Effect of non-steroidal anti-inflammatory drugs (aspirin and naproxen) on inflammation-associated proteomic profiles in mouse plasma and prostate during TMPRSS2-ERG (fusion)-driven prostate carcinogenesis

Abstract

Recent pre-clinical studies have shown that the intake of non-steroidal anti-inflammatory drugs (NSAIDs) aspirin and naproxen could be an effective intervention strategy against TMPRSS2-ERG fusion-driven prostate tumorigenesis. Herein, as a follow up mechanistic study, employing TMPRSS2-ERG (fusion) positive tumors and plasma from TMPRSS2-ERG. Pten^flox/flox mice, we profiled the stage specific proteomic changes (focused on inflammatory circulating and prostate tissue/tumor-specific cytokines, chemokines, and growth factors/growth signaling-associated molecules) that contribute to prostate cancer (PCa) growth and progression in the TMPRSS2-ERG fusion-driven mouse model of tumorigenesis. In addition, the association of the protective effects of NSAIDs (aspirin 1400ppm and naproxen 400ppm) with the modulation of these specific molecular pathways was determined. A sandwich Elisa based membrane array-proteome profiler identifying 111 distinct signaling molecules was employed. Overall, the plasma and prostate tissue sample analyses identified 54 significant and differentially expressed cytokines, chemokines, and growth factors/growth signaling-associated molecules between PCa afflicted mice (TMPRSS2-ERG. Pten^flox/flox, age-matched non-cancerous controls, NSAIDs-supplemented and no-drug controls). Bioinformatic analysis of the array outcomes indicated that the protective effect of NSAIDs was associated with reduced expression of a) tumor promoting inflammatory molecules (M-CSF, IL-33, CCL22, CCL12, CX3CL1, CHI3L1, and CD93), b) growth factors- growth signaling-associated molecules (Chemerin, FGF acidic, Flt-3 ligand, IGFBP-5, and PEDF), and c) tumor microenvironment/stromal remodeling proteins MMP2 and MMP9. Overall, our findings corroborate the pathological findings that protective effects of NSAIDs in TMPSS2-ERG fusion-driven prostate tumorigenesis are associated with anti-proliferative and anti-inflammatory effects and possible modulation of the immune cell enriched microenvironment.

1. INTRODUCTION

Prostate cancer (PCa) is the second leading cause of cancer-related mortality in the American male population.¹ TMPRSS2-ERG gene fusion is the most common somatic gene rearrangement found in nearly 50% of PCa cases.² The fusion of the ETS transcription factor coding region to an androgen-responsive promoter element of the transmembrane serine protease 2 (TMPRSS2) leads to overexpression of the ETS transcription factor, which can then turn on the expression of target genes associated with PCa initiation, growth and progression.² Thus, these identified gene fusions in PCa, specifically TMPRSS2-ERG, can be considered as a target for prevention/treatment strategies against this malignancy.³ Importantly, epidemiological studies have indicated that aspirin [a non-steroidal anti-inflammatory drug (NSAID)] use is associated with reduced risk of developing TMPRSS2-ERG (fusion) positive tumors.⁴ This observation was supported by our recent pre-clinical studies, wherein using a fusion-driven PCa mice model (TMPRSS2-ERG. Pten^flox/flox) we identified that indeed NSAIDs, aspirin and naproxen, have strong efficacy in reducing prostate tumorigenesis in the TMPRSS2-ERG driven PCa model but not in the non-TMPRSS2-ERG driven (Hi-Myc^+/− mouse) PCa model.⁵

We also established in the same study that the NSAIDs primarily affect proliferation in the TMPRSS2-ERG fusion driven cancers, which could possibly stem from the NSAIDs-mediated decrease in androgen receptor (AR) expression.⁵ This is because of the androgen sensitivity of the TMPRSS2 promoter; specifically, after TMPRSS2-ERG fusion, the androgen-bound AR binds to TMPRSS2 regions, and results in ERG overexpression which in turn causes overexpression/activation of proliferative pathways.⁶ Another important factor that cannot be overlooked is that cancer progression is a complex process; many pathways associated with cancer initiation and progression are similar to inflammation and wound healing.⁶ Tumor inflammation is characterized by the presence of inflammatory cells and cytokines/chemokines in the tumor microenvironment, which contributes to tumor growth and progression, immunosuppression, and chemoresistance.⁷ Cytokines, chemokines, and growth factors secreted by prostate tumor cells, stromal cells, and inflammatory cells in the tumor microenvironment promote the epithelial to mesenchyme transition (EMT), angiogenesis, and cancer metastasis.⁸ Given that immune infiltration and associated inflammatory features were observed in the TMPRSS2-ERG. Pten^flox/flox model, there seems to be a strong possibility that the NSAIDs due to their inherent anti-inflammatory potential have specifically more protective effect against TMPRSS2-ERG fusion driven PCa.⁶

NSAIDs are potent and cost-effective interventions against inflammation caused by tissue damage/injury.⁹ NSAIDs inhibit the activity of cyclooxygenase (COX)/prostaglandin G/H-synthase enzyme that catalyzes the synthesis of prostaglandins from arachidonic acid.¹⁰ COX1 and COX2 are two distinct isoforms of prostaglandin synthase; COX1 is constitutively expressed in many tissues while COX2 expression is induced in the tissues during inflammation, wound healing, and tumor development.¹¹ COX-2 acts as an inflammatory trigger molecule and its overexpression has been reported in PCa; its higher/aberrant expression has been implicated in PCa growth/progression and poor prognosis.¹² Importantly, in our recently completed study, we observed a significant impact on both COX-1 and 2 pathways in the NSAIDs protective efficacy against TMPRSS2-ERG driven PCa compared to non-fusion driven PCa.⁵ However, the specific molecular and inflammatory pathological considerations that contributed to the differential effect of NSAIDs in TMPRSS2-ERG (fusion) and non-fusion-derived PCa could not be identified.⁵ Thus, in the present study, we assessed the inflammatory stage-specific circulating and prostate tissue-specific cytokines, chemokines, and growth factors/growth signaling-associated molecules that contribute to tumorigenesis in the TMPRSS2-ERG fusion-driven mouse model of PCa and then determined how the protective effects of NSAIDs against tumorigenesis in this model are associated with the modulation of these molecular pathways.

