Chronic IL-1 Exposed AR+ PCa Cell Lines Show Conserved Loss of IL-1 Sensitivity and Evolve Both Conserved and Unique Differential Gene Expression Profiles

Introduction: Inflammation drives prostate cancer (PCa) progression. While inflammation is a cancer hallmark, the underlying mechanisms mediating inflammation-induced PCa are still under investigation. Interleukin-1 (IL-1) is an inflammatory cytokine that promotes cancer progression, including PCa metastasis and castration resistance. We previously found that acute IL-1 exposure represses PCa androgen receptor (AR) expression concomitant with the upregulation of pro-survival proteins, causing de novo accumulation of castration-resistant PCa cells. However, acute inflammation is primarily anti-tumorigenic, while chronic inflammation is pro-tumorigenic. Thus, using the LNCaP PCa cell line as model, we found that PCa cells can evolve insensitivity to chronic IL-1 exposure, restoring AR and AR activity and acquiring castration resistance. In this paper we expanded our chronic IL-1 model to include the MDA-PCa-2b PCa cell line to investigate the response to acute versus chronic IL-1 exposure and to compare the gene expression patterns that evolve in the LNCaP and MDA-PCa-2b cells chronically exposed to IL-1. Methods: We chronically exposed MDA-PCa-2b cells to IL-1α or IL-1β for several months to establish sublines. Once established, we determined subline sensitivity to exogenous IL-1 using cell viability assay, RT-qPCR and western blot. RNA sequencing was performed for parental and subline cells and over representation analysis (ORA) for geneset enrichment of biological process/pathway was performed. Results: MDA-PCa-2b cells repress AR and AR activity in response to acute IL-1 exposure and evolve insensitivity to chronic IL-1 exposure. While cell biological and molecular response to acute IL-1 signaling is primarily conserved in LNCaP and MDA-PCa-2b cells, including upregulation of NF-κB signaling and downregulation of cell proliferation, the LNCaP and MDA-PCa-2b cells evolve conserved and unique molecular responses to chronic IL-1 signaling that may promote or support tumor progression. Conclusions: Our chronic IL-1 subline models can be used to identify underlying molecular mechanisms that mediate IL-1-induced PCa progression.


Introduction
Prostate cancer (PCa) is the second leading cause of cancer-related death in American men behind lung cancer [1]. Metastatic PCa is responsible for this high burden of mortality, as the five-year survival rate drops from almost 100% for PCa confined to the prostate to 30% for metastatic disease [1]. Metastatic PCa is currently incurable, and with the annual burden of metastatic PCa expected to rise 42% by 2025 [2], the need to understand the underlying mechanisms of PCa disease initiation and progression cannot be overstated.
As part of immunosurveillance, immunes cells infiltrate the tumor microenvironment and secrete cytotoxic and cytostatic inflammatory cytokines [3,4]. Thus, acute inflammation is anti-tumorigenic, but if left unresolved, chronic inflammation ensues [3,4]. Chronic inflammation promotes cancer [3,4], including PCa initiation and progression [5]. While chronic inflammation in the tumor microenvironment is an established hallmark of cancer [6], the underlying mechanisms of inflammation-induced disease are not fully elucidated.
In this study, we compared our RNA sequencing and bioinformatics analysis of LNCaP and MDA-PCa-2b sublines isolated from cultures chronically exposed for several months to IL-1 to identify the conserved and divergent gene expression changes and predicted pathways that are known to promote tumorigenicity, and thus, could underlie chronic IL-1-induced PCa progression and serve as novel therapeutic targets. Both LNCaP and MDA-PCa-2b sublines evolved insensitivity to exogenous IL-1-induced cytotoxicity, cytostaticity, and intracellular signaling and show conserved basal gene expression in a relatively small subset of genes that encode for proteins implicated in cancer, including PCa. LNCaP and MDA-PCa-2b chronic IL-1 sublines share 55 basally high genes and 17 basally low genes. Many of the shared genes reportedly have functional roles in castration resistance and/or correlate with PCa disease progression. Notably, while constitutive NF-κB signaling, a canonical IL-1-induced intracellular pathway, promotes PCa castration resistance [20,21], the LNCaP and MDA-PCa-2b sublines do not show constitutive upregulation of canonical NF-κB target genes, suggesting that chronic IL-1 can drive NF-κB-independent PCa disease. Taken together, chronic IL-1 exposure selects for cytotoxic-and cytostatic-resistant cells that evolve a gene expression profile favoring disease progression.

