Abstract
Purpose: Inhibition of AKT is a key target area for personalized cancer medicine. However, predictive markers of response to AKT inhibitors are lacking. Correspondingly, the AKT-dependent chain of command for tumor growth, which will mediate AKT-dependent therapeutic responses, remains unclear.
Experimental Design: Proteomic profiling was utilized to identify nodal hubs of the Trop-2 cancer growth–driving network. Kinase-specific inhibitors were used to dissect Trop-2–dependent from Trop-2–independent pathways. In vitro assays, in vivo preclinical models, and case series of primary human breast cancers were utilized to define the mechanisms of Trop-2–driven growth and the mode of action of Trop-2–predicted AKT inhibitors.
Results: Trop-2 and AKT expression was shown to be tightly coordinated in human breast cancers, with virtual overlap with AKT activation profiles at T308 and S473, consistent with functional interaction in vivo. AKT allosteric inhibitors were shown to only block the growth of Trop-2–expressing tumor cells, both in vitro and in preclinical models, being ineffective on Trop-2–null cells. Consistently, AKT-targeted siRNA only impacted on Trop-2–expressing cells. Lentiviral downregulation of endogenous Trop-2 abolished tumor response to AKT blockade, indicating Trop-2 as a mandatory activator of AKT.
Conclusions: Our findings indicate that the expression of Trop-2 is a stringent predictor of tumor response to AKT inhibitors. They also support the identification of target-activatory pathways, as efficient predictors of response in precision cancer therapy. Clin Cancer Res; 22(16); 4197–205. ©2016 AACR.
Inhibition of AKT is a pivotal area in personalized cancer medicine. However, predictive markers of response to AKT inhibitors are lacking. Correspondingly, unclear is the AKT-dependent chain of command in growing tumors, which in turn mediates AKT-dependent therapeutic responses. Proteomic profiling of the Trop-2 network led us to the identification of AKT as a major hub of the Trop-2–triggered signaling. We then went on to show that Trop-2 signaling activates AKT to induce tumor growth. Correspondingly, Trop-2 was shown to determine response to AKT-targeting drugs. These findings pave the way for the validation and use of Trop-2 as predictor of response to AKT-focused anticancer therapies. They correspondingly open novel ways for designing clinical trials, as focused on responsive patients' subgroups.
Introduction
Effective prediction of response of tumors to anticancer therapy remains a largely unmet need. Most efforts for identifying predictive biomarkers for cancer therapy have been directed to target measurements (1). However, this misses key information on actual target activity and on target-activatory networks. One such a case is AKT, where considerable efforts have been dedicated to developing and testing inhibitory drugs in clinical settings, but reliable indicators of response are entirely lacking. Trop-2 is a transmembrane calcium signal transducer (2–6), which is overexpressed by cancer cells, and drives tumor growth (7, 8) and progression (9). In this work, we dissected the Trop-2 signalosome, through high-throughput proteomic analysis. This led us to identify AKT as a central hub of the Trop-2 activatory network. These findings were validated in human tumors xenotransplants in animal models, and in patients, where activated AKT signaling was found to associate with Trop-2 expression and tumor progression. Systematic drug screening then revealed that AKT inhibitors (10) only blocked the growth of Trop-2–expressing tumors, but were ineffective on Trop-2–null cells. Thus, Trop-2 appears as a nodal AKT activator for tumor growth and determines response to AKT-targeted therapies.
Patients and Methods
Cells
Human breast MDA-MB-231 and colon Colo205, KM12SM, and HCT-116 cancer cells were grown in RPMI1640 medium supplemented with 10% FCS. Human HEK293T kidney cells and immortalized murine thymic epithelial MTE 4-14 (7) cells were maintained in DMEM supplemented with 10% FCS. All media were supplemented with 100 IU/mL penicillin and 100 μg/mL streptomycin (Euroclone). The MDA-MB 231 and Colo205 cell lines were obtained from ATCC. The KM12SM and HCT-116 cell lines were obtained from Roche Diagnostics GmbH, Pharma Research. The KM12SM cell line was derived from KM12 hepatic metastases as described previously (11). The MTE 4-14 cell line was a kind gift by Philippe Naquet, Centre d'Immunologie de Marseille-Luminy, Marseille, and was established and characterized as described previously (12). The HEK293T cell line was obtained from Angelo Veronese (University ‘G. d'Annunzio’, Chieti, Italy) and reauthenticated before use by short tandem repeat analysis (PowerPlex 16 System, Promega). Cell lines were further verified by morphology and flow cytometry assessment. Multiple frozen master stocks were prepared within the first 2 months from receipt for each cell line. Experiments were performed on cells resuscitated from master stocks and passaged for no more than 6 months. Absence of mycoplasma contamination was routinely confirmed by PCR analyses (13).