2. MATERIAL AND METHODS

2.1. Animal source of prostate tissue/tumor and plasma

Snap frozen prostate/tumor tissues and plasma (archived and stored at −80℃) from homozygous TMPRSS2-ERG. Pten^flox/flox male mice from recently reported animal model characterization and NSAID efficacy studies were used.⁶ Briefly in the characterization study, TMPRSS2-ERG. Pten^flox/flox mice were subjected to Cre-induction at 8 weeks of age with tamoxifen (TAM). As required, samples were harvested at sacrifice from either Cre-induced TMPRSS2-ERG. Pten^flox/flox (+TAM) mice as a function of mice age (12, 20, 28 weeks of age which relates to 4-, 12-, and 20-weeks post Cre-induction) or no-TAM mice (age-matched controls) fed a control powdered diet (modified AIN-76A: Envigo# TD 94096). Accordingly, in the NSAID efficacy studies, samples were harvested after TMPRSS2-ERG. Pten^flox/flox (+TAM) mice fed either control (TD 94096) or NSAID-supplemented powder diets (human equivalent doses: aspirin 1400ppm or naproxen 400ppm which equals to 1.4 g of aspirin or 0.7 g of naproxen in 1 kg of powder TD 94096 diet) from one week after Cre-induction till 28 weeks of age.⁵ Additionally, samples from eight-week-old TMPRSS2-ERG. Pten^flox/flox (no-TAM) mice were used as a reference point (baseline samples) for data analysis. For the present study, archived samples (tissue or plasma) were individually sourced from n=3 mice per group.

2.2. Proteome profiler array reagents and methodology

Proteome profiler-Mouse XL Cytokine Array Kit (#ARY028)-a membrane-based sandwich Elisa kit (with 111 different antibodies) was purchased from R&D Systems (Minneapolis, MN). Plasma was isolated by subjecting the blood collected in sodium heparin tubes to centrifugation at 5000 x g at 4°C for 20 min; supernatant was decanted and stored at −80°C till analysis. For preparing the protein lysates, the dorsolateral prostate tissue lobe/tumor was micro-dissected, snap frozen and stored at −80°C till analysis. For membrane-based array analysis, the above defined archived tissues were homogenized in lysis buffer [PBS containing protease inhibitors (5 μg/ml aprotinin, 10 μg/ml PMSF, 1 mM EDTA, 1 mM EGTA, 5 μg/mL sodium orthovanadate per ml of buffer) including triton x-100 (1%)] and subjected to 3 rounds of freeze-thaw/centrifugation cycles [13000 x g for 20 min at 4°C] and the supernatant was collected. Proteome profiling arrays were performed using 150 μl of plasma and 500 μg of protein lysate for each sample per membrane protein array according to manufacturer recommendations. Protein expression levels were visualized by chemiluminescence and captured on X-ray film. The protein spots on X-ray film were analyzed by Quick Spots Image Analysis Software (R&D Systems). Each molecular signal was normalized by negative control and reference spots. Eight-week-old TMPRSS2-ERG. Pten^flox/flox (no-TAM) mice were used as a reference point for baseline (BL-8) to calculate fold change in expression. Statistical data were analyzed either between baseline and mice in different age groups, or between TMPRSS2-ERG. Pten^flox/flox (+TAM) and no-TAM mice (age-matched controls) or between TMPRSS2-ERG. Pten^flox/flox (+TAM) mice and NSAID-supplemented (either aspirin or naproxen) TMPRSS2-ERG. Pten^flox/flox (+TAM) mice (as relevant). Unpaired, two-tailed Student’s t-test was performed to analyze differences between two groups. Data are reported as mean values ± SEM for experiments performed with at least three replicates, with p-value < 0.05 considered significant.

2.3. Protein-protein interactions and pathways analysis

A network of protein-protein interactions was generated using the STRING database to decipher the relationships among the differentially expressed cytokines, chemokines, and growth factors/growth signaling-associated molecules.¹³ Cytoscape software was used to visualize the network and protein with an interaction score of more than 0.4.¹⁴ Ingenuity Pathway Analysis (IPA) software (version 68752261, Qiagen Inc., CA, USA) was used to determine the relevant pathways and network. The significance of pathways was determined on the basis of enrichment of differentially expressed molecules using Fisher’s exact test, Benjamini-Hochberg corrected p-value, and Z-score. Z-score cutoff > 2 represents the significant activation of pathways, and <−2 shows significant inhibition. We have used the −log (P-value) >2 and Z-score cutoff >2 for canonical pathway, upstream regulators, and disease and biological function. Differentially expressed inflammatory molecules and growth factor protein data (unpaired, two-tailed Student’s t-test, p-value < 0.05) were used to identify significant canonical pathways, upstream regulators, and disease and biological function.

3. RESULTS

3.1. Profiling of differentially expressed cytokines, chemokines, and growth factors/growth signaling-associated molecules in plasma of TMPRSS2-ERG. Pten^flox/flox mice as a function of age of mice.

