As a key component of the TME, tumor-associated macrophages (TAMs) play critical roles in the regulation of antitumor immune response. TAMs have been sub-classified into two distinct subtypes. Type I macrophages (or M1) secrete important amounts of pro-inflammatory cytokines and ROS and promote a T-helper 1 (Th1) anti-tumor immune response. On the contrary, type II macrophages (or M2) promote tumor progression and are characterized by the secretion of factors with immune-suppressive activity such as TGFβ, IL-10, Arg, and IDO, which particularly affect cytotoxic CD8⁺ T cell functions [78]. Interestingly, in oral squamous and colorectal cancers, CD163⁺/DC-SIGN⁺ M2 macrophages are the most prominent immune cells in the neighborhood of αSMA⁺, FSP1⁺, and FAP⁺ CAF-rich areas, suggesting a close relationship between these two cell populations, with important consequences on the clinical outcome for patients [79, 80]. Further evidence was provided by several studies which have demonstrated that the recruitment of monocytes into the TME and their differentiation toward M2 subtype macrophages are actively promoted by CAFs [81], especially through their secretion of CXCL12/SDF1, macrophage colony-stimulating factor (M-CSF)/CSF-1, IL-6, CCL2/MCP-1 and chitinase-3-like-1 (Chi3L1)/YKL-40 [82–92]. However, and in an intriguing way, CAFs might also alter TAMs infiltration under certain circumstances, by a FAP-mediated modification of the ECM [93]. Finally, it is important to note that reciprocal crosstalk exists between CAFs and TAMs. Indeed, several studies have suggested that M2 macrophages can regulate CAFs generation, for example by enhancing EMT progression through IL-6 and SDF1 [83], or by influencing the trans-differentiation of MSCs into CAFs [94, 95].

Recent evidence indicates that TANs represent a significant component of the TME [96, 97] and several studies have suggested that TANs can be polarized to an N1 anti-tumoral or N2 pro-tumoral subtype, as observed for TAMs. N1 neutrophils differentiate following TGFβ blockade and express immuno-activating cytokines and chemokines, low levels of Arg 1, and can kill cancer cells. On the opposite, N2 neutrophils are induced following exposure to high TGFβ levels [98], are characterized by expression of CXCR4, VEGF, and MMP9, and can inhibit CD8⁺ T cell function [99]. Of note, TANs have been linked to a poorer prognosis for patients with renal and pancreatic cancer; gastric, hepatocellular, colorectal, head and neck carcinomas, and melanoma [100, 101]. A few studies have highlighted the crosstalk between CAFs and TANs. For example, CAF-derived CXCL12/SDF1 and CXCR2 are involved in TANs recruitment within the TME and CAF-derived IL-6 stimulates signal transducer and activator of transcription 3 (STAT3) signaling pathway in TANs, potentially inducing immune tolerance through the expression of PD-L1 [102]. CAF-secreted TGFβ can also probably redirect TANs differentiation toward an N2 phenotype [98]. Furthermore, in hepatocellular carcinoma, CAF-derived CLCF1 increases CXCL6 and TGFβ secretion by tumor cells, which subsequently promotes TAN infiltration and polarization [103]. It seems that CAFs can also induce pro-tumorigenic neutrophil extracellular traps (NETs) formation in an amyloid β-dependent manner [104]. In a mouse breast tumor model, it was also shown that CAF-derived IL-33 facilitates lung metastasis by the recruitment of TANs [105]. Interestingly, the N2 polarization is also increased by vascular mimicry between CAF and cancer cells [106]. Finally, reciprocal crosstalk probably exists between CAFs and TANs. For example, neutrophil NETs can promote liver micro-metastasis in pancreatic ductal adenocarcinoma via the activation of CAFs [107] and TANs are capable to promote the differentiation of MSCs into CAFs [108]. It was also shown that the CAF marker PDPN interacts with the neutrophil protein CD177, with possible implications for CAF functions [109].

