The Akt pathway is a well-known promoter of tumor malignancy. Akt3 is expressed as two alternatively spliced variants, one of which lacks the key regulatory serine 472 phosphorylation site. Whereas the function of full-length Akt3 isoform (Akt3/+S472) is well-characterized, that of Akt3/−S472 isoform remains unknown. Despite being expressed at a substantially lower level than Akt3/+S472 in triple-negative breast cancer cells, specific ablation of Akt3/−S472 enhanced, whereas overexpression, suppressed mammary tumor growth, consistent with a significant association with patient survival duration relative to Akt3/+S472. These effects were due to striking induction of apoptosis, which was mediated by Bim upregulation, leading to conformational activation of Bax and caspase-3 processing. Bim accumulation was caused by marked endocytosis of EGF receptors with concomitant ERK attenuation, which stabilizes BIM. These findings demonstrate an unexpected function of an endogenously expressed Akt isoform in promoting, as opposed to suppressing, apoptosis, underscoring that Akt isoforms may exert dissonant functions in malignancy.
Significance: These results illuminate an unexpected function for an endogenously expressed Akt isoform in promoting apoptosis, underscoring the likelihood that different Akt isoforms exert distinct functions in human cancer. Cancer Res; 78(1); 103–14. ©2017 AACR.
Akt signaling is frequently deregulated in cancer due to PI3K pathway activation by PIK3CA, PTEN or Akt pleckstrin homology (PH) domain mutations (1, 2). PI3K activation results in Akt phosphorylation leading to transcriptional inhibition of apoptosis genes. Akts phosphorylate a plethora of substrates including GSK3, S6K, mTOR, and FOXOs, resulting in cell proliferation, migration, metabolism, and survival (1, 2, 3). In addition, Akts phosphorylate and inactivate Bad or the Forkhead transcription factors of the FOXO family (1, 4), thus causing FOXO's interaction with 14-3-3 proteins, leading to nuclear exclusion. If Akt signaling is switched off by attenuation of growth factor signaling, FOXOs are retained in the nucleus where they promote transcription of apoptosis genes (4). These include the Fas ligand and the proapoptotic Bcl-2 family member, Bim (5). Bim promotes apoptosis by triggering the conformational activation of BAX, resulting ultimately in pore formation in the mitochondrial outer membrane with the release of cytochrome c to the cytoplasm, caspase activation, and cell death (6, 7).
The Akt kinase family is comprised of three isoforms Akt1, Akt2, and Akt3, which are encoded by distinct genes located on separate chromosomes (8, 9, 10). Akt isoforms contain an N-terminal PH domain, a central kinase domain, which harbors a threonine phosphorylation site in the activation loop, and a C-terminal hydrophobic domain, which contains the serine regulatory phosphorylation site (9, 11). In response to insulin or growth factor signaling, PI3K phosphorylates the membrane associated phosphatidyl (4, 5) -bisphosphate (PIP2) to generate phosphatidyl (3, 4, 5) -trisphosphate (PIP3), which binds the PH domain and recruits Akt to the plasma membrane, where it is phosphorylated on threonine 308 in the kinase activation loop by PDK1, and subsequently on serine 473 by mTORC2 (1, 2). Despite similar effects on cell growth and survival, Akt isoforms exert distinct and even opposite functions on the malignant phenotype. Akt1 and Akt2 may have opposing effects on cell motility, which is inhibited by Akt1 and stimulated by Akt2 (12–17). The conflicting effects of Akt isoforms may explain the failure of clinical trials using pan-Akt inhibitors, which might be otherwise overcome by selective Akt isoform–based therapy (18).
Akt3 was shown to drive the growth of melanomas, ovarian, lung, prostate, and triple-negative breast cancer (TNBC) cell lines (3, 19–22). Mining The Cancer Genome Atlas (TCGA) breast cancer database showed an upregulation of Akt3 mRNA in 28% of triple-negative breast carcinomas (23). Akt3 knockdown in some TNBC cell lines was found to attenuate tumor growth, which was associated with increases in p27KIP1 levels (22). Akt3 was also shown to regulate lung carcinoma cell proliferation and invasion by phosphorylating IWS1, a splicing factor that converts epithelial FGFR2 IIIB into mesenchymal FGFR2 IIIC isoform (3). In contrast, we and others showed that Akt3 inhibits mammary and vascular tumor cell motility, while having no effect on cell growth, thus underscoring the conflicting effects of Akt3 on tumor cell behavior (24, 25).
A major feature distinguishing Akt3 from the otherwise similar Akt1 and Akt2, is that Akt3 is encoded by a gene that gives rise to two almost identical variants via differential splicing of C-terminal exons (9). Whereas the Ser472 phosphorylation site, encoded by exon 13, is present in the full-length Akt3 isoform (Akt3/+S472), it is absent from the second isoform (Akt3/−S472). This isoform excludes exon 13, and instead, encodes at its C-terminus exons 14 and 15 (9). Akt3 is phosphorylated on Thr305, thus creating a conformational change leading to phosphorylation on Ser472, a step necessary for full kinase activation, a process common to all Akts (1, 2).