We performed a comparative stage-specific analysis of plasma cytokines, chemokines, and growth factors/growth signaling-associated molecules in TMPRSS2-ERG. Pten^flox/flox (+TAM) and no-TAM mice as a function of mice age (Fig. 1). Plasma from 12-, 20- and 28-weeks old TMPRSS2-ERG. Pten^flox/flox (+TAM) mice and no-TAM- mice (age-matched controls) was analyzed (Fig. 1 and Fig. S1A–C). Eight-week-old TMPRSS2-ERG. Pten^flox/flox (no-TAM) mice (BL-8) were used as a reference point to calculate the changes in the expression of the circulating molecules in the plasma. Membrane-based array analysis showed the presence of 62 molecules in the plasma of 12-, 20-, and 28-weeks old mice (Fig. 1A). Among these, 16 molecules from the plasma of TMPRSS2-ERG. Pten^flox/flox (+TAM) mice showed a change in expression levels as compared to the age-matched controls (Fig. S1A–C). We observed 8 growth factors/growth signaling-associated molecules [Angiopoietin-1, Chemerin, fibroblast growth factor (FGF) acidic, FMS-related tyrosine kinase 3 ligand (Flt-3 ligand), insulin-like growth factor binding protein (IGFBP)-1, IGFBP-5, LDL-R, and platelet-derived growth factor-BB (PDGF-BB)] that were markedly changed from the time of prostate tumor initiation to the development of adenocarcinoma (Fig. 1B–D, left panel). The plasma levels of Chemerin and IGFBP-5 were detected to be higher in TMPRSS2-ERG. Pten^flox/flox (+TAM) mice compared to no-TAM mice, irrespective of tumor stages (Fig. 1B–D, left panel and Fig. S1). Elevated Flt-3 ligand and IGFBP-1 plasma levels were observed in mice with adenocarcinoma [(28 weeks old (+TAM) mice] as compared to their age-matched controls; though, the expression varied in the lower age groups (Fig. 1B–D, left panel and Fig. S1). Other growth factors including, Angiopoietin-1, FGF acidic, low-density lipoprotein receptor (LDL-R), and PDGF-BB, showed decline in plasma expression levels with prostate tumor development in TMPRSS2-ERG. Pten^flox/flox (+TAM) mice (Fig. 1B–D, left panel and Fig. S1). Furthermore, we observed significant change in the expression of pro-inflammatory molecules, including cluster of differentiation (CD) 93, CC chemokine ligand (CCL)17, CCL22, CF-D, CX3C chemokine ligand (CX3CL)-1, CXC chemokine ligand (CXCL)-13, CXCL-16, and macrophage stimulating factor (M-CSF) in both TMPRSS2-ERG. Pten^flox/flox (+TAM) and no-TAM mice plasma at 12, 20, and 28 weeks of age compared to baseline; however, the relative expression varied between the groups (Fig. 1B–D, right panel and Fig. S1). CD93 is a cell surface protein expressed on several inflammatory cells, including B cells, macrophages, and neutrophils.¹⁵ Notably, we observed a higher level of CD93 in the plasma of 28 weeks old TMPRSS2-ERG. Pten^flox/flox (+TAM) mice compared to age-matched controls (Fig. 1B–D, right panel and Fig. S1). Chemokines (CCL22, CX3CL1, and CXCL16) were elevated in the plasma of TMPRSS2-ERG. Pten^flox/flox (+TAM) mice compared to age-matched controls (Fig. 1B–D, right panel and Fig. S1). The plasma levels of CCL17 and CXCL13 were reduced with PCa progression (Fig. 1B–D, right panel and Fig. S1). Interestingly, we observed a higher level of M-CSF (a secreted cytokine that causes monocyte cells to differentiate into macrophages)¹⁶ in plasma of TMPRSS2-ERG. Pten^flox/flox (+TAM) mice at 20 weeks of age [prostatic intraepithelial neoplasia (PIN) stage] and 28 weeks of age (adenocarcinoma stage) compared to the plasma of age-matched controls (Fig. 1B–D, right panel and Fig. S1).

Figure 1. Comparative stage-specific analysis of plasma cytokines, chemokines, and growth factors/growth signaling-associated molecules in TMPRSS2-ERG. Pten^flox/flox (+TAM) and no-TAM mice as a function of mice age.

Membrane-based array profiling was used to analyze the expression of plasma molecules. A) Heat map of cytokines, chemokines, and growth factors/growth signaling-associated molecules differentially expressed in TMPRSS2-ERG. Pten^flox/flox (+TAM) mice and no-TAM- mice plasma at 12, 20, and 28 weeks of age. Array based immunoblots depicting relative expression of (B, left panel) growth factors/growth signaling-associated molecules, and (B, right panel) cytokine, chemokines, and cell surface molecules in mice plasma as a function of mice age. Graphical representation of age-dependent fold changes in (C, left-right panel) TMPRSS2-ERG.Pten^flox/flox (no-TAM), and (D, left-right panel) TMPRSS2-ERG.Pten^flox/flox (+TAM) mice plasma. Eight weeks old non-Cre induced mice were used as reference point (baseline, BL-8) to calculate the fold change expression. Plasma samples from 12, 20, and 28 weeks of mice age relate to 4, 12, and 20 weeks post Cre-induction (tamoxifen-induced at 8 weeks of mice age). Quantified data is represented as Columns (mean for each group); bars represent SEM (n=3 mice/group). #, P < 0.01; $, P≤ 0.05

3.2. Effect of NSAID intervention on plasma cytokines, chemokines, and growth factors/growth signaling-associated molecules in TMPRSS2-ERG. Pten^flox/flox mice.

Next, we evaluated the effect of NSAIDs (aspirin 1400ppm and naproxen 400ppm)-supplemented diets on the plasma level of cytokines, chemokines, and growth factors/growth signaling-associated molecules in the TMPRSS2-ERG. Pten^flox/flox (+TAM) mice (Fig. 2). Both NSAIDs lowered the plasma levels of many cytokines/associated receptors such as Angiopoietin-like-3, TNF ligand superfamily member 13B (TNFSF13B), Cystatin-C, M-CSF, TNF receptor superfamily 11b (TNFRSF11B), receptor for advanced glycation end products (RAGE), and Resistin in the TMPRSS2-ERG. Pten^flox/flox (+TAM) mice (Fig. 2B). On the other hand, NSAID supplementation increased the plasma levels of Pentraxin-2 (Fig. 2B). Both NSAIDs also markedly lowered the plasma levels of chemokines (CCL22, CX3CL1, and CXCL16) in the TMPRSS2-ERG. Pten^flox/flox (+TAM) mice compared to no-drug controls (Fig. 2C). A similar inhibitory effect of NSAIDs was observed on the expression of membrane proteins (CD93, CD54, P-selectin, and CD104) (Fig. 2C). Furthermore, effect of NSAID-supplementation on alteration of growth factors/growth signaling-associated molecules was analyzed and a marked reduction in plasma levels of Chemerin, FGF acidic, Flt-3 ligand, Gas 6, IGFBP-3, IGFBP-5, retinol-binding protein 4 (RBP4), pigment epithelium-derived factor (PEDF), and Wnt1-inducible signaling protein-1 (WISP-1) was observed in TMPRSS2-ERG. Pten^flox/flox (+TAM) mice (Fig. 2D) as compared to no-drug controls. Both NSAIDs reduced the plasma levels of matrix metalloproteinase (MMP) 2 [associated with the modulation of extracellular matrix]¹⁷ and Lipocalin-2 (inflammatory adipokine), while only aspirin-1400ppm was able to significantly reduce the plasma levels of Periostin [associated with epithelial to mesenchymal transition (EMT) and migration]¹⁷ in TMPRSS2-ERG. Pten^flox/flox (+TAM) mice compared to the no-drug controls (Fig.S2A).

