FAM83A confers EGFR-TKI resistance in breast cancer cells and in mice

Abstract

Breast cancers commonly become resistant to EGFR-tyrosine kinase inhibitors (EGFR-TKIs); however, the mechanisms of this resistance remain largely unknown. We hypothesized that resistance may originate, at least in part, from molecular alterations that activate signaling downstream of EGFR. Using a screen to measure reversion of malignant cells into phenotypically nonmalignant cells in 3D gels, we identified FAM83A as a candidate cancer-associated gene capable of conferring resistance to EGFR-TKIs. FAM83A overexpression in cancer cells increased proliferation and invasion and imparted EGFR-TKI resistance both in cultured cells and in animals. Tumor cells that survived EGFR-TKI treatment in vivo had upregulated FAM83A levels. Additionally, FAM83A overexpression dramatically increased the number and size of transformed foci in cultured cells and anchorage-independent growth in soft agar. Conversely, FAM83A depletion in cancer cells caused reversion of the malignant phenotype, delayed tumor growth in mice, and rendered cells more sensitive to EGFR-TKI. Analyses of published clinical data revealed a correlation between high FAM83A expression and breast cancer patients' poor prognosis. We found that FAM83A interacted with and caused phosphorylation of c-RAF and PI3K p85, upstream of MAPK and downstream of EGFR. These data provide an additional mechanism by which tumor cells can become EGFR-TKI resistant.

Figure 1. Discovery of FAM83A, a protein upregulated in breast cancer that confers resistance to EGFR-TKI AG1478.

(A) Process for the screen to identify FAM83A. Treatment of T4-2 cells with a small-molecule EGFR inhibitor, AG1478, induced reversion to a polarized acinar phenotype in 3D lrECM culture, as detected by α6 integrin staining (green) to delineate basal polarity (left); after transduction with the cDNA library, a subpopulation of cells disorganized in 3D culture after AG1478 treatment was isolated, expanded, and searched for the cDNA insert (right). (B) Protein domain structure of FAM83A. (C) Immunohistochemical analysis of FAM83A expression in normal (top) and malignant (bottom) breast tissue specimens. Whereas 0 of 16 normal cells were strongly positive for FAM83A staining, 45 of 48 malignant cells (94%) were positive. Scale bar: 50 μm. (D) FAM83A (83A) protein expression in a panel of breast cancer cell lines. (E and F) Effect of FAM83A overexpression in inhibiting AG1478-induced (AG) reversion of T4-2 cells (E) and ablating the polarized structure of S1 cells (F). Generation of T4-2 and S1 cell lines overexpressing FAM83A was confirmed by Western analyses. Scale bars: 50 μm.

Figure 2. FAM83A levels positively determine cell growth and tumorigenic potential.

(A) Generation of a T4-2 cell line stably depleted of FAM83A by shRNA (83Ash). (B) FAM83A shRNA expression reverted T4-2 cells to an acinar-like phenotype in 3D lrECM cultures. α6 integrin (red) was used to indicate basal polarity; blue, DAPI. (C) Colony size. Luciferase shRNA was used as control. (D) Number of T4-2 cells expressing either FAM83A shRNA or FAM83A-overexpressing construct versus control (vector) invaded through lrECM-coated Transwell filters after 48 hours (n = 4). (E) Generation of an MDA-MB468 cell line stably depleted of FAM83A by shRNA. (F) Control versus FAM83A-depleted MDA-MB468 cells grown in 3D lrECM cultures for 5 days. Cells were stained with ethidium bromide to indicate apoptotic cells (blue, Hoechst 33258). (G) Number of viable cells after growth in 3D lrECM for 5 days (n = 4). (H) Number of cells invaded through lrECM-coated Transwell filters after 24 hours (n = 3). (I) Growth of control versus FAM83A-depleted MDA-MB468 cells, measured by MTT assay for 9 days (n = 12). P < 0.05, 2-tailed ANOVA. (J) Clonogenic potential of control versus FAM83A-depleted MDA-MB468 cells measured by colony formation assay for 10 days. Shown are representative images of colonies formed. (K) Number of colonies formed (n = 6). (L) Tumorigenic foci formed by 3T3 fibroblasts overexpressing FAM83A through loss of contact inhibition of growth. Representative images of 3T3 cells transduced by control and FAM83A-overexpressing lentivirus and cultured for 2 weeks. (M) Quantification of 3T3 foci shown in L (n = 4). (N) Representative images of T4-2 cells in duplicates in 2 rows transduced by empty vector, FAM83A-overexpressing, or FAM83A shRNA lentivirus and cultured on soft agar for 6 weeks. (O) Number of colonies formed on soft agar (n = 6). *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001, Student’s t test. Scale bars: 50 μm.

