BRCA-IRIS overexpression triggers paclitaxel resistance, in vitro
We recently showed BRCA1-IRIS overexpression promotes intrinsic as well as acquired cisplatin resistance in ovarian tumor and HOSE cells, respectively [37]. We also showed that BRCA1-IRIS silencing or inactivation using a novel inhibitory peptide reversed both [37]. Thus, we hypothesized that BRCA1-IRIS overexpression could also be involved in paclitaxel resistance developed in TNBC patients [4,9].
To study BRCA1-IRIS overexpression on the intrinsic or acquired paclitaxel resistance, HME/IRIS and three TNBC cell lines; MDA-MB-231, MDA-MB-468 and BT549 or HME cells were used. All cell lines were exposed to increasing concentrations of paclitaxel and cells viability and BRCA1-IRIS expression were measured. HME cells were the most sensitive cell line, showing an IC50 of approximately 5 μM (black line in Figure 1A, left). MDA-MB-468 cells were the most sensitive TNBC cell line [43], showing an IC50 of approximately 20 μM (orange line in Figure 1A, left), whereas BT549 and MDA-MB-231 were more resistant showing IC50 of approximately 35 μM (blue line in Figure 1A, left) and >50 μM (green line in Figure 1A, left), respectively. Interestingly, HME/IRIS cells were as resistant as MDA-MB-231 cells showing an IC50 of approximately 50 μM (red line in Figure 1A, left). Moreover, resistance level was well correlated with BRCA1-IRIS expression in each cell line. Cell lines showing high basal level of BRCA1-IRIS and maintaining a high level after treatment, such as BT549, HME/IRIS and MDA-MB-231, were more resistant than those that had high basal of BRCA1-IRIS but failed to maintain a high level after treatment, such as MDA-MB-468 (see inset in Figure 1A, left). Taken together, these data suggest a direct relationship between the level of BRCA1-IRIS and the intrinsic paclitaxel resistance in TNBC cells.
Direct role for BRCA1-IRIS overexpression in TNBC cells paclitaxel resistance
To show the direct role of BRCA1-IRIS overexpression in TNBC cells paclitaxel intrinsic resistance, MDA-MB-231, MDA-MB-468 and BT549 cells were transfected with luciferase (siLuc, negative control) or BRCA1-IRIS (siIRIS)-specific siRNAs for 48 h before they were exposed to increasing concentration of paclitaxel and survival was measured by cell counting 24 h later. Alternatively, silenced cells were treated with 5 μM of paclitaxel and proteins isolated 24 h later were interrogated by western blot analysis.
Compared to siLuc-transfected cells, BRCA1-IRIS-silenced cells survival was significantly decreased when treated with paclitaxel. Indeed, in the more resistant TNBC cell lines, BT-459 or MDA-MB231, IC50 dropped from ≥50 μM (light blue and green line, respectively Figure 1A, right) to approximately 10 μM (dark blue and green line, respectively Figure 1A, right) when BRCA1-IRIS was silenced in them. Moreover, even the more sensitive MDA-MB-468 cells became more sensitized and the IC50 dropped from approximately 20 μM (light-brown line, Figure 1A, right) to 5 μM when BRCA1-IRIS was silenced in them (dark-brown line, Figure 1A, right).
On the molecular level, compared to vehicle-treated cells, paclitaxel treatment induced BRCA1-IRIS, p-AKT, p-FOXO, survivin and Cyclin D1 in MDA-MB-468 cells (compare siLuc on the right to the left in Figure 1B). In the absence of BRCA1-IRIS, this induction as well as the basal level of these survival factors was significantly decreased (compare siIRIS on the right and the left in Figure 1B). Taken together, these data clearly show that some if not all paclitaxel-intrinsic resistance in TNBC cells depends on BRCA1-IRIS overexpression.
Signaling loops involved in BRCA1-IRIS-induced intrinsic and acquired paclitaxel resistance
Although, HME cells were the least resistant to paclitaxel (Figure 1A, left), BRCA1-IRIS expression increased in them following paclitaxel treatment, especially at lower concentrations (inset in Figure 1A, left). This made us wonder whether in vivo too naïve HME cells exposed to low paclitaxel concentrations respond by upregulating BRCA1-IRIS expression in order to survive. Now genetically altered and BRCA1-IRIS overexpressing, these cells could pose an imminent risk later if they survived and grow as a paclitaxel-resistant TNBC clone.
