The Tumor-Suppressor Protein OPCML Potentiates Anti–EGFR- and Anti–HER2-Targeted Therapy in HER2-Positive Ovarian and Breast Cancer

This article is featured in Highlights of This Issue, p. 2045

Opioid-binding protein/cell adhesion molecule-like (OPCML) is a glycosyl phosphatidylinositol (GPI) anchored molecule that is expressed in virtually all normal adult and fetal tissues (1). We first identified OPCML as a putative tumor-suppressor gene (TSG) at chromosome 11q25, a site highlighted by frequent loss of heterozygosity (LOH), and further demonstrated that epigenetic silencing of the remaining OPCML allele occurred in 83% of epithelial ovarian cancers (EOC; ref. 2). OPCML methylation in EOC has also been confirmed by other studies (3, 4). Furthermore, OPCML is inactivated in numerous other types of cancers, such as brain, nasopharyngeal, esophageal, gastric, hepatocellular, colorectal, breast and cervical cancers, as well as lymphomas (1, 4–6). In lung adenocarcinoma, OPCML is one of the four most commonly methylated genes compared with normal lung tissue (7, 8). All these data indicate that OPCML expression is very frequently lost across different types of cancers. Recently, we have defined that OPCML acts as a TSG by negatively regulating a set of receptor tyrosine kinases (RTK), including HER2, HER4, EphA2, FGFR1, and FGFR3 (9). In particular, OPCML binding to the extracellular domain (ECD) of HER2 promotes HER2 ubiquitination and proteasomal degradation via clathrin-independent endocytosis (9). Conversely, silencing of OPCML causes upregulation of this same repertoire of RTKs and increased signaling (9). OPCML therefore induces a system level switch-off of a spectrum of RTKs, which eventually leads to a strong decrease in phosphorylation and activation of key downstream signaling molecules such as ERK and AKT, thus reducing cell proliferation in vitro and in vivo (9).

HER2 is an RTK whose overexpression drives cell proliferation in many human cancers. Studies have shown that ovarian tumors with high malignant potential overexpress HER2 in 22% to 66% of cases (10), although recent consensus reduces the number of cases with genomic amplification of HER2 to 11% (11). In breast cancer, HER2 amplification is reported in 15% to 20% of cases (12). In comparison with the above, EGFR overexpression can be as high as 28% and amplified in up to 20% of ovarian cancer patients (13) and in 13% to 52% of triple-negative breast cancers (14).

The concept of “oncogene addiction,” by which the growth of some tumors relies heavily on the amplification of a single oncogene (15), as well as its downstream pathway, led to the development of targeted therapies directed against HER2 and EGFR. Trastuzumab (a monoclonal antibody targeting HER2) and lapatinib (a small-molecule inhibitor against HER2 and EGFR) have been used successfully in the treatment of metastatic HER2-positive breast cancer along with chemotherapy, resulting in increased response rate, prolonged progression-free survival and extended overall survival compared with chemotherapy alone (16, 17). However, only 20% to 30% of HER2 positive breast tumors respond to single HER2-targeted therapy (18) and those that initially respond eventually develop recurrence due to drug resistance (19). Erlotinib (Tarceva, an EGFR-specific inhibitor) has been approved for the treatment of recurrent non–small cell lung cancer (NSCLC; ref. 20) with a specific EGFR mutation (21, 22). Despite the positive initial response, also these patients develop resistance after 10 to 13 months of treatment with erlotinib (23). Mechanisms of resistance to tyrosine kinase inhibitors (TKI) include the activation of alternative pathways through RTK cross-talk (24–29) and mutations in the tyrosine kinase domain of the receptors (30–32). Several alternative approaches have been proposed to overcome drug resistance, such as the concomitant use of lapatinib and trastuzumab upfront and the creation of new small-molecule inhibitors (33), and these are currently being evaluated.