Figure 3. FAM83A levels correlate directly with tumor growth rates and EGFR-TKI resistance in cultures and in vivo, as well as with patients’ prognosis.

(A) Growth of tumors derived from T4-2 cells treated with scrambled or FAM83A siRNA and xenografted in nude mice (n = 8). (B) Growth of tumors derived from MDA-MB468 cells treated with control (luciferase) or FAM83A shRNA and xenografted in nude mice (n = 8). (C and D) Tumor growth of vector control and FAM83A-overexpressing T4-2 cells treated with oral gavage of sham (vehicle; B) or 30 mg/kg lapatinib (LP; C). FAM83A-overexpressing T4-2 cells were resistant to lapatinib treatment (n = 8). (E) FAM83A staining of sham- or lapatinib-treated vector control tumors as in C and D. Note the upregulation of endogenous FAM83A in lapatinib-treated tumors. (F) Survival of vector control, FAM83A-overexpressing, and FAM83A-depleted T4-2 cells and luciferase-sh control or FAM83A-depleted MDA-MB468 cells treated with AG1478 at different concentrations and measured by clonogenic assay after 10 days. Values are expressed relative to untreated. IC₅₀ of AG1478 was as follows: T4-2 Ctrl, 10 nM; T4-2 FAM83A, 80 nM; T4-2 FAM83Ash, 6 nM; MDA-MB468 Ctrl, >1,000 nM; MDA-MB468 FAM83Ash, 7.5 nM. (G) Kaplan-Meier curve for a cohort of 159 breast cancer patient samples. High FAM83A expression correlated with poor prognosis (P < 0.05). *P < 0.05, **P < 0.01, ***P < 0.001, 2-way ANOVA with Bonferroni post-test.

Figure 4. FAM83A interacts with c-RAF and PI3K upstream of MEK1/2 activation.

(A) Control and FAM83A-overexpressing T4-2 cells were tested for response to AG1478 (300 nM), LY294002 (LY; 8 μM), or PD98059 (PD; 20 μM) in 3D lrECM cultures in comparison to vehicle (DMSO) treatment. α6 integrin staining (green) was used to indicate basal polarity; blue, DAPI. Scale bars: 50 μm. (B) Size of colonies formed by control versus FAM83A-overexpressing cells after drug treatment as in A (P < 0.05, Kruskal-Wallis test). (C) Detection of tyrosine-phosphorylated FAM83A after EGF stimulation (0, 30, or 60 minutes) by reciprocal IP with an antibody against FAM83A or phosphotyrosine (pY). Rabbit preimmune serum was used as control. (D) Interaction of endogenous FAM83A with c-RAF and PI3K p85 subunit after EGF stimulation (0, 30, or 60 minutes) by reciprocal co-IP with an antibody against FAM83A or c-RAF. Preimmune serum was used as control. (E) Levels of phosphorylated PI3K p85 subunit (p-Y458) and c-RAF (p-S338) in control, FAM83A-overexpressing, and FAM83Ash-expressing T4-2 cells after treatment with EGF (1 hour) or AG1478 (2 hours). (F) Levels of phosphorylated AKT (p-S473), MEK1/2 (p-S217/221), and ERK1/2 (p-T202/Y204) in control, FAM83A-overexpressing, and FAM83Ash-expressing T4-2 cells after growth in 3D lrECM culture for 5 days. (G) Levels of phosphorylated AKT, MEK1/2, and ERK1/2 in control, FAM83A-overexpressing, and FAM83Ash-expressing T4-2 cells after treatment with AG1478 for 2 hours. (H) FAM83A interacts with c-RAF and PI3K upstream of MEK activation.