Stable knockdown of endogenous Trop-2 was obtained using pLKO.1-based lentiviral plasmids as described in ref. 14. HEK293T packaging cells were cotransfected with pLKO.1-Trop-2-shRNA, pLP1, pLP2, and pLP/VSVG vectors, and lentivirus-containing supernatants were harvested 48 hours later and used for cell infection. Stably infected cells were selected with 3 μg/mL puromycin for 15 days. A scrambled short hairpin (shRNA) plasmid was used as negative-control shRNA.
Antibodies
The 162-46.2 (ATCC HB 187; ref. 15), 2E7 (manuscript in preparation), and T16 (2) anti-Trop-2 mAbs were purified and fluorochrome-conjugated with Alexa-488 or Alexa-546 as described previously (16). Trop-2 was revealed in tumor sections also with the goat anti-Trop-2 polyclonal antibody (AF650) from R&D Systems. AKT was revealed with rabbit polyclonal anti-human phospho-AKT (Ser-473; 736E11, Cell Signaling Technology), or anti-human phospho-AKT [(Thr 308)-R:sc-16646-R, Santa Cruz Biotechnology].
Antibodies directed against murine targets were: goat polyclonal anti-AKT (sc-1618); rabbit polyclonal anti-phospho-AKT (Thr308; sc-16646-R); rabbit monoclonal anti-phospho-AKT (Ser473; D9E; Cell Signaling Technology). Secondary Alexa Fluor–conjugated antibodies (633-goat anti-rabbit IgG, 546-donkey anti-goat IgG) were from Invitrogen.
Cell growth assays
MTE 4-14 and MDA-MB-231 transfectants were seeded at 1.5–5.0 × 103 cells/well in 96-well plates (five replica wells per data point). Cell numbers were quantified by staining with crystal violet (7). Cell numbers were normalized against standard reference curves of 2-fold serially diluted cell samples. Cell numbers were quantified in parallel by image analysis/nuclei count of DAPI-stained 96-well plates (submitted for publication). MTE cells transfected with empty vector (control) or TROP2 were seeded at a density of 103 cells/well in normal medium (untreated) or medium containing cytochalasin D at different concentrations, as indicated. Actual numbers of cells/well at time 0 were measured and used to normalize cell counts for each experimental group. Growth curves for each treatment were compared by two-way ANOVA.
Small-molecule signaling pathway inhibitors
In vitro treatments.
Cells were treated with signaling molecule inhibitors: 1–10 μmol/L BAY 11-7082 (for NFKB), 10, 100, 1,000 nmol/L PP242 (for MTOR), 1–10 μmol/L SP 600125 (for MAPK8 [JNK]), and 50, 100, 200, 400 nmol/L A6730 (Sigma Technical Company) or 100, 200, 400, 800, 1,600 nmol/L MK-2206 (Selleck Chemicals) to inhibit AKT. Stock solutions were prepared in DMSO and diluted in ethanol or water. Control cells received vehicle alone. A second treatment was performed 24 hours after seeding. The final concentration of DMSO and ethanol in the cell medium never exceeded 0.1% and 1%, respectively. Half-maximal inhibitory concentrations for signaling molecule inhibitors were determined through dose–response growth assay using MTE 4-14 cell transfectants. Growth was assessed as described, after 48 hours in the presence of the inhibitor at concentrations ranging from 1 nmol/L to 3.2 μmol/L. Dose–response curves were plotted and IC50 were determined using nonlinear regression analysis with a sigmoidal fit model (GraphPad Prism Version 6.0, GraphPad Software, Inc.).
In vivo treatments.
A stock solutions of MK-2206 was prepared in DMSO and diluted in apyrogenic physiologic solution. Mice intraperitoneal administration (0.35 mg/die) begun when palpable tumors developed and continued for the time indicated. Mice in control groups received vehicle alone.
Flow cytometry
Cell staining for flow cytometry was performed as described previously (17). Fluorescence analysis was carried out on fluorescence-activated flow cytometers (FACSCalibur and FACSCanto II, Becton Dickinson). Enrichment for expressing transfectants was performed by sorting with FACSAria III (Becton Dickinson) or by enrichment with MAgnetic Cell Sorting (MACS) from Miltenyi Biotec GmbH. To improve the detection of transfectants stained with FITC-mAb, subtraction of cell autofluorescence and displacement of FITC-stained cells in the red channel were performed as described previously (18, 19). All TROP2 transfectants were selected for expression levels comparable with those of endogenously expressing human cancer cells (5, 15, 16).
