SH3BGRL confers innate drug resistance in breast cancer by stabilizing HER2 activation on cell membrane

HEK293T, MCF-7 and MDA-MB-453 cells were purchased from the Cell Bank of Chinese Academy of Sciences (Shanghai, China). HEK293T and MCF-7 cells were maintained in DMEM HIGH GLUCOSE (Hyclone, Los Angeles, USA) supplemented with 10% fetal bovine serum (Hyclone, Los Angeles, USA) and 1% Penicillin-Streptomycin solution (Hyclone, Los Angeles, USA). MDA-MB-453 cells were maintained in RPMI 1640 (Hyclone, Los Angeles, USA) supplemented with 10% fetal bovine serum and 1% Penicillin-Streptomycin solution. All cells were incubated at 37 °C in a humid incubator containing 5% CO₂. The medium was changed at alternate days, and cells were split at ration 1:3 when reached 90% confluence. SH3BGRL specific monoclonal antibody was purchased from Santa Cruz. Antibodies, including HER2, p-HER2 (Y1196), p-HER2 (Y877), p-HER2 (Y1221/1222), Akt, p-Akt, ERK, p-ERK, Ki-67, GAPDH, γ-Tubulin and Na+/K + -ATPase were purchased from Cell Signaling Technology (Danvers, USA).

Twenty three pairs of breast cancer tissues and the corresponding adjacent counterparts were collected with the informed consents, according to Sun Yat-Sen University health regulation, and the study was further approved by the Research Ethics Committee. Tissue microarrays including 76 HER2-positive breast cancer tissues were purchased from Alenabio Company (Xi’an, China) under the ethics regulation.

The EGFP-SH3BGRL, sh-SH3BGRL and HER2-YFP plasmids were constructed as previously described [16]. To construct HA-tagged SH3BGRL mutants, the mutagenesis PCR was performed with the listed pairs of primers (Supplemental Table S1). Cell transfection was performed with Lipofectamine 2000 (Invitrogen, Waltham, USA) in accordance with the manufacturer’s instructions. SH3BGRL overexpression in MCF-7 cells was achieved with EGFP-SH3BGRL plasmids, and the empty vector was used as control. SH3BGRL knockdown in MDA-MB-453 cells was accomplished with two SureSlencing TM shRNA plasmids mixture for human SH3BGRL (TG309466; Origene, Beijing, China) with GFP label, and the scramble shRNA as control [16]. After transfection, cells were selected with 400 μg/ml G418 (Sigma, St. Louis, USA) (for SH3BGRL overexpression) or 1 μg/ml Puromycin (Sigma, St. Louis, USA) (for SH3BGRL knock-down) for 3–4 weeks, and the single stable cell clones were picked under fluorescence Microscope (Nikon, Tokyo, Japan) to make the stable cell pools.

After the indicated treatments, cells were lysed in RIPA buffer (50 mM Tris (pH 7.4), 150 mM NaCl, 1% Triton X-100, 1% sodium deoxycholate, 0.1% SDS). Equal amount of protein was loaded and simultaneously subjected to electrophoresis in SDS-polyacrylamide gel and transferred to 0.22 μm pore-sized PVDF membranes (Roche, Basel, Switzerland). Membranes were briefly blocked with 5% skim milk and incubated with the primary antibodies overnight at 4 °C, followed by incubation with the species-matched secondary antibody conjugated with HRP (Cell Signaling Technology, Danvers, USA) for 1 h at room temperature prior to chemiluminescence detection.

Cells were seeded in 96-well plates at a density of 1000 cells per well. The growth rate of cells was evaluated by using the CCK-8 cell proliferation kit (Dojindo Laboratories, Kumamoto, Japan), according to the manufacturers’ instructions. Cell-cycle analysis was carried out by flow cytometry (Beckman CytoFLEX, California, USA) after propidium iodide staining.