Thus, to define on the molecular level the role of BRCA1-IRIS in paclitaxel-acquired and intrinsic resistance, we analyzed HME and HME/IRIS cells exposed to 0, 10, 20 or 30 μM of paclitaxel for 24 h. Paclitaxel-acquired increase in BRCA1-IRIS, EGFR and ErbB3 (not epidermal growth factor receptor 2 (ErbB2)), epidermal growth factor (EGF) and neurogulin 1 (NRG1) expression in HME cells was documented (Figure 1C, left). All were intrinsically high in HME/IRIS and remained unchanged after treatment (Figure 1C, left). In keeping with that, a paclitaxel-acquired increase in EGFR activation detected as an increase in p-Y1173-EGFR, which is induced by EGF binding to the EGFR-ErbB2 complex and ErbB2 activation detected as increase in p-Y1248-ErbB2, which is induced by NRG1 binding to the ErbB2-ErbB3 complex in HME cells, were detected (Figure 1C, left). Both were intrinsically high in HME/IRIS cells and remained unchanged after treatment (Figure 1C, left).
Furthermore, a paclitaxel-acquired decrease in PTEN level in HME cells, while intrinsically low in HME/IRIS cells, remained unchanged after drug treatment was documented (Figure 1C, right). Concurrently, a paclitaxel-acquired increase in AKT phosphorylation on S308/T473 and FOXO3a phosphorylation on T32 (AKT target) in HME cells, while intrinsically high p-AKT and p-FOXO levels in HME/IRIS cells remained unchanged after treatment were documented (Figure 1C, right). In keeping with that, a paclitaxel-acquired decrease in total FOXO3a and FOXO1 levels in HME cells, while already low and remained so in HME/IRIS cells following paclitaxel treatment were observed (Figure 1C, right). In line with that, a paclitaxel-acquired increase in FOXO targets; for example, Cyclin D1 and survivin (also known BRCA1-IRIS targets, see [34-37]) in HME, whereas intrinsically high levels remained unchanged in HME/IRIS cells were detected (Figure 1C, right).
To ascertain that even further, we generated paclitaxel-resistant HME and HME/IRIS cells by exposing them to 10 μM of paclitaxel for 1 or 3 weeks (cells were then propagated in paclitaxel-free media). Consistent with the data described above, these paclitaxel-resistant HME cells also showed a paclitaxel-acquired increase in BRCA1-IRIS, EGFR, ErbB2, ErbB3, p-AKT, p-FOXO3a, concurrent with decrease in total FOXO3a and FOXO1 and increase in Cyclin D1, survivin and Bcl-xL in HME (Figure 1D). All were intrinsic in HME/IRIS cells and remained unchanged after drug treatment (Figure 1D). Taken together, these data suggest that intrinsic or paclitaxel-acquired resistance in TNBC cells is due to high expression level of BRCA1-IRIS.
BRCA1-IRIS overexpression promotes FOXO3a degradation
Phosphorylation by AKT and/or ERK promotes FOXO3a nuclear exclusion and degradation [28]. Total AKT levels were not changed in the cytoplasm or the nucleus of paclitaxel-resistant HME or HME/IRIS cells (Figure 1E). In contrast, the level of p-AKT was slightly higher in the cytoplasm, while slightly lower in the nucleus of paclitaxel-resistant HME cells than HME/IRIS cells (Figure 1E). While similar in the cytoplasm, total ERK decreased in the nucleus of paclitaxel-resistant HME and HME/IRIS cells (Figure 1E). The level of p-ERK increased in the cytoplasm of paclitaxel-resistant HME cells, decreased in the cytoplasm of paclitaxel-resistant HME/IRIS cells, while remained largely unchanged in the nucleus of resistant HME and HME/IRIS cells (Figure 1E). The level of p-FOXO increased in the cytoplasm of paclitaxel-resistant HME cells, while it decreased in the cytoplasm of resistant HME/IRIS cells (Figure 1E). No p-FOXO in the nucleus of paclitaxel-resistant HME cells, while relatively high level in paclitaxel-resistant HME/IRIS cells was detected (Figure 1E). MDM2 and Skp2 have been previously implicated in mono- and poly-ubiquitylation, respectively of FOXO proteins. We detected increased level of MDM2 and Skp2 in the cytoplasm and nucleus of the paclitaxel-resistant HME and HME/IRIS (Figure 1E). Taken together, these data suggest that intrinsic or paclitaxel-acquired increase in MDM2 and Skp2 expression induced by BRCA1-IRIS overexpression cooperatively promote FOXO3a ubiquitylation and proteasomal degradation in HME/IRIS and HME cells, respectively.