Here, we show that Akt3/-S472 isoform exerts powerful suppression on mammary tumorigenesis. CRISPR-mediated knockout of Akt3/-S472 in TNBC cells enhanced, whereas overexpression suppressed, mammary tumor growth. By contrast, Akt3/+S472 isoform had no effect on malignant cell growth. Mechanistically, the suppressive effect of Akt3/−S472 was due to dramatic induction of apoptosis that was mediated by increases in Bim abundance, leading to Bax and caspase-3 activation. Increases in Bim expression were, in turn, caused by ERK attenuation due to marked EGFR endocytosis.
Materials and Methods
All procedures involving mice were conducted in accordance with NIH regulations concerning the use and care of experimental animals. This study was approved by an Animal Protocol reviewed by the Albert Einstein College of Medicine Animal Use Committee.
Breast cancer cell lines were purchased from the ATCC and maintained in DMEM (Invitrogen) supplemented with 10% FBS and 1% penicillin/streptomycin (Invitrogen). These include MDA-MB-231, 3475, MDA-MB-436, Hs578T, except for SUM149 that was cultured in DMEM-F12 with 5% FBS 1% penicillin/streptomycin, 5 μg/mL insulin/1 μg/mL hydrocortisone. The 3475 cell line is a lung metastatic MDA-MB-231 subline line that was obtained from Dr. Joan Massagué (Memorial Sloan Kettering Cancer Center, New York, NY). SUM 149 was obtained from Asterand Bioscience.
Antibodies and reagents
Anti Akt3 antibody was from Millipore or ProSci. β-Actin antibody was from Sigma. Antibodies against Akt1, Akt2, Bcl-2, Bim, Bid, Puma, cleaved caspase-3 or PARP, p-ERK, p-Akt S473, p-Akt T308, p-p53, were from Cell Signaling Technology. Bad, Bax, active Bax (6A7) were from Abcam. Anti-ERK, EGFR or EEA1 antibodies were from Santa Cruz Biotechnology. EGF or IGF were from PeproTech. PD 0325901 was from Tocris Bioscience. Drugs were dissolved in DMSO to yield a 1 mmol/L stock solution and stored at −80°.
Subcloning and production of lentiviral particles of hAkt3 isoforms
To subclone the hAkt3 splice variant 1 (Akt3/-S472) into lentiviral expression vector pLVX (Clontech), PCR was performed using 5′-forward: 5′-TCGCTCGAGATGAGCGATGTTACCATTGTG-3′ and 3′ reverse: 5′-CCCTCTAGATTATTTTTT CCAGTTACCCAGC-3′ primers using hAkt3-variant 1 plasmid from Origene as template. Using Xho1/Xba1 restriction sites hAkt3-variant 1 was subcloned into lentiviral vector expression vector pLVX-puro (Clontech). Akt3/+S472 plasmid (HA-hAkt3-WT-pcDNA3) was a kind gift from Dr. Yiling Lu (MD Anderson Cancer Center, Houston, TX). HA-hAkt3-WT clone was transferred from this construct into lentiviral pLVX-puro plasmid, using these primers 5′-GGGCTCGAGATGTACCCATACGATGTTCCAGATT ACGCT-3′ and 3′ region primer: 5′-GGGTCTAGATTATTCTCGTCCACTTGCAGA-3′. Lentiviral particles generated from these constructs were infected into cells. Other constructs include constitutively-active MEK1 retroviral vector, obtained from Dr. Stuart Aaronson at Mount Sinai School of Medicine (New York, NY).
Akt3 isoform CRISPR-Cas9 design and construction
To silence Akt3/+S472 or Akt3/−S472 isoform expression, we performed CRISPR-Cas9 (GeCKO) of human PKB gamma (hAkt3-WT) by targeting exon 13 or 14 respectively using the MIT CRISPR server (http://crispr.mit.edu/). As control, a non-targeting sequence was used (see Supplementary Methods).
Genomic validation of CRISPR-Cas9–mediated Akt3 knockout
To validate the genomic deletion induced by CRISPR-Cas9 to knock out the Akt3 isoforms, we designed primers flanking the gRNA recognition site and used Sanger sequencing as well as Deep Sequencing analysis (see Supplementary Methods).
Total RNA was extracted from cell lines and mammary tumors using RNeasy Mini Kit and RNase-free DNase set (Qiagen). Real-time RT-PCR was carried out using TaqMan RNA-to-Ct 1-Step Kit and gene-specific TaqMan probes on a StepOnePlus Real-time PCR system (Applied Biosystems; see Supplementary Methods).