For analysis the role of SH3BGRL in anticancer drug-induced apoptosis, parental MCF-7 and SH3BGRL-overexpressing MCF-7 cells were treated with 10 μM Cisplatin (TOCRIS, Abingdon, OX, UK) for 12 h or with 150 ng Herceptin (Roche, Basel, Switzerland) for 96 h, respectively. Likely, MDA-MB-453 and MDA-MB-453 SH3BGRL knockdown cells were treated with 10 μM Cisplatin for 12 h, with 150 ng Herceptin for 96 h, PI3K/AKT inhibitor LY294002 (CST, Danvers, MA,USA) for 24 h, or Herceptin and ATK inhibitor combination for 48 h, respectively. Cell apoptosis was analyzed using the Annexin V/7-AAD Apoptosis Detection kit (Keygen Biotech, Nanjing, China), and the percentage of apoptotic cells was analyzed by flow cytometer (Beckman CytoFLEX, California, USA). The detailed procedures of all above assays were described as previously [16].

Cells were grown at low density on coverslips overnight, washed with PBS three times and fixed with 4% paraformaldehyde for 15 min. After permeabilization with 0.5% Triton X-100 for 20 min, cells were blocked with 5% BSA for 30 min at room temperature. Then the cells were incubated with the specific primary antibody overnight at 4 °C, followed by incubation with secondary antibody: goat anti-mouse Alexa Fluor 488 and anti-rabbit Alexa Fluor 594 (Invitrogen, Waltham, USA) for 1 h at room temperature in the dark. Finally, the samples were mounted with Anti-fade reagent with DAPI (Invitrogen, Waltham, USA). Imaging was processed with confocal fluorescence microscope (Nikon, Tokyo, Japan).

For co-immunoprecipitation, HEK293T cells were co-transfected with each SH3BGRL mutant along with YFP-HER2 plasmids for 48 h. The transfected cells were harvested and lysed with lysis buffer (20 mM Tris (pH 7.4), 150 mM NaCl, 1% NP-40, 20 mM EDTA). For endogenous interaction of SH3BGRL with HER2, MDA-MB-453 cell lysates were directly used for co-immunoprecipitation. All mentioned cell lysates were incubated with SH3BGRL, HER2 specific antibody or normal mouse IgG overnight at 4 °C respectively, followed by mixing with protein A or protein G Magnetic beads (ThermoFisher, Waltham, USA) for another 3 h on a rotator. All beads were washed with lysis buffer for 6 × 5 min, and the precipitated protein complexes were eluted with loading buffer for western blotting analyses with proper antibodies.

After washing with 1 ml of ice-cold PBS, cells were harvested and lysed with the membrane protein extraction reagent A (Plasma Membrane Protein Extraction Kit, Beyotime, China). After homogenization, the mixture was centrifuged at 700 g for 10 min at 4 °C and the supernatant was further centrifuged at 10,000 g for 30 min at 4 °C. The sediment contains the total membrane protein while the supernatant contains cytosol protein. Further, the pellet was mixed with membrane protein extraction reagent B for 20 min and centrifuged at 10,000 g for 30 min at 4 °C to collect the supernatant.

Protein tertiary structures of SH3BGRL(PDB:1U6T) and ERBB2 (PDB:1MFG) were downloaded from the protein structure database, and ZDOCK 3.0.2 was used to perform online molecular docking (http://​zdock.​umassmed.​edu/​). Many possible docking modes of SH3BGRL and HER2 were obtained. Based on the ZDOCKScore ranking, the docking mode with the highest score was chosen as the interaction simulation between SH3BGRL and HER2, while the energy stability ranking was also considered. The Pymol software was used to analyze the possible binding sites of amino acids in the two proteins.