Additional signaling pathways induced by BRCA1-IRIS overexpression
We also observed that compared to HME, HME/IRIS cells contained lower ERK and p38 while higher JNK levels in the cytoplasm and nucleus (Figure 1F). In contrast, compared to HME cells, HME/IRIS cells contained lower level of p-ERK in the cytoplasm but higher level in the nucleus (Figure 1F), higher levels of p-JNK in the cytoplasm and nucleus (Figure 1F), and similar level of p-p38 in the cytoplasm but no p-p38 in the nucleus in both cell lines (Figure 1F). Taken together, these data suggest additional signaling pathways induced by BRCA1-IRIS overexpression [34-36] that could also be involved in promoting FOXO3a degradation.
Selective sensitivity of BRCA1-IRIS-overexpressing cells to drugs against these pathways
The above data imply BRCA1-IRIS-overexpressing cells sensitivity to inactivation of these pathways. To evaluate that, similar numbers of HME or HME/IRIS cells were grown in the presence of 10 μM of ERK (PD98059), JNK (SP600125), p38 (SB203580), PI3′K/AKT (LY294002), EGFR (Erlotinib), ErbB2 (CP-724714), EGFR/ ErbB2 (Lapatinib) or EGFR/ErbB2/ErbB3 (Sapitnib) inhibitors for 24 h.
First, because BRCA1-IRIS overexpression significantly increased proliferation of HME cells (Figure 1G and see [34-36]) the data obtained with the inhibitors were normalized to untreated cells of each cell line, separately. None of the drugs at the concentration used had an effect on HME cells survival (white bars in Figure 1H). In contrast, ERK, JNK, PI3′K/AKT, EGFR and ErbB2 inhibitors decreased HME/IRIS cells survival by approximately 50% compared to untreated HME/IRIS cells (black bars in Figure 1H). More dramatic effect was noticed using the dual EGFR/ErbB2 inhibitor and even more dramatic using the inhibitor that targets EGFR, ErbB2 and ErbB3 at the same time (Figure 1H). Taken together, these data suggest that BRCA1-IRIS-overexpressing cells are more sensitive to drugs that block EGFR-ErbB2 and ErbB2-ErbB3 complexes and their downstream signaling pathways. Schematic representation of all the above data is presented in Figure 1I.
A BRCA1-IRIS mimetic inhibitory peptide abolishes BRCA1-IRIS expression and functions
The above data seem to argue that inactivating BRCA1-IRIS could sensitize TNBC cells to paclitaxel. Since BRCA1-IRIS-specific inhibitor is not yet available, the following reasons motivated us to pursue the inhibitory mimetic peptide approach instead. First, previous results suggested that BRCA1-IRIS connect with partners using a domain in its C-terminus [32,34]. Second, unlike full-length BRCA1-IRIS, overexpression of an intronless BRCA1-IRIS (Δint11, missing the 34 amino acid domain encoded by intron 11) failed to induce expression of targets such as; Cyclin D1, survivin and vimentin in HME cells (see Figure 2A and [34-38,44]). Based on these facts, we hypothesized that BRCA1-IRIS forms an oncogenic complex(es) by connecting with other factor(s) (Figure 2B) and that transduction of BRCA1-IRIS intron 11 domain could in a dominant-negative fashion disrupt this interaction leading to loss of BRCA1-IRIS oncogenic effect (Figure 2B).
