Somatic mutations are present in all members of the AKT family in endometrial carcinoma

Sir, 
 
The activating E17K mutations recently discovered in the pleckstrin homology domain of AKT1 in about 2% of endometrial cancer patients (Shoji et al, 2009) suggest a new mechanism for PI3 kinase pathway activation in these patients, as previously described in breast, colorectal and ovarian cancers (Carpten et al, 2007). Additional mechanisms of PI3 kinase pathway activation in endometrial cancer include the somatic mutation of PTEN and PIK3CA (Kong et al, 1997; Oda et al, 2005), amplification and overexpression of PIK3CA (Miyake et al, 2008; Salvesen et al, 2009), and decreased expression of PTEN (Kanamori et al, 2001; Kappes et al, 2001). 
 
We found an additional four mutations in AKT family members (Table 1). Two of these (mutations in the catalytic domain of AKT2 (D399N) and the regulatory domain of AKT3 (E438D)) were previously reported in a sequencing screen of 123 genes in 41 primary endometrial cancers (Dutt et al, 2008). Manual reinspection of these data in light of the report on activating AKT1 mutations in endometrial cancer (Shoji et al, 2009) revealed an additional mutation, AKT1 E17K, identical to the one reported by Shoji et al (2009), and a novel mutation in the catalytic domain of AKT2 (R368C) in two additional samples. We validated these mutations as somatic by mass spectrometric genotyping of the tumour and matched normal DNA, after an independent PCR amplification. We also found a novel candidate mutation in the pleckstrin homology domain of AKT2 (D32H), which we could not validate because of insufficient DNA. All these mutations occurred in cancers of the endometrioid subtype that had no signs of metastasis either at primary treatment or during follow-up. Taken together, we find that 5 out of 41 endometrial cancers have mutations in AKT family members for a 12% rate. 
 
 
 
Table 1 
 
AKT family mutations found in endometrial cancer 
 
 
 
Confirmation that these novel mutations activate the PI3 kinase pathway awaits their functional characterisation. Notably, all the AKT family member mutations found in our data occur at residues conserved across multiple species (see Supplementary Figure 1). However, three of these five mutations were identified in samples harbouring mutations of PTEN, one of which also had amplification of and a mutation in PIK3CA (Table 1); the AKT1 E17K mutation is not associated with either PTEN or PIK3CA genomic alteration. It is therefore possible that these AKT family mutations have different functional effects from mutations of PTEN and PIK3CA. Given the importance of the PI3 kinase pathway in endometrial cancer oncogenesis (Salvesen et al, 2009), and the emerging therapeutic options for PI3 kinase inhibition (Garcia-Echeverria and Sellers, 2008), the functional effects of all these AKT family mutations should be investigated in appropriate model systems of endometrial cancer.


Sir,
We appreciate the attention given by Drs Dutt, Salvesen, Greulich, Sellers, Beroukhim and Meyerson to our recent publication 'The oncogenic mutation in the pleckstrin homology domain of AKT1 in endometrial carcinomas' (Shoji et al, 2009). In this article, we report a 2% mutational frequency of AKT1 (E17K) among 101 endometrial carcinomas. We also described that these two AKT1 mutant tumours do not possess any mutations in PIK3CA, PTEN and K-Ras. Our report has proposed two issues to be clarified: (1) Are there any 'oncogenic' mutations in other AKT family members in endometrial carcinomas? (2) Are all the AKT family mutations mutually exclusive with other PI3 kinase-AKT-activating mutations?
Including the data in their earlier report, Dutt et al (2008) revealed mutations in AKT2 and AKT3, as well as in AKT1. As for AKT1 (E17K) mutations, their data and our data are compatible, showing that AKT1 mutations were detected in 2% of the endometrioid subtype in endometrial cancer. Compared with the accumulative data in AKT1 (E17K) mutations, mutations in AKT2 and AKT3 are not well characterised. Carpten et al (2007) and Kim et al (2008) reported no E17K mutations in AKT2 and AKT3 in breast, colorectal, gastric, hepatocellular, lung carcinomas and acute leukaemias. Davies et al (2008) first reported AKT3 (E17K) mutations in melanoma at 1.5% frequency. Parsons et al (2005) found two mutations of AKT2 (S302G and R371H) in 204 colorectal cancer samples, and Soung et al (2006) reported one missense mutation (A377V) and two possible splice-site mutations in intron 11 of AKT2 in gastric and lung adenocarcinoma. However, the physiological role of AKT2 mutations, including those (D399N, R368C and D32H) reported by Dutt et al (2008), has not been validated yet. In addition, the AKT3 (E438D) mutation has not been reported in any type of tumours. It is important to clarify whether these rare AKT2/3 mutations cause an oncogenic effect in cancer.
We previously reported that PIK3CA mutations frequently coexist with mutations in PTEN and/or KRAS, and suggested that the PIK3CA mutation might require another upstream input to fully activate the PI3K kinase -AKT pathway (Oda et al, 2005(Oda et al, , 2008. Their data of coexistent mutations in AKT1 (E17K) and KRAS (G13D) suggest that the KRAS mutation alone is insufficient for a full activation of the PI3 kinase -AKT pathway. Their data are not inconsistent with the hypothesis that the AKT1 (E17K) mutation and the PIK3CA mutation are mutually exclusive. The coexistent mutations of AKT3 and PIK3CA in one sample suggest that some types of AKT mutations (non-E17K) might coexist with PIK3CA mutations to enhance the activation of the PI3 kinase -AKT pathway. Further analyses are necessary to clarify the frequency of coexistent mutations in AKT1 (E17K), AKT2/3 and other mutations.
Their current data and our data suggest that the PI3 kinase -AKT pathway is prevalently activated in endometrial cancer through various genetic alterations and their combinations. Further analyses in AKT, PIK3CA and other PI3 kinase-AKTrelated genes would be helpful to comprehensively understand the mechanism of activation in this pathway and to use PI3K-targeted therapies in various types of cancer.
We thank Dr Dutt and colleagues for their recognition of our work.