Figure 2. Effect of NSAID intervention on plasma cytokines, chemokines, and growth factors/growth signaling-associated molecules in TMPRSS2-ERG, Pten^flox/flox mice.

A) Heat map of cytokines, chemokines, and growth factors/growth signaling-associated molecules differentially expressed in plasma of TMPRSS2-ERG.Pten^flox/flox (+TAM), no-TAM, and NSAIDs (aspirin 1400ppm and naproxen 400ppm) treated TMPRSS2-ERG.Pten^flox/flox (+TAM) mice. (B-D, left panel) array based immunoblots depicting relative expression, and (B-D, right panel) graphical representation of fold changes in (B) cytokines, (C) chemokines and immune cell associated cell surface proteins, and (D) growth factors/growth signaling-associated molecules. Plasma from 28 weeks old mice (study end point) was used. Eight weeks old non-Cre induced mice were used as reference point (baseline, BL-8) to calculate the fold change expression. NSAID-supplementation was done from 8 till 28 weeks of mice age. Quantified data is represented as Columns (mean for each group); bars represent SEM (n=3 mice/group). $, P≤ 0.05; #, P < 0.01; *, P < 0.001

3.3. Profiling of differentially expressed cytokines, chemokines, and growth factors/growth signaling-associated molecules in prostate tumor tissues of TMPRSS2-ERG. Pten^flox/flox mice.

Next, we analyzed the prostate tissues in TMPRSS2-ERG. Pten^flox/flox (+TAM) and no-TAM mice (age matched controls) to assess the changes in the expression of cytokines, chemokines, and growth factors/growth signaling-associated molecules during tumorigenesis (Fig 3). Eight-week-old TMPRSS2-ERG. Pten^flox/flox (no-TAM) mice (BL-8) prostate tissues were used as a reference point to calculate the changes in the expression of these molecules in the prostate tissue; Fig. 3A represents the relative expression levels of these molecules. Several chemokines, including CCL6, CCL12, CX3CL1, CXCL1, CXCL2, CXCL16, and lipopolysaccharide-induced CXC chemokine (LIX)/(CXCL5), showed significantly increased expression in the prostate adenocarcinoma stage [28 weeks old TMPRSS2-ERG. Pten^flox/flox (+TAM)] mice compared to age-matched controls (Fig. 3B). The expression of other cytokines/associated receptors, including interleukin (IL)-1F3, IL-33, and TNFRSF11B were also upregulated in cancerous prostate tissues compared to the non-cancerous prostate tissues (age-matched controls) (Fig. 3C). CD93, CD14, and CD40 are reportedly expressed on the surface of macrophages, neutrophils, leukocytes, and tumor endothelial cells^15,18,19; analysis of these proteins showed increased expression in PCa tissue as compared to age-matched controls (Fig. 3C). The levels of growth factors/growth signaling-associated molecules including HGF, IGFBP-3, IGFBP-5, IGFBP-6, PEDF, VEGF were also significantly increased in PCa tissues compared to the non-cancerous prostate tissues (Fig. 3D). Though the relative expression of low-density lipoprotein receptor (LDLR) and RBP4 was significantly higher in TMPRSS2-ERG. Pten^flox/flox (+TAM) prostate tissues compared to non-cancer tissues, the overall expression was less. Extracellular matrix (ECM) remodeling is crucial for cancer cell migration and infiltration of immune cells into the tumor microenvironment.²⁰ Proteins associated with ECM modification, including MMP2 and MMP9, were also upregulated in PCa tissues of TMPRSS2-ERG. Pten^flox/flox (+TAM) mice compared to the non-cancerous tissues in the age-matched no-TAM mice (Fig.S2 B).

Figure 3. Comparative stage-specific analysis of prostate tissue cytokines, chemokines, and growth factors/growth signaling-associated molecules in TMPRSS2-ERG. Pten^flox/flox (+TAM) and no-TAM mice.

Membrane-based array profiling was used to analyze the expression of molecules in prostate tissues. A) Heat map of cytokines, chemokines, and growth factors/growth signaling-associated molecules differentially expressed in TMPRSS2-ERG. Pten^flox/flox (+TAM) mice and no-TAM- mice prostate at 28 weeks of age. (B-D, left panel) array based immunoblots depicting relative expression, and (B-D, right panel) graphical representation of fold changes in (B) chemokines, (C) cytokines and immune cell associated cell surface proteins, and (D) growth factors/growth signaling-associated molecules. Eight weeks old non-Cre induced mice were used as reference point (baseline, BL-8) to calculate the fold change expression. Dorso-lateral prostate tissue lysates from 28 weeks old mice (study end point) were used in the study [which relates to 20 weeks post Cre-induction (tamoxifen-induced at 8 weeks of mice age)]. Quantified data is represented as Columns (mean for each group); bars represent SEM (n=3 mice/group). $, P≤ 0.05; #, P < 0.01; *, P < 0.001

3.4. Effect of NSAID intervention on cytokines, chemokines, and growth factors/growth signaling-associated molecules in prostate tissue of TMPRSS2-ERG. Pten^flox/flox mice.

