Fibroblasts in Orchestrating Colorectal Tumorigenesis and Progression

Abstract

Cancer-associated fibroblasts (CAFs) are an abundant component of the tumor microenvironment and have been shown to possess critical functions in tumor progression. Although their roles predominantly have been described as tumor-promoting, more recent findings have identified subsets of CAFs with tumor-restraining functions. Accumulating evidence underscores large heterogeneity in fibroblast subsets in which distinct subsets differentially impact the initiation, progression, and metastasis of colorectal cancer. In this review, we summarize and discuss the evolving role of CAFs in colorectal cancer, highlighting the ongoing controversies regarding their distinct origins and multifaceted functions. In addition, we explore how CAFs can confer resistance to current therapies and the challenges of developing effective CAF-directed therapies. Taken together, we believe that, in this rapidly evolving field, it is crucial first to understand CAF dynamics comprehensively, and to bridge existing knowledge gaps regarding CAF heterogeneity and plasticity before further exploring the clinical targeting of CAFs.

Summary.

This article explores the diverse roles of cancer-associated fibroblasts (CAFs) in colorectal cancer, emphasizing their impact on initiation, progression, and metastasis. It addresses controversies surrounding CAF origins, their multifaceted functions, and the challenges associated with developing targeted therapies, underscoring the importance of comprehending CAF dynamics for future clinical advancements.

Colorectal cancer (CRC) is the third most frequent cancer and second leading cause of cancer-related deaths worldwide.¹ The progression of colorectal tumors is characterized by a series of well-defined genetic alterations that accompany the malignant transformation of epithelial cells within the colorectal cancer mucosa.² Nevertheless, mounting evidence has suggested that CRC development is not driven solely by cancer cells, but is influenced profoundly by the surrounding tumor microenvironment (TME), including cells located in the tumor stroma. Moreover, the consensus molecular subtypes of colorectal tumors highlighted the clinical significance of stroma-related genes as vital prognostic markers, with the mesenchymal (high stroma signature) CRC subtype (consensus molecular subtype 4) showing the most unfavorable survival outcomes.³ Remarkably, stromal gene signatures rather than epithelial ones showed a robust correlation with CRC patient outcomes.⁴

The TME in CRC includes endothelial cells, infiltrating immune cells, and a large number of cancer-associated fibroblasts (CAFs). CAFs are a heterogenous population of mesenchymal cells that reside within and around the tumor mass, and are particularly abundant at the invasive front of CRC. CAFs play an essential role in supporting cancer cell growth by secreting growth factors, cytokines and chemokines, and governing extracellular matrix (ECM) remodeling that regulates angiogenesis, invasion, therapy resistance, and immunosuppression.⁵ As such, CAFs increasingly are recognized as a promising target for cancer therapy. Nevertheless, the endeavor to target CAFs therapeutically has been hampered by the incomplete understanding of their diverse origins, multifaceted functions, and their dual roles in either promoting or inhibiting tumor progression.⁶

In this review, we provide an overview of the current knowledge regarding the evolving role of CAFs in CRC progression, address the ongoing controversies involving the origin and functions of CAFs, highlight the existing knowledge gaps, and explore potential avenues for future research.