BALB/c-nude mice (female, 4–5 weeks of age, 18-20 g in body weight) were purchased from Beijing Vital River Laboratory Animal Technology Co., Ltd. (China). All experimental procedures were approved by the Institutional Animal Care and Use Committee of Sun Yat-sen University. Briefly, cells were trypisinized, washed and resuspended in phosphate buffered saline at a density of 10⁷ cells/ml. Seven mice were inoculated subcutaneously with 5 × 10⁶ cells of MD-MBA-453 Vector cells into the right dorsal flanks, and MD-MBA-453 SH3BGRL knockdown cells into the left dorsal flanks, respectively. Similarly, another three mice were inoculated subcutaneously with 5 × 10⁶ MCF-7 parental (Vector) and MCF-7 SH3BGRL cells, respectively. For the growth of the MCF-7 derived tumor, 50 μg estrodiol cypionate (MedChemexpress, NJ, USA) was injected in each mouse once 7 days to sustain the stable estrogen level, till the end of experiment. Mice were sacrificed 30 days after tumor implantation, and the tumors were removed and weighed. For drug treatment, 5 × 10⁶ cells of MD-MBA-453 parental cells (Vector) into the right dorsal flanks, and MD-MBA-453 SH3BGRL knockdown (KD) cells into the left dorsal flanks, respectively. Lapatinib, Herceptin or LY294002 were dosed as 40 mg/kg, 20 mg/kg or 50 mg/kg via i.p. injection twice a week. PBS was used as a negative control. After injection for one week, volumes of the formed tumors were measured once a week. Mice were sacrificed in the fourth week after tumor cell implantation, and the tumors were removed and weighed. GraphPad Prism 5 and paired t test were used for statistical analysis.

All xenografted and patient tissues were fixed with 4% paraformaldehyde overnight at 4 °C, and then embedded in paraffin. The samples were subsequently sectioned into thin slices and mounted on slides, followed by deparaffinization in xylene and rehydration through a series of ethanol-water solutions. Antigen retrieval was carried out by immersing the sections in citrate acid buffer with heating with Microwave oven. Slides were then blocked with 3% hydrogen peroxide to block nonspecific activity. After rinsing, slides were blocked with 5% BSA, and then incubated with SH3BGRL antibody or p-HER2 (Y1196) antibody overnight at 4 °C. Immunohistochemical staining kit (BOSTER Biological Technology, Wuhan, China) was used for color development. Images were captured and confirmed by professional pathologist under microscope (Nikon, Tokyo, Japan). The color intensity of slides was divided into four grades (points) to score SH3BGRL and p-HER2 (Y1196) expression level for statistical analysis.

Previous results indicate that human SH3BGRL functions as a tumor suppressor [16, 17], and only the mutated ones promote tumor metastasis and progression. However, accumulating evidence shows that SH3BGRL is highly upregulated in breast tumors [15, 18, 19], implying its possibly oncogenic role in breast cancer progression. However, the exact function of SH3BGRL in breast cancer remains elusive. To investigate whether SH3BGRL is related to breast cancer progression, we analyzed the gene dataset, GSE15852 on breast cancer and observed that the SH3BGRL expression in Invasive Ductal Carcinoma (IDC) is obviously higher than in adjacent normal tissue (P = 0.034) (Fig. 1a). Similar analysis of another gene expression datasets, GSE45827 on breast cancer also manifested that SH3BGRL is more highly expressed in various types of breast cancers than that in normal tissues, suggestive of an oncogenic role of SH3BGRL in breast cancers (Fig. 1b). To examine if there is a direct interaction between SH3BGRL and HER2, as HER2 was predicted to interact with SH3BGRL, we firstly co-transfected HEK293T cells with GFP-SH3BGRL and YFP-HER2 plasmids. Immunoprecipitation with SH3BGRL-specific antibody showed that YFP-HER2 was detected in the precipitated complex compared with the control IgG (Fig. 1c), indicating the exogenous interaction of SH3BGRL and HER2. To confirm the intrinsic interaction of SH3BGRL with HER2, we performed the reciprocal immunoprecipitations with anti-SH3BGRL antibody as well as the anti-HER2 antibody in MDA-MB-453 breast cancer cells, respectively. Immunoblotting results clearly demonstrated that the endogenous HER2 was precipitated by the endogenous SH3BGRL; inversely, SH3BGRL was detected in the precipitated complex with HER2-specific antibody, compared to those with the normal control IgG (Fig. 1d). Structurally, SH3BGRL contains two important binding domains, SH3 and EVH1 motifs, which endow SH3BGRL as a scaffold protein in signaling transduction. The tertiary structure of SH3BGRL comprises 6 α helixes and 4 β sheets [21]. In order to characterize which specific binding motif in SH3BGRL bound to HER2, we deleted all 6 α helixes and β3 sheet (core region of SH3B domain) individually to construct 7 SH3BGRL mutant plasmids (Fig. 1e). After co-transfection of HEK293T cells with YFP-HER2 construct, co-immunoprecipitation assays demonstrated that deletion of either α2 helix or β3 sheet in SH3BGRL evidently abolished the interaction between HER2 and SH3BGRL, and deletion of the α1 helix attenuated the interaction to some extent (Fig. 1f), indicating that besides the recognized SH3 domain (β3 sheet), the other two α1, 2 helix are also critical for the interaction of SH3BGRL with HER2.