Because we have shown that the expression of OPCML is a good prognostic factor for progression-free survival in ovarian and breast cancer patients and that it induces HER2 downregulation in ovarian cancer cells (9), OPCML restoration could represent a novel approach to reduce the level of activity of this oncogene and thereby sensitize cancer cells to anti-HER2 therapy. Furthermore, as HER2 is the preferred binding partner of EGFR (34, 35), OPCML restoration could also prove effective for anti-EGFR therapy in HER2-positive cancer cells.

We therefore investigated this hypothesis in a panel of breast and ovarian cancer cell lines that do not express OPCML. Here we show that OPCML restoration interferes with the HER2-EGFR heterodimer formation by interacting directly with HER2 but not with EGFR, and that stably expressing OPCML in ovarian and breast cancer cells results in sensitization to HER2/EGFR (lapatinib) and EGFR-only (erlotinib) targeted therapies. Finally, we show that high endogenous OPCML expression is associated with better response to lapatinib in breast cancer patients and with better progression free survival in ovarian cancer patients with high HER2 expression.

SKOV3 cells (ATCC HTB-77) were maintained in RPMI medium (Sigma-Aldrich) supplemented with 10% FCS (First Link), 2 mmol/L l-glutamine (LG; Invitrogen), 50 U/mL Penicillin/Streptomycin (Invitrogen). OV90 cells (ATCC CRL11732) were cultured in a 1:1 mixture of MCDB 105 medium (Sigma-Aldrich) and Medium 199 (Sigma-Aldrich) supplemented with 15% FCS, 2 mmol/L LG, 50 U/mL Penicillin/Streptomycin. SKBR3 and MDA-MB-231 (ATCC HTB-26) cells were cultured in DMEM medium (Sigma-Aldrich) supplemented with 10% FCS, 2 mmol/L LG, 50 U/mL Penicillin/Streptomycin. Puromycin was used at 1 μg/mL as a selective antibiotic for SKOV3 cell lines, at 0.1 μg/mL for OV90 cells and at 0.2 μg/mL for SKBR3.

COS-1 simian kidney fibroblast cells (CV-1) were maintained in DMEM, supplemented with 10% FCS, 2 mmol/L LG, 50 U/mL Penicillin/Streptomycin.

All cells were tested regularly to exclude mycoplasma infection. COS-1 cells were tested and authenticated as described previously (9). All the other cell lines were purchased from the ATCC in January 2013.

Polyclonal goat anti-human OBCAM (OPCML) antibody was purchased from R&D Systems. Total and phospho-HER2 (Y1248), total and phospho-EGFR (Y1068), total and phospho-AKT (S473) antibodies were purchased from Cell Signaling Technology. Total HER2 antibody (OP16L) used for Duolink was purchased from Merck Millipore. Calnexin antibody was purchased from ENZO Life Sciences.

Before collection, cells were washed in PBS. Whole-cell lysates were collected in RIPA buffer (Sigma-Aldrich) supplemented with protease inhibitor cocktail (Roche) and phosphatase inhibitor cocktail II (Calbiochem) at the manufacturers' recommended concentration. Protein concentration was estimated using the BCA assay (Pierce) according to the manufacturer's recommendations. Lysates were incubated at 95°C for 5 minutes and then separated into 10% sodium dodecyl sulphate polyacrylamide gel electrophoresis (SDS-PAGE). Gels were then transferred onto nitrocellulose membranes (VWR) that were then blocked with 5% milk/TBST. Membranes were incubated with primary antibodies in 5% BSA/TBST or 5% milk/TBST at 4°C overnight. HRP-conjugated secondary antibodies (Dako) were used at a 1:2,000 dilution in either 5% BSA/TBST or 5% milk/TBST for incubation for 1 hour at room temperature. Proteins were detected using Immobilon Western Chemiluminescent HRP Substrate system (Millipore) and developed on X-ray film using Kodak SRX2000 (Rochester) developer machine. Quantitative densitometry analysis was carried out using ImageJ software (National Institute of Health).