To test this hypothesis, we synthesized a peptide (hereafter IRIS peptide) consisting of the BRCA1-IRIS intron 11 peptide (red sequence in Figure 2C) fused to a cell/nuclear penetrating signal at the N-terminus (black sequence in Figure 2C). Equal numbers of HME, HME/IRIS, MDA-MB-231, MDA-MB-468 and BT549 cells were exposed to increasing concentrations of scrambled or IRIS peptides for 48 h. Compared to scrambled peptide, treatment with IRIS peptide decreased survival in all cell lines. HME cells showed an IC50 of approximately 0.5 μM, whereas HME/IRIS and the TNBC cell lines an IC50 of 5 to 7.5 μM (Figure 2D). On the molecular level, treatment with 5 μM of IRIS peptide for 24 h almost completely abolished BRCA1-IRIS expression in all these cell lines compared to treatment with 5 μM scrambled peptide (inset in Figure 2D). Taken together, these data support our hypothesis that IRIS peptide acts as a dominant negative in TNBC cells suppressing BRCA1-IRIS oncognecity by destabilization of the protein (Figure 2B).
To ascertain the specificity of IRIS peptide even further, HME, HME/IRIS, MDA-MB-231, MDA-MB-468 and BT549 cells were first transfected with siLuc or siIRIS for 48 h before exposing them to increasing concentrations of IRIS peptide for another 48 h. Gradual decrease in all cell lines was detected in siLuc-transfected cells following treatment with increasing concentrations of IRIS peptide (Figure 2E, left). In contrast, BRCA1-IRIS silencing significantly decreased survival of all cell lines (Figure 2E, right) and treatment with IRIS peptide had no additive effect on these cells even at higher concentrations (Figure 2E, right). These data, suggest that IRIS peptide targets BRCA1-IRIS specifically leading to significant growth retardation and cell death in TNBC cells, in vitro.
BRCA1-IRIS inhibition sensitizes breast cancer cells to low paclitaxel concentrations, in vitro
Whether inactivating BRCA1-IRIS could sensitize TNBC cells to paclitaxel was investigated next. HME cells exposed to 0.5 μM or HME/IRIS, MDA-MB-231, MDA-MB-468 and BT549 cells exposed to 5 μM of IRIS peptide were exposed or not to 5, 10 or 20 μM of paclitaxel for an additional 24 h. At these concentrations, IRIS peptide completely abolished BRCA1-IRIS expression from all cell lines (Figure 2F, inset) and reduced survival by approximately 50% in all cell lines (Figure 2F). While paclitaxel also decreased survival of all cell lines that gradually increased with increasing the concentration, the combination had even more pronounced effect on the survival of all cell lines (Figure 2F).
To identify the lowest paclitaxel concentration that could synergize with IRIS peptide, MDA-MB-231 and MDA-MB468 cells were treated with 1 μM of paclitaxel, increasing concentrations of IRIS peptide or both for 24 h. As shown above, survival gradually decreased with increasing IRIS peptide concentration in both cell lines (light-colored lines in Figure 2G, upper), which correlated very well with gradual increase in activated caspase 3 and 7 in both cell lines (Figure 2G, lower). Although at this very low concentration of paclitaxel we detected no cell death in either cell line (white bar in Figure 2G, upper), addition of 1 μM additively increased the effect of IRIS peptide on the survival of both cell lines (dark-colored lines in Figure 2G, upper). Taken together, these data suggest synergy between BRCA1-IRIS inhibition and very low concentrations of paclitaxel on the survival of TNBC cells, in vitro.
Molecular comparison between the effect of BRCA1-IRIS silencing or inactivation in TNBC cells
To further explore IRIS peptide mechanism of action on cell survival and at the same time generate further evidence for its specificity, MDA-MB-231 and MDA-MB-468 cells were silenced from BRCA1-IRIS for 72 h, treated with 5 μM of IRIS peptide for 24 h and for comparison cells were exposed to 10 μM of the PI3′K/AKT inhibitor; LY294002 (hereafter LY) or the ERK inhibitor; PD98059 (hereafter PD) for 24 h. Surprisingly, LY and PD beside inactivating AKT and ERK, respectively with no effect on the total proteins, both also significantly decreased BRCA1-IRIS expression (Figure 2H). In contrast, BRCA1-IRIS silencing or inactivation not only decreased BRCA1-IRIS expression level, but also total AKT and p-AKT levels in both cell lines (Figure 2H). Concurrently, LY, IRIS peptide and BRCA1-IRIS silencing all significantly decreased p-FOXO3a level and increased total FOXO3a level in both cell lines, which again was correlated with significant decrease in MDM2 and/or Skp2 expression (Figure 2H). Surprisingly, LY and IRIS peptide (and to a lesser degree BRCA1-IRIS silencing) had little effect on total ERK expression in this experiment but increased the level of p-ERK in both cell lines (Figure 2H). Taken together, these data, in addition to confirming the specificity of the peptide, they define a molecular mechanism of action for IRIS peptide in TNBC cell survival, a positive-feedback loop between BRCA1-IRIS and AKT signaling in both cell lines (see model in Figure 2I).
