Fibroblast heterogeneity in prostate carcinogenesis

Abstract

Our understanding of stromal components, specifically cancer-associated fibroblasts (CAF), in prostate cancer (PCa), has evolved from considering these cells as inert bystanders to acknowledging their significance as players in prostate tumorigenesis. CAF are multifaceted- they promote cancer cell growth, migration and remodel the tumor microenvironment. Although targeting CAF could be a promising strategy for PCa treatment, they incorporate a high but undefined degree of intrinsic cellular heterogeneity. The interaction between CAF subpopulations, with the normal and tumor epithelium and with other cell types is not yet characterized. Defining these interactions and the critical signaling nodes that support tumorigenesis will enable the development of novel strategies to control prostate cancer progression. Here we will discuss the origins, molecular and functional heterogeneity of CAF in PCa. We highlight the challenges associated with delineating CAF heterogeneity and discuss potential areas of research that would assist in expanding our knowledge of CAF and their role in PCa tumorigenesis.

1. Introduction

Prostate cancer (PCa) is the most prevalent non-cutaneous malignancy among men with an estimated 191,930 new cases and 33,330 deaths in 2020¹. PCa has a strong association with age, with increasing incidence rates in men over 50 years². Clinical presentation of PCa is often asymptomatic. Assessment of prostate specific antigen (PSA) levels in circulating blood allows for the screening of populations over the age of 50 to focus subsequent diagnosis of localized disease. This allows for appropriate management of PCa patients developing aggressive or lethal forms of cancer. Decrease in disease-specific death rate has resulted from advances in the diagnosis and treatment of localized disease. While most PCa patients with localized disease respond well to standard clinical care, the mechanisms responsible for aggressive disease, failure of standard androgen deprivation therapy, radiation and chemotherapy are not fully understood. As a result, some tumors progress to an advanced stage. Management of this aggressive and metastatic disease remains a major clinical challenge.

The prostate stroma comprises the bulk of the human gland. Well-coordinated interactions between stromal and epithelial tissues are crucial for prostate organogenesis and for the maintenance of its normal structure and function^3–5. The stroma in the normal prostate is predominantly composed of smooth muscle cells while in cancer this is partially replaced by fibroblasts and myofibroblasts⁶. These fibroblastic cells are responsible for collagen and extracellular matrix (ECM) deposition, changing local organ stiffness. Concomitant interactions with local nerves and changes to the local inflammatory infiltrate and microvasculature occur as tumors grow. All of these changes are characteristic of a “reactive stroma”. Several decades of basic research support a role for this reactive stroma as a component of the tumor microenvironment (TME) and as a key contributor to PCa progression^7,8. Stromal changes occur early in the pathogenesis of prostate cancer with subtle phenotypic modifications seen immediately adjacent to pre-neoplastic lesions⁶. As PCa develops, the TME undergoes a series of modifications of its cellular and non-cellular elements often contributing to tumor growth. Immediately outside the prostatic capsule a layer of periprostatic fat provides an energy rich environment for invading cells. The specific role of each of these components of TME in PCa have been discussed in detail elsewhere^9–11, here we offer an overview focused on the origin, molecular and functional heterogeneity of normal fibroblasts and carcinoma associated fibroblasts (CAF), the most abundant cell type in the TME.

2. Origin and function of normal fibroblasts

Normal prostate stroma is composed mainly of smooth muscle cells and occasional fibroblasts derived from the fetal urogenital sinus mesenchyme (UGM)⁵. These cells are found in close association with epithelial cells and their constant interaction has been shown to be crucial for the development and maintenance of homeostasis of the organ (Fig. 1A). Development of the prostate is androgen-dependent and activation of androgen receptors (AR) in UGM is required for prostatic epithelial cell proliferation and differentiation¹². Differentiated epithelial cells in turn induce smooth muscle cell differentiation resulting in the major component of the prostate stroma¹³. It has been proposed that once an organ is fully developed, a few mesenchymal cells persist by averting apoptosis and remain in a resting state unless activated by a stimulus¹⁴. These mesenchymal cells are fibroblasts that reside in the stroma of organs, including the prostate gland. In organs such as heart¹⁵, kidney¹⁶ and breast¹⁷, resident fibroblasts have been shown to be in a resting state until activated by stimuli including secreted factors, mechanical stress, or injury. During tissue repair, fibroblasts are activated to acquire a myofibroblast phenotype¹⁸. Fibroblast specific protein-1 (FSP-1), fibroblast activation protein (FAP), platelet derived growth factor receptor α (PDGFRα) and β(PDGFRβ), and smooth muscle alpha-actin (α-SMA) are commonly used markers for identifying activated fibroblasts. However, these proteins are not unique to such cells and can be expressed by various other cell populations.

