Per- and Polyfluoroalkyl Substances Target and Alter Human Prostate Stem-Progenitor Cells

Abstract

Per- and polyfluorinated alkyl substances (PFAS) are a large family of widely used synthetic chemicals that are environmentally and biologically persistent and present in most individuals. Chronic PFAS exposure have been linked to increased prostate cancer risk in occupational settings, however, underlying mechanisms have not been interrogated. Herein we examined exposure of normal human prostate stem-progenitor cells (SPCs) to 10 nM PFOA or PFOS using serial passage of prostasphere cultures. Exposure to either PFAS for 3–4 weeks increased spheroid numbers and size indicative of elevated stem cell self-renewal and progenitor cell proliferation. Transcriptome analysis using single-cell RNA sequencing (scRNA-seq) showed 1) SPC expression of PPARs and RXRs able to mediate PFAS effects, 2) the emergence of a new cell cluster of aberrantly differentiated luminal progenitor cells upon PFOS/PFOA exposure, and 3) enrichment of cancer-associated signaling pathways. Metabolomic analysis of PFAS-exposed prostaspheres revealed increased glycolytic pathways including the Warburg effect as well as strongly enrichment of serine and glycine metabolism which may promote a pre-malignant SPC fate. Finally, growth of in vivo xenografts of tumorigenic RWPE-2 human prostate cells, shown to contain cancer stem-like cells, was markedly enhanced by daily PFOS feeding to nude mice hosts. Together, these findings are the first to identify human prostate SPCs as direct PFAS targets with resultant reprogrammed transcriptomes and metabolomes that augment a preneoplastic state and may contribute to an elevated prostate cancer risk with chronic exposures.

Graphical Abstract

1. INTRODUCTION

Per- and polyfluoroalkyl substances (PFAS) are a family of chemicals containing long hydrophobic (8-carbon) chains fully saturated with fluorine atoms (i.e., perfluoroalkyl chains) and a hydrophilic polar functional group [1]. These chemicals are thermally stable and lipid- and water-repelling, prompting their extensive use since the 1970s in household products such as carpets, Gortex products, and non-stick cookware. Such widespread usage has led to accumulation of these persistent organic pollutants in the environment [2–4]. Two PFAS, perflurooctanoate acid (PFOA) and perfluorooctanesulfonic acid (PFOS), are the most abundant in the environment and are a major source of exposure to humans through food and drinking water. Unfortunately, these and newer PFAS substitutes (e.g., GenEx) do not readily degrade and are commonly referred to as “forever chemicals”. An estimated 99% of Americans are exposed to and/or have these chemicals in their body [5–7]. Occupational PFAS exposures and those living near production facilities or contamination sites, including military bases, are associated with the highest body burdens [8]. Wastewater treatment plants are a major sink for PFAS by receiving PFAS-impacted wastewater. Landfill leachate, wastewater effluent, biosolids application, and groundwater recharge using surface runoff re-introduces PFAS to water sources including drinking water. Crucially, PFAS exposures can exert adverse health outcomes in humans. In a recent large report from the Center for Disease Control, diseases linked to chronic PFAS exposures include liver damage, cardiovascular abnormalities, altered immune function, and developmental and reproductive disorders based largely on epidemiology studies supported by laboratory research [9]. Additional studies have linked PFAS exposures to type 2 diabetes [10, 11], thyroid disruption [12] and cancers, particularly testicular and kidney cancer [13, 14]. The U.S. Environmental Protection Agency recently added PFAS to the 2021 Toxic Release Inventory [15], and the International Agency for Research on Cancer [16] has listed PFOA as a Class 2B carcinogen.

Prostate cancer is the most commonly diagnosed non-cutaneous cancer in American men and the second leading cause of cancer-related deaths [17], although its etiology remains elusive. While cancers are primarily a result of germline and/or somatic mutations, considerable evidence has identified environmental contributions to carcinogenesis in multiple organs including prostate cancer [18]. As the prostate and prostate cancer are androgen-dependent and influenced by multiple other hormones, it is logical that endocrine disrupting chemicals (EDCs) that mimic or interfere with hormone action may influence prostate growth and carcinogenic risk. Indeed, this has been shown for several EDCs including bisphenol A (BPA), arsenic, cadmium, dioxin, several pesticides, and PCBs, among others [19]. Of relevance to the present research, several epidemiology studies have indicated that prostate cancer risk and mortality increase with chronic PFAS exposure in both occupational settings and in men living near high contamination zones [20–23]. Men with a familial history of prostate cancer were particularly affected, suggesting a gene-environment interaction in disease promotion [23]. Despite this evidence, there are no laboratory studies that identify cellular targets or mechanistic data on the molecular underpinnings of PFAS chemicals in the prostate.

