Physical Exercise-Induced Hyperinsulinemic Hypoglycemia Is an Autosomal-Dominant Trait Characterized by Abnormal Pyruvate-Induced Insulin Release

In addition to the published index case of family 2 (Fig. 1B), another obvious EIHI patient was submitted to our clinic (index case of family 1; Fig. 1A). Exercise testing of family members revealed 10 additional EIHI patients in these two families in a manner typical for autosomal-dominant inheritance, with cases in three consecutive generations, and ∼50% of the siblings affected (Fig. 1A and B). Tables 1 and 2 summarize the major features of the patients and their healthy control subjects. The age of the patients varied widely between 1.5 and 60 years (median 27). The youngest patient could naturally not be tested with the physical exercise test, and his diagnosis was thus based on a clearly positive insulin response in the intravenous pyruvate test. He had also suffered from mild nonketotic prolonged neonatal hypoglycemia. In general, the adult patients reported recurrent hypoglycemic symptoms associated with strenuous exercise, particularly swimming. However, aerobic mild exercise was well tolerated, and most subjects learned to adjust their activities so that they did not develop episodes of serious hypoglycemia. The severity of the condition was clearly variable; therefore, some affected individuals had barely any symptoms, whereas others suffered from recurrent severe hypoglycemia. The index cases of both families had benefited from medical treatment with diazoxide, which acts by inhibiting insulin release. Diazoxide relieved their symptoms but did not totally prevent hypoglycemic episodes. The index case of family 1, who also had epilepsy, used diazoxide for several years. First-phase insulin responses to calcium, glucose, and tolbutamide were studied in this patient as previously described (5), revealing responses within the normal range of obese control subjects to glucose and tolbutamide and a normal negative response to calcium. Repeated abdominal ultrasound examinations and computed tomography scans had been normal. The detailed history of the index case of family 2 has been published previously (3). Her abdominal MRI scan did not reveal tumors. Blood ammonia and plasma amino acids were normal in both index cases. Finally, Table 1 shows that the basal and exercise-induced levels of the stress hormones cortisol and growth hormone are similar in patients and control subjects, whereas the plasma glucagon levels were slightly lower in patients both before and after exercise.

The study protocols were approved by the Ethics Committee of the Hospital for Children and Adolescents, University of Helsinki. Informed consent was given by all subjects studied. After an overnight fast, the patients received an intravenous bolus injection of sodium pyruvate (300 mmol/l) during 1 min. The dose was 46.3 ml (13.9 mmol)/1.7 m² body surface area. This dose was initially chosen based on a previous study in which this total dose had been used without any side effects when infused intra-arterially for 15 min (6). Samples for blood glucose, pyruvate, and serum insulin were taken at 0, 1, 3, 5, 10, and 30 min.

The test was based on a 10-min bicycle exercise, which began 60 min after the intravenous pyruvate test. The patients were asked to perform moderately strenuous bicycle exercise for 10 min. The aim was to increase the pulse rate to an age-adjusted target of 220 minus age in years. Blood test for glucose, insulin, pyruvate, and lactate were taken at −10, 0, 5, 10, 15, 20, 25, 30, 35, 40, 50 and 60 min. Samples for stress hormones (growth hormone, cortisol, and glucagon) were taken at −10 and 20 min. Heart rate and respiratory frequency were recorded throughout the test.

Two patient samples (female index case of family 2 and the previously reported male case [3]) and one control sample were used for screening mutations in the promotor region of the MCT1 gene and coding regions of the genes MCT1–MCT8 and CD147. PCR assays of genes MCT1, MCT3, and MCT6–MCT8 (with known genomic organization) were carried out in 50-μl reactions containing 30–50 ng genomic DNA from lymphoblasts, 1× PCR buffer (10 mmol/l Tris-HCl, 1.5 mmol/l MgCl₂, 150 mmol/l KCl, and 0.1% Triton X-100), 160 mol/l dNTPs, 0.6 mol/l of each primer, 0.6 units DNA polymerase (DyNAzyme II; Finnzymes, Espoo, Finland), and 0–4% DMSO. The primer sequences used for sequencing of MCT family gene exons are available from T.O. upon request. Some poorly amplifying parts of coding regions were amplified by using DNA polymerase AmpliTaq Gold (Perkin Elmer, Roche Molecular Systems) and the high-fidelity enzyme DyNAzyme EXT (Finnzymes). To amplify the MCT2, MCT4, and MCT5 cDNAs, RNA was extracted (RNeasy Mini Kit; Qiagen, Valencia, CA) from lymphoblasts, and RT-PCR (Perkin Elmer) was carried out using the manufacturer’s protocol. The primers were designed using the Primer3 program (http://www-genome.wi.mit.edu/genome_software/other/primer3.html). The samples were denaturated for 2 min at 94°C, followed by 35–40 cycles each of 35 s at 94°C, 35 s at 55–62°C, and 1 min at 72°. Elongation was performed for 8 min at 72°C. Purified (PCR purification kit, gel extraction kit; Qiagen) PCR products were sequenced using ABI 377 and ABI 3100. Coding regions of the genes were inspected by direct sequencing of the genomic and RT-PCR products, and some parts of coding regions were cloned (TOPO TA Cloning Kit, pCR 2.1-TOPO vector; Invitrogen) before sequencing. Sequencing was routinely done in both directions, with some rare exceptions in which the unidirectional results left no uncertainty.