TCGA-based Kaplan–Meier curves
RNASeqv2 level 3 RSEM isoform expression and clinical data were downloaded from TCGA Firehose (Broad Institute TCGA Genome Data Analysis Center (2014): Firehose stddata 2014 06 14 run, Broad Institute of MIT and Harvard. doi:10.7908/C1ZC821V). Invasive ductal carcinoma samples were selected and expression of isoforms uc001hzz.1 and uc001iab.1 used to identify Akt3/-S472 and Akt3/+S472, respectively. SPSS was used to perform Kaplan–Meier Analysis.
RNA extraction from FFPE TNBC specimens
TNBC formalin-fixed and paraffin-embedded (FFPE) tumors were selected by Dr. Susan Fineberg, following approval of the IRB protocol by the Committee of Clinical Investigation of the Albert Einstein College of Medicine (IRB# 2010-366). Ten unstained 10-μm sections from each tumor were microdissected using a pathologically reviewed duplicate hematoxylin and eosin (H&E) section as guide. RNA from microdissected carcinoma tissue, was analyzed and quantified on an Agilent 2100 Bioanalyzer. Akt3 isoform mRNA expression was measured by TaqMan gene expression custom assays and Taqman RNA-to-CT 1-Step Kit, with 50 ng RNA in triplicates. Analysis was performed per the 2(ΔΔCt) (Loudig and colleagues; 2011).
Cell monolayers were trypsinized and single cells replated in suspension in DMEM containing 0.5% methylcellulose solution/0.5% FBS on petri dishes coated with 1% agarose, which kept cells in suspension. Cells were incubated for 48 hours in incubator, harvested, spun down, lysed.
TUNEL assay was performed using the In Situ Cell Death Detection Kit (Roche). Briefly, after deparaffinization, rehydration, and antigen unmasking, tumor sections were incubated with proteinase-K for 15 minutes at room temperature. Sections were incubated with TUNEL reaction buffer for 1 hour at 37°C, washed three times in Tris buffer for 5 minutes and mounted with media containing DAPI to visualize nuclei. Images were taken by Zeiss Axioskop 2 microscope (Zeiss).
Akt nonradioactive kinase assay
We used the nonradioactive kinase assay kit (Cell Signaling Technology). Cell lysate (0.5-mg protein) was incubated overnight at 4°C with rabbit monoclonal anti phospho-Akt (Ser473) beads conjugate. Beads were washed in lysis buffer and resuspended in kinase buffer including 0.2 mmol/L ATP and 0.5 μg of kinase substrate (GST fusion protein of GSK-3α/β peptide containing the Ser21/9 residue). Beads were incubated for 30 minutes at 30°C; reaction stopped with 25 μL 3× SDS sample buffer and boiling. Bead supernatant was analyzed by SDS-PAGE and immunoblotting using anti phospho-GSK-3α/β (Ser21/9).
EGFR internalization assay using cell surface biotinylation
Cell monolayers were incubated with 100 ng/mL EGF in culture medium at 37°C for various intervals to allow EGFR internalization. Cells were then treated Sulfo-NHS-Biotin (0.5 mg/mL, Thermo Fisher Scientific) for 30 minutes on ice. Excess biotin was quenched with 50 mmol/L NH4Cl in ice-cold PBS-CM (calcium and magnesium). Cells were scrapped, and lysed in RIPA buffer (25 mmol/L Tris-HCL, pH 7.4, with 150 mmol/L NaCL, 0.1% SDS, 1% Triton X-100 supplemented with protease and phosphatases inhibitors. Cell extracts were centrifuged, and the supernatants were incubated with streptavidin beads (Sigma) to collect the biotinylated proteins. After washing the beads, bound proteins were eluted by boiling at 95°C for 10 minutes in 2× sample buffer and analyzed by immunoblotting with anti-EGFR antibody.
All additional procedures, including immunoblotting, immunoprecipitation, immunostaining, tumor formation, and lung metastasis are included as methods in Supplementary Information.
Akt3 isoform expression in TNBC cell lines and tumors
Akt3 was shown to be expressed as two isoforms, which interestingly differ in the carboxyl terminal Ser472 regulatory phosphorylation site, that is present in the full-length Akt3 (Akt3/+S472), whilst absent in the second isoform (Akt3/−S472; ref. 9). In the mammary gland, Akt3/-S472 splice variant comprises approximately 5% of all Akt3 transcripts, implying a low abundance relative to Akt3/+S472 (9). qRT-PCR analysis of six clinical TNBCs using Akt3 isoform specific TaqMan primers revealed lower expression of Akt3/−S472 compared with Akt3/+S472 mRNA in these tumors (Fig. 1A). Differential isoform expression was also observed in TNBC cell lines such as HS578T, MDA-MB-231, 3475 (MDA-MB-231 metastatic subline), MDA-MB-436 and SUM149, where the expression of Akt3/-S472 mRNA was substantially reduced relative to Akt3/+S472, likely due to upregulation Akt3/+S472 in cultured cell lines (Fig. 1B). Furthermore, TCGA mining of invasive duct carcinomas, pointed to a more significant association of Akt3/−S472 mRNA (P = 0.08) relative to Akt3/+S472 (P = 0.99), with patient survival duration (Fig. 1C and D), implying this isoform may negate malignancy.