The prostate epithelium contains three cell types: luminal (majority), basal, and neuroendocrine (infrequent). These cells derive from rare stem cell population(s) [24, 25] that, during adult tissue replenishment, undergo asymmetric cell division to self-renew and generate bipotent progenitors that lineage commit and, upon amplification, differentiate to mature epithelial lineages. Further, prostate cancer contains cancer stem-like cells that continuously seed and maintain tumor growth [26, 27]. EDCs have been shown to target stem cells, which are compelling cancer targets in organisms during development and cancer progression [18, 28, 29]. Previous work from our laboratory has identified the prostate epithelial stem-progenitor cell (SPC) population as direct hormone and EDC targets, with environmentally relevant doses of BPA perturbing self-renewal and lineage commitment, altering gene transcription and reprogramming the SPC epigenome to increase carcinogenic risk [30–33] and environmentally relevant levels of arsenic increasing stem cell numbers and impairing the autophagy-lysosome pathway with subsequent transformation of SPCs [34].

In this context, the goals of the present study were to determine if PFAS exposures can target the human prostate SPC, modify their behaviors and in so doing, influence prostate cancer risk. For these studies, we employed primary prostate epithelial cells from normal organ donors as well as the cancerous human prostate cell line RWPE-2. We interrogated the human SPC for peroxisome proliferator-activated receptors (PPARs) and their obligate partners, retinoid X receptors (RXRs) known to mediate many PFAS effects [35–37], and serially passaged prostate SPC using spheroid culture in the presence of 10 nM PFOA or PFOS to evaluate their self-renewal capacity and proliferative responses to chronic exposure at environmentally relevant levels. Lineage commitment alterations were evaluated by scRNA-seq and PFAS-targeted gene pathways were identified. The PFAS-exposed metabolome of the SPC populations were next interrogated using gas chromatography - mass spectroscopy (GC/MS). Finally, the effects of PFOS exposure on RWPE-2 derived prostate tumors in vivo were examined. Together, the findings implicate prostate SPCs as direct PFAS targets and raise the intriguing possibility that PFAS-induced reprogramming contributes to their transformation and tumor-initiating capacity.

2. MATERIALS AND METHODS

2.1. Prostate Epithelial Cells and Prostasphere Culture

Primary normal human prostate epithelial cells (PrECs) were obtained from young (19–21 years of age) disease-free organ donors (Lonza, Walkersville, MD, USA) at passage 2–3 and cultured at 37 °C, 5% CO₂ in ProstaLife Epithelial Cell Growth Medium (PrEGM) (LifeLine Cell Technology, Frederick, MD, USA) on fibronectin-coated flasks. Prostaspheres were cultured from PrECs as previously described and confirmed to be clonally derived spheroids of stem/progenitor cells [38–40]. Briefly, 1 × 10⁵ PrEC cells were resuspended in 1:1 PrEGM:Matrigel (Cat# CB356238, Corning/BD Biosciences, Corning, NY, USA) and the suspension plated around the bottom well edge of 12-well plates. After the cell-Matrigel mix solidified, 1 mL PrEGM with or without 10 nM of PFOS or PFOA (Sigma-Aldrich, St. Louis, MO, USA; Cat #77282 for PFOS; Cat#171468 for PFOA) was added and replenished every 48 hr. Prostaspheres were cultured for 1 wk at 37 °C in 5% CO₂ and serially passaged for 3–4 wks to provide extended PFOS/PFOA exposure. For serial passage, prostaspheres were harvested from Matrigel by digestion with 2 mL dispase (StemCell Technologies, Vancouver, BC, Canada) for 30 min and the spheres dispersed into single cells by digestion with 1 mL 0.05% trypsin-EDTA (Cat# 25300120, Gibco/Fisher Scientific, Waltham, MA, USA). Similar cell numbers for each treatment group (4 × 10⁵) were re-plated in Matrigel culture to form the next generation spheroids. PFAS-exposed and control prostaspheres were serially passaged up to 6 generations. Prostasphere number and size at the end of each 7-day culture period were assessed using an automated digital image processing algorithm as previously detailed [31].

The human prostate cancer cell line RWPE-2 (RWPE-kRAS) was obtained from ATCC (American Type Culture Collection, Manassas, VA, USA) and cultured at 37 °C, 5% CO₂ in Gibco Keratinocyte SFM 1X growth media with glutamine (Cat# 17005042, Gibco/Fisher Scientific, Waltham, MA, USA). Prior to reaching confluence, the 2D cultured cells were transferred to 3D culture in Matrigel as described above to enrich for the cancer stem cell population. After 7 days in spheroid culture, some spheroids were switched to organoid culture medium (PrEGM containing 5% charcoal-stripped fetal bovine serum (Gibco/Fisher Scientific, Waltham, MA, USA), 1 nM Y-27632 (Apebio Technology, Houston, TX. USA) and 1 nM R1881 (Sigma-Aldrich, St. Louis, MO, USA)) and continued for 14 days.