MCs are tissue-resident sentinel cells that, upon activation, release a wide spectrum of chemokines and cytokines. MCs are mostly known for their role in an allergy but can also modulate tumor initiation and progression. Depending on their localization or cancer type, MCs exert dual effects on tumor progression [110]. As such, it seems that MCs display two subtypes, anti-tumorigenic MC1 and pro-tumorigenic MC2, which produce different mediators with opposite roles in tumorigenesis. In particular, MC1 produces IL-9, and histamine, which induces DC maturation and inhibits tumor growth in murine models. In contrast, MC2 produces a variety of angiogenic and metastatic substances, including VEGF, FGF, MMP9, TGFβ, and cytokines (IL-1β, IL-6, and IL-13) [111, 112]. Importantly, MCs can also alter the anti-tumor immune response. For example, the release of free adenosine [113] or IL-13 by MCs can respectively inhibit T cell function and promotes M2 polarization [114–116]. MCs can also favor the generation of highly suppressive MDSCs and Tregs in the TME [117, 118]. To date, research on the cooperation between MCs and CAFs in tumors is still in its infancy, with only a few studies addressing this question. For example, in odontogenic lesions that affect the jaw or neurofibroma, a large number CAFs and MCs in tumor islets are associated with the aggressiveness of the disease [119, 120]. In pancreatic tumors, stellate cells (a CAF precursor) can activate MCs which in turn enhance CAF proliferation by their secretion of IL-13 and tryptase. This process results in the formation of a fibrotic TME and ultimately suppresses the antitumor immune response [121]. Finally, in an in vitro three-dimensional (3D) microtissue model of prostate cancer, a recent study has revealed cooperation between MCs and CAFs, which enhances the transition from a benign to an anormal epithelia via a tryptase-dependent mechanism [122].

In the TME, important antigen-presenting cell subpopulation, known as DCs, have a pivotal role in the activation of T cell-mediated, adaptive, anti-tumor immunity [123] and their global biology can be affected by the CAFs, even if in-depth mechanisms remain poorly understood. As a major source of TGFβ in the TME, CAFs can probably affect DC functions, in particular through the inhibition of MHC class II molecules, co-stimulatory molecules (CD40, CD80, and CD86), and cytokines (TNF-α, IFNγ, and IL-12) expression/secretion [124], which alter CD8⁺ cytotoxic T cell activation and Th1 polarization of CD4⁺ Th cell populations, and also promote the formation of CD4⁺ forkhead box protein P3 (FoxP3)⁺ Treg cells that potently inhibit the function of other T cells [125, 126]. Similarly, in hepatocellular carcinoma, CAFs have been described as a major source of IL-6 that affects DC functions through the activation of the STAT3 pathway leading to the generation of regulatory DCs, characterized by low expression of costimulatory molecules and high secretion of immune-suppressive cytokines, which impair T-cell proliferation and promote Tregs expansion [127]. Furthermore, CAF-produced IL-6 can also favor the emergence of pro-tumorigenic TAMs from monocytes at the expense of DCs [82]. Interestingly, in lung tumors, galectin1-driven secretion of TDO2 and IDO by CAFs promotes tryptophan degradation in kynurenines that inhibits DCs differentiation and functions [128]. In pancreatic tumors, the secretion of TNF-α and IL-1β by tumor cells promotes CAFs activation and their secretion of TSLP, which favor the generation of DCs with Th2-polarizing capabilities, associated with reduced patient survival [129]. In mouse esophageal squamous cell carcinoma, CAFs-secreted Wnt family member (WNT2) has been linked to suppression of the DC-initiated antitumor T-cell response via the suppressor of cytokine signaling 3 (SOCS3)/phosphorylated Janus kinase 2 (p-JAK2)/phosphorylated STAT3 (p-STAT3) signaling pathway. On the opposite, anti-WNT2 monoclonal antibodies (mAbs) can significantly restore T-cell responses and enhance the efficacy of anti-programmed cell death 1 (PD-1) therapy by increasing active DCs [130]. Furthermore, in ovarian cancers, CAFs can secrete wingless-type mouse mammary tumor virus integration site 16B (WNT16B) in response to DNA damage-associated treatment, which promotes the secretion of IL-10 and TGFβ by DCs [131]. Finally, as a major source of VEGF, CAFs might inhibit DC generation, maturation, and functions through this pathway [132].