Small skin biopsies were taken from the ventral arm of a patient with EIHI (index case of family 1) and an age- and sex-matched healthy control subject. Fibroblasts were grown out of the biopsies in RPMI-1640 medium (Gibco, Paisley, Scotland) supplemented with 30% FCS (Gibco) at 37°C in a CO₂ incubator. Transport of radioactively labeled pyruvate was studied in batches of 10⁴ cells plated in 96-well plates between passages 3 and 5. The cells were preincubated for 30 min in Krebs-Ringer bicarbonate buffer (KRB; 140 mmol/l NaCl, 3.6 mmol/l KCl, 0.5 mmol/l NaH₂PO₄, 0.5 mmol/l MgSO₄, 1.5 mmol/l CaCl₂, 2 mmol/l NaHCO₃, 10 mmol/l HEPES [pH 7.4], and 0.1% BSA). Experiments were initiated by replacing the buffer with fresh KRB containing 1 μCi/ml (0.05 μCi/well) of [1-¹⁴C]l-pyruvate (NEN Life Science Products, Boston, MA). After 1–30 min incubation at 37°C, the cells were immediately washed twice with KRB, detached with trypsin-EDTA (Gibco), and mixed with Optiphase Hisafe liquid scintillation solution (Wallac, Turku, Finland). The incorporated radioactivity was counted in a β-counter (Wallac). Alternatively, the cells were incubated for 5 min in KRB containing different concentrations (0.125–10 mmol/l) of cold Na-pyruvate (Gibco) and 1 μCi/ml of [1-¹⁴C]l-pyruvate.

Lactate dehydrogenase (LDH) activity was measured from the confluent fibroblast cultures. Cells were washed with PBS, harvested, and, after three freeze-thaw cycles, centrifuged for 15 min at 15,000g. Supernatants were filtered with PD-10 columns (Pharmacia Amersham Biotec), and enzymatic activity was measured with a spectrophotometer in 1-ml cuvettes containing 100 mmol/l lactate, 600 mmol/l Tris [pH 8.5], 0.1 mmol/l NAD+, and sample. Protein concentration was measured with a Bio-Rad DC Protein Assay kit.

Significance of the observed differences between patients and control subjects was tested with one-way ANOVA for repeated measures and Student’s unpaired t test. Logarithmic transformation was used to normalize the distribution of serum insulin values. P < 0.05 was regarded as significant.

All individuals were normoglycemic after the overnight fasting and before the test (blood glucose 3.5–5.2 mmol/l). However, the pretest blood glucose levels were slightly lower in patients than in control subjects (4.1 ± 0.4 vs. 4.6 ± 0.2 mmol/l, means ± SD, P = 0.016) (Table 1). Fasting insulin levels were not significantly different. The level of physical stress obtained during the 10-min exercise was similar in patients and control subjects, as judged by the age-dependent target heart rate (90.1 vs. 91.8% of target) (Table 1). Also, the circulating levels of the stress hormones (growth hormone and cortisol) were similar both before and 10 min after the end of exercise. However, plasma glucagon levels were slightly but significantly lower both before and after exercise (Table 1). During exercise, and particularly immediately after, blood glucose levels decreased in the patients, whereas a slight increase was observed in the control subjects (Fig. 2). A clearly hypoglycemic level (<3.0 mmol/l) still remained in 5 of 11 patients at 60 min after the onset of exercise. The patients reported mild hypoglycemic symptoms during and after the test.