2.2. Prostate Cancer Xenografts

RWPE-2 2D cultures were used for tumor xenografts in 4-week-old athymic nude male mice (Jackson Laboratory, Bar Harbor, ME, USA). Animal experiments and protocols were approved by the University of Illinois at Urbana-Champaign (IACUC Protocol #20159) and NIH standards for the use and care of animals were followed. Mice were fed AIN93M diet (Envigo Teklad Diets, Madison, WI, USA) ad libitum. RWPE-2 cells (2 × 10⁶/graft) mixed in Matrigel were injected to the left and right flanks of the mice (N=8 mice/group) under anesthesia. Silastic tubes (1 cm, inside diameter 1.02 mm × outer diameter 2.16 mm; Dow Corning Corp, Midland, MI, USA) packed with testosterone (Sigma-Aldrich, St. Louis, MO, USA) were implanted to provide sufficient hormone for graft growth in the hypogonadal nude mice [38]. Animals were administered PFOS (Sigma-Aldrich, St. Louis, MO, USA) daily by oral gavage at 10 mg/kg BW. Animal weights were monitored twice weekly and tumor size measurements were performed 3x/week using a digital caliper and calculated using the formula V=0.5 × length × width² [41]. Animals were euthanized at 5.5 wks according to IACUC regulations for maximum allowable tumor size.

2.3. Single-Cell Capture by the 10X Genomics Platform and Single-Cell RNA-Sequencing (sc-RNA-seq) Using NovaSeq 6000

PFAS-exposed and control passage 3 spheroids were dispersed into a single-cell suspension as described above. Single cells were separated and captured at a 10X Genomics station using Chromium Next GEM Single Cell 3ʹ v3 Reagent Kits (10X Genomics, CG000183, Pleasanton, CA, USA). After GEM generation and barcoding, mRNA was reverse transcribed into cDNA, and further amplified for library construction followed by RNA-seq using NovaSeq 6000 SP (Illumina, San Diego, CA, USA) at the University of Illinois Keck Center. Each lane of the sequencer generated 1.6 billion reads, 0.4 billion reads per sample, 0.1 million reads (paired-end, 28 nt + 91 nt) for each single-cell. We validated ~3500 single-cell cDNA libraries from each sample and ~7,500 genes per cell for data analysis.

ScRNA-seq data was aligned against hg38 reference genome sequence using 10X Cellranger pipeline. The aligned dataset was imported into Seurat package (Satijalab.org). Pre-processing of raw data included filtering out cell debris, unhealthy singlets, and potential doublets by setting cutoff based on the distribution of the feature RNA abundance (cutoff range was from 2000 to 7500; each data point indicates a singlet or single cell) and the percentage cutoff of mitochondrial genes within each data point was 10%. Post-filtered single cell feature expression values were normalized by the total expression through a global-scaling normalization method “LogNormalize”. The scale factor was set at the default value of 10,000. Subsequent feature selection was performed with default setting (nfeature = 2000) to gain a subset of features that reflect high cell-to-cell variation of the dataset. The expression value of each gene was scaled so that data points were centralized around the origin, and the variance across these data points equals to 1. Linear dimensional reduction was performed on the subset of feature genes selected by principal component analysis (PCA). Next, a resampling test inspired by the Jackstraw procedure was utilized to determine the PC number to include. The data points or single cells were finally clustered based on PCs determined previously using a graph-based clustering approach [42]. Pre-defined PCs were introduced to the clustering analysis to plot uniform manifold approximation and projection (UMAP) for improved cluster visualization. Stem cell cluster identification was based on previously identified biomarkers [40].

Gene set enrichment analysis (GSEA) was performed on normalized gene expression data generated by RNA-seq using the Broad Institute’s Molecular Signatures Database (gsea-msigdb.org) [43]. Analyses were run using the gene expression data against two gene sets: C2: curated gene sets and CP: BIOCARTA: BioCarta gene sets. The false discovery rate (FDR) was calculated by comparing the actual data with 1000 Monte Carlo simulations. The familywise error rate is a conservative correction that seeks to ensure that reported results do not include any false-positive gene sets.

2.4. GC/MS-based metabolic profiling of prostaspheres

Cell metabolites were extracted from prostaspheres serially passaged for 4 weeks in vehicle, PFOA or PFOS (10nM) using a 1:2:1 mixture of acetonitrile (Sigma-Aldrich, St. Louis, MO, USA):isopropanol (Fisher Scientific, Waltham, MA, USA):water. Extracts were sent to the University of Illinois at Urbana-Champaign’s Metabolomics Core Facility to detect and quantify metabolites using GC/MS. Metabolic profiles were obtained from an Agilent GC-MS system (Agilent 7890 gas chromatograph, Agilent 5975 MSD, and HP 7683B autosampler, Lexington, MA, USA).