BRCA1-IRIS overexpression induces in HME cells, while silencing or inactivation inhibits aggressiveness in TNBC cells, in vitro
Next, we asked whether BRCA1-IRIS depletion or inactivation impacts TNBC cells aggressiveness. To evaluate that, we used one of the best assays, ‘growth in three-dimensional culture’. HME, HME/IRIS, MDA-MB-231 or MDA-MB-468 cells stably expressing shcontrol or shIRIS were layered on matrigel-coated wells. Additionally, parental cell lines layered on matrigel-coated wells were grown in the presence of scrambled or IRIS peptide (added every third day). Ten days later, HME cells formed acini that were small/round/organized, composed of 40 to 50 cells and hollow in the middle (see Figure 3A and N1). In contrast, at day 10, acini formed by HME/IRIS (Figure 3D), MDA-MB-231 (Figure 3G) or MDA-MB-468 (Figure 3J) were much larger, non-round and unorganized with large protrusions, and were filled with cells on the inside resembling DCIS (Figure 3N2 and N3).
Moreover, BRCA1-IRIS depletion (Figure 3B) or inactivation (Figure 3C) completely abolished formation of acini by HME cells and converted HME/IRIS (Figure 3E and F, respectively), MDA-MB-231 (Figure 3H and I, respectively) and MDA-MB468 (Figure 3K and L, respectively) acini into smaller/rounder/organized acini. These acini also lacked protrusions and most importantly were hollow on the inside (see for example Figure 3N4 for MDA-MB-468 silenced cells). Quantitatively, BRCA1-IRIS overexpression led to more than threefold increase in HME cells aggressiveness (defined as generation of non-round, non-organized, large filled with cells acini, Figure 3M, upper), whereas depletion or inactivation reduced that phenotype in TNBC cells by more than threefold (defined as generation of round, organized, small and hallow acini Figure 3M, lower).
BRCA1-IRIS overexpression induces in HME cells, while silencing inhibits aggressive biomarkers expression in TNBC cells, in vitro
To correlate these data with expression of aggressiveness biomarkers, similar acini were processed for immunofluorescence (IF) labeling with EGFR, ErbB2, ErbB3 and cortactin antibodies. HME acini showed low expression levels of all biomarkers that when found were, as expected, restricted to the apical side of the acini (see Figure 3O1, P1, R1 and S1). In contrast, HME/IRIS acini showed increase expression of all biomarkers, which were expressed at the apical as well as the basolateral side of the acini (Figure 3O2, P2, R2 and S2). MDA-MB-468 acini showed high levels and unrestricted expression of all aggressiveness biomarkers (Figure 3O3, P3, R3 and S3), whereas BRCA1-IRIS silencing in MDA-MB-468 cells not only decreased the expression of all aggressiveness biomarkers, but also restricted it to the apical side (Figure 3O4, P4, R4 and S4). Taken together, these data suggest that BRCA1-IRIS overexpression triggers a polarization, epithelial to mesenchymal transition (EMT) and aggressiveness in HME cells, which could explain the invasive, aggressiveness and unrestricted growth of HME/IRIS cells in three-dimensional matrigel and SCID mice [38], and that BRCA1-IRIS silencing in TNBC cells reverse these phenotypes.