The GST-OPCML construct was described previously (9). For mammalian 2-hybrid studies, the domains 1, 2, and 3 of OPCML were amplified by PCR from a human cDNA clone (pcDNA3.1-OPCML) and sub-cloned into the pM vector using as forward primer 5′-GggatccGTGCCACCTTCCCCAAAGCTATG-3′ and as reverse primer 5′-aagcttCCAGGCCCATACAATGTGATGC-3′. The full ECD of p185^neu/HER2 (NEX) was amplified by PCR from the expression plasmid pNex (36) kindly provided by Prof. Mark Greene, University of Pennsylvania, using the forward primer 5′-gaattcCCCGGAATCGCGGGCACCC-3′ and the reverse primer 5′-gtcgacCGTATACTTCCGGATCTTCTG-3′. The full ECD of EGFR was amplified by PCR from a human cDNA clone (EGFR-pcDNA3.1zeo, kindly provided by Prof. Bill Gullick, University of Kent) using the forward 5′-gtcgacGCCCGGCGAGTCGGGC-3′ and the reverse primer 5′-aagcttGGACGGGATCTTAGGCCC-3′. The p185^neu/HER2 and EGFR constructs were sub-cloned into the pVP16 vector.

For virus production, p8.9 (gag/pol) and pmol/LDG (VSVG envelope) plasmids were kindly provided by Dr. Ari Fassati, University College London, UK; the full-length OPCML sequence was cloned into pLenti CMV Puro Dest that was purchased from Addgene.

Assays were performed with recombinant GST-OPCML fusion proteins bound to magnetic glutathione beads (Promega) as described previously (9).

The three Ig domains of OPCML were cloned in the pM vector downstream of the SV40 promoter and GAL4 DNA–binding domain, whereas the ECDs of EGFR and HER2 and the mutated HER2, lacking the ECD 4 (Δ4), were cloned in the pVP16 vector downstream of the SV40 promoter and VP16 activation domain. COS-1 cells were transiently transfected at 80% confluence in 24-well plates for 2 to 3 hours with 50 ng of the pGAL4 Luc reporter, 50 ng of each of the different pM and pVP16 constructs and 50 ng of the β-gal transfection control plasmid using Effectene Transfection Reagent (Qiagen) at a 1:10 DNA to Effectene reagent ratio. Cells were then washed with 1X PBS and incubated in full medium overnight. Cells were collected in 1X Cell Culture Lysis Reagent (CCLR) from the Promega Luciferase Assay System and incubated at −80°C for 30 minutes. Luciferase activity was measured by adding lysates and Luciferase Assay Substrate in Assay buffer (Promega) in equal volumes to white flat-bottomed 96-well plates (PerkinElmer) followed by reading with a VICTOR XLight Luminescence Plate Reader (PerkinElmer) using the emission filter slot A7.

Transfection efficiency was measured using the Galacto-Light Plus beta-Galactosidase Reporter Gene Assay System (Applied Biosystems) according to the manufacturer's guidelines. Each experiment was carried out in quadruplicate and the luciferase measurements were corrected by dividing them for the β-galactosidase readings. Data presented are the mean of each repetition normalized to the empty pM and pVP16 condition for each experiment and are expressed as mean ± standard error of mean (SEM). Student's unequal variances t tests (two-tailed) of the data were performed to determine statistical significance of a positive protein–protein interaction using the OPCML-pM and RTK-pVP16–negative controls as baseline. All statistical analysis was carried out in Microsoft Excel.

For proximity ligation assay (PLA assay; Duolink, OLink Biosciences, Sigma-Aldrich), the various cell lines were seeded on 13 mm glass coverslips. After the required treatments, cells were fixed with ice-cold methanol for 5 minutes at −20°C, washed with PBS and fixed with 3% BSA in PBS for 30 minutes at room temperature. Cells were then labeled with primary antibodies diluted 1:100 in 1% BSA/PBS at room temperature for 1 hour, taking care to use two different species (e.g., Mouse anti-HER2 and Rabbit anti-EGFR). Incubation with PLA probes MINUS and PLUS, ligation and amplification steps were carried out as per the manufacturer's instructions. Quantification of the number of Duolink particles was performed using the ImageJ software.

