PI3K/AKT pathway activation in acute myeloid leukaemias is not associated with AKT1 pleckstrin homology domain mutation

Summary

Despite its’ central role, the precise mechanisms of the phosphoinositide 3-kinase/Akt (PI3K)/Akt pathway activation in acute myeloid leukaemia (AML) have not been elucidated. Recently, a recurrent novel AKT1 pleckstrin homology domain (PHD) mutation leading to membrane translocation, constitutive AKT activation and leukaemia development in mice was described. To assess AKT1 PHD mutations in AML, we sequenced 57 specimens from 49 AML patients, all of whom showed PI3K/AKT pathway activation by analysis of total and phospho-protein expression for AKT, mTor, p70S6Kinase, S6ribosomal protein and PTEN. No mutations in AKT1 PHD were identified, making this mutation an unlikely cause of PI3K/AKT pathway activation in AML.

The phosphoinositide 3-kinase/Akt (PI3K/AKT) pathway is a central regulator of cellular proliferation, survival and apoptotic signalling and a major contributor to malignant transformation. Proteins in this pathway are often constitutively phosphorylated/activated in up to c. 60–80% of primary acute myeloid leukaemia (AML) blast samples (Min et al, 2003; Xu et al, 2003; Martelli et al, 2006). Activation contributes to chemo-resistance (Min et al, 2003; Xu et al, 2003; Martelli et al, 2006) and correlates with poor clinical outcome (Min et al, 2003; Xu et al, 2003; Martelli et al, 2006), although this has become controversial recently (Tamburini et al, 2007). Conversely, PI3K or AKT inhibitors sensitize to varying agents (Martelli et al, 2006). Inhibition of the downstream target mammalian target of rapamycin (mTOR) has been clinically validated as a potential site for therapeutic intervention (Recher et al, 2005; Yee et al, 2006). However, the question remains as to the causes of activation of the PI3K and AKT kinases, which, for the most part, are activating phosphorylation events with subsequent increased kinase activity. Growth factors and upstream (mutated) signalling molecules, such as FLT3, Ras and c-Kit are able to activate PI3K and AKT (Martelli et al, 2006). In solid tumours, reduced protein expression and loss or mutations of PTEN can lead to AKT activation. In AML, PTEN mutations are rare (Cheong et al, 2003) and paradoxically high levels of expression and phosphorylation of PTEN seem to negatively affect outcome (Cheong et al, 2003; Martelli et al, 2006), conceivably through a reduced activity of phosphorylated PTEN (Martelli et al, 2006). Activating PI3K mutations have been described in solid tumours (Carpten et al, 2007) but, except in single cases, not in AML (Bousquet et al, 2005).

Until the recent paper by Carpten et al (2007) was published, no major activating mutations in AKT had been identified. Given the importance of PI3K and AKT in leukaemogenesis, we examined AML samples for the occurrence of a novel AKT1 pleckstrin homology domain (PHD) mutation (AKT1 E17K), possibly explaining constitutive activation of AKT and downstream effectors in AML.

Materials and methods

AML specimens

Samples were collected and assayed under Institutional Review Board approved protocols at the U.T. MD Anderson Cancer Center. Specimens included a wide range of French–American– British (FAB) (predominance of M4, M5, M2 and M1) and cytogenetic subgroups (Table I). Analytes (DNA and protein) were extracted using standard protocols as described and referenced (Tibes et al, 2006). Eight patients had blood and marrow samples sequenced in parallel. Two cell lines, Molm13 and THP-1, were also assessed.

Table I.

Baseline characteristics of the patient samples.

Number of samples (n) (all newly diagnosed AML)
Patients/specimens	49/57
With protein data	46/49
AHD	15
Prior cancers	12
Cytogenetics
Diploid	24
chr. 5/7	6
Complex	8
inv 16	1
inv 9, t(6;9), +8, +21	5
Non-evaluable/IM	5
FAB
M0	6
M1	3
M2	12
M4	14
M5	7
M6/M7	2
Unknown	5

Samples used for mutation screening and protein expression. Fifty-seven samples from 49 patients, eight patients had specimens from blood and marrow. For 46 of 49 patients, protein expression data was available.

AML, acute myeloid leukaemia; AHD, antedecendent haematological disorder; FAB, French–American–British Classification; chr. 5/7, any abnormality of chromosome 5 or 7; +8/+21, trisomie 8 or 21; IM, insufficient metaphases.

Reverse phase protein array analysis

Reverse phase protein array analysis (RPPA) is a protein array that quantitatively measures total and corresponding phosphoprotein expression (Kornblau et al, 2006; Tibes et al, 2006) of proteins, thus more comprehensively assessing protein expression and activation of an entire pathway (Petricoin et al, 2005; Kornblau et al, 2006; Tibes et al, 2006). Methodological details have been published and presented (Kornblau et al, 2006; Tibes et al, 2006). The present study reports protein data on 10 proteins, supporting the evidence of protein expression, phosphorylation, thus activation of the PI3K/AKT pathway in most samples.

Mutational analysis

Polymerase chain reaction (PCR) primers were designed to amplify genomic DNA encompassing coding sequences and adjacent intronic boundaries for exon 3 of the human AKT1 gene (AKT1, NM_005163), in addition to exon 9 and exon 20 of PIK3CA (NM_006218). PCR primers were tailed with universal M13 sequences to standardize cycle sequencing. Primer sequences and appropriate annealing temperatures are available upon request. PCR amplicons were purified using the AMPURE® magnetic bead system (Agencourt, Beverly, MA, USA) and sequenced using BIGDYE® Terminator chemistry (Applied Biosystems, Foster City, CA, USA). Cycle sequencing products were separated by capillary electrophoresis on ABI DNA Analyzers. Raw sequencing data was imported into Sequencher 4.7 (GeneCodes) for alignment and variant analysis. For each exon, a normal germline DNA from Centre d’Etude du Polymorphisme Humain (CEPH, Paris, France) sample 1347-02 was sequenced and used as a normal reference along with publicly available RefSeq consensus sequences.