BRCA1-IRIS silencing or inactivation inhibits tumor-initiating phenotype in TNBC cells, in vitro
It has been suggested recently that a population with tumor-initiating (that is breast cancer stem-like) capabilities exists within TNBC tumors. To evaluate whether BRCA1-IRIS inactivation suppress stemness phenotype in TNBC cells, MDA-MB-231 and MDA-MB-468 cells stably expressing shcontrol or shIRIS were plated in ultra-low binding dishes. Additionally parental cells were also plated in ultra-low binding dishes and treated with scrambled or IRIS peptide readded to cells every third day. Mammospheres - a hallmark of breast cancer stem-like cells - formed within 10 days were then counted and photographed. Control shRNA expressing or scrambled peptide-treated MDA-MB-231 or MDA-MB-468 cells formed many large-size mammospheres (Figure 4A, for MDA-MB-468, identical result was documented for MDA-MB-231). BRCA1-IRIS silenced (Figure 4B, for MDA-MB-468, identical data was documented for MDA-MB-231) or inactivated (Figure 4C, for MDA-MB-468, identical data was documented for MDA-MB-231) showed much smaller mammospheres. Quantitatively, BRCA1-IRIS silencing or inactivation reduced the ability of both cell lines to form mammospheres by approximately threefold (Figure 4D) and whenever formed their size was approximately threefold smaller than control cells (Figure 4E).
BRCA1-IRIS induces in HME cells, whereas silencing inhibits stemness biomarkers expression in TNBC cells, in vitro
To investigate whether altering BRCA1-IRIS alters stemness biomarkers expression, HME cells were induced to overexpress BRCA1-IRIS or TNBC cells were silenced from BRCA1-IRIS, and expression of stemness biomarkers, Oct-4, Sox-2 and Nanog was analyzed. BRCA1-IRIS overexpression significantly increased expression of these biomarkers in HME cells (Figure 4F). Conversely, BRCA1-IRIS silencing significantly decreased the expression of these biomarkers in MDA-MB-231 and MDA-MB-468 cells (Figure 4G). Taken together, these data suggest that inactivating BRCA1-IRIS suppresses the tumor-initiating phenotype, implicated in aggressiveness and recurrence of TNBC cells.
BRCA1-IRIS silencing or inactivation inhibits migration and invasion in TNBC cells, in vitro
TNBC tumor cells often possess enhanced migration and invasion capabilities. To assess whether BRCA1-IRIS plays a role in these phenotypes as well, we plated MDA-MB-468 cells stably expressing shcontrol or shIRIS inside inserts placed in the middle of wells of 6-well plates. When confluent, inserts were removed, dying and dislodged cells were washed away and cells were allowed to migrate for 24 h. MDA-MB-468 cells expressing shcontrol showed robust migration capabilities as evident by filling the empty space left after the removal of the insert (see white arrow in Figure 4H) as well as the outward migration (see arrowheads in Figure 4H). BRCA1-IRIS-silenced MDA-MB-468 cells, on the other hand, almost completely lost their migratory ability (see white arrows and arrowheads in Figure 4I).
To study whether BRCA1-IRIS silencing or inactivation affects TNBC cells invasion ability, MDA-MB-231 and MDA-MB-468 cells expressing shcontrol or shIRIS were layered on matrigel-coated Boyden chambers transwells. Additionally, parental cells were layered on transwells and grown in the presence of scrambled or IRIS peptide. To migrate to the lower side of the chamber or the bottom well, cells must first digest the matrigel. Four days later, a significant number of control cells (expressing empty vector or treated with scrambled peptide) invaded the matrigel and migrated to the lower side of the chambers (Figure 5A for MDA-MB-231, 5D for MDA-MB-468) and the bottom well (Figure 5H for MDA-MB-231, 5K for MDA-MB-468) in both cell lines. BRCA1-IRIS-silenced (Figure 5B and I, for MDA-MB-231, 5E and L for MDA-MB-468) or -inactivated (Figure 5C and J for MDA-MB-231, 5F and M for MDA-MB-468) cells almost completely lost their ability to invade and migrate. Quantitatively, BRCA1-IRIS silencing or inactivation reduced the invasive/migratory ability of these TNBC cells by approximately 100-fold (see Figure 5G for cells on the reverse side of the transwell, and Figure 5N for cells on the bottom well).
