To start assessing the role of PLEKHA7 in IBC, we determined its expression pattern by immunohistochemistry (IHC) in IBC patient samples. Archival surgical pathology material from 62 patients with a diagnosis of IBC was recovered from the Institutional Tissue Registry from Mayo Clinic Minnesota, Florida, and Arizona campuses and evaluated for adequacy. Sixteen samples were excluded from analysis either due to lack of appropriate tissue (e.g. small representation of neoplastic population) or due to IHC technical issues (e.g. loss of neoplastic population on IHC slides, or failure of IHC staining, as was shown by lack of staining in normal ducts that served as our internal control). Interpretation of hematoxylin and eosin (H&E) and IHC slides was performed by an independent pathologist. Two main morphological patterns were observed: solid and glandular. The predominant pattern of growth was solid, with sparse glandular formations (see Supplemental Fig. 1 for examples). Five tumors demonstrated only solid pattern of growth with no glandular formations. Tumors were divided into predominantly solid (75–100% of tumor exhibits solid pattern of growth), solid (25–74% of tumor exhibits solid pattern of growth), and sparsely solid (0–24% of tumor exhibits solid pattern of growth). The distribution of IBC samples based on the predominance of the solid pattern of growth is displayed in Fig. 1A. Tumors were also categorized based on the percentage of glandular pattern of growth into five categories: 0%, 1–5%, 6–25%, 26–50%, and 51–100%.The distribution of IBC samples based on the glandular pattern of growth is displayed in Fig. 1B.
PLEKHA7 expression was distinct between the solid and glandular patterns. For each morphological pattern, the average PLEKHA7 staining pattern across all tumor samples was calculated and categorized based on localization (apical, lost, cytoplasmic, or basal). In solid areas, PLEKHA7 was lost in 56.4%, while cytoplasmic pattern was observed in 26.4% and localized to the basal membrane in 24.1% across all IBC tumor samples (See Fig. 1C). Apical staining was not observed in solid areas of the tumor. In contrast, PLEKHA7 staining in glandular areas was either lost (68.6%) or apical (31.4%) (See Fig. 1D). Examples of PLEKHA7 staining for each location are shown in Fig. 1E-I. For complete breakdown and analysis of PLEKHA7 expression in IBC samples, see Supplemental Table 1. It is notable that PLEKHA7 must properly localize to the apical AJs to maintain its tumor suppressing function (19). Therefore, we anticipate that PLEKHA7 would not be functional in the overwhelming majority of these patient tumors.
To further interrogate the function of PLEKHA7 in IBC, we utilized two frequently used cell line models: SUM149 and SUM190. SUM149 cells belong to the triple negative basal molecular subtype, while SUM190 to the hormone receptor negative, erbB2/Her2 positive molecular subtype. Western blot experiments indicated that both SUM149 and SUM190 cell lines express very low levels of PLEKHA7 protein compared to Caco2 cells, an often utilized epithelial model for studying the AJs (see Fig. 1J). Collectively, these data suggest a consistent loss of functional PLEKHA7 in IBC, despite normal to high expression of p120 and E-cadherin (Fig. 1J and previous studies) (5, 17).
Next, we used viral transduction to examine the effects of PLEKHA7 re-expression in SUM149 cells. We found that exogenously expressed PLEKHA7 localizes to and strengthens the AJs, as evidenced by increased junctional accumulation of p120, E-cadherin, α-catenin, and β-catenin (see Fig. 2A-B). This junctional strengthening is similar to previous reports (19, 20, 24). Notably, PLEKHA7 re-expression altered the location, but not the overall levels of junctional proteins (see Fig. 2C), also consistent with previous publications (19, 20). In agreement, we observed decreased cytoplasmic localization of p120 and β-catenin in PLEKHA7-expressing cells (see Fig. 2B).
p120 regulates the activities of RhoGTPases, including RhoA, Rac1, and Cdc42 (25–27), as well as RhoGEFs and RhoGAPs (26, 28). The ability of p120 to regulate RhoGTPase signaling is thought to be regulated by p120’s junctional vs. cytoplasmic localization, and is essential for EGFR, HER2, Rac1, and Src-mediated induction of tumorigenesis (29–31). Depletion of p120 suppressed the growth and emboli formation of SUM149 cells in vitro, suggesting an important role for p120 in IBC (17). In this study, we did not explore further the possibility that 1) p120 promotes tumor growth by regulating RhoGTPase signaling in SUM149 cells, or that 2) restoring PLEKHA7 decreases the capacity of p120 to function in this manner.