CAFs can also alter the activity of NK cells, which are a major participant in the early immune response through their cytotoxic functions, and contribute to the adaptive immune response through their secretion of cytokines and the promotion of DC maturation. The detailed mechanisms of the complex relationship between CAFs and NK cells are still emerging and most likely involve multiple molecules. As such, TGFβ released in the TME by CAFs most likely plays an important role in the alteration of NK cell activation and cytotoxic activity [133], for example by reducing NK-activating receptor expression [134–136]. Furthermore, more direct evidence of the effect of CAFs on NK cells has been provided during the past few years. Independent studies involving melanoma, colorectal, and hepatocellular carcinoma-derived fibroblasts have shown that CAFs, through the secretion of PGE2 and IDO, can decrease the expression of several natural cytotoxicity receptors [NCRs, e.g., NKp30, NKp44 and NK receptor DNAX accessory molecule (DNAM)] at the NK cell surface, as well as perforin and granzyme B [137–139], leading to attenuated cytotoxic capabilities of NK cells. We also demonstrated that melanoma-associated CAFs decrease the sensitivity of melanoma tumor cells to NK cell-mediated killing through the secretion of MMPs which cleave MHC class I-related chain (MIC)-A and MIC-B [two ligands of NK group 2D (NKG2D)], at the surface of the tumor cells and consequently decrease both NKG2D-dependent cytotoxic activity of NK and their secretion of IFNγ [140]. In pancreatic ductal models, the high expression of the glutamatergic pre-synaptic protein netrin G1 (NetG1) in CAFs is also linked to their ability to inhibit NK cell-mediated killing of tumor cells [141]. Furthermore, in endometrial cancer, CAFs can decrease NK cells’ lytic potential through their downregulation of poliovirus receptor (PVR/CD155), a ligand of the activating DNAM-1/CD226 [142]. Finally, in the context of radiotherapy, CAFs isolated from non-small cell lung cancer inhibit NK cell activation and cytotoxic functions [143].

In summary, due to their secretion of cytokines, chemokines or other soluble factors, and possibly other mechanisms, CAFs shape the TME and favor the recruitment of innate immune cells and their acquisition of an immunosuppressive phenotype like M2 macrophages, N2 neutrophils, possibly MC2, but also affect DC functions or cytotoxic potential and cytokine production of NK cells.

As mentioned above, CAFs are an important source of TGFβ in the TME which acts on both CD8⁺ and CD4⁺ T cells [124, 126] and consequently hamper the antitumor T cell-dependent immune response and the response to immunotherapies. For example, in breast cancer, one of the cellular clusters identified among FAP^HIGH CAFs is characterized by TGFβ signaling and is linked to immunosuppression and resistance to immunotherapy [70]. Similarly, poor response to immunotherapies in the metastatic urothelial, lung, and colon cancer and melanoma have been linked to TGFβ signature in CAFs [145, 146]. Furthermore, in an ovarian cancer cohort, it has been shown that the key determinant of T cell exclusion is the up-regulation of TGFβ in the activated stromal compartment [147]. Mechanistically, TGFβ is known to have pleiotropic “bad” effects on the T cell-dependent immune response. This includes the alteration of effector CD8⁺ T cell survival through the inhibition of the pro-survival protein B cell lymphoma-2 (Bcl-2) expression [148], the reduction of CD8⁺ T cell cytolytic functions through the reduction of perforin, granzymes A and B, Fas ligand (CD95L) and IFNγ expression [149, 150], the reduction of CD8⁺ T cells infiltration [145], the alteration of the acquisition of effector function by memory CD8⁺ T cells [149, 151] or the promotion of Tregs recruitment and differentiation [152]. As such, TGFβ-secreting myCAF is very abundant in immune-excluded ovarian tumors [153], and αSMA⁺FAP⁺ CAFs from head and neck tumors have been shown to inhibit CD8⁺ T cell proliferation and to promote the recruitment of Tregs in a TGFβ-dependent manner [154]. Of note, it has been suggested that CAFs and Tregs enter into a reciprocal cross-talk via their mutual expression of TGFβ, increasing in parallel CAFs activation and Tregs activity [155].

Furthermore, CAFs are also an important source of cytokines and chemokines in the TME, with once again a potential pleiotropic effect on T cells. For example, in αSMA⁺FAP⁺ CAFs from head and neck tumors mentioned above, IL-6 secretion cooperates with TGFβ to inhibit CD8⁺ T cell proliferation and promote the recruitment of Tregs [154]. Similarly, in murine PDAC models, IL-6 depletion specifically in αSMA⁺ CAFs synergizes with anti-PD-1 immunotherapy to significantly improve the survival of tumor-bearing mice [156]. In breast cancer, CAF-derived IL-33 has been identified as a driver of the Th2-polarized immune response [25]. Furthermore, in lung and pancreatic tumors, the secretion of CXCL12/SDF1 by CAFs contributes to the exclusion of T cells from the cancer cell proximity [157, 158]. Similarly, in high-grade serous ovarian cancers, CAF-S1 increases the attraction, survival, and differentiation of Tregs via microRNA-141/200a (miR-141/200a)-dependent secretion of CXCL12β [61]. Similarly, recent scRNAseq analysis in breast cancer has also demonstrated that CXCL12 is highly secreted by iCAFs [70]. In the TME, CAFs-secreted CCL2, CCL5, CCL17, IL-1, IL-6, IL-13, and IL-26 can also promote a Th2 and Th17 CD4⁺ polarization, at the expense of anti-tumor Th1 response [33, 159–161]. Consequently, in vivo elimination of CAFs in a murine model of breast cancer using a vaccine targeting FAP can shift CD4⁺ T cell polarization from a Th2 to a Th1, increase expression of IL-2, increase CD8⁺ T cell functions, and hamper Tregs recruitment [161].