Chronic IL-1 subline generation and maintenance
MDA-PCa-2b cells were maintained in HPC1/20% FB Essence (FBE) containing 0.5 ng/ml IL-1α (Gold Bio, St. Louis, MO; 1110-01A-10) or IL-1β (Gold Bio, St. Louis, MO; 1110-01B-10) for approximately 4 months; and the proliferative colonies that emerged were expanded and termed MDA-PCa-2b IL-1α subline (MDA-αs) and MDA-PCa-2b IL-1β subline (MDA-βs). We generated two sets of sublines (MDA-αs1, MDA-αs2, MDA-βs1, MDA-βs2). During subline expansion, the sublines were maintained in 0.5 ng/ml IL-1 up to an additional 4 months. During subline generation and expansion, MDA-PCa-2b parental cells were cultured in vehicle control (0.1% BSA) in 1 × PBS) alongside the sublines. Similar passage number for parental and subline cells were characterized. For phenotypic characterization, established sublines are cultured in the absence of IL-1 for at least one month to ensure the sublines are stable. Cell line authentication was previously performed for LNCaP parental, LNas1, LNbs1 [16]. Cell line authentication was performed for MDA-PCa-2b parental 1, MDA-αs1 and MDA-βs1 was performed by STR profiling by the DNA Genotyping Core, UTSW Medical Center. All cell lines used in the study are commercially available and purchased from American Tissue Culture Collection (ATCC). We hereby confirm that none of the used cell lines require any ethics approval for their use.

RNA isolation and reverse transcription quantitative PCR (RT-QPCR):
Total RNA was extracted, reverse transcribed, and analyzed by RT-QPCR as previously described [19]. Primer sequences for genes of interest are listed below. Gene of interest cycle times (CT) were normalized to the β-actin. Relative mRNA levels were calculated using the 2-ΔΔCT method. Primer

Over representation analysis
Over representation analysis (ORA) was performed using the WEB-based Gene SeT AnaLysis Toolkit (WebGestalt) site (http://www.webgestalt.org/) [28]. Gene lists were analyzed using the geneontology biological process functional database. The parameters for the enrichment analysis are minimum/maximum genes in a category = 5/2000, FDR method = BH, number of categories to report = top 40. Induced and repressed gene lists were uploaded separately to identify, respectively, upregulated and downregulated biological processes.

Western blot
Protein was isolated from cells using NP40 lysis buffer (0.

