Most activated CAFs originate from resident fibroblasts, which can recruit and activate many growth factors and cytokines, such as transforming growth factor β, fibroblast growth factor-2, and platelet-derived growth factor (PDGF). It has been found that these growth factors and cytokines are abundant in TME (Räsänen and Vaheri, 2010). CAFs can also be derived from bone marrow mesenchymal stem cells (Figure 1), transforming from resident epithelium or endothelial cells in the tumor stroma via EMT or endothelial–mesenchymal transition (EndMT), respectively (Kidd et al., 2012). The functions of activated CAFs include the synthesis and secretion of ECM and the release of proteolytic enzymes, such as heparanase and matrix metalloproteinases (MMPs), leading to ECM remodeling (Kessenbrock et al., 2010; Wu and Dai, 2017).

Cancer-associated fibroblasts can interact with tumor cells through direct contact and can also secrete a variety of cytokines through paracrine methods to promote the occurrence and development of cancer (Kalluri, 2016; Salimifard et al., 2020). Orimo et al. (2005) have shown that CXCL12 (stromal cell-derived factor-1, SDF-1) secreted by CAFs directly stimulates tumor growth by acting through the cognate receptor, CXCR4, which is expressed by carcinoma cells. In addition, CAF-secreted vascular cell adhesion molecule-1 (VCAM-1) also promotes the proliferation, migration, and invasion of tumor cells by activating the AKT and MAPK signals of lung cancer cells (Zhou et al., 2020). Recently, Seino et al. (2018) found that CAFs can provide a Wnt-producing niche to support the in vivo growth of the Wnt-deficient pancreatic ductal adenocarcinoma (PDAC) organoid mode. CAFs are also an important source of growth factors and cytokines [including hepatocyte growth factor (HGF), vascular endothelial growth factor (VEGF), PDGF, etc.], which can stimulate the growth of tumor cells in vitro and lead to therapeutic drug resistance (Straussman et al., 2012; Erez et al., 2013; Paraiso and Smalley, 2013).

Angiogenesis in tumor tissues can provide oxygen and nutrients for tumor cell metabolism and promote tumor growth and metastasis. Many studies have shown that CAFs can release a variety of stimulating factors that promote angiogenesis and play an important role in the recruitment and proliferation of tumor vascular endothelial cells and the formation of vascular sprouts (Benyahia et al., 2017). CAFs promote angiogenesis by recruiting endothelial progenitor cells (EPCs) into carcinomas, an effect mediated in part by CXCL12 (Orimo et al., 2005). CXCL12 can activate the PI3K/AKT signaling pathway in tumor cells, upregulate the expression of VEGF in tumor tissues, and promote angiogenesis (Wen et al., 2019). VEGF activates the main signaling pathway in tumor angiogenesis by binding to its cognate receptor, VEGFR (Claesson-Welsh and Welsh, 2013). Mirkeshavarz et al. (2017) found that CAFs can secrete interleukin-6 (IL-6) and VEGF to induce angiogenesis in oral cancer, and that IL-6 can induce the secretion of VEGF in CAF cell lines. CAF can also release active growth factors from the ECM by expressing MMPs, which indirectly promotes angiogenesis (Najafi et al., 2019b) and serves as one of the sources of MMP9 (Boire et al., 2005) and MMP13 (Vosseler et al., 2009). Both these substances have been shown to release VEGF from the ECM to increase angiogenesis in tumors (Lederle et al., 2010).