Of note, and as mentioned earlier in this review, the presence of CAFs in the TME profoundly affects the ECM through the deposition of several components (e.g., fibronectin or type I collagen) and proteolytic degradation of normal ECM structure in an MMPs-dependent manner. This remodeling has important consequences on both tumor behavior [23, 162–165] and the efficacy of the antitumor immune response [166]. Indeed, this modified ECM is presumed to restrict access of immune cells to cancer cells, serving as a physical barrier [166, 167]. The perfect example is PDAC, where fibrosis is extensive and the “scar-like” ECM acts as a barrier for cytotoxic T cell infiltration into tumor cell areas [168, 169]. This also occurs in other cancer types such as lung tumors, where T cells poorly migrate in dense ECM areas [158, 170]. Similarly, the presence of FAP^HIGH ECM-secreting CAFs has been linked to the exclusion of CD8⁺ T cells from the tumor and their accumulation in the collagen-rich peritumoral stroma [70]. Furthermore, in tumors with the accumulation of matrix proteins in the ECM, tumor tissues are often poorly oxygenated, resulting in the presence of areas with a low oxygen pressure called “hypoxic zones” [16, 171, 172]. Interestingly, several studies indicated that hypoxia is involved in the process of CAFs activation and in their functionality within the TME [173–177]. In parallel, in melanoma, our group recently provided evidence that hypoxia increases CAFs TGFβ, IL-6, IL-10, VEGF, and PD-L1 expression and/or secretion and demonstrates that hypoxic CAF exerts a more profound effect on T cell-mediated cytotoxicity than their normoxic counterpart [178].

In addition, CAFs can also impair T cell proliferation and effector functions through the metabolic reprogramming of the TME. In particular, the secretion by CAFs of IDO1 [179, 180], an immuno-regulatory enzyme that catabolizes tryptophan degradation [181, 182], or Arg 2, an enzyme involved in the deprivation of arginine in the TME [183], have a potentially important effect on T cells. In this regard, a poor clinical outcome for PDAC patients has been linked to the presence of CAFs expressing Arg 2 in hypoxic zones [184]. In addition, CAFs can use aerobic glycolysis as a source of energy, which results in the production of pyruvate and lactate that switch T cell polarization, reducing the percentage of Th1 CD4⁺ T cells and increasing Treg recruitment [185–187]. Furthermore, stromal cells from cervical tumors express high levels of CD39 and CD73, two molecules known to hydrolyze ATP, generating free adenosine, which possesses important immunosuppressive properties [188]. Similarly, FAP^HIGH CAFs from breast, colorectal and ovarian cancer express high levels of CD73, which potentially promotes immunosuppression in, at least in part, a Tregs-dependent manner [61, 189, 190]. Interestingly, it was also recently shown that CAFs upregulate CD39 expression on T cells, and in turn, T cells upregulate CD73 expression on CAFs [191]. CAFs highly express cyclooxygenase-2 (COX2) and are consequently a major source of PGE2 [192, 193], with important implications in the field of immunosuppression [194], especially by shifting the balance between Th1 and Th2 responses, by suppressing CD8⁺ T cell-mediated cytotoxic activity and by promoting Treg recruitment. As such, PDAC-derived CAFs strongly inhibited T-cell proliferation in a PGE2-dependent manner, and its inhibition by indomethacin, a non-steroidal anti-inflammatory molecule, partially restored the proliferative capacities of both CD4⁺ and CD8⁺ T-cells [195]. Finally, in breast cancer, NO secretion by FAP^HIGH PDPN⁺ CAFs can suppress T cell proliferation [196].

Finally, CAF can hamper T-cell-mediated immune response through many other miscellaneous mechanisms, many of them being under investigation or probably not yet elucidated. For example, it was suggested that CAFs can trigger cytotoxic T cell apoptosis via their expression of FasL/CD95L [197] and that CAFs secretion of galectins that have a high affinity for β-galactosides [198, 199], alters T cell functions [200–202].

