Basement membrane, composed of collagens, laminins, PGs, and FN (Haraida et al., 1996), sites at the interface between parenchyma and connective tissue, providing an anchoring sheet-like layer for parenchymal cells in order to be held together preventing them from being torn (Halfter et al., 2015). In epithelial cancers, BM acts as a structural barrier of invasion, intravasation and extravasation of cancer cells. During the development and progression of cancer, changes in BM are often observed. Generally, cancer cells invade the BM through such means as producing ECM remodeling enzymes (e.g., Matrix Metalloproteinases), using natural pores in BM, or forcing their way through those pores (Chang and Chaudhuri, 2019). In addition, a more complicated phenomenon is observed, that is, invasion also depends on the contact of cancer cells with collagen fibers in the underlying stroma (Nissen et al., 2019). In areas assembled by sheet-like arrangements of laminins (Gordon-Weeks et al., 2019), FN (Lin et al., 2019) as well as collagens I, III, and IV, the BM was excessively thickened (Nissen et al., 2019). This structural arrangement effectively divides the tumor into cancer cells nests and stromal regions, thereby producing the malignant behavior of cancer cells (Eble and Niland, 2019) and regulating tumor immune response (Vaday and Lider, 2000).

Under physiological condition, IM is a loose ECM that goes deep into the BM, which is composed of collagens I and III, elastin fibers and glycoproteins. Fibroblasts, resident immune cells, vasculature, and lymphatics are all embedded within it (Bruckner-Tuderman et al., 2010; Mecham, 2012). However, in some tumors, collagen fibers in IM are thicker, more organized and denser due to the increased deposition of collagen fibers (Drifka et al., 2016b; Nissen et al., 2019). With the development of cancers, stromal collagen fibers become increasingly aligned, particularly at the edge of cancers, thus promoting the invasion of cancer cell (Conklin et al., 2011; Burke et al., 2013; Han et al., 2016). Furthermore, lysyl oxidase (LOX) family catalyzes the formation of collagen cross-linking (Laczko and Csiszar, 2020). In tumors, the increased expression of LOX leads to excessive cross-linking of collagen, and at the same time, collagen deposition increases, increasing the stiffness, and leading to solid stress in the tumor (Choi et al., 2017). Besides the change of collagen, various glycoproteins in the ECM are also expressed pathologically, such as versican (VACN) and FN. All of them form a niche to facilitate cells migration, adherence and metastasis (Theocharis et al., 2016).

Extracellular matrix in tumors is approximately at least 1.5-fold stiffer than that in the surrounding normal tissue (Najafi et al., 2019). By applying physical forces through the stiffened ECM on host tissues, tumors can enhance cell-ECM adhesion and break cell-cell contacts, leading to its growth (Dupont, 2016; Gkretsi and Stylianopoulos, 2018) and survival (Piersma et al., 2020). Inducing collagen cross-linking in a stiff ECM can enhance phosphoinositide 3-kinase (PI3K) activity, which improves the viability of cancer cells (Gao et al., 2020). Survival of cancer cells is also affected by the release of metalloproteinases (MMPs) from cells (Grzelczyk et al., 2016). MMPs, a family of proteolytic enzymes, degrade multiple components of ECM (Winer et al., 2018) and interact with integrins, thus promoting the survival of intracellular signal transducer and activator of transcription 3 (STAT3) (Najafi et al., 2019). HA activates RhoGTPase signaling, mTOR pathway and CDC42 signaling by binding CD44 to maintain proliferative signaling (Mattheolabakis et al., 2015; Skandalis et al., 2019). Moreover, ECM stiffness is indirectly involved in activation of ERK pathway, which contributes to the proliferation of cancer cells (Paszek et al., 2005).

As we all know, chemotherapeutic agents are always used to endow tumor cells with immunogenicity and/or increase their immunogenicity by inducing the death of immunogenic cell, thus increasing the release of cancer cell antigens (Garg et al., 2017; Wang et al., 2018). However, in addition to promoting survival of cancer cells by ECM, the stiffness of ECM is an obstacle to the effective uptake or delivery of drugs to the intratumoral area (Kaylan et al., 2016). In short, the enhancement of survival potential of tumor cells leads to their freedom from growth inhibition, which also reduces cell death and the release of cancer cell antigens (Gubin et al., 2015). As the first and key step for initiating anticancer immunity (Kroemer et al., 2013), a reduction in cancer cell antigen release will cripple the Cancer-Immunity Cycle.