Figure 1. Fibroblast heterogeneity in normal prostate and prostate cancer tissues.

(A) The stromal tissue of normal prostate is predominantly composed of smooth muscle cells derived from the fetal urogenital sinus mesenchyme, a few resting fibroblasts and two distinct fibroblast subpopulations associated with tissue repair (Sca1+/CD90-) and cell survival (Sca1+/CD90+). Green arrow indicates interaction between epithelial and stromal compartments required for maintenance of organ homeostasis. (B) Stromal composition in prostate cancer (PCa) is dominated by fibroblasts with few smooth muscle cells. These activated fibroblastic cells are collectively known as cancer associated fibroblasts (CAF). CAF are in constant communication with each other, with cancer cells and with the other cells in the local tumor microenvironment (Red arrows). Bone marrow derived stem cells, adipose tissue, circulating fibroblasts, and resident fibroblasts can contribute to the heterogeneity of CAF. CAF in TME secrete growth factors and cytokines that impact cancer development and progression. CAF subpopulations with therapeutic potential are indicated in the box.

The normal wound healing process involves inflammation, proliferation and remodeling. Under normal physiological conditions, activated-fibroblasts commonly referred to as myofibroblasts are recruited to the wound site. These cells produce ECM and provide active contractile action, resulting in wound closure allowing re-epithelialization and resolution^19,20. After the wound completely heals, myofibroblasts either undergo apoptosis or return to a more quiescent or dormant state²¹. To better understand normal stromal composition, several studies have characterized different subpopulations of fibroblasts in normal tissues. A recent study using single cell RNA sequencing (scRNA-seq) analysis of prostatic stromal cells from adult mouse identified three distinct subpopulations that include smooth muscle cells and two subsets of fibroblasts based on the expression of two cell surface markers Sca1 and CD90. Functional characterization of these subsets revealed that Sca-1+/CD90+ fibroblasts have elevated expression of genes associated with epithelial cell survival and growth while the Sca-1+/CD90- myofibroblast-like cells were associated with tissue repair and immune responses²². Although the presence of equivalent subsets in human prostate fibroblast populations remains to be explored, recent comprehensive studies using young adult human prostate samples revealed, in a similar fashion to the murine studies, the presence of a limited number of cells from the mesenchymal lineage consisting of myosin 11 enriched smooth muscle cells, decorin enriched fibroblasts, peri-epithelial fibroblasts (APOD+) observed around epithelial structures and interstitial fibroblasts (C7+) found dispersed in ECM^23,24. The localization of these different subsets in the stroma may be related to their defined functions in tissue homeostasis and immune surveillance. These observations in murine and human tissues suggests that fibroblast heterogeneity in normal prostate is composed mainly of myofibroblasts and fibroblasts (Fig. 1A).

3. Origin of CAF heterogeneity

Malignant tumors have been described as “wounds that do not heal”²⁵ because they can acquire key aspects of the wound healing process but do not undergo a resolution phase. More specifically, activated fibroblasts remain in the “proliferative phase” of the wound healing response constantly fighting to “heal” the tissue by releasing a number of pro-proliferative factors that instead support cancer growth²⁰. The fibroblasts found in the TME are broadly referred as CAF and represent a heterogenous population of cells with potentially different functions^26–28 (Fig. 1B). In tumors, resting fibroblasts are potentially activated in response to a variety of stimuli, present in the TME. Fibroblasts responses in such an environment may be site-specific resulting in different subsets of fibroblasts. These different subsets would potentially express various combinations of markers contributing to, and delineating, fibroblast heterogeneity in normal prostate tissues¹⁴. Several markers have been proposed to be used for the identification, or isolation of CAF populations including FAP, FSP-1, vimentin, CD90 (Thy1), and α-SMA^14,29,30. None of these markers has been validated as CAF-specific, and there is significant overlap with the proteins seen in myofibroblasts in injury, as described above. Examples of expression in non-fibroblasts include, FSP-1 which is present in a specific subset of inflammatory macrophages in liver³¹ and FAP expressed by macrophages in pancreatic ductal adenocarcinoma (PDAC)³². Expression of some combination of these proteins with the lack of any epithelial-or leukocyte-specific cell markers are the classic features employed for the identification and isolation of CAF. Lack of specific markers for CAF (generic, organ-specific, or tumor-specific) poses challenges to identify the different cell sub-populations and their origins.

