Identification of an IL-1 conferred gene signature in hormone receptor positive BCa and PCa cell lines that mimics a basal gene expression pattern in hormone receptor negative BCa and PCa cells. We previously found that IL-1 represses hormone receptors in ERa+/PR+ BCa19 and AR+ PCa18,62 cell lines concomitant with p62 upregulation, while ERa-/PR- BCa and AR- PCa cell lines intrinsically have high basal p62. This led us to speculate that IL-1 elicits similar changes in gene expression in hormone receptor positive (HR+) BCa and PCa cells that mimic intrinsic gene expression patterns in hormone receptor negative (HR-) BCa and PCa cells. Such changes would enable BCa and PCa cells to elicit compensatory survival pathways in the absence of hormone receptor activity and, thus, these changes in gene expression could confer resistance to hormone therapy for BCa and PCa tumor cells. To identify the conserved gene signature conferred by IL-1 in HR+ BCa and PCa cells that mimics intrinsic gene expression patterns in HR- BCa and PCa cells, we performed RNA sequencing (RNA-seq) followed by differential gene expression analysis (log2 CPM ≥ 0, log2 FC ≥ 0.6 or ≤ -0.6, and FDR ≤ 0.05) for IL-1a- or IL-1b-treated ERa+/PR+ MCF7 BCa cell line, IL-1b-treated AR+ LNCaP PCa cell line, vehicle control-treated ERa-/PR- MDA-MB-231 BCa cell line, and vehicle control-treated AR- PC3 PCa cell line. LNCaP and PC3 RNA-seq data was previously reported18 (GSE105088). For BCa cell line RNA-seq, RNA was isolated from MCF7 cells treated with vehicle control, 25 ng/ml IL-1a, or 25 ng/ml IL-1b for 5 days and MDA-MB-231 cells were treated with vehicle control only. For PCa cell line RNA-seq, as previously reported, RNA was isolated from LNCaP cells treated with vehicle control or 25 ng/ml IL-1b for 3 days and PC3 cells were treated with vehicle control only. Five sets of differential gene expression analysis were performed: (Set 1) MCF7 cells treated with vehicle control versus IL-1a (“MCF7_VEH-VS-MCF7_IL1A”); (Set 2) MCF7 cells treated with vehicle control versus IL-1b (“MCF7_VEH-VS-MCF7_IL1B”); (Set 3) MCF7 cells treated with vehicle control versus MDA-MB-231 cells treated with vehicle control (“MCF7_VEH-VS-231_VEH”); (Set 4) LNCaP cells treated with vehicle control versus IL-1b (“LNCaP_VEH-VS-LNCaP_IL1B”); and LNCaP cells treated with vehicle control versus PC3 cells treated with vehicle control (“LNCaP_VEH-VS-PC3_VEH”) (Fig. 1A). We identified 2,735 genes in the intersection of Set 1, 2, and 3 and of those genes, 1,707 are consistent in fold change direction (Set 6). Set 6 are the genes that are induced or repressed by both IL-1a and IL-1b in MCF7 cells which are, respectively, basally high or low in MDA-MB-231 cells. We identified 2,786 genes in the intersection of Set 4 and 5 and of those genes, 1,900 are consistent in fold change direction (Set 7). Set 7 are the genes that are induced or repressed by IL-1b in LNCaP cells which are, respectively, basally high or low in PC3 cells. Finally, we identified the 420 genes in the intersection of Set 6 and Set 7 and of those genes, 350 are consistent in fold change direction (Set 8; Supplemental Table 1, Tab 1). Set 8 are the genes that are induced or repressed by IL-1 in MCF7 and LNCaP cells which are, respectively, basally high or low in MDA-MB-231 and PC3 cells. Thus, the 350 gene set represents conserved genes in both BCa and PCa cells that are expected to promote cell survival and tumorigenicity when hormone receptor signaling is lost.
Validation of select genes from the 350 gene signature. We arbitrarily selected six upregulated genes (CCL20, CD68, IL-8, p62, SOX9, Zeb1) and four downregulated genes (CDK2, CXCR7, MMP16, PLK1) from our 350 gene set to validate by RT-qPCR. RNA-seq was perform for HR+ MCF7 and LNCaP cells treated with 25 ng/ml IL-1 and for untreated HR- MDA-MD-231 and PC3 cells; therefore, MCF7 and LNCaP cells were treated with 25 ng/ml IL-1a or IL-1b and MDA-MB-231 and PC3 cells were treated with vehicle control for RT-qPCR analysis. In addition to MDA-MB-231 and PC3 cell lines, we also performed RT-qPCR for basal gene expression in an additional ERa-/PR- BCa cell line, BT549, and an additional AR- PCa cell line, DU145.