The spectra of all chromatogram peaks were evaluated using the AMDIS 2.71 (chemdata.nist.org) and a custom-built database with 460 unique metabolites. All known artificial peaks were identified and removed before data mining. Individual metabolomic data sets for each treatment were separated and grouped into files to make comparisons between treatment conditions using MetaboAnalyst software (metaboanalyst.ca) [44]. Sample class annotations consisted of Veh vs. PFOA, Veh vs. PFOS and Veh vs. PFOA/PFOS combined (PFAS). Files were uploaded to the Enrichment Analysis tool of MetaboAnalyst software version 5.0. Data were not normalized, transformed, or scaled but were compared to the SMPDB reference metabolome, which represents metabolite values from normal metabolic human pathways. The top 25 enriched metabolic pathways and associated metabolites were retrieved along with their p-values and enrichment ratios.

2.5. Statistics

Spheroid data were analyzed using GraphPad Prism 6.0 (GraphPad Software Inc., La Jolla, CA, USA) and presented as mean ± SEM. Statistically significant differences were assessed by Student’s t-test or, with multiple groups, ANOVA followed by Tukey-Kramer post-hoc tests. For xenograft tumor growth rate data over time, a two-way ANOVA followed by Tukey’s multiple comparison test was used. For all data, P < 0.05 was considered significant.

3. RESULTS

3.1. Chronic PFAS exposures augment prostate stem and progenitor cell self-renewal and proliferation

To determine the effects of PFAS on primary SPCs grown from disease-free human prostate primary epithelial cell cultures, we studied two unique properties of stem and progenitor cells: self-renewal and differentiation. The sphere formation assay enriches for stem cells based on their unique capability to survive and clonally expand in 3D serum-free culture while differentiated primary cells undergo apoptosis or senescence. In addition, spheroid serial passage demonstrates their ability to maintain and expand a self-renewing stem cell population [45]. The number and size of prostaspheres represents stem cell self-renewal and daughter progenitor cell proliferation capacity, respectively [38, 39]. Prostaspheres were exposed to an environmentally relevant level (10 nM) of PFOA or PFOS through 3–4 serial passages to model chronic exposure. With 3–4 weeks of PFAS exposure, a significant increase in total sphere number was observed in response to either PFAS indicating augmentation of stem cell symmetric self-renewal (Fig. 1A, B). Size analysis by digital imaging revealed significant increases in small (40–60 μm), medium (60–80 μm) and large sized (>80 μm) spheres, with the large-size increases indicative of increased progenitor cell proliferative. This together suggests that both prostate stem and daughter progenitors are direct PFAS targets.

Figure 1: Effects of PFAS exposures on prostasphere formation and size.

A: Human PrEC-derived prostaspheres were culture in 3–4 serial passages (1 week each) in the absence or presence of 10 nM of PFOS or PFOA or in vehicle as controls. White bar = 50μm. B: Total spheroid numbers and sizes across 20 μm increments were quantitated by an automated digital image processing algorithm [38]. Treatment with either PFAS compound markedly increased prostasphere numbers formed as well as their size across the size spectrum, indicating that prostate stem/progenitor cells are the direct PFAS cell targets. N=4; bars represent the mean ± SEM. * Indicates P<0.05 and ** indicates P<0.01 versus vehicle.

3.2. RNA-sequencing identifies PPAR and RXR expression in prostate stem and progenitor cells

PFAS can activate cellular responses in many target tissues by engaging PPARs with PFOA and PFOS predominantly activating the alpha isoform (PPARα) compared to the gamma or delta isoform (PPARγ, PPARδ) [35–37]. As such, we examined prostaspheres for gene expression of PPARs and RXRs, their obligate binding partners, using scRNA-seq and bulk RNA-sequencing. Transcriptome analysis identified two cell clusters in spheroids: a smaller one enriched in prostate stemness genes, PITX1 and SOX2 [40], and a larger cluster of prostate progenitor cells (Fig. 2A). scRNA-seq of prostaspheres revealed that both stem and progenitor cell populations solidly expressed PPARα while PPARγ and PPARδ expression were comparatively lower (Fig. 2B) suggesting PPARα as the dominant PPAR. We next interrogated a previously derived [40] deep RNA-seq data set from FACS-sorted prostasphere stem and progenitor populations to identify average reads for the PPARs and RXRs. We confirmed that all three PPARs were expressed in stem and progenitor cells, although using this approach, the expression levels were similar for PPARα, PPARγ and PPARδ (Fig. 2C). Importantly, RXRα and RXRβ, but not RARγ, were present in both populations. Levels of PPAR and RXR were just below expression levels for glucocorticoid receptor (GR), the highest expressed nuclear receptor family member in prostasphere cells. Together, these data indicate that the prostate stem and progenitor cell populations have the necessary receptors to mediate PFAS actions within the cells.