Moreover, immunofluorescence also manifested the co-localization of HER2 with some endogenous SH3BGRL on cell membrane of MDA-MB-453 cells (Fig. 1g). Our results at least revealed the direct interaction between SH3BGRL and HER2 via α2 helix and β3 sheet.

To address the potential pathological consequence and the impact of the interaction between SH3BGRL and HER2 on HER2-mediated signaling in breast cancer, we established SH3BGRL-overexpressing MCF-7 cells and SH3BGRL-depleted MDA-MB-453 cells (Fig. S1) and performed immunofluorescence staining to investigate the HER2 localization status in MDA-MB-453 cells. With EGF stimulation for 5 min, HER2 was mainly stained on cell membranes in both parental control and MDA-MB-453 SH3BGRL knockdown cells. However, with EGF stimulation for 30 min, HER2 endocytosis was significantly accelerated in SH3BGRL knockdown MDA-MB-453 cells, in contrast, the majority of HER2 still located on cell membrane in parental MDA-MB-453 cells (Fig. 2a). Subcellular fractionization analysis further consolidated the observations of the respective HER2 localizations (Fig. 2b). As HER2 is putatively believed to undergo endocytosis once activated, we used an endosome marker, caveolin-1, to investigate if SH3BGRL depletion affects HER2 location and stability on cell membrane. Staining results revealed that in SH3BGRL-silencing MDA-MB-453 cells, HER2 dispersed into cell cytoplasm along with caveolin-1, whereas in the SH3BGRL highly expressing parental cells, HER2 was still located on the cell membranes and had less endocytosis upon EGF stimulation for 30 min (Fig. 2c). Even without EGF stimulation, HER2 still underwent endocytosis in the SH3BGRL-silencing cells (Supplemental Figure S2). Results here together indicated that the association of SH3BGRL with HER2 sustains the residence of HER2 on cell membrane, even as dimers after growth factor stimulation. Therefore, binding of SH3BGRL to HER2 could block HER2 internalization and prolong HER2 activation on cell membrane.