3.1. Potential sources of CAF:

3.1.1. Mesenchymal stem cells:

Several studies have demonstrated that CAF may be derived from multiple cell types. In an in vivo model of PCa, using human and mouse cells, it was shown that, mesenchymal stem cells (MSC) are recruited from other organs by the primary tumor (Fig. 1B). This process uses CXCL16-CXCR6 mediated chemotaxis followed by conversion into CAF and secretion of high levels of stromal-derived factor (SDF-1/CXCL12)³³. TGFβ can also act as chemoattractant for MSC to the tumor site³⁴. MSC that acquire CAF-like phenotype express α-SMA, FAP, PDGFRβ and vascular endothelial growth factor receptor 2 (VEGFR-2)³⁴. In another study, human bone marrow-derived embryonic stem cells exposed to tumor-conditioned medium from breast cancer cells exhibited higher expression of α-SMA, vimentin, and FSP-1, along with sustained expression of SDF-1, suggesting the acquisition of CAF characteristics. When these activated MSC were xenografted with cancer cells, they were shown to promote tumor growth in nude mice³⁵.

3.1.2. Adipose tissue:

Adipose tissues can be a source for adipose- or stromal- stem cells that may be incorporated into the CAF populations in different tumors^36–39. As tumors breach the prostatic capsule the major adipose depot, they will encounter is the adjacent peri-prostatic fat. Extracapsular invasion of PCa tumors into the peri-prostatic fat increases the clinical stage and is associated with post-interventional recurrence⁴⁰. The interaction of cancer cells with the stromal components of the peri-prostatic fat could provide a favorable niche to further stimulate tumor progression. A recent study showed increased reactive oxygen species (ROS) in PCa cells exposed to free fatty acids. High levels of NOX5 with further activation of the HIF1/MMP14 pathway stimulated PCa invasion in an obesogenic environment⁴¹.

3.1.3. Endothelial cells:

TGFβ-induced endothelial cells from lung and heart undergo endothelial to mesenchymal transition (EndMT) acquiring an spindle-shaped morphology similar to fibroblasts with an increase in FSP-1 expression, providing another source of CAF^42,43. An investigation of whether prostate endothelial cells can also undergo EndMT and differentiate into a CAF-like phenotype is warranted. In addition to these cells, local or resident fibroblasts likely acquire characteristics of CAF during the course of carcinogenesis^44,45.

3.1.4. Senescent fibroblasts:

There is growing evidence indicating that senescent stromal cells including fibroblasts play a role in age-related cancers such as PCa⁴⁶. Shorter telomere length in prostate stromal cells was reported in a prospective study conducted in 596 surgically treated men for localized PCa⁴⁷. This reduction in telomere length in normal stromal cells was further shown to be associated with increased PCa risk in the prostate cancer prevention trial (PCPT)⁴⁸. Senescent prostate fibroblasts have been shown to promote proliferation of prostate epithelial cells both in co-culture and through secretion of paracrine factors⁴⁹. Hypoxia-induced senescent prostate fibroblasts were shown to display CAF characteristics such as increased collagen production and α-SMA⁵⁰. These senescent fibroblasts can induce epithelial to mesenchymal transition (EMT) in prostate cancer epithelial cells, support tumor angiogenesis; and recruit endothelial precursor cells, implicating a role in tumorigenesis. Whether mesenchymal cells become senescent prior to arrival or after recruited to the TME to participate as a subpopulation in the stroma of PCa; are important questions that needs to be addressed with functional studies.

