Overexpression of β1 integrin contributes to polarity reversal and a poor prognosis of breast invasive micropapillary carcinoma

Invasive micropapillary carcinoma (IMPC) of the breast is a highly aggressive breast cancer. Polarity reversal exemplified by cluster growth is hypothesized to contribute to the invasiveness and metastasis of IMPC. In this study, we demonstrate that levels of β1 integrin and Rac1 expression were greater in breast IMPC than in invasive breast carcinoma of no specific type and paraneoplastic benign breast tissue. We show that silencing β1 integrin expression using the β1 integrin inhibitor AIIB2 partially restored polarity in IMPC primary cell clusters and downregulated Rac1. Thus, overexpression of β1 integrin upregulates Rac1. Univariate analysis showed that overexpression of β1 integrin and Rac1 was associated with breast cancer cell polarity reversal, lymph node metastasis, and poor disease-free survival in IMPC patients. Multivariate analysis revealed that polarity reversal was an independent predictor of poor disease-free survival. These findings indicate that overexpression of β1 integrin and the resultant upregulation of Rac1 contribute to polarity reversal and metastasis of breast IMPC, and that β1 integrin and Rac1 could be potential prognostic biomarkers and targets for treatment of breast IMPC.


INTRODUCTION
Invasive micropapillary carcinoma (IMPC) is a tumor with a propensity for metastasis that occurs in various organs [1][2][3]. It was first reported in the breast by Fisher et al. [4] in 1980, and the term was adopted by the World Health Organization (WHO) classification of breast tumors in 2003 [5]. We have previously shown that breast IMPC exhibits polarity reversal in cell clusters, which increases the risk of invasion and metastasis [6]. Polarity reversal can be identified by immunohistochemistry (IHC) for E-cadherin (E-cad), epithelial membrane antigen (EMA), mucin family protein-1 (MUC-1), and sialyl-Lewis X (SleX) [7][8][9][10]. Liu et al. [11] reported that presence of IMPC in breast mucinous carcinoma promotes tumor metastasis and that patients with mixed

Research Paper
IMPC/mucinous carcinoma have worse recurrence-free survival and overall survival (OS) than patients with pure mucinous carcinoma. We theorized that polarity reversal of tumor cell clusters contributes to invasion and metastasis of IMPC and thus to its poor prognosis [12].
Integrins are a family of transmembrane receptors. They are heterodimers composed of α and β subunits. β1 integrin is mainly expressed in normal cells. Lee et al. [13] found that β1 integrin helps maintain polarity of normal epithelial cells and assists in the formation of glandular lumen. Aberrant expression of β1 integrin in human breast carcinoma has been linked to cell adhesion, angiogenesis, tumor progression, and metastasis [14,15]. Overexpression of β1 integrin has been reported in several solid tumors [14,16], and inhibition of β1 integrin expression in breast cancer cell lines restores the polarity of tumor cells to a status similar to normal mammary epithelial cells [17,18]. Other studies showed that treatment of normal epithelial MDCK cells with β1 integrin inhibitor resulted in polarity disorder and malignant phenotype transformation [19,20]. Thus, balanced expression of β1 integrin is required to maintain normal polarity.
Rac is a member of the Rho family of small GTPases that is regulated by integrin and affects a variety of actin-dependent processes including cell-cell adhesion, cell migration, and cellular transformation [21]. Studies have demonstrated that Rac1 induces epithelial polarity in cells adhering to extracellular matrix [22] and contributes to cell migration, loss of adhesion, invasion, and metastasis of tumors [23]. However, its expression and regulatory relationship with β1 integrin in IMPC have not been reported.
Here, we examined β1 integrin and Rac1 expression and assessed their effects on polarity at the cytologic level. We then validated our results in breast cancer cell lines and primary breast cancer cells. We further correlated our findings with patients' clinical outcomes.