As an orphan receptor, HER2 undergoes dimerization with other EGFR family members to initiate varied downstream intracellular signals after growth factor stimulation [4]. To understand the binding consequence of SH3BGRL to HER2, we first determined the dynamic phosphorylation state of HER2. Results showed that with SH3BGRL expression level change, HER2 expression did not present any evident alteration in both MCF-7 and MDA-MB-453 cells, although MCF-7 cell is considered as a HER2-negative cell, but the basal HER2 expression was detected (Fig. 3a). We then treated all those cells starved for 24 h with EGF in a time course manner of 0, 5, 15 and 30 min to characterize the dynamic phosphorylation situation of HER2 along with SH3BGRL expression change. Immunoblotting results showed that SH3BGRL overexpression manifestly prolonged HER2 phosphorylations at Y1196 and Y877 sites in MCF-7 cells (Fig. 3b, left panel). In both the endogenous HER2 and SH3BGRL positive MDA-MB-453 cells, prior to EGF stimulation, HER2 already showed the Y1196 phosphorylation, which further increased by EGF stimulation for long time; in contrast, when knocking down the endogenous SH3BGRL with specific shRNAs, the endogenous HER2 was not phosphorylated at this site, but could be re-phosphorylated with EGF stimulation for 15 min and then be inactivated even with EGF stimulation for 30 min (Fig. 3b, right panel). Meanwhile, HER2 phosphorylation at Y1221/1222 sites was not impaired with SH3BGRL expression level change in both MDA-MB-453 and MCF-7 cells (Fig. 3b).

We further verified the possible influence of SH3BGRL mutation on HER2 phosphorylation through transient co-expression of HA-SH3BGRL and GFP-HER2 in HEK 293 T cells. Consistent with the binding capacity of SH3BGRL mutants with HER2 above, SH3BGRL overexpression could promote HER2 activation, regardless of EGF stimulation. But destroying the SH3 domain in SH3BGRL by deleting the helixes α1, α2, or β3 clearly abolished the wild-type SH3BGRL function on HER2 activation (Fig. 3c). Further molecule docking of SH3BGRL with C-terminal of HER2 revealed that helixes α1, α2, or β3 in SH3BGRL were critical for their interaction, and this interaction may sustain a particular HER2 conformation (Fig. 3d), leading to the tyrosine site-specific HER2 phosphorylation, for instance, at tyrosine 1187.

HER2 activation is closely with the downstream signaling activation, typically the MAPK and PI3K/AKT pathways. We then evaluated the activation status of PI3K/AKT and MAPK pathways in the correspondingly SH3BGRL-modulated HER2 activation. As expected, SH3BGRL overexpression prolonged activations of AKT and ERK in MCF-7 cells in comparison with the parental MCF-7 cells; however, SH3BGRL knockdown compromised the activation periods of AKT and ERK in MDA-MB-453 cells (Fig. 4a). Taken together, these results indicated that once binding to HER2, SH3BGRL could initiate HER2 activation, prolong the activation duration and the functioning time of the downstream signaling.

To seek whether SH3BGRL involving in the HER2-stimulated cell growth and survival, cell proliferation, cell cycle progression and apoptosis were examined in both MCF-7 and MDA-MB-453 cells. Results showed that SH3BGRL ectopic expression significantly promoted MCF-7 cell proliferation. On the other side, SH3BGRL knockdown suppressed cell proliferation in MDA-MB-453 cells (Fig. 4b). Likely, SH3BGRL overexpression in MCF-7 cells notably promoted cell cycle progression, manifesting the reduced cell population in G0/G1 phase and the increased in G2/M phase; while SH3BGRL silence in MDA-MB-453 cells conversely arrested cell cycle progression, showing the higher cell population at G0/G1 phase and the less in G2/M phase (Fig. 4c, Supplemental Figure S3). Thus, SH3BGRL was verified to promote breast tumor cell cycle progression. To further investigate the physiological effect of SH3BGRL in tumorigenesis, we subcutaneously inoculated SH3BGRL-overexpressing MCF-7 cells with their parental cells into immune-deficient mice. In vivo results showed that SH3BGRL overexpression could efficiently promote tumorigenesis (Fig. 4d). In contrast, when knocking down the endogenous SH3BGRL in MDA-MB-453 cells, the tumorigenesis was significantly repressed, compared to the control cells (Fig. 4e), verifying the oncogenic role of SH3BGRL in breast tumor progression.