Collectively, these studies highlight the potential of cells from various sources to contribute to CAF populations. Several questions remain, such as the specific signals involved in recruitment of these cells to the TME. The source of such molecular signals: whether originating from cancer cells or other cells of the TME, remain to be identified. Also, once cells arrive at the tumor site, the factors involved in triggering CAF conversion or activation are currently incompletely understood. While the enrichment and contribution of CAF to tumorigenesis is well established, studies exploring the mechanism of action and signaling pathways employed by CAF to promote tumorigenesis are still an active and ongoing field of research.

Despite morphologic and gene expression profile limitations, identification of functionally pro-tumorigenic fibroblasts isolated from PCa patients can be achieved based on their ability to stimulate tumor progression in initiated prostatic epithelial cells in immunocompromised mice (SCID) using a tissue recombination approach^51,52 (Fig. 2). In this in vivo assay, prostate fibroblasts isolated from benign tissues do not stimulate tumorigenesis and are considered normal prostate fibroblasts or NPF. However, CAF induce a range of malignant changes in vivo with growth somewhat correlated with tumor invasion. Differences in tumor size and invasion may reflect inter-patient diversity (host) and/or the nature of the tumor from which these cells were isolated. These effects are the results of a collective effort by the different fibroblast subsets present in CAF. Development of similar functional in vitro or in vivo assays in combination with CAF enrichment methods using techniques such as cell sorting would help with the characterization of the biological roles of each CAF subpopulations in prostate carcinogenesis.

Figure 2: Isolation and “functional” in vivo characterization of tumor-promoting CAF from human PCa tissue.

(A) Post-surgical removal of prostate tissue it is subjected to enzymatic and mechanical digestion followed by fibroblast enrichment in cell culture to isolate the stromal components. Isolated fibroblasts are maintained in culture followed by xenografting (under the kidney capsule) in mice with benign initiated epithelial cells (BPH1). (B) Fibroblasts that induce the malignant transformation of pre-malignant BPH1 cells and formation of invasive tumors (+growth/invasion) are characterized as cancer associated fibroblasts (CAF). Tumor growth and invasion was calculated as previously described⁷⁶. The degree of growth/invasion (small tumors vs large tumors) reflects inter-patient variability. Normal prostate fibroblasts (NPF) isolated from benign prostate tissue result in small to no tumor formation (-growth/invasion). Blue arrows indicate tumor (gross picture of the whole kidney) and asterisks the corresponding areas in histological images (H&E) surrounded by kidney parenchyma. (C) Analysis of the resultant tumors indicates a significant increase in graft size in CAF xenografts compared to that of NPF. Once tumors are formed under the influence of CAF, growth and invasion show a linear correlation.

3.2. Epigenetic modifications

In PCa, the relevance of genetic changes or aberrations associated with cancer biology have been widely studied in epithelial cells⁵³. Stromal cells are considered to be relatively stable with respect to genetic alterations. However, other mechanisms may account for the activation of these cells. Epigenetic changes are modifications such as DNA methylation or post-translational modifications of histones that alter chromatin structure thereby regulating access for transcription factors to genes. Increased levels of DNA methyltransferase enzymes was found to be associated with PCa tissues and PCa cell lines compared to benign prostate epithelial cells^54,55. While interactions with surrounding cancer cells or the host-environment most likely direct the functional heterogeneity of CAF, several reports suggest a role for epigenetic modifications in stabilizing the molecular and functional characteristics of CAF subsets in various cancers^56,57. Our group has previously reported on the ability of CAF but not fibroblasts derived from normal prostate, to promote tumor progression^51,52,58 (Fig. 2). In vivo tumor promoting ability of CAF isolated from human tissue samples suggests that these cells retain characteristics from the tissue of origin and promote tumorigenesis even following culture and in the absence of the primary tumor.

Upregulation of DNA methyltransferases (DNMT) consistent with aberrant promoter hypermethylation of tumor-suppressor genes in multiple cancer types including PCa has been reported^59,60. Prostate CAF demonstrate a distinct methylation pattern representative of epigenetic modifications when compared to patient-matched non-malignant fibroblasts (NPF)⁶¹. Genome-wide DNA methylation studies in 18 matched CAF and NPF from patients with moderate- to high-grade PCa revealed differential methylation profiles in more than 1000 DNA regions⁶². One particular hypomethylation in the EDARADD promoter, an adaptor protein involved in development of ectodermal tissues such as hair, teeth and mammary gland was found in high grade tumors and induced increased transcript levels and protein expression in tissues. EDARADD methylation and gene expression were associated with poor prognosis⁶². Similar hypomethylation profile of EDARADD in tumor-associated fibroblasts was also reported in non-small cell lung cancer⁶³.