As recently reported, NSAID supplementation in diet (aspirin 1400ppm and naproxen 400 ppm) showed protective effects against TMPRSS2-ERG fusion-driven PCa.⁵ Therefore, we examined the impact of NSAID treatments on the prostatic/tumor expression of the cytokines, chemokines, and growth factors/growth signaling-associated molecules during prostate tumorigenesis (Fig. 4). Fig. 4A shows the expression profiles of these molecules in NSAID-fed TMPRSS2-ERG. Pten^flox/flox (+TAM) mice, no-drug controls (untreated prostate adenocarcinoma tissue) and no-TAM mice (age matched controls) relative to the baseline controls (BL-8). Aspirin treatment led to a decrease in prostatic expression of CCL12 compared to the no-drug controls. In contrast, naproxen treatment demonstrated a more substantial effect when compared to the aspirin dose. (Fig. 4B). Interestingly, both NSAIDs increased the prostatic expression of CXCL16 compared to untreated prostate adenocarcinoma tissue (no-drug controls). IL-33 interacts with the IL1RL1 receptor of helper T cells and mast cells and activates the production of type 2 cytokines.²¹ Both NSAIDs led to a significant decrease in the prostatic expression of IL-33 as compared to the no-drug controls (Fig. 4B). CD93 is a membrane-associated protein expressed on immune and endothelial cells.²² Prostate tissues in naproxen 400ppm-fed mice showed a significantly decreased CD93 expression as compared to untreated prostate adenocarcinoma tissue; though its expression also decreased in the aspirin1400ppm group, the change was not significant (Fig. 4B). Notably, NSAIDs treatment further increased the overexpression of growth factors such as HGF and VEGF while the expression of IGFBP-5 and PEDF was significantly decreased by both NSAIDs in the prostate tissues compared to no-drug controls (Fig. 4C). Furthermore, both NSAIDs also significantly reduced the expression of Chitinase-3 like-1 (CHI3L1), MMP9, and Serpin E1 in prostate tissues as compared to the no-drug controls (Fig. 4D). This is a notable outcome as CHI3L1 [which belongs to the glycoside hydrolase family 18 and interacts with IL-13 receptor subunit alpha-2 (IL13RA2), transmembrane protein 219 (TMEM219), and CD44 receptor] signaling plays an essential role in cancer growth, metastasis, and activation of tumor-associated macrophages²³; MMP9 is an extracellular matrix remodeling protein, while Serpin E1 is an inhibitor of urokinase plasminogen activator and tissue-type plasminogen activator.²⁴

Figure 4. Effect of NSAID intervention on prostate tissue cytokines, chemokines, and growth factors/growth signaling-associated molecules in TMPRSS2-ERG. Pten^flox/flox mice.

A) Heat map of cytokines, chemokines, and growth factors/growth signaling-associated molecules differentially expressed in prostate tissues of TMPRSS2-ERG.Pten^flox/flox (+TAM), no-TAM, and NSAIDs (aspirin 1400ppm and naproxen 400ppm) treated TMPRSS2-ERG.Pten^flox/flox (+TAM) mice. (B-D, left panel) array based immunoblots depicting relative expression, and (B-D, right panel) graphical representation of fold changes in (B) chemokines, cytokines, and immune cell associated cell surface proteins, (C) growth factors/growth signaling-associated molecules, and (D) extracellular matrix modifying proteins. Eight weeks old non-Cre induced mice were used as reference point (baseline, BL-8) to calculate the fold change expression. Dorso-lateral prostate tissue lysates from 28 weeks old mice (study end point) were used in the study [which relates to 20 weeks post Cre-induction (tamoxifen-induced at 8 weeks of mice age)]. NSAID-supplementation was done from 8 till 28 weeks of mice age. Quantified data is represented as Columns (mean for each group); bars represent SEM (n=3 mice/group). $, P≤ 0.05; #, P < 0.01; *, P < 0.001

3.5. Analysis of enriched canonical pathways, upstream regulators, diseases and functions associated with differentially expressed molecules in TMPRSS2-ERG, Pten^flox/flox mice (with and without NSAID supplementation) using Ingenuity Pathway Analysis.

Analyzing the trend in the expression of the cytokines, chemokines, and growth factors/growth signaling-associated molecules changed during prostate tumorigenesis, we identified six differentially expressed molecules that were consistent in both plasma and prostate tissues of mice with prostate adenocarcinoma in TMPRSS2-ERG. Pten^flox/flox (+TAM) mice. On the other hand, it was also inferred that there were 10 differentially expressed molecules in the plasma and 23 differentially expressed molecules in the prostate tissues that are exclusively expressed in the TMPRSS2-ERG. Pten^flox/flox (+TAM) mice compared to no-TAM age-matched controls (Fig.S3A). To elucidate the effect of NSAIDs (aspirin1400ppm and naproxen 400ppm) on these molecules circulating in the plasma, we compared the significantly expressed molecules from NSAID-fed TMPRSS2-ERG. Pten^flox/flox (+TAM) mice. Total 23 inflammatory molecules and growth factors/growth signaling-associated molecules were exclusively expressed in NSAIDs-fed mice plasma, while 12 molecules (M-CSF, CCL-22, CX3CL-1, CXCL-16, CD93, E-Selectin, Chemerin, FGF acidic, Fitz-3 Ligand, IGFBP-5, LDLR, and Lipocalin-2) showed altered expression during prostate tumor development (with and without NSAID-supplementation) compared to no-TAM age matched controls (Fig. 5A). We further compared the differentially expressed cytokines, chemokines, and growth factors/growth signaling-associated molecules in prostate tissues of TMPRSS2-ERG. Pten^flox/flox (+TAM) mice (with and without NSAID-supplementation). Here, we did not find any NSAID treatment-specific proteins in the prostate tissue. However, aspirin1400ppm and naproxen 400ppm treatments altered the expression of 11 molecules (CCL12, CXCL16, IL-33, CD93, HGF, IGFPB-5, PEDF, VEGF, CHI3L1, MMP9, and Serpin E1) in the prostate tissue compared to no-drug controls (untreated prostate adenocarcinoma tissues) (Fig. 5B).

Figure 5. Analysis of enriched canonical pathways and upstream regulators associated with differentially expressed cytokines in TMPRSS2-ERG, Pten^flox/flox (+TAM) mice (with and without NSAID supplementation) using Ingenuity Pathway Analysis.

(A-B) Venn diagram depicting differential response of NSAIDs (aspirin 1400ppm and naproxen 400ppm) on the expression of cytokines, chemokines, and growth factors/growth signaling-associated molecules in mice (A) plasma and (B) prostate tissue samples of TMPRSS2-ERG, Pten^flox/flox (+TAM) mice with and without NSAID supplementation. C) STRING-based protein-protein interaction diagram, depicting the relevant clusters among differentially expressed inflammatory molecules and growth factors/growth signaling-associated molecules from plasma and prostate tissue. Green color circles represent the identified differentially expressed molecules in mice plasma and prostate tissue and gray color circles represent interacting protein. (D) Heatmap of canonical pathways associated with cytokines, chemokines, and growth factors/growth signaling-associated molecules in response to prostate cancer development and treatment with NSAIDs. Enrichment of canonical pathways was performed using Ingenuity Pathway Analysis (IPA) [−log (P-value) >2 and Z-score cutoff =2]. E) Heatmap of significant upstream regulators predicted using IPA based on differentially expressed cytokines from plasma and prostate tissue [−log(P-value) >2 and Z-score cutoff >2].