β1 integrin positively regulates Rac1 expression
We first evaluated the silencing effects of siRNA-β1 integrin and siRNA-Rac1 in MCF-10A normal breast epithelial cells. siRNA-ctrl was used as negative control. As shown in Figure 1A and 1B, β1 integrin mRNA was significantly decreased with siRNA-β1 integrin transfection. Rac1 mRNA was also significantly decreased with siRNA-Rac1 transfection. At the protein level, both β1 integrin expression and Rac1 expression were decreased ( Figure 1C). When cells were transfected with siRNA-Rac1, Rac1 expression was downregulated, but no significant decrease in β1 integrin expression was noted ( Figure 1D). The results indicate that Rac1 expression is positively regulated by β1 integrin. β1 integrin silencing leads to disordered polarity of MCF-10A cell clusters To assess polarity changes induced by β1 integrin, we planted MCF-10A cells in collagen gel for threedimensional (3D) culture and then silenced β1 integrin expression using siRNA. MUC-1, the marker of cell polarity, was detected by rhodamine-conjugated affinipure goat anti-rabbit IgG (red). Immunofluorescence analysis of the siRNA-ctrl control group showed that MUC-1 was expressed on the inner side of the cell clusters, indicating that MCF-10A cell clusters displayed normal polarity. As shown in Figure 1E, after treatment with siRNA-β1 integrin, MUC-1 staining was predominantly located on the stromafacing surface of the cell clusters, demonstrating polarity reversal. The cell clusters were transformed from having a hollow growth pattern to irregular clusters. When cells were treated with siRNA-Rac1, cell cluster polarity also became disordered ( Figure 1E). These observations indicate that loss of β1 integrin and/or Rac1 can lead to disordered cell polarity in 3D culture.

AIIB2 downregulates Rac1 in breast cancer cell lines and IMPC primary tumor cells
In 3D cell culture, we treated MCF-7 and MDA-MB-231 breast cancer cell lines, as well as primary tumor cells of IMPC and invasive ductal carcinoma of no specific type (IDC-NST), with AIIB2, a β1 integrin inhibitor used in previous studies [24][25][26]. After 72 hours of incubation with AIIB2, Rac1 protein levels decreased in both cancer cell lines ( Figure 2A) and primary tumor cells ( Figure 2B, 2C), indicating that AIIB2-induced inhibition of β1 integrin downregulates Rac1. These findings support the earlier conclusion that β1 integrin positively regulates Rac1 expression.

AIIB2 causes polarity change in IMPC cell clusters
To assess whether β1 integrin induces polarity changes, we planted IMPC primary tumor cells in collagen gel for 3D culture and then treated the cells with AIIB2. Rhodamine-conjugated affinipure goat anti-rabbit IgG was used to label MUC-1 (red), and rhodamine-conjugated affinipure goat anti-mouse IgG was used to label E-cad (green). Immunofluorescence analysis identified polarity reversal in IMPC cell clusters without AIIB2 treatment. After AIIB2 treatment, polarity was partially restored. Similar changes were observed in IDC-NST cells ( Figure 3). These observations suggest that β1 integrin overexpression causes tumor cell polarity reversal.
Our results indicate that polarity reversal is associated with increased LNM in IMPC.

Overexpression of β1 integrin correlates with poor prognosis of IMPC patients
To assess whether β1 integrin expression correlates with the prognosis of IMPC patients, DFS and OS analyses were performed in all patients. Kaplan-Meier survival curves showed that DFS was significantly shorter in patients with overexpression of β1 integrin or Rac1 or with polarity reversal than in patients with low β1 integrin or Rac1 expression or without polarity reversal (P = 0.017, P = 0.026, and P = 0.003, respectively) ( Figure 5A-5C). The DFS of patients with both β1 integrin and Rac1 overexpression was     significantly shorter than in patients with overexpression of only β1 integrin or Rac1 (P = 0.029 and P = 0.019 separately), whereas no significant difference in DFS was noted between patients with overexpression of both β1 integrin and Rac1 and patients with low expression of both of these markers ( Figure  5D). No significant OS differences were observed between patients with and without high expression of β1 integrin and/ or Rac1 or patients with and without polarity reversal (data not shown). Univariate Cox regression analysis revealed that increased tumor size, presence of LNM, overexpression of β1 integrin and/or Rac1, and polarity reversal are associated with worse DFS in patients with IMPC (P < 0.05) ( Table  8). However, multivariate Cox regression analysis only confirmed that increased tumor size (P = 0.010) and polarity reversal (P = 0.031) were independent predictors of poor DFS in IMPC patients (Table 8).