GSTP1 is another gene that is epigenetically modified in PCa⁶⁴. Comparison of epithelium and stroma isolated from tumor and normal regions from same patients revealed GSTP1 methylation in both tumor-associated epithelium and tumor-associated stroma^65–67. Quantification of GSTP1 methylation using pyrosequencing followed by three-dimensional reconstruction of the prostate revealed that tumor-associated stromal cells were methylated only in distinct anatomical sub-fields of the tumor⁶⁸. Whether GSTP1 promoter methylation represents a heterogeneous CAF subset in PCa remains to be explored. Similar analysis of PCa samples showed that while GSTP1 methylation was observed in both epithelial and stromal cells, TGFβR2 methylation was prevalent in stromal cells⁶⁹. In the same study, pharmacologic and transgenic knockout of Tgfβr2 induced Gstp1 promoter methylation in mouse prostatic stromal cells⁶⁹. This suggests that disruption of Tgfβr2 gene expression in fibroblastic cells may be a precursor to stromal methylation of the promoter. Such activity could support cancer progression through silencing of reactive oxygen metabolizing and DNA damage repair genes, suggesting a sequence of stromal evolution in its association with cancer epithelia⁶⁹. The loss of TGFβR2 in a subpopulation of human prostate stromal cells is well documented and is associated with hypermethylation of GSTP1 and TGFβR2 promoters^69–71. These studies demonstrate that close association and cross communication of CAF with PCa cells results in the evolution of a CAF-subpopulation that supports or promotes cancer progression. Promoter hypermethylation and silencing of the RasGAP and/or RASAL3 genes results in the activation of Ras signaling in CAF with increased micropinocytosis-mediated glutamine synthesis. CAF-secreted glutamine causes PCa epithelial cell proliferation and promotes transdifferentiation into a neuroendocrine phenotype observed in advanced disease⁶⁴. The continuous stress from various stimuli in a cancer environment could potentially contribute to epigenetic changes in CAF by silencing or activating genes, microRNAs or long non-coding RNA sequences promoting a tumor-supportive TME.

3.3. Functional heterogeneity of CAF subpopulations

The stroma of PCa is heterogeneous with respect to the cell types present and contains functionally diverse populations of fibroblasts^29,73. CAF are involved in increased ECM production and remodeling to induce, either through direct cell-cell contact or via soluble factors, motility and proliferation of PCa cells⁷⁴. Evidence of functional heterogeneity in prostate CAF subpopulations was studied using few mouse models and fibroblasts isolated from PCa patients.