Cancer-associated fibroblasts interact with tumor cells through inflammatory signals, thereby affecting tumor cell migration and invasion. The CAF-mediated CXCL12/CXCR4 axis plays a key role in tumor cell proliferation, invasion, and migration. The CXCL12/CXCR4 axis can activate the MEK/ERK, PI3K/AKT, and Wnt/β-catenin pathways to promote EMT, thereby promoting tumor invasion and metastasis (Guo et al., 2016; Zhou et al., 2019; Mortezaee, 2020). It also activates the PI3K, MAPK, and ERK1/2 signaling pathways, promotes the secretion of MMPs, reduces the adhesion of tumor cells, and increases their invasion and metastasis ability (Wu and Dai, 2017). In addition, a recent study found that CAF-secreted CXCL-1 can stimulate the migration and invasion of oral cancer cells, that there is an interdependent relationship between CAFs and cancer cells in the oral squamous carcinoma microenvironment, and that CXCL-1 can upregulate MMP-1 in CAF expression and activity (Wei et al., 2019). In addition, CAFs can change the structure and physical properties of the ECM, thereby affecting tumor cell migration and invasion (Egeblad et al., 2010).

The fight against drug resistance remains a major challenge in tumor treatment. CAFs mediate a variety of tumor resistance to chemotherapeutic drugs. CAFs can act on tumor cells by secreting cytokines, activating downstream signaling pathways in tumor cells, and promoting tumor resistance (Chen and Song, 2019). Studies have shown that CAFs can enhance EMT and cisplatin resistance in non-small cell lung cancer induced by transforming growth factor β by releasing high levels of IL-6, while cisplatin, in turn, promotes cancer cells to produce transforming growth factor β, resulting in CAF activation (Figure 1). CAFs can also promote chemotherapy resistance in tumor cells by secreting exosomes. Gemcitabine (GEM) is currently a chemotherapy drug that is commonly used in the treatment of pancreatic cancer. Fang et al. (2019) found that exosomal miR-106b derived from CAFs plays an important role in GEM resistance in pancreatic cancer. Recently, Zhang et al. (2020) showed that exosomal miR-522 secreted by CAFs prevents the death of cancer cells by targeting ALOX15 and blocking the accumulation of lipid-ROS. In addition, a new mechanism for obtaining gastric cancer drug resistance through the intercellular signaling pathways of USP7, hnRNPA1, exo-miR-522, and ALOX15 has been observed.

Direct ablation of CAF can promote the regression of immunogenic tumors (Feig et al., 2013), which has been explored in several recent studies, where these cells are cleared by injection of diphtheria toxin or targeting FAP-specific chimeric antigen receptor T cells; direct ablation of CAF, however, can lead to significant side effects due to lack of specificity, such as cachexia and anemia (Roberts et al., 2013; Tran et al., 2013). Because of the lack of specific markers for CAF, this method is not feasible at present, so the need to know more about the mechanism by which CAF works remains important for the development of more targeted treatments.

In a parallel study, pharmacological stimulation of the VDR was successfully performed in activated pancreatic stellate cells (PSCs). VDR is the main genomic inhibitor that is activated by PSCs. In addition, treatment with the VDR ligand calcipotriol induced matrix remodeling, which can inhibit tumor-related inflammation and fibrosis, and also improves the transport of gemcitabine to the tumor area, thus reversing chemotherapy resistance in the pancreatic ductal adenocarcinoma model (Sherman et al., 2014). Due to the complex interaction between CAF and other cells in the tumor environment, targeting some CAF subsets may cause multiple responses in the TME, which may have multiple effects depending on the individual. To eradicate cancer, the synergistic combination of CAF-targeted therapy and other effective treatments (such as immunotherapy) should also be considered.

Furthermore, the CXCL12/CXCR4 axis activates multiple signaling pathways to promote tumor cell proliferation, invasion, distant metastasis, and inhibit apoptosis. Therefore, the screening of antagonists targeting the CXCL12/CXCR4 signaling pathway is a promising target for tumor therapy. Lecavalier-Barsoum et al. (2018) found that the CXCR4 inhibitor AMD3100 can inhibit the CXCL12/CXCR4 axis in the treatment of patients with advanced disseminated high-grade serous ovarian cancer, and the combination of AMD3100 and low-dose paclitaxel can inhibit the growth of ovarian cancer cells. In osteosarcoma, AMD3100 blocks the invasion and metastasis of osteosarcoma to the lung by inhibiting the JNK and AKT pathways (Liao et al., 2015). Another CXCR4 antagonist, AMD3465, can inhibit the proliferation, colony formation, invasion, and migration of bladder cancer cells through the CXCL12/CXCR4/β-catenin axis (Zhang et al., 2018).