CAF Origin and Heterogeneity

To date, the cellular origin of CAFs, independent of whether they are tumor-promoting or tumor-suppressing subsets, is poorly understood. This uncertainty largely is owing to the absence of universally accepted, fibroblast-specific markers, coupled with the diverse and often-conflicting definitions of the markers that fibroblasts are expected to express. Recently, a consensus statement was published defining CAFs as elongated, spindle-shaped cells within a tumor that are negative for the expression of endothelial, immune, and epithelial cell markers; and are positive for mesenchymal markers, such as α-smooth muscle actin (α-SMA), fibroblast activation protein (FAP), and vimentin.⁶ Many years of work, summarized in Figure 1, have shown that CAFs can originate from the following: (1) the proliferation and activation of local resident fibroblasts; (2) circulating bone marrow–derived mesenchymal stem cells5, 6, 7; and (3) epithelial or endothelial cells adopting a mesenchymal phenotype, often indicated as epithelial/endothelial-to-mesenchymal transition. The prevailing theory currently supports that the majority of CAFs in tumors likely originate from the activation of local tissue-resident fibroblasts.⁶ In intestinal homeostasis, fibroblasts typically are quiescent and play a role in providing structural support and inducing differentiation of epithelial cells. Upon damage to the colorectal mucosa, however, fibroblasts readily are activated by multiple factors, including mechanical stress, inflammatory cytokines and chemokines such as interleukin (IL)6, and growth factors such as transforming growth factor-β (TGF-β).⁷ These activated fibroblasts, so-called myofibroblasts, which are positive for α-SMA, proliferate and facilitate wound healing. When the tissue is repaired, those either undergo apoptosis or revert to a quiescent state. However, during CRC carcinogenesis, the activation and accumulation of CAFs, which begins early during carcinogenesis, co-evolves with the malignant cells as the tumor progresses. Recently, it was reported that many α-SMA⁺ CAFs emerge from the proliferation of intestinal pericryptal leptin receptor cells in colitis-associated carcinogenesis in mice (azoxymethane/dextran sodium sulfate model for CRC).⁸ Yet, the sequential changes within the fibroblast compartment and the extent to which early changes in CAF behavior influence the course of tumor development have yet to be fully elucidated. In a recent work, we showed that already at the earliest stages of CRC (T1-CRC), T1-CAFs can regulate cancer cell invasion and show enhanced ECM remodeling.⁹ Moreover, spatial gene expression profiling of T1-CRC lesions showed an increasing predominance of the fibroblast compartment during malignant transformation.¹⁰ Nevertheless, we still have much to learn about how fibroblast composition in the human intestine changes before adenoma formation and whether these changes accelerate or protect against the progression to invasive carcinomas.

Figure 1.

The origins and heterogeneity of CAFs. CAFs form a heterogeneous cell population with diverse origins. Activation of tissue-resident fibroblasts is the largest source of CAFs. Circulating bone marrow–derived mesenchymal stem cells (MSCs) can be recruited to the tumor to become a part of the CAF population. Nonfibroblastic cells, such as epithelial and endothelial cells, can differentiate into CAFs via epithelial-to-mesenchymal transition (EMT) and endothelial-to-mesenchymal transition (EndoMT), respectively. Pericytes also are recognized as a potential source of CAFs through transdifferentiation. Single-cell sequencing has identified 2 major subsets of CAFs in human primary CRC, designated as CAF-A and CAF-B. CAF-A is characterized by marker genes such as FAP, matrix metalloproteinase (MMP)-2, and COL1A2, and is reported to be involved in ECM remodeling. On the other hand, CAF-B expresses marker genes such as α-SMA, transgelin, and platelet-derived growth factor subunit A (PDGFA), but its specific function remains unknown. Additionally, a CAF subset expressing MHC-II, known as antigen-presenting CAFs, may exist in CRC, but their role in T-cell activation and immunosuppression has yet to be elucidated.

Figure created with BioRender.com.

It should be emphasized that heterogeneity is a significant factor impacting both quiescent and activated fibroblasts. This heterogeneity results in a pronounced, context-dependent diversity in their phenotype and functionality. In this regard, a landmark study has shed light on the remarkable cellular heterogeneity within human colonic mesenchyme compartment.¹¹ Alongside myofibroblasts, the same study identified 4 additional fibroblast-like populations (stromal 1–4 cells), each characterized by distinct expression profiles. Intriguingly, patients with chronic inflammation of the colon (colitis) showed a decrease in stromal 2 cells (enriched for Wingless-related integration site (WNT) signaling and an increase in stromal 4 cells (associated with proinflammatory features such as the expression of IL32 and CD74).¹¹ Additionally, in human primary CRC, single-cell profiling uncovered the existence of 2 main CAF subsets, termed CAF-A, expressing high levels of FAP, matrix metalloproteinase-2, and collagen (COL)1A2; and CAF-B, which express myofibroblastic markers such as α-SMA, transgelin, and platelet-derived growth factor subunit α.¹² Although the precise role of these CAF subsets in CRC progression remains uncertain, evidence from clinical trials has indicated that tumor-promoting and tumor-suppressing CAFs are very likely to coexist rather than exclusively dominate the tumor microenvironment.^5,7