Immunofluorescence
Cells plated on glass coverslips were fixed with 4% paraformaldehyde in PBS for 20 minutes. Permeabilization and blocking were performed in PBS with 10% FCS and 0.1% saponin (20). Live cells on glass coverslips were stained in medium with 10% FCS at 37°C for 5 minutes, and fixed after staining. Slides were analyzed with an LSM-510 META (Zeiss) confocal microscope.
Experimental tumors
Tumor cell lines and TROP2 transfectants (15, 16) were injected subcutaneously (5–10 × 106 cells) into 8-week-old female athymic Crl: CD1-Foxn1 nu/nu mice (Charles River Laboratories). The tumor longest/shortest diameters (D/d) were measured every 5 to 7 days. Tumor volumes were calculated using a formula for ellipsoid volumes (D × d2/2; ref. 21). AKT inhibitors were administered as described. All reagents were prepared and used as endotoxin-free, to prevent nonspecific activation of natural immunity. All procedures involving animals and their care were approved by the Interuniversity Animal Research Ethics Committee (CEISA) of Chieti-Pescara and Teramo Universities and by the Italian Ministry of Health (Prot.26/2011/CEISA/PROG/16 and no. 723/2015-PR), and were conducted in compliance with consensus international protocols [D.L. No.116, G.U., Suppl. 40, Feb. 18, 1992; No. 8, G.U., July, 1994; UKCCCR Guidelines for the Welfare of Animals in Experimental Neoplasia; EEC Council Directive 86/609, OJ L 358. 1, Dec. 12, 1987; Guide for the Care and Use of Laboratory Animals, United States National Research Council, 1996; ARRIVE guidelines, 2013; ref. 22].
Tumor patient case series
Patients (N = 414) with unilateral breast cancer, who had undergone surgery with axillary lymph node dissection were analyzed. Patients were selected according to the following criteria: tumor diameter: T1, T2; absence of nodal diffusion (23, 24); absence of distant metastases at diagnosis. Clinical and pathologic status, tumor histotype, histologic grade, levels of expression of estrogen receptor α and progesterone receptor (PGR), and fraction of proliferating cells (% Ki-67) were recorded (25). Tumor samples were analyzed for correlated expression of Trop-2, AKT, and TP53 (26, 27) by immunohistochemistry (IHC) as described (Supplementary Methods). This project, including case collection and procedures of analysis, was approved by the Italian Ministry of Health (RicOncol RF-EMR-2006-361866). Informed consent was obtained from all subjects and the experiments conformed to the principles set out in the World Medical Association Declaration of Helsinki (28).
Statistical analysis
The Student t test was used for comparisons between mean protein levels in control and TROP2 transfectants in the antibody microarrays. Normality of distribution of assay values was verified. Two-tailed Fisher exact tests were used to compare protein expression levels in normal versus tumor samples. Spearman nonparametric correlation coefficients were computed for protein expression levels in human cancer samples. ANOVA (21) and post hoc Bonferroni t test were used for comparison of tumor growth curves. Data were analyzed using Sigma Stat (SPSS Science Software UK Ltd.), GraphPad Prism, and SISA (29).
Results
The Trop-2 signaling network
Trop-2–driven growth-inducing pathways were dissected through differential proteomic profiling (Supplementary Methods). Comparative assessment of wild-type Trop-2–expressing versus parental cells, across 258 different phosphorylation sites, 240 protein kinases, and 111 other cell signaling proteins that regulate cell proliferation, stress response, and apoptosis (Supplementary Table S1), revealed 109 Trop-2–modulated signaling molecules/activatory modifications (Supplementary Table S1A; 45 upregulated, 64 downregulated). Target molecules included transmembrane receptors, cytoplasmic Tyr and Ser/Thr kinases, phosphatases, cell cycle, and apoptosis-regulatory molecules.