Overall, the plasma and prostate tissue samples analyses identified 54 significant and differentially expressed cytokines, chemokines, and growth factors/growth signaling-associated molecules between PCa afflicted mice [TMPRSS2-ERG. Pten^flox/flox (+TAM)] and age-matched non-cancerous controls and NSAIDs-supplemented and no-drug controls (untreated prostate adenocarcinoma afflicted mice). A protein-protein interaction network was created using the STRING database and Cytoscape to examine the correlation among these differentially expressed molecules. Identified proteins were clustered into four major groups: growth factors/growth signaling-associated molecules, cytokines, chemokines, and tumor microenvironment remodeling pathways (Fig. 5C). Canonical pathway analysis was performed using IPA to explore the signaling pathways associated with differentially expressed cytokines/chemokine/growth factors (growth signaling-associated molecules) during PCa pathogenesis in the TMPRSS2-ERG. Pten^flox/flox (+TAM) model. Based on IPA algorithms and knowledge-based analysis differentially expressed inflammatory molecules and growth factors/growth signaling-associated molecules were enriched in 15 canonical pathways (Table S1). The identified active pathways (Z score >2) (ranked according to p-values) in the plasma of mice afflicted with PCa were associated with hepatic fibrosis, tumor microenvironment, wound healing, IL-17, IL-8, EMT, and inflammation; on the other hand, the plasma from NSAID-supplemented mice groups showed the loss of activation of these pathways (Z score <−2) (Fig. 5D). However, cytokines, chemokines, and growth factors/growth signaling-associated molecules from PCa tissues were enriched in additional canonical pathways, including HIF1-alpha, dendritic cell maturation, acute phase response, and IL-6. Based on the Z-score values, identified pathways overall represented inhibitory patterns in the prostates of NSAID-supplemented mice compared to the no-drug controls (untreated prostate adenocarcinoma tissue) (Fig. 5D).

Next, we analyzed the expression of upstream regulators of the differentially expressed molecules/pathways in the plasma and prostate tissues (Fig. 5E). We found 22 upstream regulators; some molecules, including JUN, tumor necrosis factor (TNF), hypoxia inducible factor-1 alpha (HIF1alpha); nuclear factor kappa-light-chain-enhancer of activated B cells (NF-kB), signal transducer and activator of transcription 3 (STAT3), IL1A/B, SP1, and extracellular signal-regulated kinase, (ERK)1/2 appeared crucial for activation of inflammation, EMT, and wound healing pathways (Fig. 5E). We further analyzed the disease and biological functions associated with the differentially expressed molecules and identified pathways of interest (Fig. S3B). We found 21 biological functions; identified functions were associated with cell survivability, migration, inflammation, immune-infiltration, and immune cells recruitment (Fig.S3B). We further compared the differentially expressed molecules identified in plasma and prostate tissue to the public database ‘The Cancer Genome Atlas’ (TCGA) using IPA analysis match tools to identify the occurrence of similar pathways in other malignant conditions. Gastric adenocarcinoma, glioblastoma, sarcoma, ovarian serous cystadenocarcinoma, and lung adenocarcinoma showed similar canonical pathways, upstream regulators, disease, and biological functions as the TMPRSS-ERG fusion driven PCa model (Fig.S4A–C).

4. DISCUSSION

Alterations in the expression profiles of oncogenes and tumor suppressor genes are characteristic of cancer initiation, growth, and progression.²⁵ However, crosstalk between transformed cells and the surrounding microenvironment influences the development and progression of various stages of cancer pathogenesis.²⁶ Cytokine, chemokines, and growth factors are produced in the tumor microenvironment by inflammatory cells, tumor cells, and surrounding normal cells.²⁷ These secretory molecules play critical roles in inflammation, tumor growth, immunosuppressive niche development, and cancer metastasis.^26,28 In this study, we assessed the circulating and prostate tissue-specific cytokines, chemokines, and growth factors/growth signaling-associated molecules in the TMPRSS2-ERG fusion-driven mouse model of PCa. Next, we also determined the association of the protective effects of NSAIDs against tumorigenesis in this model with the modulation of these identified molecular pathways.

CCL22 is a member of the CC chemokine family, prominently secreted by dendritic cells, and binds to the CCR4 receptor expressed on T cells (Th2 and Treg) and tumor cells.^29,30 Increased expression of CCL22 promotes the recruiting of Tregs cells to the tumor microenvironment and cancer metastasis by activating the Akt pathway.^31,32 CCL6 is produced by macrophages, eosinophils and neutrophils and binds to the CCR1 receptor expressed on tumor cells.³³ Increased expression and activation of CCL6-CCR1 promotes tumor cell migration and metastasis.³⁴ CCL12 is chemotactic protein, mainly secreted by M1 like macrophages and astrocytes, and attracts eosinophils, monocytes, and lymphocytes.³⁵ Studies have shown that higher expression levels of CCL12 results in the recruitment of CCR2 positive myeloid-derived suppressor cells in tumors.³⁶ In the present study, increased level of CCL22 in plasma, and CCL6 and CCL12 in the prostate tissue of mice afflicted with prostate adenocarcinoma compared to non-cancerous controls, suggests a role of these proteins in prostate tumor growth and escape of tumor cells from immunosurveillance which could contribute to PCa growth and progression in TMPRSS2-ERG. Pten^flox/flox (+TAM) mice.