CRISPR-Cas9–mediated knockout of Akt3/-S472 in TNBC cells enhances mammary tumor growth
To investigate the role of Akt3 isoforms in mammary tumorigenesis, we under or overexpressed each of the isoforms in MDA-MB-231 and 3475 breast cancer cell lines, which are endowed with tumorigenic and metastatic potential (26). Although Akt3/-S472 mRNA was weakly expressed in these cell lines, we tested whether silencing this isoform affects the tumor phenotype. We used the CRISPR-Cas9 technology to knock out endogenous expression of the two Akt3 splice variants (27, 28). The CRISPR-Cas9 genome deletion technique introduces a double-strand break, which leads to DNA repair by introducing small insertions or deletions (indels) at the target region. Small guide RNA (gRNA) sequences targeting Akt3/+S472 on exon 13 or Akt3/−S472 on exon 14, were generated using the CRISPR gRNA design tool at MIT, that excludes off-target genome modification. Twenty nucleotides gRNA sequence for each of the Akt3 isoforms was cloned into Cas9 containing lentiviral constructs that were used for virus production and expression into cancer cells. To validate the targeting efficiency of CRISPR-Cas9, we measured alterations in genomic DNA from bulk cancer cell populations by next-generation sequencing. A 200-bp amplicon spanning the gRNA-binding site on the targeted locus was subjected to multiple sequence alignment to validate CRISPR-Cas9 induced mutations and knockout efficiency in the bulk tumor cell population. Our data confirmed mutations were located close to the targeted sequence in 86% of Akt3/+S472 and 77% of Akt3/−S472 knockout 3475 cell populations (Supplementary Fig. S1A and S1B). In contrast, control 3475 cells showed no modifications in the nucleotide sequence of either isoform, confirming that genomic locus was unaffected by the nontargeting CRISPR sequence (Supplementary Fig. S1C and S1D). Thus, next-generation sequencing of amplicon obtained around the gRNA recognition site clearly shows that Akt3-targeting CRISPR-Cas9 constructs were capable of introducing deletion or insertion (indels) at the genomic locus with high efficiency, thereby rendering defective Akt3 isoform protein.
Limitation in protein detection due to lack of antibody that discriminates the two highly homologous Akt3 splice variants, prompted us to perform TaqMan-based qRT-PCR using isoform-specific primers to confirm the CRISPR-mediated knockout. Akt3/-S472 KO cells exhibited a 4-fold reduction in mRNA relative to control cells, compared with unchanged expression of Akt3/+S472 mRNA (Fig. 2A). Similarly, knockout of Akt3/+S472 led to a 2.2-fold reduction in mRNA, relative to unchanged Akt3/−S472 mRNA (Fig. 2B), implying CRISPR-mediated knockout of Akt3 isoform led to impaired transcript expression.
The effect of Akt3 isoform deletion on tumor cell growth in vitro was evaluated by comparing cell growth kinetics of control and isoform-deficient cells over 5 days in culture; but no differences were noted between the two groups and controls, implying Akt3 isoforms do not regulate cell proliferation in vitro (Fig. 2C). However, evaluation of tumor growth, 60 days postinjection of cancer cells into mammary fat pads of athymic nude mice, revealed a striking difference between the isoforms. Namely, Akt3/−S472 ablation in 3475 cells caused a 2.5-fold increase in mammary tumor volume as compared with control 3475 cells (P < 0.05; Fig. 2D). In contrast, Akt3/+S472 knockout had no effect, as shown by similar sizes of 3475-Akt3/+S472 KO and 3475 control tumors (Fig. 2D). Given the polyclonal nature of the CRISPR-ed cells, we measured whether Akt3 isoform knockout was sustained throughout tumor growth using qRT-PCR analysis. Indeed, Akt3/-S472 KO or Akt3/+S472 KO tumors maintained very low levels of the respective isoform (Fig. 2 E and F). Furthermore, deep sequencing analysis of tumors using genomic DNA, showed multiple indel events occurring near the targeted sequence on several thousands of sequencing reads in 91% of Akt3/+S472 KO and 96% of Akt3/−S472 KO tumors (Supplementary Fig. S2A and S2B). As expected, control tumors exhibited 100% matching of multiple reads with NCBI reference sequence of Akt3/+S472 or Akt3/−S472, with no indels on any of the sequences (Supplementary Fig. S2C and S2D).