Association analysis through NetworkAnalyst identified global protein–protein interaction maps, which included 3,396 nodes, 7,369 edges, and 95 seed proteins. Minimum interaction networks were subsequently generated through the Trim algorithm, for a core network which included 391 nodes and 2,029 edges from 94 seed proteins (Supplementary Tables S1B–S1D). This final network was mapped over the KEGG database, and highest scores were obtained by the Cancer Pathway (N hits = 69; P, 1.06E−26; Supplementary Table S1C). Pathways in the Cancer Reactome were utilized to dissect the most relevant signaling chains. The largest contribution was provided by transmembrane Tyr-kinase growth factor receptors (IGF1R, FGFR, KIT, ERBB2, ERBB4, PDGFR, TGFß), with signaling through SOS, GRB2, RAS, SHC1, IRS, toward ERK/MAPK/MAPK14, JNK, and cyclin D1, for regulation of the cell cycle and apoptosis. ERK triggering correlated with phosphorylation of MEK1 at the S298 activatory site (30) and hypophosphorylation at the T292 inhibitory site (Supplementary Table S1A), as a feedback mechanism driven by PAK1/ERK1/ERK2 (31). Downstream targets included NFKB and TP53 (32), with AKT-mediated MDM2 phosphorylation for TP53 ubiquitination. Seven of the identified pathways were found to include members of the PTEN/PIK3CA/AKT signaling network (Supplementary Table S1D).
Identified downstream effectors were challenged by high-throughput multiplex Western blotting (Supplementary Fig. S1). One hundred and sixty five proteins were validated as Trop-2–regulated molecules. At least two-thirds of these (115/165) were shown to take part to a network of direct protein–protein interactions (Supplementary Table S1E). This was shown to map over cancer growth signaling paths (P = 6.19 × 10−45; Supplementary Table S1F), with PTEN/PIK3CA/AKT/GSK3ß as a major activated pathway.
Akt is a required mediator of Trop-2 downstream signaling
The AKT signaling pathway is initiated by upstream activatory phosphorylation. At least 13 AKT phosphorylation sites have been identified, 11 of which appear involved in regulating function in transformed cells (Supplementary Table S2). Among these, phosphorylation at T308 and S473 was shown to play a pivotal role in AKT activation and kinase catalytic activity, in apoptosis, cell cycle/cell growth, and cell differentiation. Growth factors binding to receptor tyrosine kinases leads to activation of PIK3CA. PIK3CA then translocates to the cell membrane where it phosphorylates the PI inositol ring at the D3 position to form PI (3, 4, 5) triphosphate (PIP3). PIP3 serves to anchor AKT to the plasma membrane, where AKT is phosphorylated at T308 by the phosphoinositide-dependent kinase 1 (PDK1; ref. 33). AKT is subsequently fully activated by phosphorylation at S473 by CRTC2, PRKDC (34), or PRKCA (PKCα) (35). PKCα was shown to positively regulate T-cell-receptor (TCR)-induced AKT S473 phosphorylation and phosphorylated AKT at S473 in vitro upon TCR stimulation (36). ILK can also phosphorylate AKT at S473, but is downregulated by PKCα (37). Consistently, Trop-2 was found to induce AKT phosphorylation at T308 and S473 (Fig. 1A and Supplementary Fig. S2), both in in vitro models and in human cancer.
AKT phosphorylation at T308 was associated with a Trop-2–driven stoichiometric downregulation of the Ca++-dependent PP2A phosphatase (ref. 34; Supplementary Table S1A; Supplementary Fig. S1), and with diminished levels of the PDK1 kinase, suggesting diminished dephosphorylation of AKT, rather than induction of activatory kinases. Our data on a central role of the AKT and PKCα pathways in Trop-2 signaling are supported by reports showing that PKCα phosphorylates PP2A at the regulatory subunit and inhibits its activity, consequently reducing the rate of AKT dephosphorylation/inhibition (38). The AKT signaling cascades converge on GSK3 (39), promoting cell-cycle progression and survival. GSK3 is generally expected to be inhibited by AKT-mediated phosphorylation. However, it also exerts a regulatory feedback on AKT. Phosphorylation of GSK3 inhibits proteasomic degradation of cyclin D1, and GSK3 was found to be hyperphosphorylated upon Trop-2 expression in tumor cells for enhanced levels of cyclin D1.
Cell–cell contacts play an inhibitory role on cell growth in vitro and in vivo. Consistently, cell confluency was found to abolish Trop-2–induced hyperphosphorylation of AKT at S473, supporting a central role of Trop-2 in the regulation of transformed cell growth. Corresponding findings were obtained for T308, but much to less dramatic extents, consistent with selective regulation of the two AKT phosphorylation sites. Antibody cross-linking is utilized to mimic cell–cell contact signaling by molecules engaged in cell adhesion (5, 40). Hence, we went on to generate the 2E7 anti-Trop-2 murine mAb as an acute modulator of Trop-2 signaling. Consistent with inhibition of AKT S473 activatory phosphorylation by cell confluency, acute cross-linking of Trop-2 with the 2E7 mAb strongly diminished phospho-AKT S473 levels (Fig. 1A).