CX3CL1 and CXCL16 are members of the CX3C and CXC chemokines family, secreted by lymphatic, vascular endothelium, and tumor cells, either membrane-anchored or soluble form. CX3CL1 binds with the CX3C chemokine receptor (CX3CR) 1 expressed on monocytes and lymphocytes, while CXCL16 interacts with CXCR6.^37–39 CX3CL1-CX3CR1 signaling plays crucial roles in chemotaxis movement and infiltration of monocytes and lymphocytes at the inflammatory site.⁴⁰ CXCL16-CXCR6 signaling is correlated with M2-macrophage infiltration, angiogenesis, and cancer metastasis.^41,42. In the plasma and PCa tissues of TMPRSS2-ERG. Pten^flox/flox (+TAM) mice, increased expression of CX3CL1 and CXCL16 were observed. CXCL13 is a B-cell attracting chemokine secreted by tumor cells, binds with only the CXCR5 receptor.⁴³ CXCL13-CXCR5 signaling shows a dualistic effect on cancer progression: CXCL13 has been shown to promote infiltration of the B-cell in the tumor microenvironment and correlated with increased survival of tumor cells⁴⁴; however, in mouse colorectal cancer, CXCL13 treatment significantly reduced the growth of the tumor.⁴⁵ In the present study, we observed decreased plasma levels of CXCL13 in the early stage of tumor formation (hyperplasia and PIN stages), while no significant change was observed in plasma and tissue samples of TMPRSS2-ERG. Pten^flox/flox (+TAM) mice afflicted with prostate adenocarcinoma as compared to age-matched non-cancerous controls.

CXCL1, CXCL2, and CXCL5 (all three chemokines) bind to the CXCR2 receptor expressed on neutrophils.⁴⁶ CXCL1 expression is associated with tumor growth, invasion, and metastasis.⁴⁷ CXCL1 along with pro-inflammatory cytokines Lipocalin-2 promotes PCa progression via Src activation.⁴⁸ Increased expression of CXCL1 and CXCL2 has been linked with cancer metastasis and chemoresistance.⁴⁹ CXCL5 expression has been also reported to be high in PIN, adenocarcinoma and metastatic cancer stage compared to normal or benign prostatic gland.⁵⁰. CXCL1, CXCL2, and CXCL5 chemokines were highly expressed in PCa tissue of TMPRSS2-ERG. Pten^flox/flox (+TAM) mice compared to normal prostate tissue.

IL-33 is a member of the IL-1 cytokine family and displays both pro-tumor and anti-tumor functions, and activates effector cells, basophils, ILC2, and eosinophils.⁵¹ In the present study, increased expression of IL-33 was observed in prostate adenocarcinoma tissue of TMPRSS2-ERG. Pten^flox/flox (+TAM) mice, which signifies its pro-tumorous function in TMPRSS2-ERG fusion-driven PCa. Chemerin is an adipokine, and its expression is associated with increased adipogenesis, inflammation, and glucose homeostasis.^52,53 We observed an increased plasma level of Chemerin during hyperplasia, PIN, and prostate adenocarcinoma stages in TMPRSS2-ERG. Pten^flox/flox (+TAM) mice compared to the normal prostate. M-CSF is a secreted molecule associated with increased proliferation, survivability, and differentiation of monocyte into macrophages.⁵⁴ Increased serum M-CSF level is associated with bone metastatic of PCa⁵⁵. We observed significantly increased plasma levels of M-CSF in higher stages of PCa in TMPRSS2-ERG. Pten^flox/flox (+TAM) mice compared to controls. CD93, CD14, CD40, and TNFRSF11B are transmembrane proteins expressed on the tumor cells, dendritic cells, and other myeloid cells and regulate the tumor growth, vascularization, infiltration, and recruitment of anti-tumor T-cells, monocytes/macrophages in the tumor microenvironment to acquire immune-suppressive characteristics.^{15,18,19,56,57} We observed an increased CD93, CD14, CD40, and TNFRSF11B expression in prostate adenocarcinoma tissue of TMPRSS2-ERG. Pten^flox/flox (+TAM) mice.

Growth factors are secreted extracellular molecules; mainly these polypeptides maintain tissue homeostasis under normal physiological conditions.⁵⁸ However, functions of growth factors are mostly subverted in cancer.⁵⁸ HGF and VEGF are potent angiogenic factors secreted mainly by tumor cells under hypoxic conditions, promoting angiogenesis, tumor growth and proliferation, and metastasis.⁵⁹ LDLR, RBP4, and PEDF are known to promote tumor growth, proliferation, and migration.^60–62 IGF is a mitogen and is involved in regulating cell proliferation, apoptosis, and transformation by interacting with the IGF-I receptor.⁶³ The mitogenic effect of IGF on cancer cells is determined by bioavailability and interaction with IGF-IR, which IGFBPs regulate.⁶⁴ IGFBP-1, IGFPB-2, and IGFPB5 show the dual regulatory effect (promoting and inhibitory), while IGFPB-3, IGFPB-4, and IGFPB-5 have been shown to have inhibitory effect on mitogenic actions of IGF.⁶⁵ In the present study, we found an upregulated expression of HGF, VEGF, LDLR, RBP4, PEDF, IGFPB-3, IGFPB-5, and IGFBP-6 in prostate adenocarcinoma tissue of TMPRSS2-ERG. Pten^flox/flox (+TAM) mice compared to non-cancerous controls, which signifies their pro-cancerous role in development of PCa in this model. MMPs are the major proteases responsible for the degradation and remodeling of ECM.¹⁷ Controlled modification in ECM is crucial for the invasion and migration of tumor cells. We observed an increased expression of MMP2 and MMP9 in prostate adenocarcinoma tissue in the TMPRSS2-ERG. Pten^flox/flox (+TAM) mice, which suggests their possible role in PCa progression in this model.