Finally, these findings were reproduced in the MDA-MB-231 parental cell line, which is heterogeneous compared with the clonal 3475 subline (Supplementary Fig. S3). Akt3/−S472 or Akt3/+S472 CRISPR-mediated knockout in MDA-MB-231 cells led to a significant reduction in isoform mRNA (Supplementary Fig. S3A–S3B). However, while Akt3/-S472 knockout did not have any effect on cell growth in culture (Supplementary Fig. S3C), it caused a 2-fold increase in tumor burden in vivo (Supplementary Fig. S3D). In contrast, Akt3/+S472 ablation in these cells did not inhibit tumor cell growth in vitro or in vivo (Supplementary Fig. S3C–S3D). Of, note mammary tumors resulting from knockout cells maintained low levels of isoform expression, as indicated by substantially reduced levels of Akt3/−S472 or Akt3/+S472 mRNA (Supplementary Fig. S3E–S3F). These data underscore a robust tumor suppressive effect of Akt3/−S472, especially in light of low basal expression of this spliced variant relative to Akt3/+S472.
Akt3/−S472 overexpression in TNBC cells suppresses the malignant phenotype
To confirm the suppressive effect of Akt3/−S472 on mammary tumorigenesis, this isoform was overexpressed in 3475 cells. Akt3/−S472 overexpression did not affect Akt1 or Akt2 expression, and had interestingly no effect on cell growth in vitro (Fig. 3A and B). However, orthotopic injection of tumor cells into athymic nude mice, revealed a dramatic reduction in mammary tumor volume in mice inoculated with Akt3/−S472–overexpressing 3475 cells relative to 3475 controls (Fig. 3C). This was consistent with the tumor-enhancing effect of Akt3/−S472 knockout (see Fig. 2D). By contrast, Akt3/+S472 overexpression in 3475 cells did not influence tumor growth (Fig. 3D; Supplementary Fig. S4A). As control, overexpression of the prosurvival Akt1 in 3475 cells, enhanced tumor burden (Fig. 3E; Supplementary Fig. S4B).
Limitations in primary tumor growth do not necessarily impair metastasis, which is dependent on invasiveness, that may not be inhibited by Akt3/−S472. We therefore measured the effect of Akt3/−S472 on distant metastasis, 4 months following surgical removal of the primary tumor, to mimic the clinical setting. In addition, we measured lung colonization following intravenous injection of tumor cells, which is independent of primary tumor growth. In both instances, metastasis burden, as gauged by the number and distribution of pulmonary foci, was dramatically reduced in mice inoculated with 3475-Akt3/-S472 OE cells relative to controls (rank-sum analysis; P < 0.001; Fig. 3F–H). These findings demonstrate that Akt3/−S472 suppresses the tumor phenotype by inhibiting primary and metastatic tumor growth.
Akt3/−S472 suppresses mammary tumorigenesis by stimulating apoptosis
The mechanism whereby Akt3/−S472 inhibits tumor growth remained to be addressed. Although Akt3/−S472 had no effect on cell growth in vitro, it suppressed tumor burden in vivo (see Fig. 3B and C). To rule out effects on tumor cell proliferation in vivo, we measured Ki67 staining in mammary carcinomas, which showed no difference between 3475-Akt3/-S472 OE and control tumors (Fig. 4A). Because the balance between cell proliferation and cell death regulates tumor size, we measured apoptosis in non-necrotic carcinoma areas. This revealed a substantial increase in TUNEL staining in 3475-Akt3/-S472 OE tumors relative to controls (Fig. 4B). Consistent with this effect, CRISPR-mediated knockout of Akt3/−S472 in 3475 cells, resulted in mammary tumors expressing lower TUNEL activity, yet similar levels of Ki67 staining relative to control carcinomas (Supplementary Fig. S4C–S4D). By contrast, Akt3/+S472 or Akt1-overexpressing tumors did not exhibit any change in TUNEL staining relative to controls (Supplementary Fig. S4E–S4F).
Akt3/−S472 promotes apoptosis via Bim upregulation, leading to Bax and PARP/caspase-3 activation
The proapoptotic effect of Akt3/−S472 overexpression was associated with robust caspase-3 activation and PARP cleavage in several tumors, an effect that was independent of Akt1 or Akt2 expression, which was unchanged (Fig. 4C). Moreover, Akt3/−S472 caused a dramatic upregulation in the levels of proapoptotic Bim protein in several independent tumors (Fig. 4C), but had no effect on the level of Bax apoptotic effector protein, which is normally activated by Bim (Fig. 4C). Interestingly, Bim protein upregulation was matched by higher expression of Bim mRNA in tumors overexpressing Akt3/−S472 relative to controls (Fig. 4D). By contrast, immunoblotting analysis of mammary tumors showed that Akt3/−S472 overexpression led to Bid downregulation, but had no effect on the expression of other Bcl-2 family genes including the proapoptotic Bad and Puma, or the antiapoptotic Bcl-xL (Fig. 4E). In addition, Akt3/−S472 did not affect p53 phosphorylation, which results from DNA damage induced apoptosis (Fig. 4E; ref. 29).
