The Extracellular Matrix: A Key Accomplice of Cancer Stem Cell Migration, Metastasis Formation, and Drug Resistance in PDAC

Simple Summary This review takes the extracellular matrix (ECM) as the starting point, and describes its influence and related mechanisms on the biological behavior of pancreatic ductal carcinoma (PDAC). It focuses on how the ECM regulates cancer stem cells and thus affects the metastasis and drug resistance of PDAC. Finally, current and ongoing treatment strategies for ECM are presented. ECM-related factors and mechanisms are both the focus and difficulty of current therapeutic strategy research, but further exploration is crucial for effectively eradicating the disease and improving patient survival. Abstract Pancreatic ductal adenocarcinoma (PDAC) is rich in dense fibrotic stroma that are composed of extracellular matrix (ECM) proteins. A disruption of the balance between ECM synthesis and secretion and the altered expression of matrix remodeling enzymes lead to abnormal ECM dynamics in PDAC. This pathological ECM promotes cancer growth, survival, invasion, and alters the behavior of fibroblasts and immune cells leading to metastasis formation and chemotherapy resistance, which contribute to the high lethality of PDAC. Additionally, recent evidence highlights that ECM, as a major structural component of the tumor microenvironment, is a highly dynamic structure in which ECM proteins establish a physical and biochemical niche for cancer stem cells (CSCs). CSCs are characterized by self-renewal, tumor initiation, and resistance to chemotherapeutics. In this review, we will discuss the effects of the ECM on tumor biological behavior and its molecular impact on the fundamental signaling pathways in PDAC. We will also provide an overview of how the different ECM components are able to modulate CSCs properties and finally discuss the current and ongoing therapeutic strategies targeting the ECM. Given the many challenges facing current targeted therapies for PDAC, a better understanding of molecular events involving the interplay of ECM and CSC will be key in identifying more effective therapeutic strategies to eliminate CSCs and ultimately to improve survival in patients that are suffering from this deadly disease.


Introduction
Pancreatic ductal adenocarcinoma (PDAC) is one of the most aggressive and lethal cancers in the world, with a dismal overall 5-year survival rate of about 9% [1]. Some studies predict PDAC will be on track to develop into the second leading cause of cancer death by 2030 [2,3]. Unfortunately, a large proportion of patients have locally advanced stages or regional/distant metastatic spread at the time of initial diagnosis, making the majority of patients not suitable for curative surgery [4]. Although great progress has been made in the treatment of PDAC, such as chemotherapy, radiotherapy, and molecular targeting, it is still challenging that many patients show intrinsically or acquired resistance fate and interacts with cells to regulate proliferation and differentiation, but also controls the host tissue response [40][41][42]. Cells that are in contact with the ECM sense the properties of the ECM through receptors and focal adhesion complexes in order to maintain tissue homeostasis. At the same time, cells in turn regulate the expression of ECM components and enzymes in response to signals from the ECM. This creates a feedback mechanism in which cells can also influence the ECM, leading to a balance between deposition and degradation of ECM components [43].

ECM in PDAC
Abnormal remodeling of the ECM is associated with a variety of pathologic conditions, such as inflammation, fibrosis, and cancer [40]. In particular, a growing number of studies have highlighted the active role of dysregulated ECM dynamics in tumor progression and invasion [18,44]. The expression of many ECM remodeling enzymes is frequently deregulated in human cancers, most notably Matrix metalloproteinases (MMPs) [45]. MMPs are secreted by cancer-associated fibroblasts (CAFs) in PDAC [46]. Fibroblasts in the tumor microenvironment, collectively referred to as CAFs, are associated with tumor aggressiveness and reduced survival [47,48]. The CAF population, due to the different cellular origins, exhibits marked heterogeneity and functional diversity [49]. One of the most common cellular origins of CAFs is pancreatic stellate cells (PSCs) [50]. After pancreatic injury, or TGF-β stimulation, PSCs were activated, resulting in changes in cell morphology from stellate to spindle, increased nuclear volume, and decreased vitamin A droplets [51][52][53]. CAFs can also originate from tissue-resident fibroblasts, which can be activated under the control of growth factors (such as TGF-β) [54]. Additional studies have demonstrated that CAFs may also arise from trans-differentiation of non-fibroblast lineages or epithelial cells, as well as the recruitment and differentiation of bone marrow-derived mesenchymal stem cells [46,55]. Markers that were used to distinguish different CAF subsets include α-SMA, platelet-derived growth factor receptors α and β (PDGFRα/β), FAP, and Ca 2+ binding protein (S100A4), but none of them are exclusively expressed by CAFs [55,56]. In terms of functional diversity, CAFs can secrete MMP, which promotes the degradation of the ECM and the release of various factors, and leads to the recruitment of some specific cells and various cytokines. On the other hand, CAFs are mainly responsible for the deposition of dense tumor stroma, which can serve as a structural scaffold for cell interactions and a physical barrier against immune infiltration. Furthermore, many growth factors and proinflammatory cytokines, including interleukin 6 (IL-6), TGF-β, and vascular endothelial growth factor (VEGF) can be produced by CAFs, which can promote tumor growth, angiogenesis, and assists in immune escape [46,57]. Therefore, CAFs, combining the above characteristics, are often programmed into three different subgroups: "antigen-presenting", "myofibroblastic", and "inflammatory" CAFs [58].