Cells were seeded in triplicate in 96-well plates and treated with increasing concentrations of targeted therapy (0–100 μmol/L) for 48 hours before measuring cell viability by CellTiter 96 AQ_ueous One Solution Cell Proliferation assay (Promega) following the manufacturer's recommendations. GI50 values were calculated using the GraphPad Prism software.

For Western blot analysis, cells were treated with targeted therapy for 3 hours and then stimulated with 50 ng/mL rhEGF (R&D systems) for 30 minutes before the cells were lysed for 20 minutes on ice.

Lapatinib (Sigma-Aldrich) and erlotinib (VWR) were reconstituted in dimethyl sulfoxide (DMSO; Sigma-Aldrich) and used at the concentrations shown. Each experiment was repeated three times and a representative figure from each experiment is included in the result section.

Cells were seeded in triplicate in 96-well plates and treated with increasing concentrations of targeted therapy (0–120 μmol/L) for 24 hours before assessing apoptosis by measuring caspase 3/7 activation using the Caspase-Glo 3/7 assay (Promega) following the manufacturer's instructions. Caspase activity was normalized to cell density obtained by CellTiter 96 AQ_ueous One Solution Cell Proliferation assay (Promega) for each treatment. Each experiment was repeated three times.

SKOV3 cells were transfected in 24-well plates with 20 nmol/L of siRNA 24 hours post trypsinization. After cells were harvested, 2,000 cells per well were seeded in triplicate in 96-well plates for measuring cell viability by CellTiter 96 AQ_ueous One Solution Cell Proliferation assay (Promega) and 30,000 cells per well were seeded in 24-well plate for western blot analysis. siRNAs containing a pool of siRNAs against EGFR2 (SMARTpool ON-TARGETplus L-003114) and HER2 (SMARTpool ON-TARGETplus L-003126) were transfected either alone or in combination, non-targeting siRNA #2 was used as control. The siRNAs were transfected using Oligofectamine (Invitrogen, Life Technologies Corporation) in Opti-MEM media (GIBCO, Thermofisher Scientific) according to the manufacturer's protocol. All siRNAs were purchased from Dharmacon (GE Healthcare Life Science).

Histopathological samples were obtained from 8 patients with advanced recurrent breast cancer that were treated previously with adjuvant chemotherapy and following treatment with lapatinib (Tyverb). The following information was extracted from the histopathology reports: tumor type, grade, stage, estrogen receptor (ER), progesterone receptor (PR), as well as HER2 immuno-scores. Sections of 2-μm thickness from each tumor were cut from paraffin wax-embedded blocks, floated onto polysine slides, and dried overnight at 37°C. The next day, sections were de-waxed, rehydrated, and incubated with 0.6% peroxide block for 15 minutes to block endogenous peroxidase activity. Antigen retrieval was performed by microwaving the samples for 20 minutes in EDTA (pH 9.0) at 1,000 W. Background staining was blocked using Power block (protein block supplied by Biogenex) for 10 minutes. The sections were then incubated with the monoclonal anti-human OPCML antibody (R&D Systems) at 1 μg/mL for 1 hour at room temperature and detected using Super-Sensitive Polymer-HRP detection kit with diaminobenzidine substrate at the manufacturer's instructions (Biogenex, Launch Diagnostics). Counterstaining was carried out with hematoxylin. For negative controls, duplicate sections were used in which the primary antibody was omitted and replaced with PBS. Sections were examined by light microscopy and staining was assessed with respect to presence, localization and intensity of staining. Staining was scored as weak (+), moderate (++) or strong (+++). Ethical approval to carry out expression analysis on snap-frozen samples from breast cancer patients was approved by Hammersmith and Queen Charlotte's and Chelsea Research Ethics Committee (REC reference 05/Q0406/178).