DISCUSSION
Pure breast IMPC is rare, constituting <2% of breast cancers; however, IMPC has a high rate of metastasis. Currently, there are no universally accepted criteria for   pathologic diagnosis of IMPC, particularly regarding the required proportion of micropapillary clusters in a nonhomogeneous tumor. In this study, we selected tumors consisting of at least 50% IMPC component. Based on previous studies, we hypothesized that tumor cell polarity reversal is a critical factor facilitating metastasis. The integrins constitute a family of polarity-related proteins, and their upregulation in breast cancer cells has been reported [27], although conflicting results have also been described [28]. We chose to study the expression of β1 integrin and Rac1 in breast cancer cell lines and IMPC and correlated our findings with the clinicopathologic features of patients. To our knowledge, this is the first study to investigate the role of β1 integrin and Rac1 expression in IMPC. We showed that when β1 integrin expression was blocked in the MCF-10A cell line by siRNA-β1 integrin, the expression of Rac1 was also significantly downregulated. In contrast, when Rac1 expression was blocked in the same cell line by siRNA-Rac1, no significant decrease in β1 integrin expression was found. The results confirm that, at least in the MCF-10A cell line, β1 integrin positively regulates Rac1 expression, but not the reverse, and indicate that Rac1 is in a downstream position to β1 integrin in the chain of molecular processes that maintain epithelial cell polarity.
Using 3D culture, we demonstrated that the polarity of MCF-10A clusters became disordered after treatment with siRNA-β1 integrin; the cell clusters were transformed from normal clusters with a hollow growth pattern to irregular cell clusters, as reported previously by others [19,20]. Weaver et al. [17] reported that treatment of breast cancer cells with inhibitory β1 integrin antibody leads to a striking morphologic and functional reversion to a normal phenotype. We confirmed these results in our study, in which treatment with the β1 integrin inhibitor AIIB2 partially abrogated the polarity reversal seen in IMPC clusters. These results indicate that upregulation of β1 integrin causes polarity reversal of IMPC cells, followed by The DFS of patients with high β1 integrin and Rac1 expression was shorter than that of patients with high expression of only one of these markers (both high vs β1 integrin high, P = 0.029; both high vs Rac1 high, P = 0.019; and both high vs both low, P = 0.15). www.impactjournals.com/oncotarget formation of tumor cell clusters as a result of upregulation of Rac1. This could explain how morula-like clusters of IMPC are formed and could support our previous findings that IMPC tumor cells divorce from primary tumor and then invade and metastasize through tumor cell clusters [12].
Our immunohistochemical analysis revealed that β1 integrin and Rac1 expression was higher in IMPC than in paraneoplastic benign breast tissue and IDC-NST. IMPC patients with high expression of both β1 integrin and Rac1 had a significantly higher incidence of LNM than patients with low expression of both markers or with high expression of only one marker. In patients with IMPC that showed weak expression of β1 integrin and/or Rac1, the incidence of LNM was significantly greater than in patients with no expression. A similar trend was observed in patients with IDC-NST. A study in patients with IDC-NST showed that increased Rac1 expression was associated with partial reversed cell polarity and LNM, which supports the idea that IDC-NST with partial reversed cell polarity may be part of the IMPC spectrum [8,29]. Univariate analysis showed that β1 integrin and Rac1 overexpression was associated with tumor cell polarity reversal, presence of LNM, and decreased DFS in IMPC patients. Multivariate analysis indicated that polarity reversal was an independent predictor for poor DFS in IMPC patients. These results suggest that upregulated β1 integrin may promote LNM of IMPC through formation of tumor clusters with polarity reversal. Perhaps this proposed model could be applied to other metastasis-prone tumors with IMPC growth patterns, including tumors of the lungs, ovaries, and colon [1][2][3].
Of note, patients with IMPC with high expression of both β1 integrin and Rac1 had a worse DFS than patients with high expression of either β1 integrin or Rac1. One explanation for this result is that tumor cell polarity reversal might require Rac1 to be activated by β1 integrin. Both β1 integrin and Rac1 could be involved in multiple molecular pathways, and activation of Rac1 by other unknown factors might lead to unrelated biologic events. This might explain why high expression of Rac1 alone had virtually no impact on DFS and why high expression of β1 integrin alone had minimal impact ( Figure 5D). In addition, the prognostic difference between high expression of both markers and low expression of both markers did not reach statistical significance. However, the number of IMPC patients included in the clinical assessment was small, and it is possible that low expression of both β1 integrin and Rac1 represents another abnormality unrelated to polarity reversal. Investigations using greater numbers of patients are required to validate these findings.
In conclusion, this study indicates that that high metastatic potential of breast IMPC is associated with polarity reversal of tumor cell clusters. β1 integrin positively regulates Rac1, an important factor involved in polarity reversal in IMPC. Overexpression of β1 integrin and the associated upregulation of Rac1 are associated with cancer cell polarity reversal, presence of LNM, and poor DFS, and polarity reversal is an independent predictor of poor prognosis of breast IMPC patients. Additional studies are warranted to validate these findings and to further explore the mechanisms involved in tumor growth and metastasis of breast IMPC.