The dysregulation of ECM remodeling can cause mutant cells to escape apoptosis due to the pro-and anti-apoptotic effects of various ECM components or their functional fragments [59]. Abnormal ECM remodeling affects the behavior of not only cancer cells, but also those of stromal, endothelial, and immune cells of the local microenvironment as well [60,61]. Angiogenesis that is induced by abnormal ECM remodeling often results in an abnormal vasculature that is characterized by tortuous and immature leaky vessels in malignancies [62]. Vascular abnormalities, including the absence of a lymphatic network within the tumor, lead to increased interstitial fluid pressure, acidosis, and hypoxia, which underlie metastasis and drug resistance [63]. IL-6 that is secreted by stromal CAFs recruits tumor-associated macrophages and promotes their transition to an immunosuppressive phenotype (M2) [64]. Stromal CAFs recruit and induce Treg differentiation to suppress antitumor immunity and similarly enhance the tumor-promoting function of CD4+ helper T (Th) lymphocytes [65]. Moreover, the dense fibrotic ECM that is induced by CAFs can also physically prevent immune cells from effectively infiltrating tumors, greatly limiting T-cell contact with cancer cells [66].
A prominent feature of PDAC is a strong pro-fibrotic response leading to a hyperfibrotic stroma. Sparse vascularization and abundant deposition of extracellular components lay the groundwork for the stroma of PDAC [67]. Collagen, produced mainly by CAFs (especially derived from PSCs), is the most representative component of PDAC connective tissue [68]. Fibrinous collagen Type I and III account for more than 90% of the total and protein levels of these collagen proteins that increased nearly three-fold during the progression of PDAC [69]. However, compared with normal tissue, the ratio of these two kinds of collagen in tumor connective tissue did not change significantly, so one should pay attention to other differentially expressed collagen in the process of PDAC [70]. Other collagen proteins, including types IV, V, VI, VII, XII, XIV, and XV, are also key participants in PDAC tumorigenesis and can play both beneficial and harmful roles. Collagen IV, produced by cancer cells, helps cancer cells to proliferate, migrate, and reduce apoptosis. Therefore, the high level of Type IV collagen in patients after surgery often indicates the possibility of recurrence and low survival rates [71,72]. Moreover, ECM collagen interacts with integrins (α-and β-subunits), expressed on the surface of PDAC cells to promote the proliferation and migration of tumor cells [73,74]. Conversely, the overexpression of Type XV collagen during pancreatic tumorigenesis is detrimental to the ability of cancer cells to migrate into the matrix, which is rich in Type I collagen [75]. Surprisingly, in addition to directly promoting tumor progression and signaling, collagen can also serve as a nutritional source for tumor cells. PDAC cells can metabolize collagen molecules and produce proline that enables cancer cell proliferation in the special hypoxic and low-nutrient environment of tumors [76].