Our previous co-immunoprecipitation (co-IP) experiments suggested that OPCML interacts with HER2, but not with EGFR, in the ovarian cancer cell line SKOV3 (9). Here, we have investigated more in depth this difference by using a recombinant GST-OPCML to perform pull-down experiments in two breast cancer cell lines with no detectable OPCML expression: SKBR3, which overexpresses HER2 and is positive for EGFR, and MDA-MB-231, positive for EGFR and expressing very low levels of HER2. In SKBR3 cells, we found that HER2 was efficiently pulled-down in the cell lysate by GST-OPCML (Fig. 1A), but no EGFR could be detected (Fig. 1B), thus validating our previous findings. The pull-down assay was then repeated in the HER2-negative MDA-MB-231 breast cancer cell line and again no EGFR could be detected (Fig. 1B), thus providing further evidence that OPCML does not interact with EGFR.

To further confirm these results, we carried out mammalian 2-hybrid studies. The ECDs of HER2 and EGFR were cloned to determine their ability to interact with OPCML (Supplementary Fig. S1) and transiently transfected in Cos-1 cells, which have a low basal expression of both HER2 and EGFR. The luminescence was normalized by co-transfection with a plasmid expressing the β-galactosidase gene and the empty pM and pVP16 vectors were used as negative controls. As presented in Fig. 1C, a positive signal was obtained only when cells were co-transfecte with OPCML and HER2, confirming again that OPCML interacts with HER2, but not with EGFR. Furthermore, we investigated which portion of the receptor was important for the binding to OPCML and deletion of domain 4 of the ECD of HER2 (HER2 Δ4, Supplementary Fig. S1) was shown to disrupt the interaction (Fig. 1D). As the membrane-proximal portion of domain 4 of EGFR and HER2 has been proposed to mediate part of the interaction between the two molecules (37), we then analyzed the ability of the two receptors to heterodimerize in the presence or absence of OPCML. In SKOV3 and SKBR3 cells, an ovarian and a breast cancer cell line, respectively, that express high levels of both HER2 and EGFR, HER2 was co-immunoprecipitated by an anti-EGFR antibody three times less in the presence of OPCML expression compared with OPCML non-expressing control cells (Fig. 1E). We then confirmed these results further and quantified the interaction by using a proximity ligation assay (PLA). The Duolink signal given by the close vicinity of EGFR to HER2 in non-stimulated cells was maintained in control cells stimulated with EGF for 30 minutes and dropped after one-hour stimulation, while in OPCML-expressing cells it was rapidly and strongly reduced by 30 minutes (Fig. 1F). This suggests that OPCML interaction with HER2 induces a reduction of the HER2/EGFR dimer through dimer formation interference.