The effect of TGFβ signaling activation in fibroblast-subpopulations on prostate epithelial malignant transformation was first demonstrated in 2004 by conditional inactivation of TGFβ type II receptor (Tgfβr2) in mouse fibroblasts using an Fsp-cre construct to drive stromal deletion of TGFβR2 in a floxed mouse model⁷⁵. This impaired ablation (about 50% of fibroblasts showed Cre-recombination) of TGFβ signaling was shown to be associated with the development of preneoplastic lesions in the prostate epithelia potentially mediated by hepatocyte growth factor (HGF) signaling pathway as one of a number of phenotypes, the most marked being cancer in the mouse forestomach⁷⁵. Prostates from mice with incomplete stromal inactivation of Tgfβr2 were grafted to wild type hosts and progressed to adenocarcinoma after seven months⁷⁰. Conditioned media from Tgfβr2 inactivated-stromal cells induced proliferation of PCa cell line, LNCaP in a Wnt3a mediated pathway underlining the importance of stromal heterogeneity in regulating epithelial phenotypes. These observations also focus the previously noted changes in TGFβR2 expression observed to be absent in around 60% of human PCa stroma⁷⁰. Manipulation of TGFBR2 signaling in a subpopulation of human prostate fibroblasts altered their secretome, including higher levels of TGFB1 and CXCL12/SDF1 and promoting tumorigenicity⁷⁶. Inhibition of TGFβ in CAF was shown to impair their tumor promoting ability⁷⁷. These data indicate that a heterogenous stromal environment with its altered secreted factors can promote the growth of robust and invasive tumors^76,78. This suggests that a fine balance of stromal heterogeneity and the associated secretome dictates the degree of cancer progression promoted by stromal cells (Fig. 2B). It has been shown that SDF-1/CXCL12, secreted by CAF, acts via TGFβ–mediated upregulation of CXCR4 in epithelial cells to activate Akt signaling associated with tumorigenesis. This interplay of TGFβ, SDF-1/CXCL12, and CXCR4 contributing to PCa progression makes them potential targets for therapeutic strategies^76,77. Overexpression of SDF-1/CXCL12 and TGFβ1 in benign human prostate fibroblasts has been shown to induce malignant transformation of benign prostate epithelial cells and formation of highly invasive tumors in vivo⁷⁶. These engineered fibroblasts parallel the action of patient-derived CAF in this model. Consistent with this observation, suppression of the SDF-1/CXCL12 receptor, CXCR4, in epithelial cells significantly impaired CAF-induced tumor formation in vivo⁷⁷. Using scRNA-seq analysis of cultured human prostate-derived CAF, we recently demonstrated that differential gene expression identified six different CAF subsets, among which CCL2 and SDF-1 were highly expressed in different subpopulations²⁹. PCa patients with elevated CCL2 levels were shown to have higher risk of metastasis and poor prognosis^79,80. Consistent with high CCL2 expression in CAF, in vivo tissue recombination of CAF and BPH1 cells revealed a significant increase in myeloid cell recruitment to the tumor site. Neutralization of CCL2 using a monoclonal antibody significantly impaired CAF paracrine induction of THP-1 migration²⁹.

In other words, the consequences of exposure to CAF cells may be a function of interactions between subpopulations of these cells. Studies including PCa patients with various grades of disease provided an insight into interpatient CAF heterogeneity. Functional heterogeneity among CAF isolated from different patient samples was documented in a study co-culturing PCa cells (LNCaP) with fibroblasts isolated from ten different PCa patients. This study revealed that only a subset of patient-derived fibroblasts were capable of inducing an invasive phenotype and increased cadherin 2 (CDH2) mRNA expression in LNCaP cells⁸¹. This highlights the need for further research to better understand inter- and intra-patient CAF heterogeneity and its relevance to PCa progression in different human subjects.

Functional heterogeneity of CAF has been reported in other cancers with most detailed work using inbred mouse models. While CAF are most commonly associated with a pro-tumorigenic role, some populations may exert beneficial effects. For example, depletion of α-SMA+ myofibroblasts in a PDAC mouse model resulted in more invasive tumors associated with poor survival⁸². In addition to demonstrating the protective role of α-SMA+ myofibroblasts in this model, this study also reported no change in FAP+ fibroblasts that have been shown to promote PDAC progression indicating functionally heterogenous CAF subpopulations^82,83. Multiple CAF subpopulations were reported in other diseases such as PDAC, breast cancer and cholangiocarcinoma including but not limited to myogenic fibroblasts (myCAF), inflammatory CAF (iCAF) and antigen-presenting CAF (apCAF)^84–90. In PDAC, apCAF express MHC class II protein and display an immunomodulatory role avoiding optimal T cell response while iCAF display an immunosuppressive role⁹¹. In patient samples with cholangiocarcinoma, myCAF was shown to express hyaluronan synthase 2 and promote intrahepatic cholangiocarcinoma (ICC). Another CAF subtype, iCAF displayed hepatocyte growth factor expression and enhanced ICC growth via tumor-expressed mesenchymal to epithelial transition (MET)⁸⁶. It still needs to be defined whether these CAF subtypes are organ-specific or disease-specific, and compared with the subtypes identified in PCa. It is notable that no formally standardized terminology for these cells has been developed and observations between groups may reflect different functions of the same populations of cells. Whether α-SMA+ or FAP+ populations display similar behaviors in PCa is unknown. In human, the balance between benign and malignant cues from the stroma was shown previously in an in vivo model with varying proportions of CAF isolated from PCa patients and rUGM (rat urogenital sinus mesenchyme) combinations and the BPH1 reporter cell line. In this study, the well-organized benign epithelial cords formed under the influence of rUGM were shown to decrease along with formation of less organized epithelium ultimately resulting in an invasive phenotype as CAF proportion increased⁷⁶. These observations suggest that a constant interplay between the pro-tumorigenic and anti-tumorigenic CAF subpopulations and their interaction with cancer cells influences the fate of cancer progression. It should also be noted that mouse models play a valuable role in establishing potential roles for CAF subpopulations because they have a consistent genetic background and tumor-initiation route. However, unlike murine models, in human patients each tumor and host genomic background is unique, suggesting the need to identify high incidence cellular phenotypes with potential translational applications.