Proteoglycans and glycoproteins are composed of core proteins that undergo posttranslational glycosylation, which largely shapes their conformation and cell signaling functions, and has important effects on tumor cells [77]. Most subunits of laminin, a small leucine-rich proteoglycan, are overexpressed in PDAC and are related to poor prognosis [78]. In particular, Lm332 and Lm511/521, which are currently common in various tumors, promote tumor progression by interacting with other proteins and cytokines [33]. The interaction of Lm332 with collagen Type VII and integrin α3β1 promotes tumor cell growth and metastasis, respectively [79,80]. Fibronectin has been shown to act as scaffolds and regulate cellular processes in the PDAC microenvironment [81]. Fibronectin stimulates tumor cell proliferation and invasion of the basement membrane and acts as a bridging molecule between ECM collagen and integrin [82,83]. Moreover, studies have shown that reduced galectin-1 (GAL1) in mouse models leads to reduced matrix activation and increased cytotoxic T-cell infiltration [84]. GAL1, as a class of glycoproteins, is expressed in multiple tumor types and is involved in proliferation, invasion, angiogenesis, metastasis, and is associated with patient survival [85,86]. The expression of GAL1 is upregulated in the PDAC tumor microenvironment and is low in long-term (≥10 years) survivors of PDAC [85,87]. The transforming growth factor β1 (TGF-β1) protein regulates cell adhesion through various integrins (αvβ3, αvβ5, and α1β1). TGF-β1 is increased in PDAC tissues and inhibits tumor cells by reducing proliferation and activation of tumor CD8+ T-cells or by promoting their migrative and invasive properties [88,89]. A previous study has confirmed that testican is a proteoglycan that promotes tumor resistance through mediated epithelial-mesenchymal transition signaling [90]. Lumican is a small leucine-rich proteoglycan that directly binds and inhibits MMP14, thereby preventing extracellular matrix collagen hydrolysis by this enzyme [91]. Therefore, testican promotes the growth and invasion of PDAC cells by influencing collagen deposition, while lumican interferes with tumor progression and prolongates the survival of patients by restricting the growth and metastasis of cancer cells [92,93]. Versican is an extracellular matrix proteoglycan that plays key roles in tumor cell invasion, metastasis, and angiogenesis [94]. In addition, versican is an immunosuppressive component that is detrimental to patient survival by reducing T-cell infiltration, while core proteoglycan is an anti-tumor component that slows down tumor cell growth [69,95]. The interaction of these proteoglycans with hyaluronic acid regulates the hydration level of the interstitial fluid and thus interstitial pressure.
Compared with normal pancreas, the total glycosaminoglycan content was significantly increased in PDAC (four-fold, p ≤ 0.001), mainly due to increased hyaluronic acid content [96]. A large amount of hyaluronic acid is already produced in the tumor microenvironment in the pre-cancerous intraepithelial neoplasia (PanIN) stage [97]. Hyaluronic acid is thought to be deposited primarily by CAFs and to some extent by PDAC cells [98,99]. Most importantly, hyaluronidase breaks down the hyaluronic acid matrix to allow interactions between growth factors and growth factor receptors, promoting glucose metabolism, tumor cell proliferation, and migration [100].
PSCs are key to maintaining the balance between ECM synthesis and degradation, and most studies showed an "activated" state in PDAC, leading to excessive deposition of ECM proteins [101]. It also leads to the increased expression of collagen, αSMA, immunomodulatory and other pro-tumorigenic genes [101]. TGF-β promotes ECM deposition and tumor progression in late stages of carcinogenesis [102]. As a potent activator of PSC, TGF-β mediates the interaction between the tumor microenvironment and tumor cells by binding to TGF-β cell surface receptors, promoting the deposition of ECM proteins including fibronectin and collagen [103]. The PSCs, once activated, further modulates ECM through various mechanisms. Previous studies have suggested that tissue stiffness can enhance the proliferation of tumor cells [104]. Collagen cross-linking by Lysyl-oxidase (LOX) and Tissue-transglutaminase 2 (TG2) enhances matrix stiffening [105,106]. Driven by TG2 and/or LOX, additional cross-linked collagen and rigid ECM activate Yes-associated protein (YAP) and the transcriptional coactivators of PDZ-binding motifs (TAZ), which enhances cell proliferation [107]. The stiffened ECM matrix in turn regulates the activity of ECM MMPs, Vimentin, and E-cadherin [107,108].

ECM and Pancreatic Cancer Stem Cells (PCSCs)
The tumorigenic ability of each PDAC cell is heterogenous, and the development and proliferation of PDAC are highly dependent on the restricted PDAC cell subpopulation, namely pancreatic cancer stem cells (PCSCs) [109]. The activity of stem cells is dependent on exogenous niche factors in normal tissues and organs. Similarly, significant changes within the tumor microenvironment in PDAC can have a similar effect on PCSCs. The tumor microenvironment of PDAC strongly promotes fibroplasia, in which the ECM interacts with integrins (including integrin subunits β1, α6, and β3) to regulate PCSCs function by influencing autocrine and paracrine signaling pathways [110]. Type 1 collagen is the main scaffold for CD133-positive CSCs and promotes cell invasion through the PI3K/AKT pathway and also increases PCSCs enrichment by activating Focal Adhesion Kinase (FAK) [111]. Abnormal collagen cross-linking creates mechanical stresses that increase the rigidity and stiffness of the cancer matrix, which are important factors that are transmitted to CSC and regulate its proliferation and plasticity [112]. The ECM can also dynamically influence CSC ecology by creating a hypoxic environment [113]. On the one hand, hypoxia can regulate the "niche" of PCSCs by directly activating Hypoxia-inducible factor (HIF) and its target genes [114]. On the other hand, HIF can also regulate signaling pathways and transcription factors that are critical for maintaining stem cell self-renewal and pluripotency, such as Oct-4 and Notch signaling cascades [115]. Furthermore, the high expression of HIF-1α under hypoxia can enhance the expression of PCSC markers, chemotherapeutic resistance, and an epithelial-mesenchymal transformation (EMT) phenotype [116].