Breast cancer tissue samples
Formalin-fixed, paraffin-embedded tissue blocks from 102 patients with breast IMPC collected from 2007 to 2008 were retrieved from the archive of the Department of Breast Cancer Pathology and Research Laboratory, Tianjin Medical University Cancer Hospital, Tianjin, China. All cases had an IMPC component constituting >50% of the tumor volume. Paraneoplastic benign breast tissue was available in 48 blocks. In addition, 100 random cases of IDC-NST were retrieved during the same period as the control group. The diagnosis was confirmed independently by two pathologists using World Health Organization (WHO) criteria [30]. None of the patients received preoperative radiation or chemotherapy. Patients were followed for 1 to 100 months, with a median followup time of 63 months. Fresh tumor tissue samples were obtained from five IMPC and five IDC-NST patients who underwent surgical resection from November 2015 to May 2016. The research protocol was approved by the Tianjin Medical University Institutional Review Board, and informed consent was obtained from all participants.
In breast cancer primary culture, after removal of adipose tissue, tumor tissue was treated as previously described [31]. Cells cultured in flask were then digested and centrifuged. Collagen gel was prepared on ice for 3D cell culture. The cells were counted and embedded in collagen gel with a cell density of 1.5 × 10 5 /mL. The collagen gel was then planted in a 24-well plate. After incubation at 37°C for 1 hour, complete medium was added and changed in 3-day intervals. After incubation for 3 to 7 days, spheroids of various sizes were harvested. For the inhibitor assessment, AIIB2 (7.5 µg/mL) was added on the day of planting, and the medium was replaced with a new one containing fresh inhibitor every 3 days. Control cultures were treated with ordinary medium only.

siRNA transfection
Cells were treated with siRNA and Lipofectamine 2000. After 6 hours of incubation, transfection medium was replaced with complete culture medium for 48 hours. To verify the siRNA efficacy, cells were lysed and β1 integrin and Rac1 expression was analyzed by real-time polymerase chain reaction (PCR) and Western blot.

Western blot
Three-dimensional cultured collagen gel was digested by type I collagenase and centrifuged before protein extraction. Total protein extracts from 3D cell cultures were prepared as described previously [24]. The protein concentration of each lysate was measured using the BCA method (Beyotime, China). Twenty micrograms of protein from each sample were separated by 10% SDSpolyacrylamide gel electrophoresis and transferred to PVDF membranes (Millipore, Billerica, MA) for 2 hours. The membranes were blocked with 5% skim milk for 1 hour. After overnight incubation with primary antibody β1 integrin (1:5000) and Rac1 (1:5000), membranes were incubated with secondary antibodies for 1 hour, and proteins were detected using the LiCor Odyssey Infrared Imaging System. Anti-β-actin antibodies were used as internal control.

Immunofluorescence analysis
Primary tumor cells in 3D culture were fixed with 4% paraformaldehyde for 20 minutes and washed three times with phosphate-buffered saline (PBS); then 0.25% Triton X-100 (PBS prepared) was added for 20 minutes. The collagen gel was blocked with 3% BSA for 40 minutes. The cells were then washed with PBS and incubated with primary antibodies E-cad (1:50) or MUC-1 (1:200) overnight at 4°C in a dark wet chamber, followed by incubation with secondary antibodies (ZF-0316/ZF-0312; Zhongshan Golden Bridge, China) for 2 hours at room temperature. After washing with PBS, the nuclei were counterstained with DAPI (blue). Immunofluorescent images were captured by the Olympus IX51 (Olympus, Tokyo, Japan) confocal microscope. Experiments were repeated a minimum of three times.

Clinicopathologic features
Tumor size was measured as the largest dimension grossly or microscopically. Lymph node status was obtained from pathologic report. The tumor grade and tumor-node-metastasis (TNM) stage were assigned according to the standard WHO classification [30]. IHC for ER, PR, and HER2 was re-evaluated at review of each case. Tumors with staining in >1% of tumor cell nuclei were defined as positive for ER and PR [32]. HER2 immunostaining was evaluated per the Dako HercepTest scoring system.

Statistical analysis
Statistical analysis was carried out using SPSS 19.0. Differences among groups were analyzed using the Mann-Whitney U test. Correlations between two variables were evaluated using Spearman's rank correlation analysis. Survival curves for DFS and OS were constructed using the Kaplan-Meier method, and the differences between two groups were assessed using the log-rank test. Cox proportional hazards models were used to perform univariate and multivariable analysis. A two-tailed P < 0.05 was considered statistically significant.