RT-qPCR confirmed that IL-1 induces CCL20, IL-8, p62, and Zeb1 in MCF7 (Fig. 2A) and LNCaP (Fig. 2B) cells. We detected a significant increase in CD68 and SOX9 mRNA (Fig. 2B) and/or protein (Fig. 3A) in IL-1-treated LNCaP cells. We did not detect an increase in SOX9 mRNA in IL-1-treated MCF7 cells (Fig. 2A) and we detected only a slight increase in SOX9 protein (Fig. 3A). CD68 mRNA levels were only slightly induced by IL-1b in MCF7 cells (Fig. 2A). IL-1-induced fold changes detected by RNA-seq for CD68 and SOX9 were much greater in LNCaP cells (CD68 log2 FC = 4.46; SOX9 log2 FC = 4.33) than in MCF7 cells (CD68 log2 FC = 0.89 (IL-1a), 0.62 (IL-1b); SOX9 log2 FC = 1.31 (IL-1a), 1.36 (IL-1b)) (Supplemental Table 1, Tab 1), which may correspond with our inability to detect induction for these genes by RT-qPCR in MCF7 cells. Finally, RT-qPCR confirmed that IL-1a and IL-1b repress CDK2, CXCR7, MMP16, and PLK1 in MCF7 and LNCaP cells (Fig. 2).
RT-qPCR confirmed that CCL20, CD68, IL-8, p62, and Zeb1 are basally high MDA-MB-231, BT549, PC3, and DU145 cells (Fig. 2). RT-qPCR and western blot confirmed that SOX9 mRNA (Fig. 2) and protein (Fig. 3A) are basally high in MDA-MB-231 and PC3 cells and SOX9 protein is basally high in BT549 (Fig. 3A). However, SOX9 mRNA is not basally high in BT549 or DU145 (Fig. 2). RT-qPCR confirmed that CXCR7 is basally low in MDA-MB-231, BT549, PC3, and DU145 cells, MMP16 is basally low in MDA-MB-231, PC3, and DU145 cell lines, and CDK2 and PLK1 are basally low in BT549 cells (Fig. 2). However, inconsistent with RNA-seq, CDK2 and PLK1 are not basally low in MDA-MB-231, PC3, and DU145 cells by RT-qPCR and MMP16 is basally high in BT549 (Fig. 2). Cell line differences among the HR- cell lines likely reflect the inconsistencies we observed for BT549 and DU145 cell lines, which were not sequenced. In addition, RNA sequencing can be more sensitive than RT-qPCR analysis of gene expression, which may also account for inconsistency. For example, MDA-MB-231 and PC3 RNA sequencing show much lower basal expression for CXCR7 (log2 FC = -7.61 and -9.02) and MMP16 (log2 FC = -2.33 and -2.70) than for CDK2 (log2 FC = -0.86 and -1.29) and PLK1 (log2 FC = -0.22 and -1.43) (Supplemental Table 1, Tab 1).
Taken together, of the 10 arbitrary genes we picked for validation of the LNCaP, MCF7, PC3 and MDA-MB-231 RNA-seq results, 10 genes were validated by RT-qPCR in LNCaP cells, 8 genes were validated in MCF7, MDA-MB-231, BT549, and PC3 cells, and 7 genes were validated in DU145 cells. Taken together, our RT-qPCR of arbitrarily selected genes provides additional evidence of the 350 gene set expression patterns identified in MCF7, LNCaP, MDA-MB-231, and PC3 cells by RNA-seq.
p62 and SOX9 are upregulated in hormone receptor-independent LNCaP and MCF7 cells. Given that IL-1 reduces hormone receptors concomitant with the upregulation of pro-survival and pro-tumorigenic genes, such as p6220,23,26,27,32,62,63, we hypothesized that IL-1 can promote resistance to hormone receptor-targeted therapy. Enzalutamide is an anti-androgen therapeutic that inhibits AR activity and fulvestrant is an anti-estrogen therapeutic that inhibits ERa activity. Enzalutamide64 and fulvestrant65 are used to treat AR+ PCa and ERa+ BCa patients, respectively. However, patients can develop treatment resistance. Therefore, labs have produced publicly available data sets investigating changes in gene expression patterns in treatment-resistant cells to better understand resistance mechanisms and identify alternative therapeutic targets.