3.4. CAF subpopulations as therapeutic targets

Exploration of the utility of targeting tumor stroma, especially CAF as a potential therapeutic option has progressed since the very first clinical trial using murine monoclonal antibodies (mAb) against FAP+ CAF in metastatic colon cancer⁹². Therapeutic strategies continuously evolve to combat the complexity presented by the molecular, functional, and spatial heterogeneity of CAF in various cancers^93,94. Even with the advent of novel technologies, many types of cancer, including PCa, can still evade treatment and progress to aggressive disease. It has been shown that CAF provide protection to tumors in multiple ways, rendering the cancer cells resistant to therapeutic options and promoting an aggressive phenotype. Mechanisms underlying CAF-mediated protection against commonly used chemotherapeutic drugs demonstrates the possibility of multiple functional mechanisms. For example, such studies on the effects of doxorubicin and taxol in PCa demonstrated specific responses⁹⁵. CAF or CAF-CM inhibited drug-induced cancer cell death by blocking the uptake of doxorubicin by LNCaP cells. In contrast, when taxol was used, a significant decrease in drug-induced cell death was also observed in LNCaP cells under the same conditions but this was achieved without impairing drug uptake by cancer cells. In this study, CAF-secreted soluble factors exerted a protective effect against DNA-damage and ROS generation in drug-treated cells⁹⁵. CAF-induced protection of cancer cells from therapy identifies TME as a valuable target for cancer-treatment, especially in resistant tumors. Unfortunately, markers used to identify CAF are shared by other cell types, therefore specific targeting of CAF or of specific CAF subpopulations still represents a significant challenge. A recent study identified CD10 and GPR77 cell-surface molecules as markers associated with a CAF subpopulation that renders chemoresistance and promotes tumor progression in breast and non-small cell lung cancers⁹⁶. Blocking GPR77 with a neutralizing mAb substantially reduced the infiltration of CD10⁺GPR77⁺ CAF, decreased tumorigenesis and enhanced chemosensitivity in a patient derived xenograft model of breast cancer⁹⁶. In patients with PDAC, upregulation of the glutamatergic presynaptic protein NetrinG1 (NetG1) was observed in CAF. Targeting NetG1 in vitro and in vivo in mice impaired tumorigenesis, revealing a potential therapeutic target for PDAC⁸⁵. Another subset of CAF that express the leucine-rich repeat containing 15 (LRRC15) protein; LRRC15+ CAF are enriched in PDAC patients. Increase in LRRC15+ CAF subpopulation was associated with poor response to immunotherapy, identifying this subset as a potential target for combination therapy for PDAC⁹⁰ (Fig. 1B). Studies looking at the presence of similar CAF subtypes in PCa needs to be explored. In addition to these fibroblast sub-populations in different types of cancer, recently developed tools such as scRNA-seq might identify novel tumor-specific marker combinations that can be targeted. Many advanced PCa tumors escape androgen-targeted approaches and may acquire a neuroendocrine phenotype reducing therapeutic options. The role of similar populations in PCa therapeutic resistance in androgen deprivation therapy (ADT) or chemotherapy remains to be determined.

