Variation in Akt protein kinases in human populations

Abstract

The three Akt kinases are related proteins that are essential for normal growth and metabolic regulation and are implicated as key signaling mediators in many physiological and pathophysiological processes. Each Akt is activated by common biochemical signals that act downstream of growth factor and hormone receptors via phosphatidylinositol-3 kinase, and each controls several downstream pathways. The importance of Akt actions in human physiology is strengthened by the rarity of modifying mutations in their genes and by the devastating impact caused by these mutations on growth and development and in disorders such as cancer. Recent advances in genomics present unique opportunities for enhancing our understanding of human physiology and disease predisposition through the lens of population genetics, and the availability of DNA sequence data from 60,706 people in the Exome Aggregation Consortium has prompted this analysis. Results reveal a cohort of potential missense and other alterations in the coding regions of each AKT gene, but with nearly all changes being uncommon. The total number of different alleles per gene varied over an approximately threefold range, from 52 for AKT3 to 158 for AKT2, with variants distributed throughout all Akt protein domains. Previously characterized disease-causing mutations were found rarely in the general population. In contrast, a fairly prevalent amino acid substitution in AKT2 appears to be linked to increased predisposition for type 2 diabetes. Further analysis of variant Akt molecules as identified here will provide opportunities to understand the intricacies of Akt signaling and actions at a population level in human physiology and pathology.

the akt or pkb family of serine/threonine protein kinases (Akt1, Akt2, and Akt3 in humans) has been subjected to intensive scrutiny since initial discoveries in the 1980s and 1990s (4, 5, 22, 23, 26, 34). The three Akt proteins are highly related to each other and share similar functions (23). For example, human Akts range in length from 479 to 481 amino acids, are 77–82% identical to one another in pairwise comparisons, and are structurally very similar, with each protein being composed of four major modules, a plekstrin homology domain, a linker region, a protein kinase segment, and a regulatory domain (Fig. 1) (23). Each Akt can be activated by common biochemical pathways that act downstream of growth factor and hormone receptors via the signaling enzyme phosphatidylinositol (PI) 3-kinase (4, 22, 23), and each controls several downstream pathways, including those involving Forkhead box (FoxO) transcription factors, the mammalian target of rapamycin complex (mTORC)1 protein kinase, and the multifunctional enzymes glycogen synthase kinases-α and -β, among others (23). Initial analyses of physiological readouts of Akt actions demonstrated shared effects on cell proliferation, survival, and intermediary metabolism (9), whereas subsequent studies identified some distinct functions for each individual protein (6, 7, 10, 27, 35). For example, in mice, global genetic deficiency of Akt1 caused diminished somatic growth (7), whereas knockout of the Akt2 gene led to insulin resistance and diabetes mellitus (6), and loss of the Akt3 gene caused a decline in brain size (10). In addition, combined deficiencies of Akt1 and Akt2 genes produced more severe growth deficits and caused neonatal death (27), and lack of both Akt1 and Akt3 caused fetal death (35).

Fig. 1.

Population variation in human Akt proteins. A–C: 3 human Akt proteins are composed of 4 identifiable domains: pleckstrin homology (PH), linker (L), protein kinase (kinase), and regulatory (Reg). Scale bar represents 100 amino acids. A: human Akt1 consists of a 481-residue mature protein. The overall population prevalence of variant alleles for each segment of the protein is listed below the map. B: human Akt2 consists of a 480-amino acid mature protein. The overall population distribution of variant alleles for each segment of the protein is listed below the map, and the prevalence of the most common variant is depicted in single-letter amino acid code. C: human Akt3 consists of a 479-residue mature protein. The overall population distribution of variant alleles for each segment of the protein is listed below the map.