AKT1 49G>A Sequenom MassARRAY analysis

Genotyping for detection of the AKT1 49G>A mutation was performed using the Sequenom MassARRAY^™ platform (San Diego, CA, USA) with iPLEX^™ chemistry according to the manufacturer’s recommendations. Briefly, an iPLEX^™ assay was designed utilizing the Sequenom Assay Design software, allowing single base extension (SBE) designs. PCR/SBE primer sequences are available upon request. Five to 10 ng of genomic DNA were amplified by PCR and treated with shrimp alkaline phosphatase (SAP) to neutralize unincorporated dNTPs. Subsequently, a post-PCR single base extension reaction was performed using concentrations of 0.625 μmol/l for low mass primers and 1.25 μmol/l for high mass primers. Reactions were diluted with 16 μl of H₂O and fragments were purified with resin, spotted onto Sequenom SpectroCHIP^™ microarrays and scanned by MALDI-TOF mass spectrometry. Individual SNP genotype calls were generated using Sequenom TYPER^™ software, automatically calling allele specific peaks according to their expected masses. A water sample served as negative control. As a positive control for the mutant allele, we genotyped DNA from a tumour with a known heterozygous AKT1 49G>A, which was previously reported (Carpten et al, 2007). Furthermore, a normal germline DNA from CEPH sample 1347-02, which is homozygous for the wildtype allele, was genotyped as a wildtype control. All samples were genotyped in duplicate.

Results and discussion

Protein data was available for almost all patients (n = 46/49) for either blood or marrow. All 46 samples arrayed by protein array expressed measurable total AKT and phospho-AKT (Ser308 and Thr473) protein levels, as well as major downstream proteins (Fig 1), indicating a general activation of PI3K and AKT in these samples. PTEN expression and phosphorylation has been described in AML (Cheong et al, 2003) and these results are confirmed here. Assessing AKT1 E17K mutation status in our cohort of 57 samples (49 patients) revealed no mutations. To more comprehensively assess the PI3K/AKT pathway, we also sequenced exons containing the two most commonly mutated codons within the helical and activating domains of PIK3CA, which are codons 545 (exon 9) and codon 1047 (exon 20) respectively. We did not find mutations in these mutational hot spots of PIK3CA. Both cell lines, Molm13 and THP-1 were negative for AKT and PIK3CA mutations.

Fig 1.

Total and corresponding phospho-protein expression of major AKT pathway proteins. Protein expression in samples from 46 of the 49 patients on which mutation analysis was performed showed detectable levels of major PI3K/AKT pathway proteins. Proteins: Akt and phospho-Akt(Ser308), phospho-AKT(Ser473) was also expressed in the samples but due to different scaling and expression intensities is not shown in this scatter plot; mTor and phospho-mTor(Ser2448); P70S6K: p70S6 kinase and phospho-p70S6 kinase (Thr389); PTEN: PTEN and phospho-PTEN (Ser380); S6: S6 ribosomal protein and phospho-S6 ribosomal protein(Ser240-244). Relative expression intensity is the expression intensity of each protein after normalization using the RPPA approach as described previously (Tibes et al, 2006).

We understand that our sample size is limited; however, based upon published power estimates, with c. 96 haploid genomes, we would have >95% power to detect a mutation in our samples with a frequency >1% (Kruglyak & Nickerson, 2001). With a mutation frequency of 2–8% as we recently described, we would have expected a high probability of detecting the AKT1 E17K mutation if it were present in minimally 1% of patients in our sample set. Furthermore, mutation frequency can be associated with multiple factors, including ethnicity, geography and gender among other characteristics. Thus, to further increase the confidence (to <1%) that this mutation is truly absent in AML, a larger and even more diverse cohort would need to be assessed. Additionally, we might have missed AKT1 mutations in certain cytogenetic or FAB subgroups given the limited numbers in some of these groups (Table I). However, the protein data clearly showed protein expression and phosphorylation in several pathway members and strongly suggests that PI3K/AKT pathway activation is most probably not due to AKT1 PHD or PIK3CA mutations in this sample set. Our sample set enabled us to dissect protein expression and phosphorylation of many proteins at various levels to identify sites of potential pathway activation. Further work is currently ongoing.

This report is the first one to assess the AKT1 PHD mutation status in AML and did not show that AML blasts harbour these mutations. The PI3K/AKT signal transduction pathway plays a central role in AML and is frequently activated (Min et al, 2003). To date, opposite to other tumour types, few mutations in the main pathway members (either activating ones in oncogenes or inactivating in tumour suppressors) have been described in AML, and new evidence for the presence of mutations (Carpten et al, 2007) in AKT1 needed to be investigated in AML. Excluding mutations is important to further characterize the PI3K/AKT pathway in terms of pathway activators. Clinically relevant, given the rather high frequency and presumed importance of the AKT1 PHD mutation (Carpten et al, 2007) in solid tumours, it is likely that drugs which specifically target this defect will be developed. Testing those drugs in AML, when available, will need to be done with caution. However, it is still worthwhile to test new AKT inhibitors in AML, as AKT is a promising target and off target effects of new compounds in vivo are difficult to predict. Screening for the mutation will probably not be of value in AML. Investigations further delineating the PI3K/AKT pathway need to look into other causes of activation of this important oncogenic network in AML.