The functional basis for ICIs response is the immunogenicity of a tumor, which is mainly determined by tumor antigenicity and the efficiency of antigen presentation (Wang et al., 2019). APCs, including macrophages, DCs and B cells, are responsible for presenting antigens and triggering immunity by different mechanisms (Blander et al., 2017). DCs are the sentinel APCs of the immune system and act as initiators of antigen-specific T cell responses (Constantino et al., 2017). Nevertheless, only mature DCs are able to induce anti-tumoral immunity, while antigens presented by immature DCs may lead to immune tolerance and fail to induce a T-cell response (Wculek et al., 2020).

It’s worth noting that the ultimate fate of DC functions is determined by signals from the microenvironment (Leenman et al., 2010). ECM components might induce a DC phenotype with low immunogenicity (Sprague et al., 2011). Studies on the angiogenic and immune profile of murine myeloid DCs upon interaction with laminins (a family of large molecular weight glycoproteins) environments have shown that murine ovarian tumors produce several types of laminins and that DCs cultured on those laminins upregulate both AKT and MEK signaling pathways and decrease the immunological capacities, which contributes to their complicity in tumor growth (Aumailley, 2013; Phillippi et al., 2020). As mentioned in the second section, some matrix components, such as HS and HA, have immunoregulatory properties in experimental models. In mice, targeting the sulfation of HS on DCs through mutation in the heparan sulfate biosynthetic enzyme N-deacetylase/N-sulfotransferase-1 (Ndst1) can endow DCs with increased maturation and reduced chemokine-dependent trafficking to draining lymph nodes (El Ghazal et al., 2016), where mature DCs present the peptide to naive T cells in a complex with MHC proteins to initiate downstream reactions (Jakubzick et al., 2017).

Besides HS, HA can also regulate the maturation of DCs in a TLR4 dependent manner (Termeer et al., 2002). It was found that the exposure of DCs to HA fragments increased the expression of activation markers as MHC II, CD80, CD86, and CD40, and promoted DCs activation (Alaniz et al., 2009). When activated, DCs migrate to the draining lymph nodes and present the antigen to naive T cells (Worbs et al., 2017). Furthermore, HA can form provisional matrices with VACN. The HA-VACN interaction is important for the recruitment of inflammatory cells including neutrophils, macrophages and T cells (Andersson-Sjoland et al., 2015; Wight, 2017). The interaction between FN (a large glycoprotein of ECM) and integrin β1 in macrophages can promote TLR 2/4 signaling pathways to enhance expression of pro-inflammatory mediators and phagocytosis by macrophages (Fei et al., 2018). When located near the proximity of dying cancer cell, DCs together with macrophages can mobilize to lymphoid organs and present tumor antigens to adaptative immune cells, thus enabling the resolution of the tumor (Gardner and Ruffell, 2016).

Generally, naive T cells are located within lymph nodes to encounter with antigen-loaded DCs and be activated (Groom, 2015). In lymphoid nodes (LNs), various stromal cell subsets create dense 3D cellular networks, which provides the opportunity for naive T cells to interact with antigens presented by DCs, resulting in the initiation of an immune response (Mempel et al., 2004). The ultimate fate of T cell in LNs is determined by interactions with the matrix framework and various other immune components (Bousso and Robey, 2003). Fibroblastic reticular cells (FRCs), lymphatic endothelial cells (LECs), and blood endothelial cells (BECs) are the main LN stromal cells (SCs) (Fletcher et al., 2015). FRCs provide the three-dimensional scaffold in the LN niche, and are characterized by expressing podoplanin (gp38) and ECM proteins, such as laminins and collagens (Martinez et al., 2019). In the LN paracortex, collagen fibers, laminins and FRCs make up microchannels known as the conduit network. The conduit center has abundant collagen type I; the BM is rich with laminin α4β1γ1 (411), laminin α5β1γ1 (511), collagen type IV, and FN (Katakai et al., 2004). Bridging the subcapsular sinus and high endothelial venules (HEVs) by the conduit allows small molecules less than 70 kilodaltons, such as soluble antigens, chemokines and cytokines, to be delivered into the paracortex (Gretz et al., 2000; Roozendaal et al., 2008).