BRCA1-IRIS overexpression induces in HME cells, whereas silencing inhibits EMT and invasion biomarkers expression in TNBC cells, in vitro
To investigate whether altering BRCA1-IRIS alters EMT and invasion biomarkers expression, HME cells were induced to overexpress BRCA1-IRIS or TNBC cells were silenced from BRCA1-IRIS. BRCA1-IRIS overexpression in HME cells significantly enhanced expression of EMT biomarkers, such as slug, snail, and twist (Figure 5O), as well as invasion biomarkers, such as N-cadherin (compare Figure 5P2 to P1), FOXC2 (compare Figure 5Q2 to Q1) and vimentin (compare Figure 5R2 to R1) and suppressed expression of E-cadherin (compare Figure 5S2 to S1). Conversely, BRCA1-IRIS silencing in MDA-MB-231 or MDA-MB-468 cells significantly reduced expression of slug, snail and twist (Figure 5O), as well as N-cadherin (compare Figure 5P4 to P3), FOXC2 (compare Figure 5Q4 to Q3) and vimentin (compare Figure 5R4 to R3) and enhanced E-cadherin expression (compare Figure 5S4 to S3). Taken together, these data suggest that BRCA1-IRIS overexpression triggers EMT, migration and invasion in TNBC cells.
Elevated BRCA1-IRIS and survivin, while lack of FOXO3a expression in aggressive human breast tumors
Previously, we showed elevated BRCA1-IRIS expression in approximately 80% of breast tumors (>800 tumor samples were analyzed) that was correlated with elevated p-AKT and survivin expression (Ref. [38]). To correlate these data to FOXO3a expression, tissue microarrays consisted of normal/cancer adjacent (n = 66), DCIS (n = 167), invasive (n = 179) and metastatic (n = 99) samples were immunohistochemically stained with anti-BRCA1-IRIS, -survivin and -FOXO3a antibodies. Semi-quantitative scoring analysis (see Methods) showed that BRCA1-IRIS (P <0.0001, Figure 6A) and survivin (P <0.0001, Figure 6B) expression significantly increased concurrently with the increase in tumor aggressiveness. Indeed, aside from the fact that more cells per field showed upregulation of BRCA1-IRIS and survivin, the intensity of the staining per cell also increased as disease progressed (data not shown).
Since p-AKT phosphorylation restricts FOXO3a nuclear translocation and promotes its degradation, we evaluated the presence as well as the cellular localization of FOXO3a in these tumors. We evaluate the nuclear (N), cytoplasmic (C), both (NC) or negative (0) expression of FOXO3a. We found that FOXO3a staining was N in the majority of BRCA1-IRIS expressing (IRIS+) and survivin-expressing (survivin+) normal tissues (all P <0.0001, Figure 6C and D). In contrast, the majority of IRIS+ or survivin+ DCIS tumors showed 0, with few tumors showing N staining (P <0.0001, Figure 6C and D). The majority of IRIS+ invasive tumors showed 0, with few tumors showing C FOXO3a staining, whereas the majority of the survivin+ invasive tumors were 0, with few tumors showing N FOXO3a staining (P <0.0001, Figure 6C and D). Finally, nearly all IRIS+ or survivin+ metastatic breast cancer tissues were 0 for FOXO3a (P <0.0001, Figure 6C and D). Taken together, these data suggest that during breast cancer progression, elevated BRCA1-IRIS expression activates AKT that decreases FOXO3a expression leading to elevated survivin expression.
Low disease-free survival in patients with elevated BRCA1-IRIS signature-expressing TNBC tumors
Based on the data presented above we concluded that BRCA1-IRIS-overexpressing TNBC cells show concurrent elevation in the expression of ‘EGFR, AKT, MDM2, Skp2 and survivin’. Considering this short list as a BRCA1-IRIS-overexpression signature in TNBC tumors, we conducted an association assessment of this signature and disease-free survival (DFS) using a combined breast cancer cohort from seven Gene Expression Omnibus (GEO) studies (GSE2034, GSE2603, GSE3494, GSE4922, GSE6532, GSE7390 and GSE12093). Patients with positive status for estrogen, progesterone and HER2 were excluded leading to a size of 576 cases. The standardized BRCA1-IRIS signature expression levels were pooled together and considered as explanatory variables in a Cox model on DFS. Based on the estimation result of the Cox model, we developed a prognostic index combining the expression levels of BRCA1-IRIS overexpression signature: ‘Prognostic index = EGFR + 2 x AKT + 2 x MDM2 + 2 x Skp2 + 2 x survivin’, which was evaluated for the entire sample. The patient cohort was further categorized into low- versus middle/high-expressing groups by using the lower tertile as the cutoff value. Using Kaplan-Meier method and group-wise comparison in DFS done using the log-rank test, this analysis showed that among the cohort tested, patients with middle/high expression levels of BRCA1-IRIS overexpression signature showed significantly (Log-rank P = 0.016) shorter DFS compared to those showing low expression levels (Figure 6E). Taken together, these data suggest that elevated BRCA1-IRIS level is a poor overall prognosis in patients with TNBC disease.