Bim is known to bind the N-terminus of Bax, inducing conformational changes that culminate in the exposure of Bax α-helix 9 containing a transmembrane domain (30). These events trigger Bax translocation to mitochondria and insertion of this protein in the outer mitochondrial membrane, where it oligomerizes and forms pores that allow the release of cytochrome c to the cytoplasm, resulting in caspase activation and apoptosis (31). To test whether Akt3/−S472 stimulates Bax activation in vivo, we measured the levels of active-Bax using immunoprecipitation of tumor extract under native conditions with an antibody that recognizes the active conformation (6A7) followed by anti-Bax immunoblotting. As control, Bax expression in tumors was unchanged by Akt3/−S472 overexpression (Fig. 4F, lanes 1–2). However, the level of active-Bax protein was dramatically increased in 3475-Akt3/-S472 OE tumors relative to controls (Fig. 4F, lanes 5–6). These findings indicate that Akt3/−S472 triggers tumor apoptosis via Bim/Bax/caspase-3 activation.
Although apoptosis was prominent in mammary tumors in vivo, it was not detected in vitro in cell monolayers (ECM attached), even in response to apoptotic stimuli. We therefore tested the effect of ECM detachment on apoptotic stimulation by stress involving growing cells in suspension in low serum containing media (0.5%) for 48 hours, thus mimicking anoikis-like conditions. This interestingly resulted in a marked increase in Bim protein and cleavage of the caspase-3 substrate PARP in Akt3/−S472–overexpressing 3475 cells relative to control cells (Fig. 4G).
Importantly, these findings were reproduced in the MDA-MB-436 cell line, which as 3475, is a triple negative and highly tumorigenic breast cancer cell line. Akt3/−S472 overexpression in MDA-MB-436 attenuates mammary tumor growth, an effect that was accompanied by increases in TUNEL, Bim, and active–caspase-3 levels, as compared with unchanged Ki67 levels in tumors (Supplementary Fig. S5), thus underscoring the role of Akt3/−S472 in tumorigenic suppression via induction of apoptosis.
The proapoptotic effect of Akt3/−S472 is not due to a dominant negative effect on pan-Akt kinase activation
To determine whether Akt3/−S472 causes apoptosis by inhibiting the prosurvival effect of other Akts, involving Akt3/+S472, Akt1 or Akt2, we measured overall Akt phosphorylation in response to IGF treatment for 10 and 20 minutes (Fig. 5A). pAkt-T308 immunoblotting of cell lysates showed phosphorylation of a single Akt protein band in control cells, and a doublet in Akt3/−S472 overexpressing 3475 cells, corresponding to endogenous Akts, and overexpressed Akt3/−S472, which is only phosphorylated on threonine 305 (Fig. 5A). By comparison, pAkt-S473 immunoblotting showed phosphorylation of a single Akt protein band, which might include all Akts, except Akt3/−S472, which is lacking the S472 phosphorylation site. These data demonstrated that Akt3/−S472 overexpression does not inhibit overall Akt phosphorylation. Importantly, pan-Akt kinase activity, as measured by exogenous phosphorylation of GSK3α/β peptide in vitro by pAkt-S473 IPs, showed similar levels of phosphorylated GSK3α/β substrate in 3475-Akt3/−S472 OE and control cells, consistent with comparable levels of pAkt-S473 (Fig. 5B). These data imply that Akt3/−S472 stimulates apoptosis via a mechanism that does not involve a dominant negative effect on pan-Akt kinase activity.
Akt3/−S472 enhances Bim expression via ERK inhibition
Our data point to Bim as the predominant proapoptotic gene upregulated by Akt3/−S472, at the mRNA and protein level (see Fig. 4C–E). Bim gene transcription is known to be often driven by FOXO activation in response to Akt inhibition, causing nuclear localization and transcriptional activation (32). However, no meaningful change in FOXO3a phophorylation or nuclear localization was observed in response to Akt3/−S472 overexpression (data not shown), consistent with our finding that Akt3/−S472 overexpression does not inhibit pan-Akt kinase activation.
MAPK/ERK pathway activation is known to destabilize Bim protein via phosphorylation, resulting in Bim ubiquitylation and degradation (33). We therefore tested whether Akt3/−S472 regulates Bim via ERK inhibition. Indeed, we found that ERK and MEK1 phosphorylation were both reduced in 3475-Akt3/−S472 OE cells (Fig. 5C). In support of this notion, expression of constitutively active MEK1 in 3475-Akt3/−S472 OE cells, which rescued ERK phosphorylation, caused a marked reduction in Bim protein (Fig. 5D). By contrast, treatment of control 3475 cells with the MEK1 inhibitor, PD0325901, which suppressed ERK phosphorylation at 0.5 and 1.0 μmol/L, markedly increased Bim expression (Fig. 5E). Interestingly, growing cells under anoikis conditions, led to further inhibition of ERK phosphorylation, which was concomitant with Bim upregulation in 3475-Akt3/−S472 OE relative to control 3475 cells (Fig. 5F). In contrast with ERK, Akt phosphorylation was not observed in anoikis/nutrient–deprived cultures of 3475-Akt3/−S472 OE or control cells (data not shown), supporting the notion that ERK, and not Akt, inhibition by Akt3/−S472 mediates Bim upregulation. In aggregate, these results clearly demonstrate a role for Akt3/−S472 in ERK attenuation leading to Bim stabilization.