In LNs, immunity or tolerance induction can affect the expression of laminin α4 and α5 in all SCs (Warren et al., 2014). In response to immunity and inflammation, laminin α5 is upregulated (Li et al., 2020). On the contrary, α4 increases for tolerance induction. Functionally, laminin 411 and laminin 511 serve as co-inhibitory and co-stimulatory ligands for CD4⁺ T cells, respectively. Laminin 411 inhibits the activation of CD4⁺T cells and the polarization of Th1, Th2, and Th17, but facilitates the induction of polarization of T regulatory cells (iTreg). Laminin 511 is recognized by CD4⁺ T cells through α6 integrin and α-dystroglycan to inhibit T cells activation, proliferation and differentiation (Simon et al., 2019).

T cells are initiated and activated in the tumor draining lymph nodes (TDLNs), and obtain their effector functions to migrate toward the tumor (Rotman et al., 2019). Interactions of T cells with major ECM structural components within TDLNs can regulate the function of T cells. In the T cell zone, FRCs support naive CD4⁺ and CD8⁺ T cell or memory precursor effector cell homeostasis by producing IL-7, IL-15, and CCL19. Moreover, the presentation of antigens linked to a stiff surface has been demonstrated to impair TCR-mediated T cell activation, suggesting that TCR is affected by mechanical force (O’Connor et al., 2012; Saitakis et al., 2017; Zhang et al., 2021).

Effector T cells from LNs to the tumor site are critical for the density and diversity of tumor-infiltrating T cells, which is closely related to the prognosis and curative effect of cancer immunotherapies (Bindea et al., 2013). This view is accepted as immune hot/cold tumors (Bonaventura et al., 2019). The process of T-cell trafficking is highly dynamic and governed by a complex array of mechanisms, involving complex interactions between T cells and endothelial cells (ECs) (Vaday and Lider, 2000; Vaday et al., 2001). In brief, during this process, T cells need toattach to the endothelium for a short time at first, then roll, firmly adhere and are activated on the endothelial surface, and finally extravasate through the blood vessel wall to tumor site (Masopust and Schenkel, 2013; Chimen et al., 2017).

T-cell trafficking is also highly dependent on microenvironmental clues. T cells utilize the porous three-dimensional ECM as a scaffold for both integrin-dependent and receptor-independent amoeboid motility (Jiang et al., 2016). Laminins can act as ligands that bind immune cell membrane receptors (mainly integrins) and initiate integrin-mediated signaling. In LNs, the vascular and lymphatic systems that support T cells trafficking are surrounded by BM which contains two laminin isoforms, laminin 411 and laminin 511 (Simon et al., 2019).

Stiff ECM may act as a physical barrier to T cells infiltration into the tumors and affect the preferential localization of T cells. For example, matrix density and architecture induced the localization and migration of T cells into the tumor stroma rather than into tumor cell nest in both pancreatic ductal adenocarcinoma (PDAC) and lung cancer model due to the impaired ability of T cells to infiltrate from the stroma to cancer cell nests (Salmon et al., 2012; Wolf et al., 2013; Hartmann et al., 2014). In contrast, loose areas of FN and collagen tend to facilitate the motility of T cells (Wolf et al., 2013).

Besides the rigidity, certain ECM components can also play a role in modulating the motility of T cells (He and Baum, 2004; O’Connor et al., 2012). The dense collagen-rich ECM has direct and indirect effects on the infiltration and function of T cells. In a three-dimensional cell culture systems without chemokine gradients, T cells move along the fibrillar collagen network in a process independent of integrin or protease activity. When activated CD8⁺ T cells are encapsulated in collagen hydrogels with distinct fiber alignment, CD8⁺ T cells move along the axis of collagen alignment, and faster and more persistently in aligned collagen fibers than in non-aligned collagen fibers (Pruitt et al., 2020). Additionally, when naive T cells are activated and become T_EFF in LNs, T_EFF express high levels of CD44 and can bind to HA. CD44 and its engagement with HA skews CD8⁺ T cells toward a terminal effector differentiation state that reduces their ability to form memory cells. As T cell activation proceeds, HA binding increases, and the ability of T cells to roll on HA substrate is improved, which is quite important for the migration and extravasation of T cells (Vogt Sionov and Naor, 1997; Johnson and Jackson, 2021).