The increase in blood pyruvate (4-fold from basal level) and lactate (10-fold) were similar in patients and control subjects up to the 15-min sample. However, both pyruvate and lactate tended to decrease faster in the patients (blood pyruvate significantly lower in patients at 20–30 min and lactate at 20 min) (Fig. 2). Serum insulin levels increased during and immediately after the exercise in the patients. The increase at 20 min was significantly higher in patients than in control subjects (5.9 ± 0.8-fold vs. 2.6 ± 0.2-fold, P = 0.003) (Fig. 2).

The intravenous pyruvate test was performed to test the hypothesis that exercise-associated hypoglycemia is associated with an abnormal β-cell response to circulating pyruvate. The injected sodium pyruvate bolus was well tolerated by both patients and control subjects. No side effects were reported. Blood glucose levels remained normoglycemic in all cases, although a slightly lower resting level and a small decrease after the injection was evident in the patients. As expected, the blood pyruvate concentration increased rapidly after the injection to levels comparable with those measured after physical exercise. However, peak pyruvate concentration at 1 min after injection was nearly twice as high in the patients as in control subjects (282 ± 43 vs. 155 ± 20 μmol/l, P < 0.05) (Fig. 3). Thereafter, pyruvate concentrations decreased similarly in both groups. Lactate levels were not different between the groups at 0, 5, and 10 min. The most dramatic difference was observed in the serum insulin levels (Table 2 and Fig. 3). There was no response whatsoever in the control subjects after the pyruvate injection. However, the patients’ insulin levels increased significantly already at 1 min after injection, and the highest concentrations were measured at 3 min. The 1 + 3-min peak response was 5.4 ± 0.9-fold in the patients, as compared with 0.94 ± 0.08-fold in control subjects (P < 0.001) (Fig. 3).

The coding regions of the candidate genes MCT1–MCT8 (XM 001306, AF 058056, NM 013356, NM 004207, NM 004696, NM 004695, NM 004694, HSXPCT1, and HSU05315) and CD147 (NM 001728) were sequenced according to Halestrap and Price (7). Altogether, 40 sequence variants, as compared with publicly available sequences, were observed in the nine genes, including 22 changes observed in all patient and control individuals sequenced. Of the remaining changes, none cosegregated with the phenotype in the families, suggesting that all were functionally silent polymorphisms. A list of all variants is available from T.O. upon request.

Transport of labeled pyruvate was studied in early-passage cultured skin fibroblasts of the index patient of family 1 and a young healthy control subject. As expected, rapid pyruvate transport could be recorded. No kinetic differences were detected between patient and control cells (data not shown). To identify a possible difference in the affinity of the pyruvate transport system, the amount of labeled pyruvate transported into the cell during 5 min was recorded in the presence of increasing concentrations of unlabelled pyruvate. Again, no differences could be detected in the responses of patient and control cells (not shown). Multiple LDH measurements were done, and the specific activity in control cells was repeatedly 10–15% higher than in patient-derived cells.

We have characterized a novel disease of β-cell dysregulation, termed EIHI. All 13 patients identified had normal fasting blood glucose concentrations, although the average levels were significantly lower than in control subjects. Their serum insulin levels were not significantly elevated, whereas plasma glucagon was slightly suppressed, consistent with relative hyperinsulinemia. Anaerobic physical exercise induced a hypoglycemic response that was preceded by an inappropriate increase in the circulating insulin levels. Insulin secretion closely followed the increase in blood pyruvate concentration during exercise. Pyruvate sensitivity of insulin secretion in these individuals was confirmed by a prompt response to an intravenous dose of pyruvate. Such a response was absent in control subjects. This indicates that the pathogenetic mechanism must involve aberrant transport, signaling, or metabolism of pyruvate in the pancreatic β-cell. However, the immediate peak concentration of pyruvate after the infusion was higher in patients, possibly indicating decreased pyruvate uptake in the liver during the first-pass metabolism. Thus, this observation could be taken as an indication of a defect common to the hepatocyte and the β-cell. The pathologic response to pyruvate was clearly inherited in an autosomal-dominant manner, strongly suggesting that the phenotype is due to mutations in a single gene.