BRCA1-IRIS silencing or inactivation blocks TNBC tumor formation, maintenance and sensitizes them to low paclitaxel concentrations, in vivo
We showed previously that BRCA1-IRIS-overexpressing HME cells form aggressive TNBC tumors lacking BRCA1 expression [38]. To evaluate whether BRCA1-IRIS overexpression is indeed required for TNBC tumor formation and/or maintenance, MDA-MB-231 or MDA-MB-468 (2 x 106) cells expressing shcontrol or shIRIS were injected in mammary fat pads (second left and fourth right) of 6- to 8-week-old female SCID mice (mice, n = 6/cell line → tumors, n = 12/cell line). Control shRNA-expressing cells formed tumors that reached approximately 750 mm3 within 4 weeks (black line for MDA-MB-231 and blue line for MDA-MB-468 in Figure 7A). BRCA1-IRIS-silenced MDA-MB-231 formed very small tumors that were approximately 100 mm3 by 4 weeks (red line in Figure 7A), and BRCA1-IRIS-silenced MDA-MB-468 cells failed completely to form any tumors (green line in Figure 7A). These data, suggest that BRCA1-IRIS overexpression is required for TNBC tumor formation.
Moreover, MDA-MB-468 or MDA-MB-231 (2 x 106) cells were injected in mammary fat pads (second left and fourth right) of 6- to 8-week-old Nu/Nu female mice (mice, n = 24 → tumors, n = 48). When tumors reached approximately 0.1 cm3 in volume, the mice were divided into four groups that were treated with vehicles (that is DMSO injected intraperitoneally (i.p.) + scrambled peptide injected directly into tumors (i.t.), mice, n = 6 → tumors, n = 12), paclitaxel (10 mg/kg, i.p., mice, n = 6 → tumors, n = 12), IRIS peptide at (10 mg/kg, i.t., mice, n = 6 → tumors, n = 12), or paclitaxel + IRIS peptide (at half the concentrations, same routes, mice, n = 6 → tumors, n = 12). Drugs were delivered every third day for total of four injections (see red arrows in Figure 7B). Tumors were measured using caliper three times per week and collected at the end of the experiment (day 12). At day 12, tumors in vehicle-treated mice became approximately 3.5 times their original size at day 0 (see black line in Figure 7B and 7C). Impressively, at day 12, tumors in IRIS peptide-treated mice, actually shrunk to approximately 25% their original size at day 0 (see red line in Figure 7B and 7C). Taken together, these data suggest that BRCA1-IRIS overexpression is important for TNBC tumor maintenance.
Moreover, at day 12, paclitaxel treatment showed some effect, but tumors still grew to approximately 1.5 times their original size at day 0 (see blue line in Figure 7B and 7C). More importantly, tumors treated with the combination ‘IRIS peptide + paclitaxel’ (at only half the above concentrations) shrunk to <15% of their original size at day 0 (see green line in Figure 7B and 7C). Taken together, these data suggest that inactivating BRCA1-IRIS sensitizes TNBC tumors to low paclitaxel concentrations, in vivo.
BRCA1-IRIS inactivation suppresses paclitaxel-induced overexpression in BRCA1-IRIS → survivin, in vivo
Finally, to correlate the above in vivo efficacy data of IRIS peptide to expression of BRCA1-IRIS and hence survivin in these tumors, we stained adjacent sections from the tumors developed above with hematoxylin and eosin (H&E) or with IHC for BRCA1-IRIS or survivin. As expected, in controls-treated tumors, BRCA1-IRIS and survivin expression was high (Figure 7D, upper row), and as described above, the expression increased even further by paclitaxel treatment (even though tumors were smaller, Figure 7D, third row). In contrast, the expression of both proteins was completely abolished in tumors treated with IRIS peptide (Figure 7D, second row) or the combination (even at only half the concentrations, Figure 7D, fourth row). Taken together, these data could explain the acquired paclitaxel resistant recurrences observed in TNBC patients.