We compared our 350 gene set to published RNA-seq data from enzalutamide-resistant LNCaP3 sublines (GSE99381; APIPC subline versus APIPC_P (parental), Supplemental Table 1, Tab 4) or fulvestrant-resistant MCF766 sublines (GSE74391; ICI182R1 or ICI182R6 subline versus fulvestrant-sensitive subline (parental), Supplemental Table 1, Tabs 2 & 3). Among the select upregulated genes we chose for RT-qPCR confirmation, p62 and SOX9 were also upregulated in both treatment-resistant subline models (Supplemental Table 1, Tabs 2-4). Downregulation of AR or ERa/PR and target gene expression (e.g. AR target gene, KLK3) in the LNCaP3 or MCF766 treatment-resistant sublines (Supplemental Table 1, Tabs 2-4) indicates these models evolved to survive without hormone receptor activity. Thus, p62 or SOX9 may promote survival in BCa and PCa cells that lose hormone receptor accumulation in response to IL-1, as well as promote survival in BCa and PCa cells that intrinsically lack hormone receptors.
p62 and SOX9 are cytoprotective for HR- BCa cell lines. To determine if p62 or SOX9 are required for viability in hormone receptor-independent cells, we siRNA-silenced p62 or SOX9 in IL-1-treated HR+ BCa and PCa cell lines and in HR- cells lines that intrinsically lack hormone receptor accumulation. AR+ LNCaP PCa cells and ERa+ MCF7 BCa cells were first transfected with p62 or SOX9 siRNA and the following the day, treated with vehicle control, 25 ng/ml IL-1a, or 25 ng/ml IL-1b for 3 days (LNCaP, MCF7) or 5 days (MCF7). siRNA efficacy was confirmed by RT-qPCR or western blot and cells were assayed for viability using MTT (Supplemental Fig. 1). siRNA-mediated loss of p62 or SOX9 reduced LNCaP viability 10-25% in the presence of IL-1a or IL-1b, but had no effect on cell viability under control growth conditions (Supplemental Fig. 1A). p62 or SOX9 siRNA had no effect on MCF7 viability in response to IL-1, nor under control growth conditions (Supplemental Fig. 1B & C).
While p62 or SOX9 down regulation had little to no effect on cell viability in IL-1-treated LNCaP and MCF7 cells, repression of basal p62 or SOX9 was significantly cytotoxic for HR- BCa and PCa cell lines. MDA-MB-231, BT549, PC3, and DU145 HR- cell lines were transfected with p62 or SOX9 siRNA (Fig. 3B) and cell viability was determined on day 1 or day 3 using MTT (Fig. 3C) or by recording cell number on day 1, 2, and 3 (Fig. 3D). The loss of basal p62 or SOX9 reduced cell viability 20-60% in HR- BCa and PCa cell lines (Fig. 3C) and caused a >30% reduction in cell number by day 3 (Fig. 3D).
Taken together, HR- BCa and PCa cell lines have evolved a survival requirement for basal p62 or SOX9 function, while HR+ BCa and PCa cell lines are less dependent on p62 or SOX9 for survival. Thus, p62 and SOX9 may be upregulated by IL-1 in HR+ BCa and PCa cells to mediate other IL-1 tumorigenic functions. In kind, other genes in the 350 gene list may encode for cytoprotective proteins that protect against IL-1-induced hormone receptor repression.
Verteporfin is cytotoxic for HR- BCa and PCa cell lines. BCa or PCa cells that acquire hormone receptor independence or intrinsically lack hormone receptor activity are not susceptible to hormone receptor-targeting drugs, such as enzalutamide or fulvestrant. Therefore, alternative therapeutic targets are needed. The p62 inhibitor, verteporfin (Visudyneâ), is an FDA-approved, photosensitizing drug used with laser light to treat leaky blood vessels in the eye caused by macular degeneration. Currently, verteporfin is being tested in a Phase I clinical trial as a photosensitizer for the SpectraCure P18 photodynamic therapy system for recurrent PCa (NCT03067051). Recently, verteporfin, alone, was shown to reduce the tumor growth of subcutaneous prostate epithelial xenografts overexpressing p62 and verteporfin was able to reduce the growth of PC3 xenografts20. Verteporfin oligomerizes p62 (Fig. 4A & B; Supplemental Fig. 2A), thereby, preventing p62 interaction with binding partners and inhibiting p62 function20,67. Given that the HR- BCa and PCa cell lines demonstrated significant dependency on p62 for survival, we treated HR- BCa and PCa cell lines with 2.5, 5, or 10 mM verteporfin for 3 days and assayed cells for viability using MTT (Supplemental Fig. 2B). HR- BCa cells were also treated with 10 mM verteporfin for 5 days and assayed cell viability using MTT (Fig. 4C) or by recording cell number on day 1, 3, and 5 (Fig. 4D). Verteporfin is cytotoxic for HR- BCa and PCa cell lines and, thus, may be a rationale therapeutic alternative.