Insulin release from the normal adult β-cell is quite unresponsive to acute changes in the extracellular concentrations of lactate or pyruvate. This has been considered to be a protective mechanism of the pancreatic β-cell, allowing the organism to prevent undesired insulin secretion induced by pyruvate and lactate metabolism during exercise or catabolic states (4). The unresponsiveness is due to at least two mechanisms: very low or absent expression of the major MCT, MCT1 on the β-cell membrane (8), and low activity of LDH in the β-cell (9,10). These metabolic features result in the selective channelling of glycolytic metabolites into β-cell mitochondria. Thus, an increased uptake of pyruvate could lead to an increased production of ATP, followed by increased insulin release. Furthermore, experimental in vitro overexpression of MCT1 in the β-cell conferred sensitivity of insulin release to exogenous pyruvate but not lactate (4). To achieve lactate-induced insulin release, LDH also had to be overexpressed. This prompted us to develop a diagnostic test for EIHI based on pyruvate-induced insulin secretion. The observed clear difference in the pyruvate response between patients and control subjects would thus support the hypothesis that the major pathogenetic mechanism in EIHI involves monocarboxylate transport over the β-cell membrane.

This hypothesis could naturally not be tested directly, since it is not possible to obtain β-cells from the patients. We approached this question by sequencing the major candidate genes, encoding the known human MCTs (MCT1–MCT8) (7) and the chaperone protein CD147, which is essential for the correct targeting of MCT proteins to the plasma membrane (11). To study the role of polymorphism in the MCT genes, we used single-strand confirmation polymorphism and direct sequencing of exons to detect sequence variation and segregation study within families as well as catalogues of known common single nucleotide polymorphisms to infer their significance. Any ambiguous results were followed-up until definitively solved. Our results showed that none of the observed polymorphisms was likely to be disease associated.

In addition, we studied pyruvate transport in cultured patient fibroblasts in order to detect a universal transport defect. The results suggested that monocarboxylate transport was normal in patients as assessed. However, this result does not rule out the possibility of a β-cell-specific transport defect, possibly caused by mutations in a gene responsible for the cell-type-specific expression of a MCT. Nevertheless, other possibilities involving defects in pyruvate metabolism must be considered. Thus, we also studied LDH activity in the fibroblasts. This was slightly higher in control than in patient cells, but the small difference is most likely unrelated to the pathogenesis of EIHI.

Although not studied systematically yet, the lack of postprandial hypoglycemia and the normal glucose-induced insulin release in at least one of our patients, together with at least partial sensitivity to diazoxide, indicate that the β-cell defect is between pyruvate and the potassium-sensitive ATP channel. If the production of pyruvate in β-cells is normal in EIHI, one has to considerer mutations resulting in gain of function in pyruvate dehydrogenase or carboxylase, Krebs cycle, or mitochondrial respiratory chain. Furthermore, other enzymes affecting ATP/ADP or NAD/NADH ratios come into question, as described for glutamate dehydrogenase in the hyperinsulinism-hyperammonemia syndrome (12). At the moment, we do not have any clues to support any of these possibilities. The mild disease phenotype, as well as the normal levels of amino acids, ammonia, pyruvate, and lactate, rule out known enzyme defects in pyruvate metabolism (13).

After the identification of the first EIHI patient in Finland, we could rapidly detect 10 additional cases in two families simply by informing the local pediatric endocrinologists. This would suggest that the disease may not be exceedingly uncommon. Exercise-related hypoglycemic symptoms are likely not to be very rare. Since exercise-induced hyperinsulinism has previously not been considered and specifically sought, it may well be that many individuals carrying this trait have simply adapted to the relatively small handicap and not been diagnosed. It will be interesting to find out the prevalence of EIHI in a larger number of individuals with exercise-associated hypoglycemia. Further pathogenetic studies will become possible if the genetic mapping and identification of the disease gene through linkage disequilibrium analysis in the affected families turns out to be successful.

These studies were supported by the Foundation for Pediatric Research in Finland (to T.O. and I.S.). J.K. is supported by the Sigrid Juselius Foundation and Academy of Finland; T.M. and E.M. are supported by a grant from the Arbeitsgemeinschaft für Pädiatrische Stoffwechselstörungen; and T.O. is supported by the Juvenile Diabetes Research Foundation International.

The authors are indebted to Professor Claes Wollheim for the pathogenetic ideas that resulted in these studies and to the European Network for Research into Hyperinsulinism (QLG1-2000-00513), supported by the European Union.