Studies in cultured cells also have identified potentially unique actions for different Akts. In human keratinocytes, engineered knockdown of Akt1 enhanced cell death and prevented differentiation, whereas loss of Akt2 had no effect (32). In human breast cancer cell lines, overexpression of a constitutively active Akt stimulated cell proliferation and tumor growth in nude mice (20) but paradoxically reduced cell motility and metastasis (20, 37). By contrast, loss of Akt1 in breast cancer cells increased motility and enhanced invasiveness by mechanisms dependent on continued expression of Akt2 (14). Thus, each Akt appears to have both distinct and overlapping functions that may vary depending on the cell or tissue type analyzed.

Human diseases caused by mutations or changes in levels of Akt family members are rare (23). Among these uncommon disorders are a series of childhood overgrowth syndromes, in which activating mutations in components of the PI3 kinase/Akt/mTORC signaling pathway are found primarily in affected tissues (2, 24, 25). Proteus syndrome [see Online Mendelian Inheritance in Man (OMIM): https://www.omim.org/, no. 176920] consists of asymmetric overgrowth of skin, bones, adipose, and other tissues and is associated with a high propensity for developing deep vein thrombosis (2, 25). The cause in the majority of the few dozen cases reported to date is a mutation in the AKT1 gene, which changes E¹⁷ > K but is expressed in a mosaic fashion only in affected cells and tissues (2). Hypoinsulinemic hypoglycemia and hemihypertrophy (OMIM no. 240900) is caused by the corresponding E¹⁷ > K alteration in AKT2, which is also expressed in a mosaic way. To date, it has been identified in just four individuals (2). Megalencephaly/polymicrogyria/polydactyly/hydrocephalus syndrome (OMIM nos. 604487, 615937, and 615938) comprises a set of progressive disorders that are also associated with developmental delay and/or intellectual disability and a variety of dysmorphic features. Of the more than 40 individuals reported to date, the most common cause is a mosaic amino acid substitution in the gene encoding the PI3 kinase regulatory subunit PI3KR2, but other affected individuals have been found with one of three alterations in AKT3 that change either E¹⁷ > K, N²²⁹ > S, or R⁴⁶⁵ > W (2). Cowden syndrome (OMIM no. 615109) is a genetic precancerous disorder that is associated with the development of carcinomas of the breast, endometrium, thyroid gland, and other organs or tissues (24). The majority of individuals with this disorder have been found to harbor inactivating germ line mutations in the gene for PTEN. This gene encodes the enzyme that dephosphorylates phosphatidylinositol-3,4,5-trisphosphate (PIP3) to PIP₂ and thus normally antagonizes activation of Akt by PIP3, the product of PI3 kinase (12). Other causes of Cowden syndrome include activating mutations in the PI3 kinase catalytic subunit PIK3CA or in AKT1, including R²⁵ > C or T⁴³⁵ > P (24). In addition to these uncommon disorders, autosomal dominant type 2 diabetes has been reported in a single family with a substitution in the AKT2 gene of R²⁷⁴ > H (11).

Acquired predicted amino acid substitution and protein truncation mutations have been found in AKT1, AKT2, or AKT3 in patients with a variety of different cancers (see the Genomic Data Commons Data Portal and the cBio portal for Cancer Genomics at https://portal.gdc.cancer.gov/ and at http://www.cbioportal.org/, respectively, and Supplemental Tables S1, S2, and S3; Supplemental Material for this article can be found online at the AJP-Regulatory, Integrative and Comparative Physiology web site). For example, an AKT1 E¹⁷ > K modification has been detected in several percentages of individuals with breast, colorectal, or ovarian cancers (36).

Population-based genome sequencing has the potential to provide new insights into human variation and disease susceptibility (1, 17, 28). The recent release of DNA sequence data from the gene exons of more than 60,500 people by the Exome Aggregation Consortium (ExAC) (3, 18, 19, 31, 38) has provided an opportunity to define the range of variation within each Akt in humans. Results reveal a remarkable paucity of potential missense and other alterations in the coding regions of each of the three human Akt genes, in contrast with data from several other protein kinase genes (29, 30), with nearly all of the few alterations identified being very uncommon. Taken together, these results will provide new opportunities to understand the range and extent of Akt signaling dynamics in different physiological and pathological contexts.