The 350 gene signature maps to pro-tumorigenic processes in BCa and PCa cells. We used the IPA Canonical Pathways module to identify signaling pathways represented in our 350 gene set and found that the gene set encodes for proteins that are predicted to activate inflammatory signaling, including interferon (z-score = 3.2, -log p-value = 9.86E+00), IL-1 (z-score = 2, -log p-value 1.26E+00), and IL-8 signaling (z-score = 1.7, -log p-value = 4.10E+00) (Supplemental Table 1, Tab 5). In addition to IPA, we also used the web-based gene ontology tool, GOrilla (http://cbl-gorilla.cs.technion.ac.il/) to perform gene enrichment analysis for biological processes represented in the 350 gene set. Similar to IPA, GOrilla also reports that the 350 gene set is enriched in genes that encode for proteins involved in inflammatory signaling including defense response (enrichment score = 3.83, p-value = 8.94E-21), immune system process (enrichment score = 2.55, p-value = 8.42E-19), and interferon signaling (enrichment score = 14.23, p-value = 7.92E-17) (Supplemental Table 1, Tab 12).
We also used the IPA Regulator Effects module to predict cancer-specific networks encoded by our 350 gene set and we selected the networks in which both p62 and SOX9 were target molecules. We found three such networks in which upstream regulators CTNNB1 (z-score 2.738, p-value 5.12E-07), FGF2 (z-score 2.091, p-value 2.21E-09) and TNF (z-score 7.192, p-value 4.87E-28) were found to be activated and predicted to promote neoplasia of cells (z-score 2.699, p-value 5.20E-12 (CTNNB1)) and malignancy (z-score 2.251, p-value 1.41E-13 (FGF2, TNF)) (Supplemental Table 1, Tab 6-9). Thus, the IL-1-conferred 350 gene set encodes multiple different pro-tumorigenic signaling pathways conserved among multiple different regulators.
Predicted p62 and SOX9 target molecules and functional networks in the 350 gene signature. Finally, given that siRNA-mediated loss of p62 or SOX9 are cytotoxic for HR- BCa and PCa cells, we used the IPA Upstream Regulators module to identify p62 or SOX9 predicted target molecules and functional networks in the 350 gene data set. p62 is predicted to induce CXCL2, IL15RA, IRF1, PLAT, RGCC, and RSAD2 expression (z-score = 2.449, p-value = 8.38E-05) (Fig. 5A; Supplemental Table 1, Tab 6) and predicted to activate HIF1A, NFkB1, NFKBIA, NFkB (complex), RELA, and SQSTM1(p62) signaling (Fig. 5C; Supplemental Table 1, Tab 6). p62 is also predicted to function upstream of active IL-1b signaling and upstream of attenuated NR3C1 function; but IPA analysis of the 350 gene set did not predict if p62 promotes or prevents the activation of IL-1b signaling or the inhibition NR3C1 function (Fig. 5C). Notably, we show that IL-1 induces p62 expression and, as stated earlier, p62 binds and poly-ubiquitinates TRAF6, leading to NFkB transactivation47,48. NFkB is the canonical mediator of IL-1 inflammatory signaling68 and IL-1b is an NFkB target gene69. Thus, p62 functions in a positive feedback loop to induce IL-1b production and signaling. NR3C1 encodes the glucocorticoid nuclear receptor which represses inflammatory gene expression, including IL-1-regulated genes70. Thus, p62 is expected to participate in the cross-talk between IL-1 and glucocorticoid signaling and, in turn, promote NR3C1 inhibition. In kind, the p62-regulated genes and interactive networks are known mediators of inflammatory signaling and immunity (CXCL2, IL15RA, IRF1, RSAD2, IL-1b, NFkB, NR3C1, RELA, SQSTM1(p62)), as well as hypoxia (HIF1A), fibrinolysis (PLAT), and cell cycle regulation (RGCC). SOX9 did not appear as a master regulator in IPA; therefore, we manually extracted its interactome from IPA and overlapped the results with the 350 gene list to identify relevant interactions. SOX9 is predicted to activate VNN1 expression and mediate FN1 and T-Cell Factor (TCF) signaling (Fig. 5B). VNN1 functions in inflammation and immunity, FN1 promotes wound healing, and TCF interacts with b-catenin to mediate WNT signaling. Disruption of any of the processes that p62 or SOX9 are predicted to regulate or mediate as part of the 350 gene signature would be expected to reduce cell viability and may explain p62 or SOX9 siRNA-mediated cytotoxicity in BCa and PCa cells.