Considering that the HER2-targeted therapy is not universally effective on HER2-possitive breast cancers, we further investigated if SH3BGRL overexpression can interfere with the anti-tumor drug sensitivity in breast cancer cells. We first tentatively determined the impact of SH3BGRL on cisplatin sensitivity. Flow cytometry analysis showed that SH3BGRL overexpression significantly compromised the cisplatin-induced cell apoptosis in MCF-7 cells, while SH3BGRL knockdown sufficiently enhanced cell apoptosis in MDA-MB-453 cells (Fig. 5a; Supplemental Figure S4, Cisplastin), indicating the dual effects of SH3BGRL in breast tumor cell proliferation and anti-cancer drug resistance.

Given the persistent activation of HER2 by SH3BGRL, we sought to figure out whether SH3BGRL can orchestrate the HER2-targeted therapy. We then treated the putative HER2-negative MCF-7 cells with Herceptin. Results manifested that Herceptin did not induce so obvious cell apoptosis in both parental MCF-7 and SH3BGRL-overexpressing MCF-7 cells, compared to that by Cisplatin (Fig. 5b; Supplemental Figure S4, Herceptin). This result hints that even the marginal HER2 is activated with SH3BGRL on cell membrane, the HER2-trageted therapy is still ineffective, implying that Herceptin-based on therapy is mainly effective to the tumors with quite high HER2 expression. To further confirm this observation, we treated MDA-MB-453 cells with both HER2 and SH3BGRL in high level and noted the evident resistance of these cells to Herceptin treatment. In contrast, with SH3BGRL knocked down, those cells became to be the HER2-high plus SH3BGRL-negative/or low MDA-MB-453 cells, and Herceptin in turn promoted these cells to undergo apoptosis as cisplatin (Fig. 5b).

To further check the effect of SH3BGRL to HER2 kinase-based therapy, we treated MDA-MB-453 and MCF-7 cells with a HER2 kinase inhibitor, Lapatinib. Results indicated that endogenous SH3BGRL knockdown in MDA-MB-453 cells sensitized cell sensitivity to Lapatinib, while SH3BGRL overexpression in MCF-7 cells significantly enhanced cell resistance to Lapatinib (Fig. 5c), indicating that SH3BGRL enhances the HER2-positive breast cancer cells resistance to Cisplatin and the HER2-targeted therapies.

As the interaction between SH3BGRL and HER2 activated the downstream cell survival signaling, we sought to know if blocking this signaling transduction pathway, whether can neutralize the effect of SH3BGRL on drug resistance. We thus treated MDA-MB-453 cells with a PI3K inhibitor, LY294002 and found a generally slight increase of cell apoptosis, which is not dependent on SH3BGRL level in MDA-MB-453 cells (Fig. 5d). As PI3K-AKT signaling plays universal roles in cell survival, inhibition this signaling should naturally renders cell death to some extent.

We next approached to the combinational drug treatments of MDA-MB-453 wild-type cells and the SH3BGRL-silencing cells and found that the combination of Herceptin+AKT inhibitor did not presented the synergetic effect on SH3BGRL and HER2 doubly positive MDA-MB-453 cells, compared to AKT inhibitor alone (Fig. 5d-Vector), whereas SH3BGRL KD sensitized those cells to the combined treatment (Fig. 5d-Vector). In the combination of Cisplastin and AKT inhibitor, as both components are not specific to HER2 inhibition, there was no difference between SH3BGRL-expressing and SH3BGRL-delpleting cells to the treatment, but presenting notable cytotoxicity, compared to the DMSO-treated control groups. Moreover, combination of Cisplatin+AKT inhibitor showed more apoptotic effect than Cisplatin+Herceptin on the wild-type MDA-MB-453 cells (Fig. 5d), suggesting the involvement of the SH3BGRL/HER2-mediated AKT activation to drug resistance. Overall, results above confirmed that SH3BGRL endows HER2-positive cells to be resistant to Cisplastin and HER2-specific drugs.