Akt3/−S472 enhances EGF receptor endocytosis
Bim upregulation following ERK inhibition was found to drive anoikis of MCF10A nontumorigenic breast cells as a result of EGFR downregulation (34). Hence, it is plausible that Akt3/−S472 inhibits ERK by suppressing EGFR, which is overexpressed in TNBC cells. EGFR protein expression was, however, unchanged in 3475-Akt3/−S472 OE cells growing either as adherent or suspension cultures (Fig. 6A). We, therefore, tested whether Akt3/−S472 regulates EGFR signaling by affecting EGFR endocytosis. We measured EGFR internalization in response to EGF stimulation using cell surface biotinylation. Cells were treated with 100 ng/mL EGF in growth media for 0 to 40 minutes, and cell surface biotinylated EGFR was captured by avidin-coupled beads, and levels assessed by EGFR immunoblotting and densitometry. Interestingly, EGF stimulated a striking attenuation of surface EGFR relative to overall EGFR expression in 3475-Akt3/−S472 OE cells as compared to control 3475 cells (Fig. 6B). Densitometric analysis revealed that EGFR was internalized by 46% and 88% in 3475-Akt3/−S472 OE cells relative to 1.4% and 35% in control 3475 cells by 5 and 40 minutes post EGF treatment (Fig. 6C). Consistent with these findings, staining of cells, showed EGFR membrane localization in ligand-untreated control 3475 and 3475-Akt3/−S472 OE cells (Fig. 6D). However, stimulation with EGF for 10 minutes caused dramatic re-localization of EGFR from the cell surface to early endosomes, as shown by costaining of EGFR with EEA1, an effect that was intensified in 3475-Akt3/−S472 OE relative to control 3475 cells (Fig. 6D). Indeed, quantification of endocytosed EGFR at 10 min post EGF stimulation showed a 3-fold increase in endocytic vesicles containing EGFR and EEA1 in 3475-Akt3/−S472 OE relative to control cells (Fig. 6E). To evaluate whether increased EGFR endocytosis affects ERK phosphorylation, we measured p-ERK levels at various time points following EGF treatment. ERK was phosphorylated to a similar extent in control and 3475-Akt3/−S472 OE cells at 10 to 40 minutes post EGF stimulation; however, at 80 minutes, there was a sharp reduction in p-ERK levels in 3475-Akt3/−S472 OE cells relative to control cells (Fig. 6F).
In support of these findings, CRISPR-mediated knockout of Akt3/−S472 caused a delay in EGFR internalization, as shown by the higher levels of surface-biotinylated EGFR in Akt3/−S472 KO cells relative to control cells, expressing a non-targeting CRISPR sequence (Fig. 7A). As control, the total levels of EGFR post EGF stimulation were similar to those in control cells (Fig. 7A). Changes in EGFR internalization were especially striking at 20 minutes post EGF treatment, but less obvious at later time points (Fig. 7A and B). Immunostaining for EGFR and EEA1 confirmed these data; EGF clearly induced a less efficient intracellular localization of EGFR in 3475-Akt3/−S472 KO cells than in control 3475 cells (Fig. 7C). Indeed, Akt3/−S472 KO cells displayed fewer EGFR/EEA1 containing endosomes relative to control cells (Fig. 7D). These findings underscore that Akt3/−S472 attenuates ERK signaling by enhancing EGFR endocytosis.
Akt isoforms regulate various cellular processes including cell proliferation, migration, and survival (12, 22, 35, 36). Akt3 has been largely associated with progression of several tumor types especially melanomas, due to widespread PTEN mutations (37, 38). In TNBC, Akt3 was found to be expressed in a third of cancers, implying a potential benefit for Akt3-based therapy (39). However, Akt3 was found to have paradoxical effects on breast and other malignancies (22, 3, 24, 25). Akt3 was shown to promote breast cancer cell proliferation as well as Akt drug resistance (22, 40). By contrast, we and others showed that like Akt1, Akt3 ablation in cancer cells enhanced cell motility, but not metastasis (16, 36, 24), likely due to events, beyond cell migration, that are not regulated by Akt3/−S472.
Most studies on Akt3 inputs into malignant progression focused on the full-length Akt3 (Akt3/+S472), but overlooked the Akt3/−S472 isoform, likely because of low abundance or potential redundancy with full-length Akt3. qRT-PCR analysis of clinical TNBCs, revealed a low expression of Akt3/−S472 relative to Akt3/+S472, consistent with the scarcity of this isoform (5%) in the mammary gland (9), which could be due to inherent proapoptotic activity that may cause cell toxicity. In support of a physiologic role for Akt3/−S472, we speculate that this isoform regulates mammary morphogenesis by causing apoptotsis or clearance of epithelial cells from the luminal cavity, thus driving lumen formation and glandular differentiation (41). Consistent with this idea, TCGA breast cancer datasets revealed a borderline association of Akt3/−S472 expression with improved patient survival, implying a potential role in maintenance of a benign phenotype.