As increased cytoplasmic β-catenin could lead to increased nuclear signaling, we tested for altered activity in the Wnt/β-catenin pathway using the dual luciferase reporter assay (Fig. 2D). Although TopFlash activity is reduced in PLEKHA7-SUM149 compared to control, we did not see consistent changes in activity when normalized by the FopFlash reporter (Fig. 2D). Therefore, while PLEKHA7 expression increases the junctional localization of β-catenin, it does not affect Wnt/β-catenin nuclear signaling under these conditions in SUM149 cells.
To test the hypothesis that PLEKHA7 loss in IBC promotes a more aggressive phenotype, we next examined whether restoring PLEKHA7 to the apical AJs suppresses cell growth. Under 2D culture conditions, PLEKHA7-expressing SUM149 cells exhibited reduced proliferative capacity, compared to SUM149 cells infected with control virus (Supplemental Fig. 2). Further, when PLEKHA7-expressing SUM149 cells were plated on Matrigel, they formed fewer and smaller colonies compared to control SUM149 cells (see Fig. 3A-B). IBC patients frequently demonstrate tumor emboli in the dermal lymphatics, and spheroid formation under ultra-low attachment conditions has been used as a model of IBC tumor emboli (7). SUM149 cells infected with control virus rapidly formed compact spheres when grown in suspension. In contrast, PLEKHA7-expressing SUM149 were more loosely connected and less compacted than control SUM149 cells (see Fig. 3C-D). Interestingly, the ability of IBC cells to form compact spheroids has been correlated directly to their tumorigenic potential (7). We also hypothesized that the less compacted spheres would be more vulnerable to chemotherapy treatment. To test this, we determined the sensitivity of control and PLEKHA7-expressing SUM149 spheres to doxorubicin, a standard neoadjuvant chemotherapy used in IBC treatment. Notably, after 72 hours of treatment, PLEKHA7-SUM149 spheres were significantly less viable than control-SUM149 spheres in response to doxorubicin treatment, particularly at the highest doses (10uM) (see Fig. 3E).
Our in vitro and IHC data argued that PLEKHA7 acts as a tumor suppressor and is frequently misregulated in IBC. Next we tested whether restoring PLEKHA7 expression in SUM149 cells would decrease tumor formation or growth in an animal model. SUM149 cells reliably form tumors in xenograft models when injected orthotopically (17, 32). PLEKHA7-expressing SUM149 cells or control SUM149 cells were injected into the 4th mammary fat pad of NOD/SCID mice and mice were monitored for 8 weeks for the presence and size of tumors formed. Mouse body weight changes were not observed in either group. After 8-weeks, mice were sacrificed and tumors were obtained for IHC. As shown in Fig. 4A and 4B, tumors in the PLEKHA7-expressing SUM149 mice were smaller and less proliferative than control. Importantly, IHC analysis revealed that PLEKHA7 expression was commonly misregulated in the PLEKHA7-expressing tumors. After 8 weeks, we found that most of the PLEKHA7-tumors had lost significant expression of PLEKHA7, retaining between 15–55% PLEKHA7 depending on the mouse. Furthermore, we frequently observed cytoplasmic PLEKHA7 staining, with only approximately 10–25% junctional PLEKHA7 remaining in most tumors. An example is shown in Fig. 4C. This indicates a negative selection of PLEKHA7-expressing tumor cells. Accordingly, when we quantified tumors for changes in Snail, Myc, and Cyclin D1, proteins that have been previously shown to be suppressed by PLEKHA7 function in Caco2 cells (19), we observed a trend towards reduced expression in PLEKHA7-SUM149 tumors that did not reach significance (see Supplemental Fig. 4A-C). We hypothesize that the tumor suppressive effects observed with PLEKHA7 expression in SUM149 xenografts occurred early in tumor formation. This early-effect hampered tumor growth sufficiently enough to observe overall differences in tumor between groups. However, PLEKHA7-positive tumors escaped these suppressive effects by deregulating ectopically expressed PLEKHA7 throughout the 8-week course of the experiment.