Statistical analysis
Statistical significance for MTT and cell counts was determined using unpaired student t-test calculated using Microsoft Excel. P-values of ≤ 0.05 were considered statistically significant and denoted by asterisks (*p ≤ 0.05; **p ≤ 0.05; ***p ≤ 0.005). Graphs are shown as the average of a minimum of n = 3 biological replicates ± standard deviation (STDEV).
To demonstrate that the MDA-PCa-2b chronic IL-1 sublines acquired insensitivity to IL-1, the parental and subline cells were given an acute IL-1 exposure of 25 ng/ml IL-1 for 3 or 6 days. Acute IL-1 exposure reduced cell viability ( Figure 1A) and proliferation ( Figure  1B), reduced full-length PARP (indicative of cell death activation), and upregulated the protein accumulation of the canonical IL-1-induced genes, Super Oxide Dismutase (SOD2) and Lipocalin 2 (LCN2) ( Figure 1C). Acute IL-1 also represses AR levels and activity in AR + PCa cell lines [19,30]. As such, acute IL-1 treatment repressed AR and NKX3.1 (canonical AR-induced gene) protein accumulation in the parental cells, but not in the subline cells ( Figure 1C). Finally, while acute IL-1 exposure represses AR accumulation and activity in MDA-PCa-2b parental cells, as observed for LNCaP [16], the chronically exposed subline cells have restored AR accumulation and activity ( Figure 1C). Taken together, these results demonstrate that chronic IL-1 exposure selects for MDA-PCa-2b cells that evolve insensitivity to exogenous IL-1. Of note, mRNA levels of the IL-1 receptor, IL-1R1, are not induced by IL-1 in parental cells (Figure 2A), indicating that basal IL-1R1 levels are sufficient to mediate IL-1 intracellular signaling in parental cells. Furthermore, IL-1 does not show a differential effect on IL-1R1 mRNA levels in parental versus subline cells ( Figure  2A) and IL-1R1 mRNA levels are comparable in parental versus subline cells ( Figure 2B), suggesting subline insensitivity to exogenous IL-1 is independent of IL-1R1 levels.  Figure 1A). We performed gene set ORA against the Gene Ontology biological process database using WebGestalt functional enrichment analysis web tool [28]. IL-1 is a cytotoxic and cytostatic inflammatory cytokine that signals through NF-κB [16,29,31,32]; thus, as expected, ORA of the shared upregulated genes predicts that cytokine and immune response pathways, including NF-κB signaling, are activated by acute IL-1 exposure in LNCaP and MDA-PCa-2b cells and ORA of the shared repressed genes predicts down regulation of the cell cycle and DNA, RNA, and protein biosynthesis (Supplementary Table 1
Of note, among the independently generated MDA-PCa-2b chronic IL-1 sublines, the basal mRNA levels of the select confirmed genes is variable. We believe this is due to clonal heterogeneity and genetic drift during IL-1 chronic subculturing. Nevertheless, we see conserved basally high mRNA levels for the selected genes which we will functionally investigate in future studies.
We performed ORA for the 55 basally high and 17 basally low genes which predicted several constitutively upregulated or repressed pathways in both the LNCaP and MDA-PCa-2b chronic IL-1 sublines (Supplementary Table 1; Supplementary Figure 2B). For example, ORA of the basally high gene set predicts that the chronic IL-1 sublines evolved constitutive expression of genes that regulate insulin signaling and control glucose homeostasis and ORA of the basally low gene set predicts that the chronic IL-1 sublines evolved constitutive repression of genes that regulate abiotic stress response (Supplementary Table 1; Supplementary Figure 2 (Figures 1 and 3), the broader response to chronic IL-1 exposure is unique among the cell lines. The divergent response to chronic IL-1 likely reflects genetic differences in LNCaP and MDA-PCa-2b cell lines, as could be expected for tumor cell response to chronic inflammation among different patients. Taken together, our models reveal potential therapeutic targets for individuals whose PCa disease progression evolves from chronic inflammation.