To validate the individual or combination therapy potential of the above assayed drugs, we first inoculated the parental MDA-MB-453 cells and the SH3BGRL knockdown cells into immunodeficient nude mice till the visible tumors formed. Therapy results showed that HER2-trageted Herceptin and Lapatinib did not show any inhibitive role in SH3BGRL and HER2 doubly positive tumors, but AKT inhibitors LY294002 showed the evident inhibition to this doubly positive type tumor progression. Therefore, combinations with AKT inhibitor with Cisplatin or Herceptin exhibited similar inhibitory effects to AKT inhibitor alone. As expected, SH3BGRL knockdown dramatically increased the cell sensitivity to single Lapatinib or Herceptin, leading to the drastic tumor growth inhibition (Fig. 5e, f; Supplemental Figure 5). Furthermore, we examined the tumor cell proliferation and apoptosis states in the various treated tumors with cell proliferation marker Ki67 and the apoptotic TUNEL stainings (Supplemental Figure 6). Staining results demonstrated the matchable trends of tumor cell proliferation and cell death to the corresponding tumor growths and tumor weights upon the treatments. Therefore, targeting SH3BGRL would be a promising strategy to treat the SH3BGRL and HER2 doubly positive breast cancers and release HER2-targeted therapy resistance. Alternatively, targeting SH3BGRL/HER2 downstream signals might be another practical strategy to SH3BGRL high breast cancers, regardless of HER2 expression level.

To verify the HER2 and downstream AKT activation events in the SH3BGRL-positive and SH3BGRL-low xenografted tumors with or without the abovementioned drug treatments, we performed the immunochemistry staining of the sections of xenografted tumors. Staining results confirmed that SH3BGRL knockdown or the effective therapies efficiently repressed HER2 phosphorylation at both Y877 and Y1196 sites, as well as the downstream AKT activation (Supplementary Figure S7), validating the authenticity of SH3BGRL/HER2 tumorigenic mechanism.

To further validate the relevance of SH3BGRL on breast cancer progression in patients, we collected 23 pairs of fresh clinical breast cancer samples and performed the immunohistochemistry analysis again, in which the protein staining intensity was divided into four grades by double-blind scoring (Supplemental Figure S8A). Quantitative expression analysis showed that SH3BGRL and p-HER2 (Y1196) expression were both significantly higher in tumor tissues than those in the adjacent normal tissues (Fig. 6a). Statistically, there was a clear correlation (r = 0.7524, p < 0.0001) between SH3BGRL and p-HER2 in paired tissues (Fig. 6b). We then utilized a breast tissue array with 72 HER2-positive samples to analyze SH3BGRL expression level and showed that SH3BGRL is highly expressed in up to 60% of HER2-positive patients, although with the limited case amount (Supplementary Figure 8B,C). In addition, patients with strongest SH3BGRL expression also overlapped with the highest HER2 level cases (Fig. 6c).

To scale up the case number, we analyzed 1075 breast cancer patients data in THE HUMAN PROTEIN ATLAS database. The relationships between SH3BGRL expression and various clinical features were summarized in Table 1). All data were divided into two groups by SH3BGRL expression level cutoff (232.4 FPKM), and the plotted Kaplan-Meier survival curve demonstrated that high SH3BGRL expression is strictly associated with the poor survival of patients (Fig. 6d, p < 0.0032). These patients were further subdivided into HER2-high and -low groups, with a 10-year survival analysis, we found high SH3BGRL expression was still a poor prognostic marker in HER2-high subgroup (Fig. 6e), but no such information to the HER2-low subgroup (Supplemental Figure 8D). Furthermore, data from TCGA Pan-Cancer Clinical Data Resource showed that high SH3BGRL expression was related to poor progression-free interval (PFI) (Fig. 6f). These results together manifested that SH3BGRL plays critical oncogenic function in HER2-high breast cancers, and SH3BGRL expression status is a potent biomarker to subgroup HER2-positive breast cancers and the suitable therapy strategy.