METHODS

Databases and analyses.

Data on variation in human AKT1, AKT2, and AKT3 were obtained from the ExAc genome browser (http://exac.broadinstitute.org/), which contains results of exome sequencing of 60,706 individuals (16). Human transcripts and genes were from the Ensemble Genome Browser (www.ensemble.org) genome assembly GRCh38. The source of human protein sequences was the National Center for Biotechnology Information (NCBI) Consensus CDS Protein Set (https://www.ncbi.nlm.nih.gov/CCDS/), and the Uniprot browser (http://www.uniprot.org/). Other databases examined included Online Mendelian Inheritance in Man (OMIM, https://www.omim.org/), the Genomic Data Commons Data Portal (https://portal.gdc.cancer.gov/), the cBio portal for Cancer Genomics (http://www.cbioportal.org/), and the Type 2 Diabetes Knowledge Portal (http://www.type2diabetesgenetics.org/).

RESULTS

Allelic variation in the Akt family in humans.

ExAC contains DNA sequencing results from the exomes of 60,706 people from different population groups throughout the world (19) and thus represents a large source for potential protein sequence variation in humans. The overview from initial analysis of the data derived from these 121,412 alleles is that there is substantial variation within coding regions of genes (19). However, most alterations were found to be uncommon, as >50% were seen in just a single allele, and >99% were found in <1% of the study population (19).

Examination of the three Akt family members in ExAC revealed limited coding variation in their exons, with most of the changes consisting of missense mutations or in-frame insertions or deletions (92–96% of modified alleles; Table 1). Second most common were alterations in the reading frame, including truncating stop codons (3–4%; Table 1). The total number of different alleles in the study population varied over an approximately threefold range from 52 for AKT3 to 158 for AKT2 or 0.11–0.33 nonsynonymous changes per codon when corrected for protein length (Table 1; note that the numbers differ from those presented in the heading for each Akt in the ExAC website but agree with the actual data). When examined for prevalence within the population, 59–75% of changes were seen in a single allele (Table 1), >98% were found in ≤0.03% of the ExAC cohort, and all were detected in ≤0.1%. These modifications were distributed throughout the different domains of the three Akts (Fig. 1), with only a single amino acid substitution variant in the plekstrin homology region of Akt2 reaching a prevalence of 0.1% (Fig. 1). Remarkably, very few changes were located near the two regulated phosphorylation sites in each protein (Thr³⁰⁸ and Ser⁴⁷³ in Akt1, Thr³⁰⁹ and Ser⁴⁷⁴ in Akt2, or Thr³⁰⁹ and Ser⁴⁷² in Akt3). A single predicted variant was found once in the population at Y⁴⁷⁴ in AKT1, a single variant was detected once in the study group at Ser⁴⁷⁶ in AKT2, and a single variant at Thr³⁰⁶ in AKT2 was seen in 17 alleles (0.015% of the population). Collectively, these results indicate that variation in Akt proteins is extremely low in humans, being even less common than the average in ExAC, as noted above (19).

Table 1.

Human population variation in AKT1, AKT2, and AKT3

Protein	No. of Codons*	Missense and In-Frame Insertions/Deletions	Frame Shifts;Stop Codons	Splicing Site Changes	Loss of Start Codon	Loss of StopCodon	Total No. of Different Changes	Variants Occurring Once	Total Variant Alleles in Population
AKT1	480	99	3	1	0	0	103	68	0.17%
AKT2	481	146	7	5	0	0	158	90	0.67%
AKT3	479	48	2	2	0	0	52	39	0.08%

For AKT2, 31 variants (∼20%) mapped to transcripts corresponding to smaller predicted proteins of 56, 69, 125, 129, 292, or 450 residues. For AKT3, 5 variants (∼10%) mapped to a transcript corresponding to a predicted protein of 465 residues.

^*All AKT1 variants mapped to the 480 codons corresponding to the full-length protein.