Discussion
We have developed novel models to explore the effect of chronic inflammation on PCa progression. Specifically, we generated LNCaP [16] and MDA-PCa-2b sublines from cultures chronically exposed to IL-1 for several months. Given that IL-1 is cytotoxic and cytostatic [16,29] (Figure 1), chronic IL-1 exposure selected for viable, proliferative PCa cells that evolved IL-1 insensitivity. Interestingly, we find that castration-resistant cell lines have reduced or no sensitivity to IL-1-induced cytotoxicity and cytostaticity [16,29] (data not shown) and we and other labs have found that chronic IL-1 exposure can select for castration-resistant PCa cells [15,16]. Chronic IL-1 sublines can also develop resistance to other cytotoxic and cytostatic inflammatory cytokines [16]. Therefore, we speculate that chronic IL-1 exposure in the PCa tumor microenvironment imposes selective pressure on tumor cells to acquire constitutive pro-survival mechanisms that provide a growth advantage against cellular stresses, such as castration or acute cytostatic and cytotoxic inflammation. Thus, we can use our subline models to elucidate strategies that PCa cells elicit to escape inflammatory cytokine toxicity and growth inhibition.
While both LNCaP and MDA-PCa-2b cell lines evolve insensitivity to chronic IL-1 exposure, the underlying mechanism of insensitivity remains unknown. LNCaP chronic IL-1 sublines also develop insensitivity to TNFα [16] which also signals through NF-κB [31], suggesting that the resistance mechanism is downstream of the IL-1/IL-1R1 receptor/ligand interaction in LNCaP cells. IL-1R1 mRNA levels are comparable in LNCaP parental and sublines cells [16] and comparable in MDA-PCa-2b parental and subline cells (Figure 2), implying that the IL-1 resistance mechanism is independent of IL-1R1 levels. However, it will be important to determine efficacy of the IL-1/IL-1R1 protein-protein interaction in MDA-PCa-2b cells and given the different genetic backgrounds of LNCaP and MDA-PCa-2b cell lines, it is possible that the mechanism(s) governing IL-1 insensitivity in the LNCaP versus MDA-PCa-2b chronic IL-1 sublines are not conserved. We will use our subline models to determine how cells can acquire resistance to chronic IL-1 exposure. Uncovering the mechanisms may help mitigate chronic IL-1-induced castration resistance or PCa progression.
The LNCaP and MDA-PCa-2b chronic IL-1 sublines share a select basal gene expression profile of 55 basally high and 17 basally low genes (Supplementary Table 1). Significantly, many of these genes have reported roles in cancer and may be critical for inflammationinduced PCa progression. For example, the shared basally high genes that we confirmed by RT-qPCR, CTAG2, FSTL1, HPGD and TARP are pro-tumorigenic. CTAG2 (cancer-testis antigen-2) encodes for the autoimmunogenic tumor antigen, LAGE-1, expressed by cancer cells, including PCa [47]. Functionally, CTAG2 was shown to promote breast cancer cell invasion [48]. Phase 1 clinical trial results are pending for a combined NY-ESO-1/LAGE-1 vaccine for patients with androgen-independent metastatic PCa (NCT00616291). Metastatic castration resistant PCa (mCRPC) patients give an NY-ESO-1 (CTAG1) peptide vaccine in a Phase 1 clinical trial showed slowed PSA doubling time [38], thus, the combined NY-ESO-1/LAGE-1 vaccine is expected to be efficacious. FSTL1 (Follistatin-like protein 1) is a proinflammatory glycoprotein [49] that has been shown to promote invasion and metastasis in colorectal cancer [50] and chemoresistance and stemness in breast cancer [51]. FSTL1 is an AR-induced gene and is elevated in CRPC [39]. HPGD (15-Hydroxyprostaglandin dehydrogenase) is an anti-inflammatory enzyme that catabolizes prostaglandins [52]. HPGD is an AR-induced gene [40] and is elevated in primary and metastatic PCa tissue [41]. HPGD siRNA silencing reduced LNCaP cell viability in vitro [41]. TARP (T-cell receptor gamma alternate reading frame protein) is an autoimmunogenic tumor antigen expressed by PCa cells [42]. TARP is an AR-induced gene [43] and TARP overexpression in the PC3 PCa cell line reduced IL-1b expression and enhanced cell proliferation [44]. A Phase 1 clinical trial showed that TARP peptide vaccination slows PSA doubling time and tumor growth rate in Stage D0 PCa patients with biochemical PSA recurrence (NCT00972309) [45]. Taken together, the conserved molecular response to chronic IL-1 exposure may have functional significance in PCa tumorigenicity which we can elucidate using our chronic IL-1 cell line models.
Indeed, the individual genes modulated by chronic IL-1 exposure that reportedly function in PCa disease can be explored as novel therapeutic targets, but it will also be important to interrogate the signaling networks that integrate the global changes in chronic IL-1-induced gene expression, including those that are unique to the different cell backgrounds. Our ORA bioinformatic analysis provides putative tumor promoting pathways to investigate that would be acquired during chronic IL-1 exposure, such as insulin and interferon gamma signaling, glucose homeostasis and metabolism and cell adhesion.
The basal changes in gene expression in parental versus chronic IL-1 subline cells suggests that chronic IL-1 exposure drives constitutive genetic or epigenetic changes over time.
LNCaP and MDA-PCa-2b intracellular signaling, gene expression patterns, ORA-predicted cell biological processes, and cell behavior in response to acute IL-1 exposure are conserved, but the gene expression patterns and ORA-predicted cell biological processes diverge as the different cell lines evolve insensitivity to chronic IL-1 signaling. This suggests that genetic and epigenetic reprogramming in response to chronic IL-1 exposure would evolve differently in different patients. Therefore, we can begin to uncover the various ways in which chronic IL-1 exposure remodels chromatin using our chronic IL-1 subline models in the different cell line backgrounds. Studies are underway to elucidate chronic IL-1-induced chromatin remodeling, the subsequent constitutive changes in signaling networks, and the contribution of these altered signaling networks to PCa progression.
As stated, while LNCaP and MDA-PCa-2b cell lines both develop insensitivity to chronic IL-1 exposure, the two cell line backgrounds evolved primarily disparate changes in gene expression, likely due to underlying genetic differences in the cell line backgrounds. Notably, however, chronic exposure to IL-1α or IL-1β both cause similar gene expression changes within a cell line background, indicating that a conserved mechanism leads to chronic IL-1 insensitivity and changes in intracellular signaling within the cell line. Therefore, we can identify master regulators of the chronic IL-1 molecular response downstream of the IL-1 ligand/receptor interaction using our subline models.
Although the LNCaP and MDA-PCa-2b chronic IL-1 sublines have lost sensitivity to IL-1induced intracellular signaling, the sublines are still sensitive to other stimuli such as IL-6 or serum starvation (data not shown). Thus, the context of the tumor microenvironment and conventional therapy will be very important in determining the cell biological and physiological consequence of IL-1-insensitive PCa cell subpopulations in PCa progression. Those studies are underway.