The cytoskeleton transmits outside-in signals upon cell–cell contact, for triggering downstream signaling events (41). Moreover, Trop-2 directly modulates the expression of β-actin, and of regulators of β-actin polymerization, such as cofilin-1, gelsolin-like capping protein and profilin-1, and induces phosphorylation of the myosin regulatory light chain at S19 and of the microtubule-associated protein tau at S687 (Supplementary Table S1A). Hence, we went on to assess the role of cytoskeletal organization in Trop-2 signaling toward AKT. Depolymerization of β-actin by cytochalasin D was shown to abolish Trop-2 cross-linking–driven AKT dephosphorylation (Fig. 1A). Trop-2–AKT signaling inhibition was also obtained by administering the myosin inhibitor blebbistatin (Fig. 1A). Hence, the actin–myosin cytoskeleton plays a mandatory role in transducing signal from Trop-2 to AKT.
As AKT activation requires recruitment to the cell membrane, we went on to assess whether this depended on Trop-2. AKT was found to tightly colocalize with Trop-2 at the cell membrane in cancer cells (Fig. 1B, top). Trop-2–colocalized AKT was then shown to be phosphorylated at both T308 and S473 activatory sites (Fig. 1B, bottom). AKT recruitment and activation at the cell membrane is mediated by binding of the pleckstrin-homology (PH) domain of AKT to PIP3 (Supplementary Table S2C). Hence, we went on to assess whether Trop-2 recruits the AKT PH domain (42) at the cell membrane. Consistently, AKT PH domain-EGFP chimeras were found to be dynamically corecruited to Trop-2 sites, and this was shown to occur in quantitative relationship with Trop-2 protein density (Supplementary Fig. S3; Supplementary Video S1).
Trop-2 determines tumor response to AKT inhibitors
We assessed the activity of growth inhibitors directed against major cancer-driving pathways, to identify which ones targeted Trop-2–induced tumor growth. MTOR, JNK, and NFKB inhibitors were found to be effective in reducing growth of transformed cells. However, they did not result in differential inhibition of Trop-2–expressing versus nonexpressing cells (Supplementary Fig. S4; Supplementary Table S2D). This was confirmed in formal dose–response assays (Supplementary Fig. S5), where the above inhibitors showed comparable IC50s between MTE/vector and MTE/TROP2 transfectants (respectively; IC50-PP242, 1.01 and 1.17 μmol/L; IC50-Bay11-7082, 1.27 and 2.08 μmol/L; IC50-SP 600125, 11.1 and 12.1 μmol/L) suggesting involvement of MTOR, JNK, and NFKB in a basal, Trop-2–independent, layer of cell growth.
Allosteric AKT inhibitors have been designed (43) that bridge the PH domain of AKT to the kinase domain. This prevents the binding of the AKT PH to plasma membrane phospholipids (42) and of the AKT kinase domain to its substrates, thus specifically inhibiting AKT activity (Supplementary Table S2E; Supplementary Fig. S2B). This two-layered inhibition provided with a distinct advantage over ATP-competitive AKT inhibitors, which showed a much higher likelihood of off-target effects (10). Allosteric AKT inhibitors were shown to dramatically inhibit the proliferation of Trop-2–expressing transformed cells, in a dose-dependent manner (Fig. 2A), but had no impact on Trop-2–negative cells. Trop-2–dependent growth sensitivity to AKT inhibition was confirmed in an independent set of experiments over a wider dose range, showing lower IC50 for the MTE/TROP2 transfectants over control MTE/vector (251.7 nmol/L and 537.1 nmol/L, respectively).
Network analysis revealed a predominant action of AKT in Trop-2 signaling through the ERK/RPS6KB1 (S6K), rather than the MTOR, arm (Supplementary Table S1G), consistent with the split scenario depicted by drug inhibition findings (Fig. 2 and Supplementary Fig. S4). Hence, we went on to assess whether the somatic knockout of AKT with siRNA would result in a Trop-2–dependent inhibition of growth. Remarkably, again we observed inhibition of growth of Trop-2–expressing cells only, whereas Trop-2–null cells were not affected (Fig. 2B). We then went on to assess the impact of downregulation of endogenously expressed Trop-2, through lentiviral shRNA targeting of TROP2 mRNA. shRNA effectively inhibited Trop-2 expression (Fig. 3A), and this abolished tumor response to AKT allosteric inhibitors (Fig. 3B). Taken together, these findings indicate that Trop-2 is an activator of AKT function.