Figure 2: Human prostate stem and progenitor cells express PPARs and RXRs.

A-B: Disease-free human primary PrECs were transferred to 3D Matrigel culture to form prostaspheres and after 3 serial passages, scRNA-seq was used to identify relative gene expression in the stem and progenitor cell populations. A: UMAP clustering was used to identify 2 major cell clusters with expression of PITX1 and SOX2, known prostate stem cell genes, enriched in the smaller cluster thus identifying the minor population of stem cells. The larger cluster expressed early differentiation genes associated with prostate progenitor cells [40]. B: Expression levels of PPARα, PPARγ and PPARδ were identified in both the stem and progenitor populations with PPARα and PPARγ at somewhat higher levels than PPARδ. C: RNA-seq reads of PPARs and RXRs in prostaspheres sorted for stem and progenitor populations by FACS followed by deep RNA-seq as previously described [63]. PPARα and PPARδ levels were similar between the stem and progenitor populations while PPARγ expression was higher in progenitor cells, similar to scRNA-seq data. RXRα was expressed at higher levels that RXRβ, and approached levels of GR, the top expressed nuclear receptor.

3.3. Chronic PFAS exposures alter luminal progenitor cell differentiation

To examine hierarchical cell clusters within the stem cell-derived prostaspheres, PrEC-derived prostaspheres were serially passaged without or with 10 nM of PFOS or PFOA and passage 3 spheroids were dispersed for scRNA-seq using the 10X Genomics platform. Transcriptome analysis by UMAP plots revealed cell sub-clusters in the PFOS (5 clusters) and PFOA (8 clusters) exposed prostaspheres integrated with respective vehicle-exposed prostaspheres (Fig. 3A, B). Importantly, UMAP plots that identify the vehicle cells vs PFOS, or PFOA-exposed cells revealed the appearance of a unique cell cluster in PFAS-exposed spheroids (Fig. 3C, D). Both the PFOS-exposed targeted cluster 3 (Fig. 3A, C) and PFOA-exposed targeted cluster 5 (Fig. 3B, D) express luminal keratin genes KRT8/18, lack stemness and basal keratin gene expression [40] and show similarities to cluster 0 in vehicle-exposed spheres (Fig. 3E, F) which are characterized as early-stage luminal progenitor cells. This suggests that PFAS-exposure results in an aberrant differentiation program of luminal cells at the progenitor cell stage.

Figure 3: PFOS and PFOA reprogram progenitor cell differentiation in exposed prostaspheres.

Prostaspheres were serially passaged in the absence or presence of 10nM PFOS or PFOA. Passage 3 spheroids were dispersed for scRNA-seq using the 10X Genomics platform. A-B: UMAP plots of integrated PFOS and vehicle (A) and PFOA and Vehicle (B) transcriptomes identified 5 and 8 unique cell clusters, respectively. C-D: The same UMAP plots as in A and B are highlighted for vehicle-exposed cells (blue) and PFOS (C) or PFOA (D) exposed cells (pink). Green arrows point to unique cell clusters that emerge in the PFAS-exposed spheroids that are not present in the vehicle treated prostaspheres. E-F: PFOS exposure (E) introduced cluster 3 (A&C) while PFOA exposure introduced cluster 5 (B&D). Dot plots of keratin gene expression analysis reveal that both PFAS-targeted cell clusters express luminal keratin genes KRT8/18, lack stemness (KRT80, 16A) and basal (KRT5) keratin gene expression and most closely align with cluster 0 in vehicle-exposed spheres, characterized as early-stage luminal progenitor cells.

GSEA of genes expressed in PFOA and PFOS-exposed prostaspheres versus their respective vehicle controls revealed significant enrichment of pathways involved in cell replication including E2F targets, G2M checkpoint and mitotic spindle in both PFAS exposure groups which supports the increased proliferative responses observed in the spheroids (Fig 4). In addition, both PFAS exposures significantly increased TNFα via NFκB pathways and k-RAS signaling in spheroids, both connected to carcinogenic pathways. The top 20 enriched signaling pathways for each chemical highlight these as well as inflammatory-associated pathways (IL-2, IL-6, TGFβ, inflammatory response) and metabolic pathways (glycolysis, oxidative phosphorylation) (Table 1).

Figure 4: The top 5 enriched gene signaling pathways in prostaspheres by PFOA and PFOS exposures.

GSEA of scRNA-seq transcriptomes were used to identify the top common gene signaling pathways enriched in prostaspheres exposed for 3 wks to PFOA (left, red) or PFAS (right, red) as compared to vehicle treatments (blue). Exposure to either PFAS significantly upregulated cell cycle pathways including E2F targets, G2M checkpoint and mitotic spindle which aligns with the increased cell proliferation observed in Figure 1. Enhancement of oncogenic pathways, including NFκB and k-RAS signaling, suggesting that these PFAS exposures have potential for initiating progenitor cell transformation. NES = normalized enrichment score, FDR = false discovery rate.