Functions of CAFs

Different types of CAF subsets have distinct functions and roles in physiology and disease. Currently, there is no consensus regarding the precise role that CAF subsets play during tumor initiation and progression, given their sometimes-opposing functions. Here, we discuss the fundamental roles of CAFs in actively shaping the development of CRC.

ECM Remodeling, Angiogenesis, and Immune Suppression

A well-described primary function of (cancer-associated) fibroblasts is to produce, deposit, and remodel ECM.¹³ By creating a mechanically stiff ECM, facilitated by collagen cross-linking, CAFs can enhance cancer cell proliferation, invasion, and metastasis.^14,15 Moreover, CAFs can produce angiogenic factors, such as CXC-chemokine ligand 12 and vascular endothelial growth factor, directly driving tumor angiogenesis.¹⁶ CAF-derived WNT2 has been shown to increase angiogenesis in CRC by inducing the secretion of ECM and pro-angiogenic molecules.¹⁷ Additional in vitro functional assays showed that CAFs secrete IL6 to polarize macrophages toward a protumoral M2-like phenotype and CXC-chemokine ligand 12 to recruit and differentiate regulatory T cells, creating an immunosuppressive microenvironment that thwarts the antitumor response.^5,18 These data show that CAFs, through the production of soluble factors, can contribute to various processes related to CRC progression, including angiogenesis and immune suppression, which also involves direct interaction, as discussed next.

Antigen Presenting

CAFs also possess the capacity to function as nonprofessional antigen-presenting cells, potentially influencing T-cell immunity through both major histocompatibility complex (MHC)-I and MHC-II antigen presentation routes. Murine CAFs can sample, process, and cross-present antigens, resulting in the antigen-specific elimination of cytotoxic CD8 T cells as shown in vitro and the B16-F10 murine melanoma model.¹⁹ Our recent findings align with these observations, indicating that human CRC-derived primary CAFs also show the ability to cross-present tumor-derived exogeneous antigens in vitro, directly affecting T-cell activation and dampening the cytotoxic CD8 T-cell function.²⁰ However, the question remains whether cross-presentation by CAFs occurs in vivo. If this would be the case, CAFs in stroma-rich colorectal tumors may hinder proper antigen uptake by professional antigen-presenting cells, indirectly hampering an effective CD8⁺ T-cell response. Additionally, a subset of antigen-presenting CAFs expressing MHC-II, termed apCAFs, lack the required costimulatory ligands for T-cell activation and therefore promote the differentiation of naive CD4⁺ T cells into regulatory T cells in murine pancreatic cancer models.²¹ Interestingly, these apCAFs in lung tumors seem to enhance rather than suppress in vivo MHC-II tumor immunity via CD4⁺ T cells.²² It is important to note that the presence and characteristics of apCAFs may vary, even within the pancreatic tumor microenvironment, where they have been clearly documented.²³ Further investigations are required to fully elucidate the existence and significance of apCAFs across different murine and human tumor types, and their potential impact on antitumor immunity.