A first-in-man phase I clinical trial of an AKT inhibitor (MK-2206) was conducted in patients with advanced solid tumors (43). AKT inhibitors were thus assessed in preclinical models for their ability to reduce the growth of Trop-2–expressing tumors. Colon cancer cells were shown to express high levels of endogenous Trop-2, of AKT and of AKT activation, as in human primary tumors (Fig. 2C and Supplementary Fig. S2). Colo205 xenografts were thus treated with the MK-2206 AKT inhibitor. Remarkably, treatment dramatically inhibited tumor growth (Fig. 2D). To formally assess the Trop-2 dependence of such a response to AKT inhibition, we tested the MTE transfectants in vivo. When inoculated in nude mice, only the MTE/Trop-2 transfectants grew as tumors, whereas the MTE/vector cells did not produce any appreciable growth. Hence, Trop-2 confers tumorigenic capabilities to the immortalized MTE cell line, further supporting the role of Trop-2 in tumor growth (7). Strikingly, AKT inhibitors significantly reduced the growth of the Trop-2–dependent tumors (Fig. 2D).
Activated AKT associates with Trop-2 and downstream growth effectors in primary breast cancer
We thus went on to assess the relevance of the Trop-2/AKT growth-activatory control module in human cancer. Trop-2 is overexpressed in breast cancer (7) and determines breast cancer survival (9). Hence, we quantitatively assessed a case series of 414 human breast cancers (44) for correlative expression parameters of Trop-2, activated AKT and downstream effectors. The clinical and pathologic characteristic of the breast cancer case series are reported in Supplementary Table S3. Premenopausal and postmenopausal patients (median age: 59.7 years) carrying breast cancers at early stages (65.7% T1 tumors) were enrolled in the study. ERα and PGR were expressed by 76.6% and 58.5% of the patients tumors, respectively. Invasive ductal carcinomas were 79.7%, invasive lobular carcinomas were 13%, other histotypes represented 7.3% of the cases, as broadly representative of the natural appearance and progression of breast cancer (25). IHC analysis identified ERα, PGR, and ERBB2 (HER2) as expressed by 76.6%, 58.5%, and 15.6% of the breast cancers, respectively.
Differential correlation of membrane and storage Trop-2 with distinct risk parameters in breast cancer was previously shown (25). IHC analysis showed that membrane Trop-2 was marginally correlated with grade. Cytoplasmic/storage Trop-2 was found to be significantly correlated with ERα, TP53, cyclin D1, cyclin E, CDKN1B (p27), and CDKN2A (p16) and showed a trend of correlation with HER2 (Table 1). Cytoplasmic Trop-2 was found tightly coexpressed with cytoplasmic-activated S473 and T308 phospho-AKT (Spearman ρ = 0.489; two-tailed P < 0.0001, Spearman ρ = 0.280; two-tailed P < 0.0001, respectively; Supplementary Fig. S6; Supplementary Table S2A). Current investigational models are focused on dynamics of the cytoplasmic fraction of Trop-2, which may contain both actively recycling molecules from the cell surface, and bona fide Trop-2 repositories. S473 and T308 phospho-AKT molecules were shown to translocate to the nucleus of Trop-2–expressing cells as competent activated forms (Spearman ρ = 0.388; two-tailed P < 0.0001, Spearman ρ = 0.355; two-tailed P < 0.0001, respectively; Table 1 and Supplementary Table S2A). S473 and T308 phospho-AKT were then shown to tightly correlate with the presence of ERα, cyclin D1, cyclin E and p16, p27 cyclin inhibitors, as nodal regulators of tumor growth (Supplementary Table S2), globally supporting a key role of AKT and Trop-2 in signaling for growth in primary human breast cancer.
