with ABCC8 mutations can use sulfonylureas for hypoglycemic treatment. Blood glucose levels were regularly monitored (every day), and the HbA_1C levels were controlled for two months.

Proband A was an eight-year-old female, who was admitted to the hospital for a stuffed nose and sleep snoring and was diagnosed with type 1 diabetes at the age of eight. Her body mass index (BMI) was14.0 kg/m². The proband’s parents (III-1 and III-2) were 38 years old and non-diabetics, their BMIs were 26.1 kg/m²and 22.0kg/m². The proband’s grandmother (II-1) was diagnosed with diabetes at the age of 51, and her BMI was 20.0kg/m².The father of the proband’s grandmother (I-1) died from diabetes-related causes.

The proband B was a 12-year-old male, who was admitted to the hospital for the following symptoms: dry mouth, polydipsia, polyuria, and fatigue. The physical examination revealed fasting blood glucose of 14.28mmol/Land a BMI of 26.9 kg/m². The proband’s father (III-1), mother (III-2), grandfather (II-1), maternal grandfather (II-4), and maternal grandfather’s sister (II-5) and brother (II-6) were diagnosed with diabetes at the ages of 43, 41, 59, 69, 59, and 61, respectively, and their BMIs were less than 28.0 kg/m².

Proband C was a 25-year-old male, who was admitted to the hospital for dry mouth, polydipsia, polyuria, and fatigue, and was diagnosed with diabetes and ketoacidosis. His BMI was 27.8 kg/m². He was prescribed insulin, which resulted in well-controlled blood sugar levels. The proband’s mother (II-4) and maternal grandfather (I-4) were diagnosed with diabetes at the ages of 41 and 60, respectively, and the proband’s mother (II-4) had a BMI of 22.8kg/m².

The BMI of all family members of the diabetes patients was in the normal range or had not reached the obese range (> 28.0kg/m²). The pedigree of the three families is presented in Figure 1 .

The clinical characteristics of the three probands are shown in Table 1 . Proband A’s laboratory analysis revealed that her HbA_1C level was 12.8%, her fasting plasma glucose (FPG) level was 18.81 mmol/L, and her fasting c-peptide was 2.35 ng/ml. She tested positive for urine glucose (2+) and urine ketones (1+), and her triglyceride levels were high (50.21 mmol/L), whereas she tested negative for all diabetes autoantibodies. Ultrasound examination of the liver, kidney, pancreas, bladder, ureter, spleen, and gall showed that organ morphology was normal. Although she was prescribed high-dose insulin and oral hypoglycemic agents (insulin glulisine, 20 units, three times a day; degludec, 30 units/night; metformin tablets, 850 mg twice a day; pioglitazone hydrochloride and metformin hydrochloride tablets, 15 mg/500mg, twice a day, and notified her guardian) for two days, her FPG levels did not drop. However, her triglyceride levels decreased to 12.6 mmol/L. The test results of proband A’s grandmother (II-1) were as follows: HbA_1C, 7.1%; FPG, 11.7 mmol/L; fasting c-peptide, 1.60 ng/ml, and positive (1+) urine glucose, whereas ketones in the urine and diabetes autoantibodies were all negative. She (II-1) had diabetic retinopathy and diabetic neuropathy and was previously prescribed acarbose and insulin. However, her (II-1) blood sugar was not well controlled.

Physical examination revealed that proband B presented with hyperlipidemia, hyperuricemia (707 μmol/L), and mild fatty liver. Additionally, HbA_1C was 12.1%, urine glucose was positive (4+), and fasting c-peptide was 1.8 ng/ml, while ketones in the urine and diabetic autoantibodies were all negative. There were no obvious abnormalities in other assessments. Although proband B’s islet function was normal, his fasting and two-hour postprandial blood glucose were high for a child. Hence, the doctor had been administering insulin, which led to the patient’s blood glucose being under control. Proband B’s father (III-1) was prescribed metformin and gliclazide sustained-release tablets, his HbA_1C levels were 7.4%, FPG was 9.87 mmol/L, and he additionally presented with diabetic nephropathy and neuropathy. Proband B’s mother (III-2) and maternal grandfather (II-4) were prescribed insulin, metformin, and linagliptin. However, they (III-2 and III-4) did not achieve optimal outcomes since their HbA_1C and FPG levels were approximately 10% and11 mmol/L, respectively. The maternal grandfather’s sister (II-5) and brother (II-6) of proband B were prescribed insulin and acarbose, respectively, their FPG levels reached 23 mmol/L at times, and both had diabetic retinopathy and neuropathy.

Proband C’s laboratory analysis revealed that his HbA_1C level was 11.2%, FPG level was 13.52 mmol/L, fasting c-peptide was 1.30 ng/ml, urine glucose tested positive (4+), as well as ketones in urine (3+), and triglyceride levels were high (16.78 mmol/L). However, diabetic autoantibodies were all negative. Ultrasound examination indicated that all organs were normal and he exhibited no diabetes-related complications. Proband C was prescribed liraglutide for three days (0.6 mg/daily). Nonetheless, FPG and triglyceride levels decreased to 9.64 mmol/L and 6.11 mmol/L, respectively. The results of proband C’s mother (II-4) were as follows: HbA_1C was 8.6%, the FPG level was 9.65 mmol/L, and she presented with complicated diabetic neuropathy. Although she (II-4) had been using metformin and acarbose, her blood sugar level was not under control.