Reading frame alterations, addition of stop codons, and splicing changes at exon-intron and intron-exon junctions can all contribute to loss of protein expression and thus absence of function. The number of alleles showing these changes was minimal among the three human AKT genes, ranging from four to 13 different instances, with all alterations being found rarely in the population (0.001 to 0.06% allelic frequency). Similarly, copy number variation, in which all or part of a gene is amplified in the genome, also was very low for AKT family members in the study population, as seen in most genes in ExAC (31), and ranged from three to eight instances.

Population variability and disease.

To date, 150 different disease-associated mutations have been reported for Akt1, with nearly all being linked with different cancers (see Genomic Data Commons Data Portal and cBio portal for Cancer Genomics and Ref. 36). Of these alterations, 38 alleles are found in ExAC, affecting 34 different sites in the protein (see Supplemental Table S1). Most were present at ultralow frequencies (1 to 11 alleles in the population), with only one change, D⁴⁶ > E, being found at an allelic prevalence as high as ∼0.03% (35 alleles; see Supplemental Table S1).

Disease-linked alterations of amino acids in Akt2 and Akt3 were nearly as common as those reported for Akt1 (93 and 121 instances, respectively; see Supplemental Tables S2 and S3), bus were also rare in the ExAC population. Twenty-nine amino acid substitutions at 25 locations have been detected in Akt2 (Supplemental Table S2) and 11 at 11 sites in Akt3 (see Supplemental Table S3). There was a single, fairly common variant in Akt2, P⁵⁰ > T, which was seen in 0.1% of all alleles (Fig. 1B and Supplemental Table S2). This modification appears to be a true polymorphism in Finland, where it is found in 1.1% of the population and where it has been associated with diminished sensitivity to exogenous insulin and an increase in fasting plasma insulin values (21).

DISCUSSION

Data from population-based exon sequencing from ExAC (19) have been analyzed here to gain insights into the population genetics of Akt family proteins in humans. Results identify a small number of possible missense alterations and other modifications in the coding regions of each of the three genes studied. These substitutions and in-frame deletions occurred throughout the protein sequences of the three Akts (Fig. 1), but with allelic frequencies that were generally so low that they are unlikely to individually have a significant population impact on human physiology. In fact, only a single predicted amino acid substitution was found in Akt2 with a frequency as high as 0.1% of alleles (Fig. 1B). These results indicate that unlike the JAK2 protein kinase or the insulin or IGF-I receptor kinases, in which fairly common protein sequence variants are present in the population at several different locations and in ≤0.7% of alleles (29, 30), signaling variants in Akt proteins are rare.

Akt signaling and human disease.

Analysis of public databases has revealed that ∼120 distinct alterations in the Akt1 protein sequence are associated with different human disorders (Supplemental Table S1), with the corresponding figures for Akt2 and Akt3 being 33 and 36, respectively (Supplemental Tables S2 and S3), with most of the changes being amino acid substitutions (Supplemental Tables S1, S2, and S3). Nearly all of these modifications have been found by DNA-sequencing studies in human cancers (see the Genomic Data Commons Data Portal and the cBio portal for Cancer Genomics), a few have been linked to type 2 diabetes (see Type 2 Diabetes Knowledge Portal, and Ref. 11), and several others are associated with several distinct and rare childhood overgrowth syndromes (2, 24, 25). The majority of these disease-associated alleles were absent from the ExAC population, with the few identified being found with ultralow allelic frequencies in the 60,706 people assessed (Supplemental Tables S1, S2, and S3). Taken together, these results further illustrate that there are likely to be a minimal number of disease-neutral variants in human AKT genes in the general population. For example, the most common modification, the predicted amino acid substitution, P⁵⁰ > T in Akt2 (allelic frequency of 0.1%; Fig. 1B and Supplemental Table S2), has very recently been shown to be associated with diminished glucose tolerance and insulin resistance in individuals in Finland, where it is present in the population at a prevalence of 1.1% (21). In preliminary cell culture studies, an Akt2 protein containing this alteration exhibited reduced activation in response to insulin and a consequent reduction in Akt kinase activity in comparison with the wild-type allele (21). Proline 50 is located in the plekstrin homology region of Akt2 (see Fig. 1), which is the segment of the protein that interacts with PIP3, the product of PI 3-kinase (4, 22, 23), and it seems probable that the change of this amino acid to threonine may negatively influence recruitment of Akt2 to the plasma membrane, where it is normally phosphorylated by PDK1 as part of its activation in response to hormonal or growth factor signaling (4, 22, 23). In Akt2, P⁵⁰ is adjacent to another proline residue at position 51, a site in which a substitution in a single allele has been reported in ExAC (P⁵¹ > L). Similarly, rare modifications are found at the equivalent of the second proline in Akt1 and Akt3 in different cancers (P⁵¹ > L in Akt1 and P⁵⁰ > L in Akt3; Supplemental Tables S1 and S3; there is no proline in Akt1 or Akt3 that corresponds to P⁵⁰ in Akt2).