Micro RNA and siRNA can silence gene expression through post-transcriptional regulatory mechanisms, which may be another viable way to inhibit CXCR4 expression. In breast cancer cells, siRNA targeting CXCR4 inhibited the migration of breast cancer cells in vitro (Burger et al., 2011). miR-126 can also inactivate the RhoA signaling pathway in colon cancer by reducing the expression of CXCR4 and inducing a tumor suppressor effect (Yuan et al., 2016). These studies show that miRNA or siRNA targeting CXCR4 is of great significance in tumor treatment research. CTCE-9908 is composed of dimers of CXCL12, which is a competitive inhibitor of CXCL12 targeting CXCR4 and can inhibit the secretion of CXCL12 (Guo et al., 2016). Huang et al. (2009) reported that CTCE-9908 can target the CXCL12/CXCR4 axis and inhibit primary tumor growth and metastasis of breast cancer. Hassan et al. (2011) also found that CTCE-9908 combined with the anti-angiogenic agent DC101 also reduced the volume of the primary tumor and distant metastasis compared with DC101 alone. Moreover, an in vitro experiment proved that CTCE-9908 can inhibit the growth, invasion, and metastasis of prostate cancer (Wong et al., 2014). This evidence supports CTCE-9908 as an efficacious novel agent to prevent and treat the spread of metastatic cancer. At present, cancer treatment methods targeting CAFs and the CXCL12/CXCR4 axis are being explored and developed rapidly.

Tumor-associated macrophages can promote tumor repair response by coordinating tissue damage and limit the anti-tumor activity of conventional chemotherapy and radiotherapy by providing a protective niche for CSCs (Mantovani et al., 2017). There is increasing evidence that macrophages play a central role in both normal and diseased tissue remodeling, including angiogenesis, basement membrane rupture, leukocyte infiltration, and immunosuppression. Therefore, TAMs have become a promising target for the development of new anticancer treatments. These methods are mainly focused on the depletion of M2-like TAMs and/or promotion of their transformation to M1-like phenotype (Cassetta and Pollard, 2018; Pradel et al., 2018). However, the effectiveness of this method may be limited by a variety of factors, such as alternative immunosuppressive cells that can compensate for TAMs, the existence of innate and acquired drug resistance mechanisms, and the emergence of strong immunosuppression after cessation of treatment (Quail and Joyce, 2017). PLX-3397 is a small-molecule inhibitor of the CSF-1 pathway. It is not only an effective tyrosine kinase inhibitor of CSF-1R, but also targeted at cKit and FLT3. Blocking CSF-1/CSF-1R can reduce TAMS and reprogramming TAMS in the TME and enhance the activation of T cells in the TME by enhancing antigen presentation. The downstream effect blocked by CSF-1/CSF-1R hinders the growth of the tumor (Zhu et al., 2014). In a mouse model of preclinical lung adenocarcinoma, PLX-3397 has been shown to change the distribution of TAMs in the TME and reduce tumor load (Cuccarese et al., 2017). In the syngeneic mouse model of BRAFV600E mutant melanoma, PLX-3397 combined with adoptive cell metastasis immunotherapy showed a decrease in TAMs (Mok et al., 2014). In similar melanoma mouse models, PLX-3397 combined with BRAF inhibitor PLX4032 significantly reduced M2 phenotypic macrophage recruitment, resulting in significant tumor growth inhibition (Ngiow et al., 2016). In addition, recent studies have shown that M2-like TAMs, which seem to be regulators of lysosomal pH, express high levels of vacuolar ATP enzymes and are expected to become a new drug target (Kuchuk et al., 2018; Liu et al., 2019). Targeting TAMs has proven to be a promising strategy, and with the deepening of preclinical development of TAM-targeted drugs and the new progress in the study of TAM mechanism, TAM-targeted therapy will become an important supplement to anticancer drugs.