. | Storage Trop-2b (162–46.2 mAb) . | Membrane Trop-2b (AF650 polyclonal) . |
---|---|---|
ERα | ||
Rho | 0.196 | 0.000 |
P | <0.0001 | 0.998 |
PGR | ||
Rho | −0.034 | 0.055 |
P | 0.486 | 0.262 |
TP53 | ||
Rho | 0.213 | −0.108 |
P | <0.0001 | 0.028 |
Ki-67 | ||
Rho | 0.086 | −0.067 |
P | 0.078 | 0.174 |
Cyclin D1 | ||
Rho | 0.435 | −0.122 |
P | <0.0001 | 0.014 |
Cyclin E | ||
Rho | 0.358 | −0.103 |
P | <0.0001 | 0.038 |
Cytokeratin 5/6 | ||
Rho | −0.009 | 0.159 |
P | 0.866 | 0.001 |
p27 | ||
Rho | 0.552 | 0.026 |
P | <0.0001 | 0.601 |
p16 (N) | ||
Rho | 0.322 | −0.078 |
P | <0.0001 | 0.116 |
p16 (C) | ||
Rho | 0.371 | 0.052 |
P | <0.0001 | 0.295 |
BCL2 | ||
Rho | 0.043 | 0.030 |
P | 0.380 | 0.541 |
pAKT (S473) (C) | ||
Rho | 0.489 | −0.084 |
P | <0.0001 | 0.098 |
pAKT (S473) (N) | ||
Rho | 0.388 | −0.084 |
P | <0.0001 | 0.097 |
pAKT (T308) (C) | ||
Rho | 0.280 | −0.084 |
P | <0.0001 | 0.207 |
pAKT (T308) (N) | ||
Rho | 0.355 | −0.181 |
P | <0.0001 | 0.008 |
HER2c | ||
Rho | 0.119 | 0.104 |
P | 0.051 | 0.101 |
. | Storage Trop-2b (162–46.2 mAb) . | Membrane Trop-2b (AF650 polyclonal) . |
---|---|---|
ERα | ||
Rho | 0.196 | 0.000 |
P | <0.0001 | 0.998 |
PGR | ||
Rho | −0.034 | 0.055 |
P | 0.486 | 0.262 |
TP53 | ||
Rho | 0.213 | −0.108 |
P | <0.0001 | 0.028 |
Ki-67 | ||
Rho | 0.086 | −0.067 |
P | 0.078 | 0.174 |
Cyclin D1 | ||
Rho | 0.435 | −0.122 |
P | <0.0001 | 0.014 |
Cyclin E | ||
Rho | 0.358 | −0.103 |
P | <0.0001 | 0.038 |
Cytokeratin 5/6 | ||
Rho | −0.009 | 0.159 |
P | 0.866 | 0.001 |
p27 | ||
Rho | 0.552 | 0.026 |
P | <0.0001 | 0.601 |
p16 (N) | ||
Rho | 0.322 | −0.078 |
P | <0.0001 | 0.116 |
p16 (C) | ||
Rho | 0.371 | 0.052 |
P | <0.0001 | 0.295 |
BCL2 | ||
Rho | 0.043 | 0.030 |
P | 0.380 | 0.541 |
pAKT (S473) (C) | ||
Rho | 0.489 | −0.084 |
P | <0.0001 | 0.098 |
pAKT (S473) (N) | ||
Rho | 0.388 | −0.084 |
P | <0.0001 | 0.097 |
pAKT (T308) (C) | ||
Rho | 0.280 | −0.084 |
P | <0.0001 | 0.207 |
pAKT (T308) (N) | ||
Rho | 0.355 | −0.181 |
P | <0.0001 | 0.008 |
HER2c | ||
Rho | 0.119 | 0.104 |
P | 0.051 | 0.101 |
Abbreviations: C, cytoplasmic staining; M, membrane staining; N, nuclear staining; P, P value; Rho, Spearman coefficient of correlation.
aAnalyses were performed on the breast cancer case series as described in ref. 44 (N = 414; T1-T2, N0, M0 cases).
bSubcellular localization of Trop-2 was assessed according to ref. 9, indexing vesicle storage sites and full posttranslational processed cell membrane molecules; see the extended analysis presented in Supplementary Table S2A.
cHER2 was assessed on 288 cases.
Discussion
There is a lack of predictive biomarkers to assign effective therapy to cancer patients (45). Most efforts are directed at inferring drug response from target measurement. However, there is key information to be gained from measuring target activity and target-activatory networks. One such a case is AKT. AKT-inhibitory drugs are currently being tested in clinical trials, but no predictive indicators of therapeutic response have been, as yet, identified (43). We performed a global analysis of the Trop-2 signaling network, through differential proteomic profiling of Trop-2–expressing cells. This led us to identify AKT as a key hub of the Trop-2–driven regulatory network of tumor cell growth, and to indicate the Trop-2/AKT signaling path as a target for therapeutic intervention against cancer.