Limitations and strengths of population-based genome sequence data for understanding Akt actions.

As with any DNA-sequencing project, ExAC contains the raw material for new biological and biomedical observations as well as both ambiguities and errors. From the perspective of the three AKT genes, potential problems include the choice of minor transcripts as reference mRNAs, with the probability that the encoded proteins may not actually be synthesized. This is particularly true for Akt2 (see Table 1). Other limitations of the data include the potential nonrepresentative nature of the study population. More than 60% of samples derive from Europeans, ∼20% are from South or East Asia, and only ∼8% are each from African or Hispanic populations (19). Thus, the true rate and extent of variation among these proteins in humans may not be established yet. This becomes clear when examining the most prevalent AKT2 allele, the change of P⁵⁰ > T. It is present in ∼0.1% of ExAC chromosomes. In a subsequent study, it was found in 1.1% of 26,306 subjects from Finland yet was virtually absent in Hispanic-American or African-American cohorts (21). In addition, there is a likely error rate associated with nucleotide changes that appear only once in the 121,412 ExAC chromosomes.

Despite these limitations, the data in ExAC provide potentially exciting new opportunities to reevaluate the actions of Akts in human physiology and pathology. From a physiological perspective, Akt proteins play major roles in the biology of nearly all specialized cell types in humans as central nodes in the PI3-kinase/Akt/mTORC and PI3-kinase/Akt/FoxO signaling pathways (23), including the immune system, metabolic pathways (liver, adipose, skeletal muscle, pancreatic islets of Langerhans), the cardiovascular system (heart, endothelial cells, blood vessels), the nervous system, and others (23). Subtle differences in the extent or duration of Akt activity because of specific amino acid substitutions or loss- or gain-of-function alleles could each cause small changes in specific outcomes that over time could alter susceptibility or resistance to disease, such as inflammatory disorders, glucose intolerance/diabetes mellitus, myocardial infarction, vascular disease, stroke, and others.

Perspectives and Significance

The extensive variability revealed by ExAC represents the legacy of our ancestors, including extinct populations such as Neanderthals, Denisovans, and others (8, 13, 15, 33), as modern humans contain DNA marks documenting past interactions with many different groups. New hypotheses based on the data in ExAC and from other genome-sequencing projects could lead to novel insights about how the biology of various signal transduction pathways has been shaped over millennia, including the many cascades in which the three Akts play a central role (23). Opportunities to develop new research questions exist for these and other signaling pathways and should incentivize investigators to analyze, critically evaluate, and interpret these new large-scale genomic data.

GRANTS

These studies were supported in part by National Institutes of Health Research Grant, 5-R01-DK-042748-27 (to P. Rotwein).

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the author.

AUTHOR CONTRIBUTIONS

P.R. conceived and designed research; P.R. performed experiments; P.R. analyzed data; P.R. interpreted results of experiments; P.R. prepared figures; P.R. drafted manuscript; P.R. edited and revised manuscript; P.R. approved final version of manuscript.