As we observed previously, OPCML binding to HER2 downregulates the receptor and its downstream signaling in ovarian cancer cell lines (9). This led us to examine the potential therapeutic benefits of OPCML when combined with anti-HER2 agents. We first evaluated the impact of OPCML expression on the effect of the dual anti-HER2/EGFR small-molecule TKI lapatinib. Two ovarian cancer cell lines (SKOV3 and OV90) and a breast cancer cell line (SKBR3) with varying levels of EGFR and HER2 expression and with low or no HER3 and HER4 expression (Supplementary Fig. S2) were stably transduced to express OPCML (Supplementary Fig. S3) and treated with increasing amounts (0–100 μmol/L) of lapatinib for 48 hours before measuring their cell viability. In cells overexpressing HER2 (SKOV3 and SKBR3), we observed a significant decrease in cell viability when OPCML was present in the cells (control GI₅₀ = 34 μmol/L vs. OPCML GI₅₀ = 15.5 μmol/L for SKOV3, and control GI₅₀ = 13.6 μmol/L vs. OPCML GI₅₀ = 1 μmol/L in SKBR3), while in cells with little HER2 expression (OV90) the significance was lost even though a similar trend was maintained (Fig. 2A). We then analyzed the downstream signaling of these cell lines treated with serial dilutions of lapatinib for 3 hours followed by stimulation with EGF for 30 minutes. Lapatinib effects on the tyrosine phosphorylation levels of HER2 were very obvious at the lowest concentration of the drug (25 nmol/L) whereas phospho-EGFR levels were mostly inhibited at the highest concentrations (100–200 nmol/L; Fig. 2C and D), with the total levels of proteins being unaffected (Supplementary Fig. S4). These results confirm previous findings, suggesting that lapatinib mainly acts through HER2 rather than EGFR (38). Overall, OPCML-expressing cells had reduced amounts of the survival downstream effector phospho-AKT at lower concentrations of the drug compared with control cells in SKOV3 and SKBR3 and, at a lesser level, in OV90 cells (Fig. 2C and D), correlating with the GI₅₀ data. The effects of lapatinib on the mitogen-activated protein kinase (MAPK) pathway were also analyzed but only OPCML-expressing SKOV3 cells showed reduction of the phospho-ERK1/2 signal compared with control cells (Fig. 2C and D). The consistent downregulation of the phospho-AKT signal at lower doses of lapatinib mediated by OPCML in SKOV3 and SKBR3 cells was also reflected in activation of the apoptotic pathway in both cell lines. Caspase 3 and 7 activation after 24 hours of treatment with lapatinib, was in fact strongly increased and detected at lower concentrations of the drug in the OPCML-expressing cells compared with control (Fig. 2B). These results suggest that OPCML sensitizes cells to lapatinib mainly through pAKT inhibition- and caspase activation- pathway.

Because OPCML binds to HER2 making it less available for hetero-dimerization with EGFR (Fig. 1), a therapy that targets only EGFR could still benefit from the presence of OPCML, even though OPCML does not affect EGFR levels directly. We investigated our hypothesis by treating the cells with the anti-EGFR kinase inhibitor erlotinib for growth inhibition studies and found that, similarly to what seen for lapatinib, SKOV3 and SKBR3 cells expressing OPCML had a significant decrease in cell viability compared with control cells, while in OV90 cells this difference was not apparent (Fig. 3A). However, in OPCML-expressing cells the degree of sensitization to erlotinib (Control GI₅₀ = 32.5 μmol/L and 24.3 μmol/L vs. OPCML GI₅₀ = 24.5 μmol/L and 19.4 μmol/L for SKOV3 and SKBR3, respectively) was much less compared with the degree of sensitization to lapatinib, highlighting that OPCML confers a more efficient sensitization of anti-HER2 therapy rather than anti-EGFR. Western blot analysis of the downstream signaling of these cell lines treated with erlotinib for 3 hours and stimulated with EGF for 30 minutes confirmed the cell viability results, in which the phospho-HER2 and phospho-AKT levels were inhibited at lower doses of erlotinib in OPCML-expressing SKOV3 and SKBR3 but not in OV90 cells (Fig. 3C and D). Here, OPCML-mediated reduction of the phospho-ERK1/2 signal was found not only in SKOV3 cells but also, at a lower extent, in SKBR3 cells (Fig. 3C and D). The ability of the SKBR3 cells to respond to EGFR-targeted therapy despite lack of downregulation of the phospho-EGFR levels (Fig. 3D) has also been found by others (39) and has been suggested as an evidence for sensitivity of HER2-overexpressing tumors to EGFR-specific TKIs (39–41). Caspase 3/7 activation assays in SKOV3 and SKBR3 cells treated with erlotinib for 24 hours confirmed the GI50 data and OPCML-expressing cells showed higher levels of the activated enzymes compared with control cells (Fig. 3B) However, this was to a much lower extent compared with lapatinib treatment, suggesting again that the combined HER2-targeted therapy is more efficient in killing cancer cells.