Table 1.

Top 20 enriched gene signaling pathways in PFOA- and PFOS-exposed prostaspheres compared to Vehicle controls.

PFOA vs Vehicle	PFOS vs Vehicle
G2M_CHECKPOINT	E2F_TARGETS
E2F_TARGETS	G2M_CHECKPOINT
MITOTIC_SPINDLE	MYC_TARGETS_V1
TNFA_SIGNALING_VIA_NFKB	MITOTIC_SPINDLE
KRAS_SIGNALING_DN	MYC_TARGETS_V2
IL6_JAK_STAT3_SIGNALING	EPITHELIAL_MESENCHYMAL_TRANSITION
SPERMATOGENESIS	DNA_REPAIR
ESTROGEN_RESPONSE_EARLY	TNFA_SIGNALING_VIA_NFKB
PANCREAS_BETA_CELLS	OXIDATIVE_PHOSPHORYLATlON
GLYCOLYSIS	KRAS_SIGNALING_UP
UV_RESPONSE_UP	SPERMATOGENESIS
IL2_STATS_SIGNALING	ALLOGRAFT_REJECTION
ALLOGRAFT_REJECTION	MTORC1_SIGNALING
MYOGENESIS	APICAL_JUNCTION
ESTROGEN_RESPONSE_LATE	ANGIOGENESIS
UV_RESPONSE_DN	INFLAMMATORY_RESPONSE
NOTCH_SIGNALING	IL6_JAK_STAT3_SIGNALING
HEDGEHOG_SIGNALING	TGF_BETA_SIGNALING
HYPOXIA	IL2_STAT5_SIGNALING
INFLAMMATORY_RESPONSE	COMPLEMENT

RNA-seq followed by GSEA was used to identify the top 20 gene pathways enriched in prostaspheres exposed to 10 nM PFOA or PFOS through 3 serial passages as compared to vehicle controls. All pathways listed were considered significant with a false discovery rate (FDR) < 0.25.

3.4. PFAS alters the prostasphere metabolome

PPARs play key roles in regulating cellular metabolism and since the transcriptome data indicated PFAS-induced shifts in metabolic pathways, we next analyzed the prostasphere metabolites following 4 weeks of exposure to PFOA or PFOS using GC/MS. The top enriched metabolite set for the PFAS exposures compared to vehicle were glycine and serine metabolism with enhancement of glucose metabolism through the Warburg effect (Fig. 5A). The top individual metabolites significantly enhanced by the combined PFAS exposures are shown in Fig. 5B and include glycerol, glutamic acid, citric acid, urea, serine, alanine, and glucose. It is important to note that PFOS and PFOA, while overall producing a glycolytic state had distinct profiles for the individual metabolites with PFOA exposure resulting in the greatest changes over a wider range of metabolites as compared to PFOS (Table 2). Further, GSEA of the scRNA-seq data identified genes involved in glycolysis as significantly upregulated by PFOA exposure (Fig. 5C). Together, these findings show that PFAS exposures drive a metabolome shift that favors altered SPC proliferation and cell fate decisions and reprograms a metabolome in prostate SPCs that can act as a determinant in the formation of a malignant state.

Figure 5: Enriched metabolites in PFAS-exposed prostaspheres.

PrEC-derived prostaspheres were cultured through 4 serial passages in the absence or presence of 10 nM of PFOS or PFOA. Passage 4 spheres were analyzed by GC-MS for all metabolites including amino acids. A: The top 25 enriched metabolite sets in PFAS-exposed spheroids compared to vehicles reveals glycine and serine metabolism as the top enriched metabolites. Other markedly enriched metabolite sets include urea, gluconeogenesis, glutathione metabolism, the glucose-alanine cycle, alanine, arginine and proline metabolism and metabolites involved in the Warburg effect. B: A heat map of the most enriched individual metabolites in PFAS vs vehicle exposed spheroids. C: GSEA of transcriptome data in PFOA vs vehicle-exposed prostaspheres reveals a significant enrichment of genes involved in glycolysis following PFOA exposure.

Table 2.

Metabolite enrichment ratios for PFOA- or PFOS-exposed prostaspheres over Vehicle controls.

