An Activating Mutation of AKT2 and Human Hypoglycemia

See allHide authors and affiliations

Science  28 Oct 2011:
Vol. 334, Issue 6055, pp. 474
DOI: 10.1126/science.1210878


Pathological fasting hypoglycemia in humans is usually explained by excessive circulating insulin or insulin-like molecules or by inborn errors of metabolism impairing liver glucose production. We studied three unrelated children with unexplained, recurrent, and severe fasting hypoglycemia and asymmetrical growth. All were found to carry the same de novo mutation, p.Glu17Lys, in the serine/threonine kinase AKT2, in two cases as heterozygotes and in one case in mosaic form. In heterologous cells, the mutant AKT2 was constitutively recruited to the plasma membrane, leading to insulin-independent activation of downstream signaling. Thus, systemic metabolic disease can result from constitutive, cell-autonomous activation of signaling pathways normally controlled by insulin.

Insulin promotes energy storage and growth through effects on glucose, lipid, and amino acid metabolism; cell division; and apoptosis. These are mediated by a transmembrane tyrosine kinase receptor that phosphorylates insulin receptor substrate (IRS) and other adaptor proteins to initiate signaling events including, critically, activation of AKT serine/threonine kinases. Murine studies suggest that the Akt2 isoform is most closely linked to insulin’s metabolic effects (1). Consistent with this, a loss-of-function mutation in AKT2 produces severe insulin resistance in humans (2).

Insulin hypersecretion from tumoral or genetically hyperactive pancreatic β cells produces hypoglycemia, weight gain, and accelerated early growth. We previously described a child with severe recurrent hypoglycemia from infancy and undetectable plasma insulin despite the classical biochemical profile of hyperinsulinism (i.e., low serum levels of ketone bodies and branched chain amino acids and no elevation of free fatty acids). He also had an increased birth weight and a left-sided overgrowth (Fig. 1A and table S1) (3). We have subsequently studied two further patients with a similar metabolic profile and a left-sided facial overgrowth [table S1, fig. S1, and supporting online material (SOM) text]. From around 6 months old, each patient exhibited symptomatic severe hypoglycemia, including reduced consciousness and seizures, when fasted beyond 3 hours, which required surgical implantation of enteral tubes for long-term overnight carbohydrate feeding.

Fig. 1

Identification of a de novo activating AKT2 mutation. (A) Patient 1, showing increased left leg size, increased left truncal adipose tissue, and surgical feeding tube. (B) c.49G>A (p.Glu17Lys) mutation in the AKT2 gene seen in either heterozygous or mosaic form. (C) Pedigrees of affected patients, showing de novo occurrence of the mutation in each case. E, Glu; K, Lys. (D) Domain structure of AKT2, showing location of Glu17 in the PH domain and the location of residues phosphorylated by PDK1 and mTORC2 during activation. (E) Plasma membrane recruitment of hemagglutinin-tagged AKT2 p.Glu17Lys but not wild type (WT) in serum-starved HeLa cells treated with either phosphate-buffered saline (PBS) or 100 nmol/l insulin. Nuclei are stained with DAPI. Scale bars, 20 μm.

All patients were born to unaffected, unrelated Europid parents. The sequences of genes encoding preproinsulin, insulin-like growth factors 1 and 2, and their receptors were normal, and no basal hyperphosphorylation of insulin receptors in fibroblasts was seen. Exome-wide sequencing of the index case was undertaken, and resulting variants were filtered to exclude those that were common or not predicted to alter amino acid sequences (4). This yielded 326 rare mutations, including the heterozygous c.49G>A mutation in the AKT2 gene, which leads to substitution of Glu17 in the pleckstrin homology (PH) domain for lysine (Fig. 1, B to D). Sanger sequencing confirmed the heterozygous mutation in lymphocytes and in left- and right-sided dermal fibroblasts from the proband, but in neither parent, consistent with a de novo germline or very early somatic mutation—and further detected the same mutation in the other two probands. One was heterozygous for the mutation, and the other showed 20% mosaicism in lymphocyte and cheek epithelial DNA (4). No canonical insulin-responsive tissues were available to assess their mutation burden. None of their parents carried the mutation, which was also absent in 1130 control genomes and exomes (470 Europid) (4).

Insulin-induced AKT activation requires type 1 phosphatidylinositol 3-kinase (PI3K), which generates phosphatidylinositol 3,4,5-trisphosphate (PIP3) in the plasma membrane. The PH domain of AKT then docks with PIP3, leading to its apposition to the stimulatory kinases phosphoinositide-dependent kinase 1 (PDK1) and mammalian TOR complex 2 (mTORC2) and thus activation (5). Overexpression of Akt2 p.Glu17Lys in 3T3-L1 adipocytes produces non–insulin-dependent membrane localization of the GLUT4 glucose transporter (6). Consistent with this, we found the mutant AKT2 to exhibit plasma membrane localization even in serum-starved HeLa cells (Fig. 1E) and to produce tonic nuclear exclusion of FoxO1 in 3T3-L1 cells (fig. S2) (4).

Somatic AKT2 mutations have been described rarely in cancer, although not, to date, p.Glu17Lys (7). In contrast, the homologous p.Glu17Lys mutation in AKT1 is commonly found in cancer (8), and while this paper was in review somatic mosaicism for AKT1 p.Glu17Lys was reported to underlie the dramatic progressive skin and skeletal overgrowth of Proteus syndrome (9). The very different, mostly metabolic, phenotype of patients with germline or mosaic AKT2 p.Glu17Lys emphasizes the distinct functions of AKT1 and AKT2 in humans. However, the accelerated growth seen in our patients suggests that AKT2, too, may have some growth promoting effect, although the observed asymmetry is unexplained.

Supporting Online Material

Materials and Methods

SOM Text

Figs. S1 to S3

Table S1


References and Notes

  1. Materials and methods are available as supporting material on Science Online.
  2. Acknowledgments: R.K.S., S.O’R., I.B., and D.B.S. are supported by the Wellcome Trust (grants 080952/Z/06/Z, 078986/Z/06/Z, 077016/Z/05/Z, and 091551/Z/10/Z), the Medical Research Council Centre for Obesity and Related Disorders, and the UK NIHR Cambridge Biomedical Research Centre. B.C. is supported by NIHR and Wellcome Trust, and M.M. by the Gates Cambridge Trust. We thank L. Boyle for confocal microscopy assistance.
View Abstract

Navigate This Article