It is generally believed that CD8⁺ T cells are the main effector cells involved in the killing of cancer cells. Furthermore, killing cancer cells always follows the “recognition” step. In various cancers, collagen fibers are thicker and more closely packed around cancer cell nests than tumor stroma. The stiff ECM can serve as a space barrier surrounding tumor cells and limiting the accessibility by the CD8⁺ T cells, which disturbs the recognition (Stromnes et al., 2015).

Spatial analysis of cancers demonstrates that cancers with excessive ECM deposition are resistant to immune checkpoint inhibition (Mariathasan et al., 2018), which is confirmed by data from multiple lung cancer patients (Peng et al., 2020). Collagen-induced CD8⁺ T cell exhaustion is due to the leukocyte-specific collagen receptor 1 (LAIR1), which suppresses lymphocytic activity through SH2 domain-containing phosphatase 1 (SHP-1) signaling and is expressed on CD8⁺ T cells following integrin beta 2 (CD18) binding to collagen (Lebbink et al., 2007). Reduction in tumor collagen deposition through Lysyl oxidase-like 2 (LOXL2) suppression increases T cell infiltration, diminishes exhausted T cells and abrogates resistance to anti-PD-L1 (Peng et al., 2020). Collagen density reduces the proliferation and tumoricidal activity of tumor-infiltrating T cells. Whole-transcriptome analysis of 3D-cultured T cells revealed that a high-density matrix induces downregulation of cytotoxic activity markers (CD101) and upregulation of regulatory T cell markers (CIP2A). These transcriptional changes have been predicted to involve autocrine TGF-β signaling and they are accompanied by an impaired ability of tumor infiltrating T cells to kill autologous cancer cells (Kuczek et al., 2019).

What’s more, the expression of programmed death-ligand 1 (PD-L1) in tumor cells plays an important role in escaping from the “killing” step (Sun et al., 2018). A stiff substrate enhances the expression of PD-L1 in lung cancer cells through the actin-dependent mechanism, indicating that stiffness as a tumor environment can upregulate the expression of PD-L1 and lead to the escape of the immune system and tumor growth (Miyazawa et al., 2018; Azadi et al., 2019). Researchers are inspired by the effect of tumor-associated ECM on the cancer-immunity cycle stated above. If the negative effect of ECM is eliminated, the anti-tumor immunity would be significantly improved, thus achieving good immunotherapy effect and finally improving overall survival. Therefore, in the next section, we make a summary of strategies to eliminate the negative effect of ECM.

As aforesaid, the components of ECM can act as endogenous mediators of cancer-associated inflammation, which emerges a novel idea that immunosuppression can be improved by targeting them. For instance, some studies demonstrates that the rationality of the modulation of PG’s activities could be a novel approach in the field of cancer immunotherapy. The administration of versikine, a VCAN fragment of bioactive damage-associated molecular pattern, facilitates immune sensing of myeloma tumors and enhances T-cell-activating immunotherapies (Hope et al., 2016). The non-glycanated mouse or human endocan polypeptide (previously known as endothelial cell specific molecule-1, ESM-1) restrains tumor growth by increasing leukocyte infiltration in vivo and enhancing innate immunity response (Yassine et al., 2015). The modulation of Syndecans-1 (SDC1) expression could facilitate immunosurveillance and the resolution of precancerous lesions (Handra-Luca, 2020). It has been proved that BGN can bridge innate immunity with adaptive immunity through TLR2/4 downstream signaling because it has the ability to regulate the activities of neutrophils, macrophages, T lymphocytes, B lymphocytes, and DCs (Moreth et al., 2012). Conjugation immune checkpoint antibodies to the heparin binding domain (HBD) of placental growth factor-2 (PIGF-2_123–144) shows exceptionally high affinity for multiple ECM proteins. Peri-tumoral injection of PIGF-2_123–144-anti-PD-L1 leads to higher retention within tumor tissue. More than the high efficacy, it reduces systemic toxicity in the B16F10 melanoma model (Ishihara et al., 2017).