These observations were further confirmed by genetic inhibition through siRNA of HER2, EGFR or both. When either HER2 or EGFR or both were silenced (Fig. 4A), the viability of SKOV3 cells decreased significantly compared with a non-targeting control (siNT) after 48 and 72 hours from transfection (Fig. 4B). However, OPCML-expressing cells showed a further decrease in cell viability at 72 hours compared with control cells when HER2, but not EGFR, protein levels were downregulated (Fig. 4B).

In view of the above findings, we evaluated the predictive value of OPCML protein expression in a series of known HER2⁺ recurrent breast cancer patients receiving capecitabine and lapatinib. We had previously demonstrated that OPCML mRNA expression is a significant good prognostic factor for recurrence free survival in a series of over 2,300 breast cancer patients (9). HER2⁺ (all cases 3+) breast cancer patients treated with capecitabine and lapatinib were selected (Supplementary Fig. S5). Following immunohistochemistry (IHC) staining for OPCML (Supplementary Fig. S6), this small series demonstrated a statistically significant correlation between OPCML protein expression and RECIST response to lapatinib/capecitabine (Fisher's Exact Test, two tailed P = 0.0286). All patients with 2+ and 3+ OPCML IHC expression responded, whereas none of the patients with 0 and 1+ OPCML IHC expression did (Supplementary Fig. S5).

We then looked at progression-free survival (PFS) of ovarian cancer patients with HER2 overexpression (top quartile) in the Tothill dataset (42) and found that patients had a significantly better survival when OPCML expression was above median (log-rank P = 0.004). In patients with low HER2 (bottom quartile) OPCML expression did not improve survival (Fig. 5).

Overall these data further confirm OPCML relevance in the regulation of HER2 signaling in a clinical setting.

OPCML is a protein that is frequently lost in tumors due to a combination of LOH and hypermethylation of the promoter of the gene (1, 2). We previously described how its tumor-suppressing function is achieved, showing that, by interacting with and downregulating a specific repertoire of RTKs, it causes a profound inhibition of the downstream ERK and AKT pathways (9). In particular, OPCML downregulates HER2 by changing the receptor localization on the cell membrane and inducing its internalization through a non–clathrin-mediated pathway (9). Here, we have confirmed by pull-down and mammalian 2-hybrid assays the interaction between OPCML and HER2 and demonstrated by co-IP experiments and Duolink assays that OPCML prevents the heterodimerization of HER2 with EGFR. We have also shown that the interference with the EGFR/HER2 heterodimer formation is mediated by the binding of OPCML to the domain 4 of the ECD of HER2, which is important for the stabilization of the heterodimer (37). Given these results, we have then investigated the ability of OPCML to sensitize cells to HER2- and EGFR-targeted therapies in HER2-overexpressing cancer cell models. By acting on HER2, OPCML affects the signaling from the EGF receptor family, as HER2 is the preferential partner of the three other members of the family (34, 35), as it is always in an open conformation (43, 44). We therefore used lapatinib (EGFR/HER2) and erlotinib (EGFR-only) RTK inhibitors in cancer cells bearing varying levels of expression of these two receptors (and no or little expression of HER3/4) in which we then re-expressed OPCML. Lapatinib acts as an ATP competitor for the kinase domain of both HER2 and EGFR causing the downregulation of the mitogenic and proliferative signaling driven by the activation of the two receptors (45). Despite the promising initial response of patients to this drug, relapse and resistance are still common events. It has been then suggested that the combination of HER2-targeting therapies can be more effective in the initial line of treatment as well as in relapsing cancers (18). In this study, we have shown how re-expression of OPCML in ovarian and breast cancer cell lines with high HER2 levels (SKOV3 and SKBR3) sensitizes these cells to lapatinib through caspase 3/7 activation. This confirms that by interfering with both HER2 dimerization (through OPCML interaction with the ECD) and kinase activation (through lapatinib action) cells die at significantly lower concentrations of the inhibitor. The analysis of survival pathways mediated by AKT confirmed that cells expressing OPCML had a significant reduction in phospho-AKT levels at lower doses of lapatinib compared with control cells in both SKOV3 and SKBR3 cells and to a much lesser extent in OV90 cells, which have low expression levels of HER2. When treating the same cell lines with the EGFR inhibitor erlotinib we obtained similar results, with cells overexpressing HER2 being sensitized to the drug through a higher activation of caspase 3/7 and the phospho-AKT signaling being strongly affected by OPCML expression in SKOV3 and SKBR3 cells and only partially in OV90 cells. However, overall the magnitude of sensitization to erlotinib was lower compared with lapatinib, suggesting that OPCML targeting of HER2 is indeed a more efficient approach to inhibit cancer cell growth.