Metabolites	PFOA/Vehicle	PFOS/Vehicle	PFAS/Vehicle
urea	4.434	1.871	3.153
serine	3.451	2.099	2.775
maltose	3.578	1.563	2.571
glycerol-3-p	2.840	2.082	2.461
diethanolamine	2.355	0.000	1.177
glucose	2.269	2.011	2.140
β-alanine	2.554	1.695	2.125
glutamine	2.081	2.077	2.079
proline	2.385	1.717	2.051
glyoxylic acid	2.607	1.475	2.041
benzoic acid, 2-methyl	2.409	1.654	2.032
threonine	2.376	1.568	1.972
pyroglutamic acid	2.296	1.581	1.938
ornithine	2.862	0.941	1.901
1 -monooctadecanoylglycerol	2.239	1.339	1.789
1 -monohexadecanoylglycerol	2.239	1.237	1.738
succinic acid	1.841	1.553	1.697
ethanolamine	2.053	1.325	1.689
palmitic acid	1.990	1.378	1.684
stearic acid	2.021	1.336	1.679
tetradecanoic acid	2.252	1.063	1.658
phenylalanine	1.885	1.382	1.633
alanine	2.121	1.080	1.601
lactic acid	1.956	1.240	1.598
leucine	1.785	1.345	1.565
glycerol	2.014	1.095	1.555
valine	1.672	1.329	1.501

Metabolite reads from GC-MS analysis of prostaspheres exposed to 10 nM PFOA or PFOS for 4 serial passages were normalized to reads from respective vehicle control groups to derive enrichment ratios for each treatment group. Enrichment ratios for the combined PFAS (PFOA + PFOS) chemicals were also normalized to the combined Vehicle control groups. Enrichment ratios for individual metabolites above 1.50 in the PFAS column are shown.

3.5. Exposure to PFOS increases RWPE-2 xenograft tumor growth in vivo

RWPE-2 cells are a tumorigenic prostate cell line derived from the benign human prostate epithelial cell line RPWE-1 through stable transfection with oncogenic k-RAS [46]. Cultured RPWE-2 cells contain a minor population of cancer stem-like cells that can be enriched in 3-D spheroid culture (Fig. 6A, a&b). When transferred to organoid culture conditions, tumoroids result with enrichment of cancer stem-like cells, identified with high KRT13 expression (Fig. 6A, c&d) [39, 40]. To evaluate the effect of PFOS exposures on tumor growth in vivo, RWPE-2 xenografts were established in athymic nude mice gavaged daily with vehicle or with 10 mg/kg BW PFOS. Over a 40-day period, tumor growth in PFOS-fed mice was significantly greater as compared to controls at each time point (Fig. 6B), with a final average volume of 710 mm³ vs 400 mm³ in the vehicle-fed mice.

Figure 6: PFOS exposures enhance RWPE-2 tumor growth in vivo.

A: Tumorigenic RWPE-2 cells grown in 3D Matrigel culture form spheroids from resident stem-like cancer cells that exist in 2D cultures (a & b). When day 7 spheres are transferred to organoid culture, tumoroids form over 14 days as evidenced by abnormal-shaped nuclei with prominent nucleoli (c). Whole-mount immunocytochemistry identifies basal-like p63+ cells in the periphery and a high proportion of cancer stem-like KRT13+ cells centrally (d). White bar = 50μm. B: RWPE-2 cells were established as xenografts in athymic nude mice given vehicle or 10mg/kg BW PFOS daily. Tumor growth measured over 40 days revealed a significant increase in tumor growth in PFOS-fed mice compared to vehicle controls. N=8 mice/group. ** P<0.0001 versus control.

4. DISCUSSION

In the present study, we identify for the first time that PFAS compounds directly target the human prostate SPC population and in so doing, may contribute to increased carcinogenesis and tumor growth in the prostate gland. Specifically, prolonged exposure of prostaspheres derived from disease-free human prostate epithelial cells to an environmentally relevant level of either PFOS or PFOA significantly increased spheroid numbers with a constant cell number seeded with each serial passage. Since prostaspheres are largely established by resident stem cells under the culture conditions employed herein [39], a consistent increase in spheroid numbers indicates a stimulation of stem-cell symmetric self-renewal in the presence of PFAS that permits stem cell expansion. This is particularly relevant since cancer risk strongly correlates with the number of normal stem cell divisions across multiple tissues, including the prostate [47, 48]. In addition, PFAS exposures led to an increase in larger sized spheres denoting enhanced proliferation of progenitor cells as they comprise the cell majority within prostaspheres [39]. Previous studies have shown that other EDCs, including BPA, dioxin and inorganic arsenic likewise target the human prostate SPCs, likely due to their expression of multiple steroid receptors that make them EDC targets [31, 34, 49]. In the present study, expression of PPARS and RXRs were identified in human prostate SPC with strong levels of PPARα and RXRα which is crucial since many PFAS effects are mediated through PPARα and an RXR binding partner [35–37]. Together, this sets the stage for initiation of downstream signaling effects that alter the SPC phenotypes and behaviors observed herein.