WES details are shown in Table 2 . WES showed that the variants of each subject ranged from 110,000 to 130,000, including SNPs and InDels in probands A, B, and C. Variants were filtered as previously described and the defined panel of MODY genes (HNF4A, GCK, HNF1A, PDX1, HNF1B, NEUROD1, KLF11, CEL, PAX4, INS, BLK, ABCC8, KCNJ11, and APPL1) were firstly evaluated (15). ABCC8 was found to be a candidate pathogenic gene in all three families, and no other known MODY pathogenic variants were identified. The details are as follows. A known mutation was found in proband A: chromosome 11, position 17464342, c.C1555T, NM_001287174, p. R519C, namely, rs1057467571. This mutation has a very low frequency (25/251142, GnomAD_exome; 11/121224, ExAC; 1/31398, GnomAD) in the public database, but has not been reported in Chinese populations, and its clinical significance is not reported in ClinVar. Despite being a known mutation, its function remains unclear. In proband B, we found a novel mutation, chromosome 11, position 17419936, c.A3706G, NM_001287174, p. I1236V. In proband C, we discovered a known mutation, chromosome 11, position 17428939, c.C2885T, NM_001287174, p. S962L, namely, rs748931549.There is no reported frequency for this mutation in the public database. Its clinical significance is not reported in ClinVar. Despite being a known mutation, its function remains unclear. The three mutation sites were then sequenced and validated in other members of their respective families ( Figure 2 ). Although we did not find them in non-related healthy controls (n=200), among the members of family A, the proband A’s father (currently 38 years old) also carried this mutation site (c.C1555T) but did not have diabetes. In family B, the mutation site (c.A3706G) was inherited from the mother (III-2), and the other patients (II-4, II-5, and II-6) all carried this mutation. Although proband B’s father may also have a genetic history of diabetes, no suspicious pathogenic site has been found so far; therefore, the pathogenic site of diabetes should come from probands B’s mother. In family C, the mutation site (c.C2885T) co-segregated with diabetes ( Supplementary Table 1 ). However, all three mutation sites were predicted to be harmful mutations by multiple bioinformatics software, and the mutation sites were highly conserved in several species ( Supplementary Tables 2 , 3). Consequently, we speculated that these three mutation sites are pathogenic mutations in these families, and that they all belong to the ABCC8-MODY12 type.

Genetic testing confirmed that the above mentioned families belong to the MODY12 subtype. This information, combined with results reported in the literature, led to recommending that these patients replace current medical treatment with sulfonylurea drugs.

The proband A was administered insulin (insulin aspart injection, 12 units, subcutaneous injection before meals, and insulin glargine injection, 12 units, subcutaneous injection before 9 pm) due to high fasting blood sugar and positive urine ketones, as he was a child. However, the proband A’s fasting blood sugar was still high after two days of treatment. Then, the insulin dose was increased to 50 units, but the fasting blood glucose still did not decrease (the fasting blood sugar levels remained between 12 and 18 mmol/L), even when acarbose and metformin enteric-coated tablets were used. The results of the genetic test suggested that proband A was classified as MODY12, and there is no evidence of sulfonylureas being used previously to treat pediatric diabetes in China. Thus, we tried to utilize the relatively safe glimepiride combined with insulin on the child. When glimepiride was added, the child’s fasting blood glucose rapidly decreased. After a week, the child’s fasting blood glucose level was in the normal range (< 6.1 mmol/L).We then slowly reduced the dose of insulin (by 5 units a week), and shortly the insulin administration ceased, and only glimepiride (2 mg/day) was used. Nonetheless, after two weeks, proband A developed diabetic ketosis, necessitating the use of glimepiride (2 mg/day) in combination with a low-dose insulin aspart injection (5 units/day, twice a day) for hypoglycemia. Then, the dosage of glimepiride (4 mg/day) was increased and insulin administration was stopped. The results from follow-up tests in the following two months showed that the child’s fasting blood sugar (< 6.1 mmol/L) and glycation levels (< 6.5%) were within the normal ranges.

Proband B, who is currently 13 years old, presented with type 2 diabetes. Therefore, we replaced the previously prescribed insulin aspart with a new type of hypoglycemic drug, liraglutide injections at 1.2 mg/day, and notified his guardian. This drug’s hypoglycemic properties produced very effective results. After two months of follow-up, the patient’s blood sugar levels were under control and within the normal range (HbA1c < 6.5%). Taking into consideration that this proband B is MODY12 and a pediatric patient, liraglutide injections yielded very positive effects in terms of blood sugar reduction. Liraglutide was then gradually reduced. When glimepiride was added (2 mg/day), until liraglutide administration ceased, the blood glucose levels remained within a controlled range. However, ketosis appeared shortly, and thus, liraglutide use was continued. Then, the dosage of glimepiride was increased (4 mg/day) and liraglutide was stopped. The patient’s blood glucose returned to the normal range (HbA1c < 6.5%).

Proband C was also very sensitive to liraglutide injections (1.2 mg/day). Thus, this hypoglycemic treatment was very effective. Since injections are more problematic, whereas oral medication is relatively simple to administer, we recommended glimepiride 4mg/day as a medication plan based on the patient’s MODY12 type. After two months of follow-up, the patient’s blood glucose returned to a reasonable level (HbA1c < 6.5%).

Currently, other adult MODY12 patients in these three families were mainly using insulin or insulin combined with metformin or acarbose, with deficient results on blood sugar control. However, according to each patient’s genotype, they were slowly switched to glimepiride (4 mg/day), and their blood glucose levels were controlled after 2-3 months and the HbA_1Clevels approached normal ranges(< 6.5%). All diabetes patients continued to be treated with glimepiride (4 mg daily).