The key roles played by MDSCs in the TME show that it is necessary to target them effectively by blocking or deleting them. Although they play a key role in tumor progression, there are no FDA-approved drugs or treatments that directly target MDSCs. At present, clinical trials are underway to target the activities of iNOS Arg1 and STAT3, metabolism through CD36, transport through CXCR2, and other mechanisms for different types of cancer (Fleming et al., 2018). The antisense oligonucleotide STAT3 inhibitor AZD9150 has been used in phase 1b clinical trials of diffuse large B-cell lymphoma in combination with immune checkpoint inhibitors. Systemic administration of AZD9150 significantly decreased granulocyte MDSCs in peripheral blood mononuclear cells (PBMCs) (Reilley et al., 2018). Current targeting strategies mainly include induction of differentiation into mature cells, inhibiting its expansion and recruitment, and blocking its immune characteristics. Studies have shown that some neutralizing antibodies or inhibitors targeting chemokine systems (CXCR4, CXCR2, and CCL2) and tumor-derived factors (CSF1, GM-CSF, and IL-6) can inhibit the expansion or recruitment of MDSC (Bayne et al., 2012; Sumida et al., 2012; Highfill et al., 2014). For example, the chemokine receptor CCR5 plays a key role in the chemotaxis of MDSCs to TME (Weber et al., 2018). However, not all MDSCs express CCR5. In melanoma mice, MDSCs expressing CCR5 have stronger immunosuppressive ability than MDSCs that do not express CCR5. Blocking CCR5 can inhibit the recruitment and immunosuppressive activity of MDSCs and improve the survival rate of melanoma patients (Blattner et al., 2018).

It has been found that some drugs, such as phosphodiesterase-5 inhibitors (sildenafil, cyclooxygenase-2 inhibitors (acetylsalicylic acid and celecoxib), vardenafil and tadalafil and bardoxolone methyl, can directly block the immunosuppressive activity of MDSCs and restore T cell response (Serafini et al., 2006; Nagaraj et al., 2010; Fujita et al., 2011; Obermajer et al., 2011). Recent studies have found that MDSC-specific peptide-Fc fusion protein therapy can completely deplete MDSCs in the blood, spleen, and tumor without affecting other immune cells, and inhibit tumor growth process (Qin et al., 2014), which provides a new idea for inhibiting tumor growth in vivo. In patients and animal models, the failure of anti-angiogenic therapy based on inhibition of the VEGF pathway is often concomitant with an increase in the number of MDSCs or TAMs infiltrating tumor tissues (Lu-Emerson et al., 2013; Gabrusiewicz et al., 2014). Along this line of thinking, anti-VEGF therapy is thought to upregulate alternative angiogenic factors (prokinin-1 and proagonin-2) produced by myeloid cells, which may accidentally produce anti-angiogenic effects and limit tumor recurrence.

Recent studies have shown that the accumulation of MDSCs in tumors limits the effect of anti-programmed death 1 (PD1) in the treatment of rhabdomyosarcoma checkpoint blockage. Inhibition of MDSC metastasis with an anti-CXCR2 antibody can enhance the efficacy of anti-PD1 (Highfill et al., 2014). In a tumor model of tolerant mice, the removal of MDSCs with gemcitabine combined with immunotherapy can effectively break the self-tolerance and induce strong anti-tumor immunity (Ko et al., 2007). Several chemotherapeutic drugs, such as anthracyclines, platinum derivatives, and doxorubicin, can induce immunogenic cell death, thus activating an effective anti-tumor adaptive response (Kroemer et al., 2013). The chemical process for enhancing the anticancer effect of these drugs includes increasing the antigen presentation ability of dendritic cells and the subsequent CD8+ T cell response (Bracci et al., 2014). Although the current targeted therapy targeting only MDSCs does not strengthen clinical outcomes, it may play an important role in anticancer immunotherapy in the future.