The successful design of target diagnostics played a key role in precision medicine strategies, for example in the case of HER2–targeting antibodies, by detection of overexpression through IHC or in situ hybridization (46). Effective molecular characterization of patients' tumors (26, 32), for example, for alterations of signal transducing molecules, further extended these approaches. For example, EGFR mutations predict better response to the kinase inhibitor erlotinib, whereas mutated KRAS associates to worse prognosis and lack of response to erlotinib (47). Use of target-specific therapies on unselected populations, on the other hand, inevitably dampened estimates of overall efficacy, or even lead to observations of apparent lack of efficacy, as it was the case for early studies on cetuximab treatment of colon cancer, before the era of analyses for KRAS mutations.
AKT is a key regulator of PIK3CA signaling, and the activation of this pathway results in increased tumor cell survival, proliferation, growth, and metabolism, for malignant progression and chemoresistance. ATP-competitive AKT inhibitors have been reported to have a higher likelihood of off-target effects (10). Allosteric AKT inhibitors have thus been designed that bind to the AKT protein PH domain, and result in a conformational change that prevents localization of AKT to the plasma membrane for subsequent activation (48). A first-in-man phase I clinical trial of an AKT inhibitor (MK-2206) was conducted in patients with advanced solid tumors (43).
Multiple signaling activators, such as receptor tyrosine kinases, for example, HER family members, can converge on AKT and activate it by PIK3CA-dependent phosphorylation at S473 (49). Notably, our analysis of Trop-2 signalosome showed highly significant overlapping with the HER2 signaling pathway. Consistently, Trop-2 loss was shown to activate HER3 (50), overall suggesting a model of interactive signaling between Trop-2 and HER family receptors. Hyperphosphorylation of AKT can also be determined by distinct PIK3CA mutations that constitutively activate PIK3CA enzymatic activity. Indeed, the colon cancer HCT-116 cells, which carry the PIK3CA-activatory H1047R mutation, were found to activate AKT (P-S473) under baseline conditions, despite expressing low levels of Trop-2 (Supplementary Fig. S2A). However, Trop-2 overexpression in HCT-116 cells was still able to induce an increase in phosphorylation at S473 (Supplementary Fig. S2A) and faster growth. These findings suggest that AKT, while acting as an integrator of broad signaling activity of heterogeneous activatory pathways, still maintain the ability to respond to a superimposed Trop-2–dependent stimulus.
Our findings provide for the first time evidence for Trop-2 as an activator of AKT, consistent with a role of Trop-2 as a pivotal driver of tumor growth (2–6). Inhibitors of AKT abolished Trop-2–induced growth, but had no impact on basal growth rates of Trop-2–null cells, indicating Trop-2 as a critical flag molecule for AKT-dependent tumor growth, and for AKT viability as a target in cancer therapy. These findings help shifting current paradigms of prediction of drug response toward the identification of drivers of the activatory networks of the target molecule, rather than of the simple target per se. Novel possibilities are correspondingly opened in therapeutic clinical trial design, through stratification of patients subgroups by likelihood of responsiveness, for more efficient testing of AKT-targeted therapeutic strategies.
Disclosure of Potential Conflicts of Interest
No potential conflicts of interest were disclosed.
Disclaimer
The sponsors had no role in the design and conduct of this study, nor in the collection, analysis, and interpretation of the data, or in the preparation, review, or approval of the manuscript.
Authors' Contributions
Conception and design: E. Guerra, M. Trerotola, M. Piantelli, S. Alberti
Development of methodology: M. Trerotola, R. Tripaldi, A.L. Aloisi, V. Relli, A. D'Amore
Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): E. Guerra, M. Trerotola, R. Tripaldi, A.L. Aloisi, P. Simeone, A. Sacchetti, V. Relli, A. D'Amore, R. Lattanzio, M. Piantelli
Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): M. Trerotola, R. Tripaldi, A.L. Aloisi, P. Simeone, A. Sacchetti, V. Relli, A. D'Amore, R. Lattanzio, M. Piantelli
Writing, review, and/or revision of the manuscript: E. Guerra, M. Trerotola, R. Tripaldi, P. Simeone, V. Relli, A. D'Amore, R. Lattanzio, S. Alberti
Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): A. Sacchetti, R. La Sorda
Study supervision: S. Alberti
Acknowledgments
The authors thank G. Vacca, F. Dini, U. Weidle, and G. Lee for help during the course of this work.
Grant Support
The support of Fondazione of the Cassa di Risparmio della Provincia di Chieti, Fondazione Compagnia di San Paolo (Grant 2489IT), ONCOXX Biotech and the Italian Ministry of Health (RicOncol RF-EMR-2006-361866) is gratefully acknowledged.
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.