These findings were also corroborated in a clinical setting, where breast cancer patients with HER2-positive tumors were found to be significantly more likely to respond to capecitabine/lapatinib therapy in the context of high tumoral OPCML protein expression levels compared with OPCML non/weakly expressing patients. Furthermore, stratification for HER2 and OPCML expression of ovarian cancer patients showed that higher co-expression of OPCML significantly increased progression-free survival of HER2-overexpressing patients.

Taken together these results show how, by acting through a key molecule such as HER2, OPCML amplifies the anticancer effect of lapatinib and to a lesser extent erlotinib and allows for a more effective targeting strategy of the EGFR-HER2 axis (Fig. 6).

Furthermore, these results support the notion of OPCML as a sensitizing agent in cancers with an activated HER2 status. Therefore, we hypothesize that selection by OPCML expression status might enrich for RTK inhibitor responders in HER2⁺ breast and ovarian cancer patients, although this requires further validation in the future. In addition, we previously demonstrated that a recombinant OPCML biotherapeutic was an active anticancer agent preclinically in-vivo (9) and therefore this biotherapeutic could be used in combination with HER2 targeting TKIs to enhance their response. Considering that OPCML also downregulates other RTKs, these data also open up the possibility that this general effect of OPCML may apply to other RTK targeting scenarios, where different RTKs might be driving cancer growth. By acting at several levels in the RTK crosstalk, OPCML could delay or even prevent the onset of resistance that often follows the use of RTK inhibitors due to compensatory mechanisms. OPCML could therefore prove valuable as a therapeutic not just in RTK addicted cancers but also in cancers that are not strictly addicted to a specific oncogene, allowing responses that would not otherwise occur.

H. Gabra is a VP head of Oncology Clinical Discovery (concurrent position with an Imperial tenured chair in medical oncology) at AstraZeneca and has ownership interest (including patents) in Patent to develop OPCML-based therapeutics. No potential conflicts of interest were disclosed by the other authors.

Conception and design: E. Zanini, L.S. Louis, E. Karali, I.S. Okon, A.B. McKie, S. Vaughan, J. Stebbing, C. Recchi, H. Gabra

Development of methodology: L.S. Louis, E. Karali, I.S. Okon, J. Stebbing, H. Gabra

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): E. Zanini, L.S. Louis, J. Antony, M. El-Bahrawy, H. Gabra

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): E. Zanini, L.S. Louis, J. Antony, E. Karali, M. El-Bahrawy, H. Gabra

Writing, review, and/or revision of the manuscript: E. Zanini, L.S. Louis, A.B. McKie, M. El-Bahrawy, J. Stebbing, C. Recchi, H. Gabra

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): E. Zanini, J. Antony, H. Gabra

Study supervision: E. Zanini, L.S. Louis, S. Vaughan, C. Recchi, H. Gabra

Other (Joint Fisrt Author with E Zanini): L.S. Louis

We wish to acknowledge the support of the NIHR Biomedical Research Centre, the Cancer Research UK Clinical Centre, and the Experimental Cancer Medicine Centre at Imperial College London.

E. Zanini, L.S. Louis, J. Antony, E. Karali, I.S. Okon, A.B. McKie, S. Vaughan, and C. Recchi were supported by the Ovarian Cancer Action Research Centre Core Funding (principal investigator: H. Gabra).

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.