4. Challenges/ Questions that remain

As new subpopulations of CAF are identified in the TME of cancers, including PCa, there are several questions that naturally emerge with new challenges that need to be addressed. In vitro cell culture models, utilizing cells isolated from patients, in vivo xenograft studies and genetically engineered mouse models (GEMM) are unquestionably valuable tools for research. However, there are significant caveats to all of these models for the study of CAF heterogeneity in PCa. Direct assessment of biopsy and/or prostatectomy samples from PCa patients using scRNA-seq analysis, for example, can provide a detailed insight into the CAF composition, degree of heterogeneity, and the presence of other cell types. However, such complexity is difficult to replicate in models. Observations comparing scRNA-seq data from freshly isolated CAF populations from patients to those that have been maintained in culture depicts a considerable variation in CAF subpopulations (Fig. 3). The continued ability of such cultured CAF to subsequently drive tumorigenesis in vivo potentially reflects epigenetic changes of the sort described in section 3.1 that may be preserved throughout the modeling cell isolation, culture, and recombination process. It is well accepted that primary cells can retain, for a short period of time, the physiological state of cells in vivo and are expected to maintain their in vivo functions under optimal cell culture conditions. Due to their intrinsic plasticity, the culture conditions (type of cell culture media, specific nutrients, 2D plastic plates vs 3D ECM) can favor the selection of stromal subclones and may not fully recapitulate the patient TME. Future studies of CAF will need to consider this variation in cultured cells isolated from PCa patients posing a challenge in studying CAF heterogeneity using standard approaches. Some major limitations of current research are the under sampling of prostate tissue. The intra-tumor composition of cells and their heterogeneity is limited to the biopsied section of the tumor and may not represent other areas of the same tumor or the presence of different patterns of stromal heterogeneity in multifocal types of PCa⁹⁷. GEMM have been shown to be useful in studies of functional properties of selected subtypes or specific signaling in CAF, yet they may not represent the diverse TME observed within and between human patients. Despite these limitations, studies investigating molecular and functional characteristics of CAF isolated from patients are relevant and continue to provide significant insights into the intra-tumoral and inter-patient CAF heterogeneity^29,81,96. We have previously shown that lipid metabolism in prostate CAF is altered compared to normal prostate fibroblasts⁹⁸. However, if this metabolic state is common to all heterogeneous CAF subpopulations or only limited to a specific subset of CAF remains to be addressed. Racial disparities in TME of several cancers have been reported⁹⁹, if these disparities reflect different CAF heterogeneity profiles is a question that needs to be explored.

Figure 3. scRNA-seq analysis of CAF freshly isolated from prostate cancer tissue versus cultured cells.

Uniform Manifold Approximation and Projection (UMAP) plot comparing freshly isolated fibroblasts from a prostate cancer patient versus cultured CAF from another prostate cancer patient (unpublished data). Fibroblasts were isolated from the prostatic peripheral zone, followed by enzymatic digestion and either fluorescence activated cell sorting (EpCAM-CD45-CD200-CalceinAM+ cells) or culturing/passaging of cells. Cells were prepared for scRNA-seq analysis using 10X Chromium (v2.0 for cultured CAF and v3.0 for freshly isolated CAF). Bioinformatics analysis was conducted using CellRanger, R, and Bioconductor, and visualization of naïve clustering was completed using Seurat.

As new data are acquired, we should be able to answer some critical questions in relation to fibroblast heterogeneity such as: what are the driving force(s) or factors that determine CAF subpopulations? Are cancer cells responsible for dictating CAF subtypes as disease progresses or are subtypes directed by the host as a tumor-suppressing mechanism? Can we define CAF subpopulations that are responsible for promoting or restraining PCa growth? Do CAF subpopulations transition from one type to another? How do we model the complex interactions between the stromal, epithelial and leukocytic components of the tumor, all of which can be playing positive and negative controlling roles? There is a real need for strong interactions between bioinformaticians and mathematical modelers simplify complex data sets without investigator bias. There is a notion that both cancer cells and the neighboring stroma co-evolve as tumor progresses. It is reasonable to believe that each clone and subclone participates in the overall CAF promotion of carcinogenesis. However, are all subpopulations functionally active at all times or are there some “dormant” cells that are awaked (or that emerge) when needed? Longitudinal studies to observe the evolution in the CAF subpopulations as disease progresses from localized to metastatic should give us some clues. The role of CAF subpopulations in chemo-resistance suggests a discrete function for each clone that relates or is influenced by a particular situation. These and other future questions should be addressed to be able to develop the most effective therapeutic approaches for PCa that target CAF in the TME.

Highlights.

• Prostate cancer stroma contains diverse populations of fibroblasts

• Carcinoma associated fibroblasts can promote prostate carcinogenesis

• Interaction among fibroblast subsets modulate stromal-epithelial paracrine signaling