Conclusion
Taken together, we propose the following model ( Figure 5) wherein acute IL-1 exposure is anti-tumorigenic, reducing cell viability and proliferation and repressing AR and AR activity in androgen dependent PCa cells. If the acute inflammation is left unresolved, the PCa cell subpopulation that survives the initial acute IL-1 assault evolves insensitivity to chronic IL-1 exposure and emerges more fit to contribute to PCa tumor growth and progression. It remains unknow if the conserved loss of sensitivity to chronic IL-1 exposure among PCa cell lines is due to the same or different intracellular mechanism; but the culmination of epigenetic and gene expression changes that result from chronic IL-1 exposure appear to diverge among PCa cell lines and likely reflects the inherent genomic difference among cell lines. To this end, while AR and AR activity are restored in both LNCaP [16] and MDA-PCa-2b chronic IL-1 sublines (Figure 1), subsequent AR dependency and therapeutic response may differ based on the differing epigenetic and gene expression landscapes that evolve. Thus, we can take advantage of our current cell line models and expand them to patient derived tumor models to understand how chronic IL-1 inflammation can affect PCa disease and treatment in individuals.

Supplementary Material
Refer to Web version on PubMed Central for supplementary material. cells, but has little to no effect on the chronic IL-1 sublines. Thus, the IL-1 sublines evolved insensitivity to IL-1. Error bars, ± STDEV of 4 biological replicates; p-value, *≤ 0.05, **≤ 0.005, ***≤ 0.005, NS = not significant. Fold MTT optical density (OD) is normalized to treatment control. β-actin is the western blot loading control.