On the other hand, in this study upon NSAID supplementation, decreased expression of CCL22, CX3CL1, CXCL16 in plasma and CCL12 in prostate tissues were observed in TMPRSS2-ERG. Pten^flox/flox (+TAM) mice, compared to untreated PCa. Herein, we also observed decreased plasma levels of M-CSF during NSAID (aspirin and naproxen) supplementation. M-CSF is crucial for the proliferation, survivability, and differentiation of monocyte into macrophages.⁵⁴ This is in line with a previous in-vitro study, wherein the use of NSAIDs on human monocytes cells suppressed chemotaxis movement and genes involved in the activation of macrophages.^66,67 Previous studies have shown that NSAID treatment impairs the production of growth factors and other inflammatory molecules such as PDGF, FGF, EGF, IGF, IL-6, IL-8, VEGF and HGF by targeting circulating inflammatory cells, platelets, and tumor cells.^68–72 Anti-inflammatory drugs also inhibit ECM remodeling by reducing the expression of MMP via suppression of ERK/SP1 mediated transcription.⁷³ In the present study, the array outcomes indicate that both aspirin and naproxen supplementation exhibit inhibitory effects on the expression of inflammatory molecules and growth factors, such as Chemerin, IL-33, CD93, FGF acidic, PDGF-BB, PEDF, IGFBPs, and Flt-3 ligand and ECM remodeling proteins MMP2 and MMP9 in the TMPRSS2-ERG. Pten^flox/flox (+TAM) mice. However, some cancer promoting molecules such as HGF, VEFG, and CXCL16 (though CXCL16 decreased in plasma), were upregulated in prostate tumors upon aspirin and naproxen treatments in TMPRSS2-ERG. Pten^flox/flox (+TAM) mice indicating that the NSAIDs supplementation has differential effect on some of these signaling pathways. Notably, we have previously reported that the protective effects of NSAID supplementation in TMPRSS2-ERG. Pten^flox/flox (+TAM) mice are associated with decreased angiogenesis; thus, it is likely that the anti-angiogenic effects of NSAID in this model are independent of modulation of VEGF expression⁵. Our results are also in line with our previous observation that the anti-PCa effects of NSAIDs are associated with a decrease in COX-1 expression specifically in the TMPRSS2-ERG. Pten^flox/flox (+TAM) mice prostate compared to non-fusion driven mice model (Hi-Myc^+/− mice).⁵ Interestingly, in that comparative study, strong inhibition of COX-2 due to NSAID supplementation was observed in both fusion-driven and non-fusion driven models.⁵ Bioinformatic analysis of the array result outcomes further corroborated the previous observation indicating that NSAID supplementation has protective effects against PCa growth and progression in TMPRSS2-ERG. Pten^flox/flox (+TAM) mice which are complimentary to NSAIDs anti-proliferative and anti-inflammatory effects and possible modulation of the immune cells enriched microenvironment. However, more detailed studies are warranted in this direction to unravel the specific immune sub-type and associated immune cell-function that is modulated by NSAIDs to reverse the inflammatory/tumor promoting signaling specifically in the TMPRSS2-ERG fusion driven PCa compared to non-fusion driven PCa.

Conclusion

Herein, using TMPRSS2-ERG. Pten^flox/flox (+TAM) mice as a preclinical model of TMPRSS2-ERG fusion driven PCa, we demonstrated that during fusion driven prostate tumorigenesis there is a stage-specific increase in the expression of inflammatory molecules (cytokines/chemokines) and growth factors/growth signaling-associated molecules in circulation and tumor/tumor microenvironment. On the other hand, the protective effect of NSAIDs (aspirin 1400ppm and naproxen 400ppm) against TMPRSS2-ERG fusion driven PCa was associated with reduced expression of a) tumor promoting inflammatory molecules (M-CSF, IL-33, CCL22, CCL12, CX3CL1, CHI3L1, and CD93), b) growth factors- growth signaling-associated molecules (Chemerin, FGF acidic, Flt-3 ligand, IGFBP-5, and PEDF), and c) tumor microenvironment/stromal remodeling proteins MMP2 and MMP9 (Fig. 6).Thus, our findings provide new insights into the altered cytokines, chemokines, and growth factors/growth signaling-associated molecules in TMPRSS2-ERG fusion-driven PCa on NSAIDs treatment, which could be potentially explored in future to develop preventive/intervention treatment protocols with NSAIDs.

Figure 6: Comprehensive network of signaling pathways associated with differentially expressed cytokines, chemokines, and growth factors/growth signaling-associated molecules in TMPRSS2-ERG, Pten^flox/flox (+TAM) mice with and without NSAIDs supplementation.

Green-highlighted molecules indicate decreased protein expression in TMPRSS2-ERG.Pten^flox/flox (+TAM) mice plasma or prostate tissue upon NSAIDs intervention. Dark pink highlighted molecules indicate increased protein expression in TMPRSS2-ERG.Pten^flox/flox (+TAM) mice plasma or prostate tissue after NSAIDs intervention. Light pink highlighted molecules show modulated protein expression during prostate tumorigenesis in TMPRSS2-ERG.Pten^flox/flox (+TAM) model; however, NSAIDs supplementation had no impact on their expression.

Abbreviations:

CCL

CC chemokine ligand

CD

cluster of differentiation

CHI3L1

chitinase-3 like-protein-1

CX3CL

CX3C chemokine ligand

CX3CR

CX3C chemokine receptor

CXCL

CXC chemokine ligand

COX-1

cyclooxygenase-1

EMT

epithelial to mesenchymal transition

ERG

Ets related gene

ERK

extracellular signal-regulated kinase

ETS

erythroblastosis virus E26

FGF

fibroblast growth factor

Flt-3 ligand

FMS-related tyrosine kinase 3 ligand

HGF

hepatocyte growth factor

HIF1alpha

hypoxia inducible factor-1 alpha

IGFBP

insulin-like growth factor binding protein

IL

interleukin

IL13RA2

IL-13 receptor subunit alpha-2

LDL-R

low-density lipoprotein receptor

LIX

lipopolysaccharide-induced CXC chemokine

LUT

lower urogenital tract

M-CSF

macrophage stimulating factor

MMP

matrix metalloproteinase

NF-kB

nuclear factor kappa-light-chain-enhancer of activated B cells

NSAIDs

nonsteroidal anti-inflammatory drugs

PCa

prostate cancer

PBS

phosphate buffered saline

PDGF-BB

platelet-derived growth factor-BB

PEDF

pigment epithelium-derived factor

RAGE

receptor for advanced glycation end products

RBP4

retinol-binding protein 4

STAT3

signal transducer and activator of transcription 3

TAM

tamoxifen

TCGA

The Cancer Genome Atlas

TMEM219

transmembrane protein 219

TMPRSS2

transmembrane protease serine 2

TNF

tumor necrosis factor

TNFRSF11B

TNF receptor superfamily 11b

TNFSF13B

TNF ligand superfamily member 13B

VEGF

vascular endothelial growth factor

WISP-1

Wnt1-inducible signaling protein 1