Accumulation of excessive HA can lead to the increase of interstitial pressure and impair the perfusion and chemotherapy delivery of tumors (Jacobetz et al., 2013). In preclinical studies, pegvorhyaluronidase alfa (PEGPH20), a pegylated recombinant human hyaluronidase, has been proved to successfully degrade HA in tumors and remodel the tumor stroma, thus improving perfusion and drug delivery. Therefore, the recent phase II HALO-202 clinical trial (NCT 01839487) showed that treatment with PEGPH20 plus gemcitabine and Nab-paclitaxel significantly improved progression-free survival (PFS) in patients with previously untreated metastatic PDAC (Hingorani et al., 2018). However, the phase III trial suggested that the addition of PEGPH20 to gemcitabine and Nab-paclitaxel increased the ORR but did not improve OS, which did not support further development of PEGPH20 in metastatic PDAC (Van Cutsem et al., 2020). The depletion of HA by hyaluronidase not only improves chemotherapy delivery, but it may also improve the success of immunotherapy, which has been demonstrated by animal models. Moreover, hyaluronidase (HAase) is employed to increase tumor tissues permeability by breaking down the HA in the ECM of tumors to enhance the tumor infiltration of the nanovaccine-generated tumor-specific T cells (Guan et al., 2018). In the presence of hyaluronidase, the delivery of both nano-vaccines and therapeutic monoclonal antibodies has been enhanced (Blair et al., 2019; Clift et al., 2019).

Similarly, collagen I targeting of cetuximab, an anti-EGFR monoclonal antibody, shows promising therapeutic efficacy in the A431 epidermoid cancer model (Liang et al., 2016). Cetuximab can retain a longer period in the collagen-rich tumors without loss of potency (Magdeldin et al., 2014). Collagen affinity also can be used to mediate targeted immunotherapy antibodies, for example, cancer-collagen-targeting immunoconjugate therapy (Katsumata et al., 2019). Fusion of the collagen-binding protein lumican and cytokines increases the efficacy of systemic immunotherapies in a melanoma model. ICIs and interleukin-2 (IL-2) by conjugation (for antibodies) or recombinant fusion (for cytokine) to the von Willebrand factor A3 domain (a collagen-binding domain) eradicated tumors and exhibited obvious safety and efficacy in a breast cancer model (Ishihara et al., 2019).

In addition to directly targeting or exploiting ECM components, other approaches also focus on targeting the upstream or downstream cellular response of the altered ECM in tumors.

Collagen, the most abundant component of ECM, is mainly secreted by fibroblasts (Pankova et al., 2016). The treatment of collagen in the current study is mainly to target the function of cancer-associated fibroblasts (CAFs) (Liu T. et al., 2019). Fibroblast depletion has primarily focused on the targeting of fibroblast populations with positive fibroblast activation protein (FAP), that improves tumor-associated ECM in lung cancer and pancreatic cancer models (Lo et al., 2015). In murine melanoma model, ablation of FAP-expressing CAFs induced a reduction in immunosuppressive myeloid cells (Zhang and Ertl, 2016). However, a phase II trial of FAP inhibition using the small molecule inhibitor Talabostat failed to demonstrate clinical efficacy for colorectal cancer (Narra et al., 2007). It’s should be noted that targeting CAFs may cause the progression of cancer in the setting of pancreatic adenocarcinoma (Ozdemir et al., 2015). The depletion of α-smooth muscle actin (α-SMA)-positive myofibroblasts reveals the potential for the dissemination and metastasis of diseases (Ding et al., 2014), which emphasizes the importance of understanding the heterogeneity of fibroblasts both within and between cancer subtypes, and is necessary before the targeting of fibroblasts in human tumors. It’s single cell sequencing that probably holds the key to this understanding and enables to identify fibroblast subsets linked to abnormal tumor immunity (Kieffer et al., 2020).

In addition to the depletion of CAFs, other approaches focus on targeting the downstream cellular response to influence ECM. Up to date, the hot targets are the major matrix binding proteins, the integrins and their downstream signaling mechanisms. The use of small molecule kinase inhibitors targeting signaling pathways regulated by ECM such as focal adhesion kinase (FAK) (Jiang et al., 2016) and Rho-associated protein kinase (ROCK) (Johan and Samuel, 2019) has been proved to be effective in targeting the ECM accompanying desmoplasia in cancer.

Based on these preclinical data, several clinical studies have been launched (Table 1). According to https://clinicaltrials.gov/, some clinical studies on targeted ECM have been withdrawn or terminated due to various reasons. However, with the deepening understanding of ECM, people are exploring strategies of transforming the ECM into a strong anti-tumor immunity, and preclinical study and clinical research will continue.