CAFs in Early Stages of CRC

In recent years there has been an increasing appreciation that fibroblast protumorigenic states can occur before malignant transformation of cells. CAFs already might be shaped by insults to the colonic mucosa or chronic inflammation, and when mutations arise in epithelial cells, those tissues already might be poised for transformation, potentially influencing the initiation and progression of CRC. Interestingly, it was reported that a rare pericryptal subset of prostaglandin endoperoxide synthase 2 (Ptgs2)-expressing fibroblasts can initiate adenoma formation directly in APC mutant mice using a COLVI-Cre–mediated deletion of Ptgs2 through paracrine signaling.²⁴ However, in healthy human colon, varying degrees of Ptgs2 expression were observed across all fibroblast subsets,¹¹ indicating this observation requires further validation in the human setting. Other work showed that COLVI⁺ fibroblast–specific deletion of the signaling molecule inhibitor of nuclear factor kappa-B kinase subunit beta (IKKβ) leads to reduced tumor growth in azoxymethane/dextran sodium sulfate tumorigenesis via decreasing IL6 production.²⁵ However, deletion of IKKβ in COL1A2⁺ fibroblasts in the same mouse model surprisingly accelerated tumor growth through enhanced hepatocyte growth factor secretion.²⁶ Distinct phenotypes observed in these 2 studies could be attributed to differences in the fibroblast subpopulations targeted in each case and the different underlying mechanisms proposed. These findings also raise the question of how to interpret data from different fibroblast-“specific” Cre driver lines to study CAF biology. Currently, there is no universal and selective fibroblast or CAF-specific Cre driver. Despite this fact, using Cre reporter strains can help assess the role of target proteins in the intended subset of fibroblasts. Overall, a comprehensive understanding of the various fibroblast subsets in the intestine is crucial for unraveling their highly complex and context-dependent roles in the initiation of CRC.

CAFs in Advanced Stages of CRC

Historically, research focus on how CAFs influence tumor progression was primarily on the advanced, invasive stages of CRC, in which CAFs were perceived as potent contributors to tumor progression and metastasis. Numerous CAF subsets as defined by various markers, for instance, α-SMA⁺, CD73⁺, CD105⁺, and IL11⁺ CAFs are found to be tumor-promoting in the preclinical cancer models, with their abundance correlating to poor prognosis in CRC patients.27, 28, 29, 30, 31 Interestingly, a recent study uncovered the tumor-restraining function of α-SMA⁺ CAFs in a genetic mouse model of metastatic CRC, in which selective depletion of proliferating α-SMA⁺ cells resulted in an immunosuppressive TME with a decrease in cytotoxic CD8⁺ tumor-infiltrating lymphocytes.³² Although the depletion strategy used in this study predominantly targeted α-SMA⁺ CAFs, it is unclear to what extent the effects of depleting other α-SMA⁺ cells in vivo may confound this observation. This highlights the ongoing need for refined and targeted approaches in elucidating the precise roles of α-SMA⁺ CAFs in CRC progression. CAFs can produce TGF-β, leukemia inhibitory factor, and hepatocyte growth factor to promote proliferative and invasive behavior of cancer cells.³³ TGF-β acts as a potent inducer of epithelial-to-mesenchymal transition through paracrine signaling, enabling premalignant cells to acquire mesenchymal properties for invasion and metastasis.³⁴ Secretion of IL11 by CAFs can trigger Signal transducer and activator of transcription 3 (STAT3) signaling in cancer cells and increase liver metastasis in CRC.²⁹ In addition, CAFs might play a key role in the colonization of distant organs in metastatic disease by creating a niche for cancer cells or co-traveling with cancer cells.³⁵ Interestingly, IL6 produced by stromal cells in a mouse primary pancreatic tumor has been shown to activate STAT3 signaling in liver hepatocytes, which, in turn, significantly increased the production of serum amyloid A proteins (SAA1/2) by hepatocytes. SAA1/2 can promote myeloid cell accumulation and prometastatic niche formation.³⁶ Importantly, the up-regulation of SAAs by hepatocytes also can be detected in both pancreatic and colorectal cancer patients with liver metastases.³⁶ Unpublished data from our group also indicates a similar mechanism involving the role of TGF-β signaling in CRC CAFs actively contributing to the establishment of a neutrophil-dependent, premetastatic niche.