In contrast with our findings, Akt3 was shown to regulate TNBC cell growth in vitro and in vivo, using MDA-MB-468 and BT-549 cell lines, which are however PTEN mutated, and are thus likely to express a constitutively active Akt3 that renders an oncogenic function (22). By contrast, MDA-MB-231 and 3475 cells are PTEN wild-type, and unlikely to express a dominant active Akt3 kinase, hence supporting our findings that ablation of Akt3/+S472 in these cell lines is dispensable for tumor growth (42).
These findings convey an unconventional role for Akt in apoptosis, which is inconsistent with its classical prosurvival function. What clearly differentiates the two Akt3 splice variants is the carboxyl terminal serine 472, which is missing in Akt3/−S472, a site that is necessary for Akt kinase activation. Namely, although T308 phosphorylation by PDK1 activates the Akt kinase, subsequent phosphorylation on S472 by mTORC1 leads to full kinase activation (43, 44). Indeed, it was shown that T308 phosphorylation causes a 10-fold increase in kinase activation, whereas subsequent phosphorylation on S472 results in a 1,000-fold activation (9). This implies that Akt3/−S472, may not efficiently compete substrates shared with other Akts including Akt3/+S472, making it unlikely to act as a dominant negative Akt. In support of this notion, exogenous GSK3 peptide phosphorylation by IGF-activated Akts in vitro, was not affected by Akt3/−S472 overexpression. Moreover, under or over expression of Akt3/+S472 isoform did not influence tumor growth, implying this isoform may not act as a driver oncogene in MDA-MB-231 or 3475 cells, and is therefore unlikely to be inhibited by Akt3/−S472. Finally, Akt3/−S472 may phosphorylate substrates that cause apoptosis, that are not shared with other Akts, and that interact with Akt3/−S472 on sequences encoded by exons 14/15.
We propose that Akt3/−S472 triggers apoptosis via Bim upregulation due to EGFR–ERK axis inhibition. These data underscore findings from the Brugge laboratory, showing that apoptosis underlying mammary gland lumen formation is mediated by Bim stabilization due to ERK inhibition caused by EGFR attenuation (41). Interestingly, Akt3/−S472–mediated apoptosis was observed in vivo, but not in vitro in adherent cultures, a limitation that was overcome by growing cells under anoikis conditions. Interestingly, ERK, but not Akt, was phosphorylated in control 3475 cells growing under anoikis- or nutrient-deprived conditions, thus supporting the contention that Akt3/−S472 targets ERK, and not Akt, to provoke apoptosis.
How Akt3/−S472 regulates EGFR endocytosis remains unknown. Others have shown that Akt facilitates EGFR trafficking and degradation by phosphorylating and activating PIKfyve (45). Therefore, one might speculate that Akt3/−S472 phosphorylates PIKfyve or other substrates involved in EGFR endocytosis, a question that may be resolved by expressing a kinase-dead Akt3/−S472. Alternatively, Akt3/−S472 may activate clathrin-mediated endocytosis, which was shown to be affected by Akt phosphorylation, resulting in enhanced EGFR endocytosis (46). In sum, this study demonstrates a highly unusual function for an Akt molecule, in promoting, and not inhibiting, apoptosis through negative regulation of the EGFR–ERK survival pathway.
Disclosure of Potential Conflicts of Interest
No potential conflicts of interest were disclosed.
Conception and design: K. Suyama, R.N. Kitsis, R.B. Hazan
Development of methodology: K. Suyama, J. Yao, H. Liang, O. Benard, S. Fineberg, R.N. Kitsis
Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): K. Suyama, J. Yao, H. Liang, O. Benard, O. Loudig, D. Amgalan, W.M. McKimpson, G.R. Phillips, Y. Wang, S. Fineberg, R.N. Kitsis
Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): K. Suyama, J. Yao, H. Liang, O. Benard, O. Loudig, D. Amgalan, W.M. McKimpson, G.R. Phillips, J. Segall, Y. Wang, S. Fineberg, R.N. Kitsis, R.B. Hazan
Writing, review, and/or revision of the manuscript: O. Loudig, D. Amgalan, J. Segall, L. Norton, R.N. Kitsis, R.B. Hazan
Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): L. Norton
We are very grateful to Dr. Yiling Lu (MD Anderson Cancer Center) for fruitful discussions and generous Akt plasmids. We thank Christina Liu for technical expertise. We thank Dr. Jonathan Backer for data discussion. We thank the Cancer Center Support Grant (P30 CA013330) for the use of shared resources.
This work was supported by grants from the National Cancer Institute R01 CA135061-01A1 (to R.B. Hazan), R01CA136854-01A1 (to R.B. Hazan), and the Breast Cancer Research Foundation (to R.B. Hazan and L. Norton).
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.