CAFs in the TME can also interfere with the adaptive immune response by facilitating the infiltration and generation of MDSCs, involved in the direct or indirect alteration of the T cell-mediated immune response through their secretion of several factors including Arg, iNOS, TGFβ, IL-10, PGE2 and IDO [203, 204]. In this regard, in pancreatic tumors, CAF-secreted IL-6 favors monocyte precursors differentiation towards an MDSC phenotype, in a STAT3-dependent manner [85, 205]. Similar results were obtained in esophageal squamous cell carcinoma, where CAF-secreted IL-6 and CAF-derived, exosome-packed, miR-21 promote MDSC differentiation via STAT3 signaling [206]. Furthermore, secretion of CXCL12/SDF1 by hepatic carcinoma-derived CAFs attracts monocytes into the tumor stroma and engages their differentiation into MDSCs in an IL-6- and STAT3-dependent manner [207]. MDSC-promoting factors (e.g., IL-6, VEGF, M-CSF, CXCL12, CCL2) can also be produced by pancreatic stellate cells, described as CAFs precursors [85]. Similar results were also obtained in murine liver tumor models, where FAP⁺ CAFs are a major source of CCL2 that enhances the recruitment of MDSCs and predicts poor prognosis of patients with intrahepatic cholangiocarcinoma [208]. In lung squamous cell carcinoma, CAFs have also been reported to promote peripheral CCR2⁺ monocyte migration via CCL2 and their reprogramming into MDSCs [209]. Another study described similar effects of CAF-secreted CXCL16 on monocytes in triple-negative breast cancers [210].

Immune checkpoint receptors and ligands respectively expressed at the surface of T-cells and tumor cells, have clearly emerged as one of the main contributors to T-cell dysfunction within the TME. In particular, PD-L1 and PD-L2, two members of the B7 family of co-stimulatory/co-inhibitory molecules expressed by a large variety of cancer cells, engage their receptor PD-1 expressed on T-cells. This interaction strongly counteracts T cell receptor (TCR) signaling and CD28-co-stimulation [211], which result in the inhibition of T cell activation, proliferation, and functions. As such, PD-L1/PD-1 blocking antibodies now receive great attention in the field of tumor immunotherapies, especially in melanoma, lung, and renal cell carcinomas [212].

Very interestingly, several studies have now demonstrated that CAFs can express some of these immune checkpoint molecules. For example, CAFs from renal, colon, and lung cancers or melanoma can express programmed PD-L1, PD-L2, or CD276 (also known as B7-H3) [178, 197, 213–216], with potential participation in T cell exhaustion. Interestingly, the expression of some immune checkpoint ligands, especially OX40 ligand (OX40L)/CD242 and PD-L2, by FAP⁺ CAFs also allows their long-term interaction with Tregs, at least in vitro [61]. In parallel, CAFs can contribute to the expression of these immune checkpoints by other cell populations present within the TME, through their various production of cytokines or exosomes. As such, in pancreatic cancer, CAFs have been reported to increase the expression of PD-1, cytotoxic lymphocyte-associated antigen-4 (CTLA-4), lymphocyte-activation gene-3 (LAG-3), and T cell immunoglobulin and mucin-domain containing-3 (TIM-3), on both CD4⁺ and CD8⁺ T cell [195]. Similarly, in breast cancer, FAP^HIGH ECM-myCAF can recruit Tregs and increase PD-1 and CTLA-4 expression at their surface [70]. Moreover, CAF-secreted IL-6 can induce PD-L1 expression on TANs in a STAT3-dependent manner [102]. Importantly, CAF can also promote the expression of immune checkpoint ligands by tumor cells. For example, CAF-secreted CXCL5 was involved in the expression of PD-L1 on mouse melanoma and colorectal tumor cells in a phosphatidylinositol 3-kinase (PI3K)/AKT-dependent manner [217]. Similarly, in lung adenocarcinoma, CXCL2 secretion by αSMA⁺ CAFs increases PD-L1 expression by tumor cells [218] and αSMA⁺ CXCL5-secreting CAFs are positively correlated to PD-L1 expression by melanoma and colorectal carcinoma tumor cells [217]. Additionally, in human breast cancer, CAF-derived exosomes containing miR-92 decrease the expression of large tumor suppressor 2 (LATS2), and secondarily promotes the nuclear translocation of yes-associated protein 1 (YAP1) and its binding to the enhancer region of PD-L1 to promote its transcription within tumor cells [219]. Nevertheless, it is crucial to emphasize that further studies are clearly needed to clarify both the mechanisms of CAF-induced immune checkpoint expression by the diverse population present in the TME and the influence of immune checkpoint ligand expression by CAF on the T cell-mediated anti-tumor immune response.

In summary, CAF can shape the adaptive antitumor immune response by switching CD4⁺ Th lymphocytes polarization from a Th1 to a Th2 phenotype, affecting Tregs and Th17 cells generation, affecting CD8⁺ T cell functions, modifying the ECM and T cell migration, by affecting MDSCs generation or through their effect on immune checkpoint receptors/ligands expression.