CAF and Therapeutic Resistance

Emerging evidence underscores the critical role of CAFs in conferring substantial resistance to conventional cancer therapies. The production of ECM proteins and matrix remodeling matrix metalloproteinases by CAFs increases matrix stiffness and intertumoral pressure in addition to forming a physical barrier, hindering the penetration of chemotherapeutics and targeted therapies.³⁷ Furthermore, CAFs secrete key soluble factors, such as IL6, IL17A, and insulin-like growth factor, mediating resistance to chemotherapy and amplifying cytokine secretion after treatment, leading to enhanced resistance.^38,39 Moreover, CAFs play a major role in immunosuppression in the tumor microenvironment. Accumulation of abnormal ECM components further exacerbates immunosuppression and interferes with the efficacy of immune checkpoint inhibitors. TGF-β signaling in CAFs also is implicated in the development of resistance to immunotherapy. Notably, the blockade of TGF-β signaling via TGF-β–receptor kinase inhibitors synergizes with anti–programmed cell death protein 1/programmed cell death ligand 1 immunotherapy in CRC mice with liver metastasis.^40,41 Encouraged by these preclinical findings, various clinical trials are ongoing in which the combination of TGF-β inhibition and checkpoint blockade is being investigated. For example, preliminary data from NCT03724851 has indicated a favorable safety profile with promising efficacy in patients with microsatellite-stable metastatic CRC.⁴² Collectively, these findings emphasize the potential of harnessing the synergy between TGF-β inhibition and immunotherapy, offering a promising avenue for the treatment of advanced CRC.

CAF-Targeting Strategies

Increasing recognition of CAFs as key players in the TME has led to the development of novel therapeutic strategies targeting CAFs. These strategies can be categorized into 3 main approaches: (1) eliminate “detrimental” CAFs⁴³; (2) reprogram CAFs into either a “normal” fibroblast or an antitumorigenic “beneficial” CAF subtype⁴⁴; or (3) block CAF-derived signals or ECM components.^45,46 These approaches require a well-defined and organized classification of CAF subsets and a greater understanding of how different CAF subsets function. Despite extensive research efforts aimed at targeting CAFs, many strategies have yet to demonstrate favorable clinical outcomes.^7,47,48 Regarding direct targeting of specific CAF subsets, various approaches have been attempted. For instance, the depletion of FAP⁺ CAFs using a DNA vaccine that induces CD8 T-cell killing of CAFs has been shown to inhibit tumor progression in a murine model for CRC.⁴⁹ Although promising preclinical data on FAP targeting, clinical trials evaluating monoclonal antibodies against FAP have not shown the expected efficacy in patients with advanced CRC.⁵⁰ The difficulty in clinical translation may be attributed to the considerable functional heterogeneity of CAFs and potential interconvertibility between CAF subsets. Therefore, it is essential to first gain a comprehensive understanding of CAF biology before targeting CAFs for therapeutic benefit of patients.

Future Perspectives

The field of CAFs, particularly CAF subset biology, is evolving rapidly, with an increasing body of evidence suggesting that specific CAF subsets play a significant role in tumor initiation, progression, metastasis, and therapeutic resistance. Powerful technologies such as single-cell sequencing and spatial transcriptomic approaches have enabled the mapping of CAF subsets at a level of detail that was not possible before. Based on these data, we now understand that CAFs from different cellular origins might have functionally distinct phenotypes and biological roles. However, the mechanisms underlying the functional and phenotypic heterogeneity, spatiotemporally dynamic evolution, and plasticity of CAFs still are poorly understood. Although single-cell technology is valuable for identifying CAF subsets, careful characterization and in-depth functional analyses are warranted to validate and contextualize the data for a better grasp of CAF roles and behaviors. Moving forward, research efforts should be put on understanding the dynamic nature of CAFs, exploring their plasticity, elucidating the factors influencing their transformation, and investigating the potential reversibility of these changes. Achieving personalized medicine relies on identifying reliable biomarkers for patient stratification and developing targeted treatments for specific CAF-defined subgroups. Overcoming challenges, including standardizing CAF classifications and addressing patient-to-patient variability, is of utmost importance. Before embarking on CAF-targeted strategies, a crucial step is determining which subtypes are most relevant for specific tumors and stages and evaluating the risks and benefits of CAF interventions comprehensively. Collaborative research and clinical trials will be instrumental in navigating these complexities.