AGR2 is overexpressed in a number of epithelial cancers [11
] and is inversely associated with patient survival in the ER-positive patient population [21
]. Although this association itself does not indicate a critical role for AGR2 in disease progression, it does warrant examination of the functional relevance of this gene. Both anchorage-dependent growth and anchorage-independent growth were inhibited after silencing endogenously expressed AGR2 in ER-positive cell lines (Figure ). This impact on growth was seen with multiple AGR2 siRNA sequences (Supplementary figure S1 in Additional file 1
), which supports the specificity of the effect of AGR2 knockdown on the inhibition of cancer cell growth. AGR2 knockdown significantly reduced colony formation in soft agar in ER-negative lines, as well (Figure ), raising an interesting discrepancy between ER-positive and ER-negative lines. The soft-agar assay represents a particularly stressful in vitro
assay, as it requires the cancer cells to grow from a single cell, whereas other assays allow for cell-cell interaction. Support exists in the literature for a role of AGR2 in stress conditions [15
]. Another possibility is that the soft-agar assay measures AGR2-dependent biologies that are not captured in the other assays, a concept consistent with distinct breast cancer models, in which AGR2 overexpression was reported to play a role in metastases and adhesion, but not growth [11
]. Aligned with the hypothesis that MDA-MB-231 cells may be more sensitive to neutralization of AGR2 activity under stress conditions, we observed that our AGR2-neutralizing mAb inhibited MDA-MB-231 cell growth under low serum conditions. As noted earlier, siRNA-mediated knockdown of AGR2 in MDA-MB-231 cells in full-serum conditions had little growth effect in anchorage-dependent conditions. The MDA-MB-231 cells were more sensitive to growth inhibition with the AGR2-neutralizing mAb than were T47 D and ZR-75-1 cells, although this may be trivially explained by the observation that MDA-MB-231 cells have significantly lower levels of AGR2 than do T47 D and ZR-75-1 cells used in the same experiment and thus may be more readily neutralized. Although this is the first time a role for AGR2 has been demonstrated in breast cancer cell growth, these results are consistent with the impact of AGR2 knockdown on cell growth reported in other non-breast cancer models.
To further evaluate the role of AGR2 in breast cancer, the impact on cell death and the cell cycle were explored. Supporting the anchorage-dependent functional data, a cell-cycle phenotype and induction of cell death was seen in both T47 D and ZR-75-1 cells (Figures and ). Thus far, neither phenotype has been reported. The BrdU phenotype is in contrast to previous reports in which stable expression of AGR2 in H1299 cells had no effect on the cell cycle or DNA synthesis [15
]. Cell context-dependent differences exist in the mechanisms of cell death, which may be related to the genetic background of the respective cell lines, such as the differences in p53 status, but cell death was observed in either case. That inhibition of AGR2 activity can induce cytotoxicity, in addition to being cytostatic, is particularly important for the potential to treat slow-growing tumors.
To provide support for the phenotypic effects shown with AGR2 knockdown and to understand AGR2 biology in greater depth, we investigated intracellular signaling downstream of AGR2. Mechanistic hints from the literature have been limited to a role for AGR2 in the wild-type p53 transcriptional response [15
]. Because T47 D cells have mutated p53, and ZR-75-1 cells have wild-type p53, and a phenotype with AGR2 knockdown was observed in both, AGR2 biology cannot be limited to the p53 pathway. Cell-cycle modulation and induction of cell death after transient knockdown of AGR2 directed us to explore other signaling pathways that may be relevant to AGR2 biology. Modulation of cyclin D1, c-Myc
, and E2F1 by AGR2 knockdown is consistent with the cell-cycle and anchorage-dependent growth phenotype seen in the ER-positive cell lines. The impact of AGR2 knockdown on cyclin D1 in ER-negative cell lines that did not translate into a cell-cycle phenotype was an intriguing result. Several possibilities could account for this observation. It could also be a threshold issue, requiring that a certain percentage of cyclin D1 be downregulated to result in a phenotype similar to that observed in ER-positive cells. It is of note that cyclin D1 can be downregulated through both ER-dependent and ER-independent pathways. In addition, ZR-75-1 cells have cyclin D1 amplified and are known to be driven by cyclin D1 [40
], which may not be so important a driver in the ER-negative cell types tested. Collectively, downregulation of cyclin D1 in all four cell types after AGR2 knockdown or treatment with an anti-AGR2 Ab supports cyclin D1 being downstream of AGR2.
We also provide evidence that AGR2 can act extracellularly. This finding is in contrast to a recent report suggesting that AGR2, a member of the protein disulfide isomerase family, is localized in the endoplasmic reticulum and plays an essential role for mucus production [39
]. Because cyclin D1 is induced with an exogenous source of AGR2 (Figure ) and reduced with an anti-AGR2 Ab (Figure ), and AGR2 is detected in the supernatant of breast cancer cell lines, it suggests that AGR2 may have an extracellular mechanism of action. Consistent with our data, other disulfide isomerases have also been associated with cancer-relevant biologies, including invasiveness and stress survival [41
To provide further evidence that AGR2 is important in breast cancer progression, silencing AGR2 in ER-positive cells downregulated c-Myc
, p-Src, and survivin. All of these molecules play critical roles in breast cancer progression by affecting growth, survival, angiogenesis, migration, and invasion [43
], and hence are individually being investigated as targeted monotherapies [44
]. Given the high level of biologic relevance of p-Src, survivin, and c-Myc in cancer, modulation of these key players after silencing AGR2 suggests that AGR2 may be the key in other cancers, beyond breast cancer.
E2 is another key driver and potent mitogen of breast cancer. Several articles in the literature indicate AGR2 is an E2-responsive gene [11
], and we similarly observed an induction of AGR2 in E2-treated ZR-75-1 cells (Figure ). The more-novel finding is that ER protein levels are reduced after AGR2 knockdown (Figure ). Because ER downregulation leads to reduced estrogen responsiveness with fulvestrant, this suggests that ER downregulation induced by AGR2 knockdown might also negatively influence the mitogenic activity of E2. The relative E2 responsiveness could not be accurately assessed in these AGR2-silenced cells, given the complexity of the relations of these molecules, because E2 itself modulates ER in T47 D cells [46
Given the similar profiles of AGR2 knockdown with E2 signaling, important considerations existed when assessing the potential crosstalk between the ER and AGR2 pathways. Initially, we asked how AGR2 knockdown would affect the effect of antiestrogens on cancer cell lines. Next, we determined whether AGR2 might have ER-independent activities. AGR2 knockdown in combination with antiestrogens did not preclude antiestrogen efficacy, and the combination enhanced the impact on growth, ER and cyclin D1 (Figure ). The kinetics of AGR2 knockdown showed that cyclin D1 downregulation occurs before ER downregulation, and therefore, AGR2 has an ER-independent pathway for downregulating cyclin D1, which is supported by the impact on cyclin D1 seen in ER-negative cells.