Two non-exclusive downstream actions were characterized in the present studies; PFAS-induced alterations in the transcriptome and metabolome. Using scRNA-seq to identify transcriptomic shifts in SPCs as a function of extended PFAS exposures, we first observe that both PFOS and PFOA result in the emergence of a unique progenitor cell cluster not found in the vehicle-treated spheres. Recent work from our laboratory identified the lineage hierarchy of human prostate SPCs based on cell-type specific shifts in KRT gene profiles during lineage commitment and differentiation [40]. Using these markers to interrogate the new cell clusters that appear upon PFOS or PFOA exposures, they identify as altered early-stage luminal progenitor cells. This is of noteworthy since aberrant luminal cell differentiation may underpin an increased carcinogenic risk in the prostate [50, 51]. Further analysis of the spheroid transcriptome ascertained a significant PFAS-induced upregulation of gene signaling pathways involved in proliferative potential (G2M checkpoint, mitotic spindle, E2F targets) and oncogenesis (k-RAS and MYC signaling, TNFα via NFκB, IL-6-JAK-STAT3, TGFβ and inflammatory signaling). Enhancement of these gene sets in the SPC pool is of special note in the context of a recent report that identified a gene signature in therapeutic-resistant metastatic prostate cancer having stemness-associated pathways including IL-6-JAK-STAT, TGFβ and inflammatory signaling [52]. Collectively, these findings support the potential for early transformation of the prostate SPC population to a pre-malignant state by PFAS and, according to the cancer stem cell hypothesis [53–55], augment the risk for prostate cancer.

Since PPARs are key signaling regulators of cellular metabolism, we interrogated the effects of PFAS exposures on the SPC metabolome. Results identified enhancement of glycine and serine metabolism and enhanced glucose metabolism through the Warburg effect, among other changes in response to PFOA and PFOS exposures. This is further supported by GSEA identification of enhanced expression of glycolysis pathway genes in SPCs exposed to PFOA. To our knowledge, these are the first studies to identify metabolome shifts in prostate SPCs in response to environmental chemical exposures. Metabolic factors are known to be critical in regulating the balance between stem cell self-renewal and differentiation [56] as well as drivers of carcinogenesis and tumor progression [57]. Cultured pluripotent stem cells are reliant on exogenous glucose and glutamine and engage in aerobic glycolysis, consuming high levels of glucose and secreting lactate [58], a phenomenon that is reduced with early differentiation. Similar properties have also been reported for cancer stem cells compared to non-stem cancer cells [59]. As such, further enhancement of glycolysis and lactate production by PFOA in cultured prostaspheres may underlie the increased self-renewal capability in prostate stem cells as well as differentiation perturbation. Another critical stem cell self-renewal/differentiation regulator recently identified in epidermal pre-malignant stem cells is amino acid availability, specifically serine synthesis from glucose [60]. There, activation of glucose flux to serine is a rate-limiting step in pre-malignant stem cell growth [61] suggesting that the metabolic profile of a stem cell is a determinant in the formation of a malignant state. Elevated serine and glycine metabolism have also been reported for prostate cancer stem-like cells [59]. This is particularly relevant since the present studies identified increased serine and glycine metabolism as the top enriched metabolic set by far in prostaspheres exposed to PFOS over 4 weeks of serial passage. Whether this can serve to promote a pre-malignant prostate SPC fate remains to be determined, however it is tempting to speculate that enhanced levels of serine and other amino acids and heightened serine/glycine metabolism due to PFAS exposures may augment stem cell self-renewal, alter normal luminal progenitor cell differentiation, and enhance the tumor-initiating capacity of prostate SPCs.

To directly test the ability of PFAS exposures to modify prostate cancer growth, human-derived RWPE-2 cells (transformed by k-RAS) were grafted into nude mice and daily exposure to PFOS significantly increased tumor growth over a 40-day period. While this treatment directly targets the cancer cells to alter their metabolome and enhance their proliferation [62], we also provide evidence that RWPE-2 cells contain a cancer-stem-like population that is enriched in tumoroid growth. Thus, it is possible that PFOS-driven enhancement of tumor growth may be in part driven through effects of PFOS on the cancer stem cell population as well.

In summary, the present study provides novel evidence that PFAS exposures can directly target human prostate stem and progenitor cells, increasing their self-renewal and proliferative potential and altering luminal progenitor cell differentiation. This is underpinned, in part, by alterations in the SPC transcriptome and metabolome towards pathways that resemble a pre-malignant stem-progenitor cell fate. Since in vivo PFOS exposure was capable of accelerating tumorous xenograft growth from RWPE-2 cells that harbor a cancer stem-like cell population, we propose that direct PFOS effects on the cancer stem cells may contribute to tumor growth and progression. Together, these findings help to understand the reported increase in prostate cancer incidence and mortality observed in PFAS-exposed populations and support the need for additional research on these epidemiologic observations balanced with mechanistic studies to delineate human prostate cancer risks from chronic PFAS exposures.