Nevertheless, proteoglycans bind to various cytokines and chemokines in the tumor microenvironment to activate multiple signaling pathways in CSCs, including Notch, Wnt, and Hedgehog [117]. Transcriptional activation, mediated by the Wnt/β-catenin signaling pathway, induces the expression of C-MYC and SOX2, thereby promoting cancer stemness [118]. Glypican-4 (GPC4), as a typical proteoglycan, is associated with various human malignancies [119,120]. A study revealed a role for GPC4 in PCSC stemness regulation and the inhibition of GPC4 attenuates the stem cell-like properties by inhibiting the Wnt/β-catenin pathway [121]. Syndecan-1, a well-known multifunctional integrator of cell surface signaling, regulates components of STAT3/NF-κB signaling and connects them with the AKT pathway in a Notch-activation-dependent manner to promote stemnessrelated phenotypic traits [122].
In addition, hyaluronic acid is a major ECM component of the stem cell niche, is frequently overexpressed in PDAC, and affects stromal cell behavior, which provides a favorable microenvironment for CSC self-renewal and maintenance [123]. This process mainly relies on the interaction of hyaluronic acid with its main receptor CD44. The hyaluronic acid-CD44 axis modulates the stemness properties of CSCs by inducing the EMT program and the secretion of extracellular vesicles. Specifically, excessive hyaluronic acid can activate AKT and ERK1/2 to induce EMT, and CD44 was found to be an important factor in TGF-β-induced EMT [124]. Intracellular hyaluronic acid-induced EMT increases centrosome abnormalities and micronucleation, creating a suitable niche for CSCs [125]. Notably, CSC-derived extracellular vesicles regulate communication between CSCs and their niche [126]. Studies have confirmed that tumor-initiating cells (CICs)-derived exosomes with CD44v6 in PDAC can transfer migration and invasion capabilities to non-CICs, and regulate the expression of integrins and proteases to promote cancer cell migration and invasion. Once CD44v6 was knocked down, it altered the composition of secreted tumor exosomes and lost the ability to promote the malignant phenotype. This indicates that the induction of the pre-metastatic niche by exosomes is dependent on CD44v6, further demonstrating the importance of the hyaluronic acid-CD44 interaction in extracellular vesicles promoting malignancy [127].

ECM, PCSCs, and Metastasis
PCSCs, one of the main forces that are responsible for cancer metastasis, have the ability to evade treatment, exosmosis, and colonization from the primary tumor site to the site of secondary metastasis [128]. The niche formation of the metastatic site drives the transfer of CSCs [7]. Interactions between CSCs and the microenvironment such as CAFs and immune cells, as well as with the cell matrix, contribute to CSC migration and the priming of metastatic sites [7]. Another major factor in the formation of metastatic niches is the presence of extracellular vesicles, which are exosomes that also contain ECM regulatory genes [112]. The cross-linking of collagen and elastin in the ECM is primarily regulated by the LOX protein family, which consists of LOX-1 and four related enzymes, LOX -like protein (LOXL1-4) [129][130][131]. LOXL2, in particular, leads to the stiffening of PDAC tissues by promoting the cross-linking of collagen fibers, and increases the secretion of related factors (such as exosomes) in primary tumor tissues, which leads to ECM remodeling or stromal cell recruitment, and promotes the formation of an ecological niche before secondary organ metastasis, and ultimately promotes metastatic formation [132]. Furthermore, the overexpression of LOXL2 in mouse models increases EMT and stemness, thereby promoting primary and metastatic tumor growth and reducing the overall survival [132].
Some specific ECM molecules, including hyaluronic acid and Tenascin-C, have become the focus of research on the mechanism of tumor metastasis to specific sites [133,134]. CD44, a PCSC marker, has been identified as a Hyaluronic acid receptor that activates the EMT pathway and promotes pancreatic cancer metastasis [134]. Tenascin-C enhances the expression of stem cell signaling components musashi homolog 1 (MSI1) and leucine-rich repeat G protein-coupled receptor 5 (LGR5). These molecules are positive regulators of the Notch signaling and target genes of the Wnt pathway, respectively, and promote the formation and growth of metastases [133]. The ECM components and their biological and physical properties are involved in the regulation of basic functions of immune cells, such as activation, proliferation and migration. The previously described M2 macrophages engineered with CAFs can drive cancer cell invasion in a CCL18-dependent manner [135].
Cytotoxic T-lymphocytes (CTLs) physiologically play an important role in eliminating cancer cells, but studies have found that they are often trapped in dense ECM compartments. Immune cells that are initially attracted to the tumor site by chemokines are prevented from migrating to the tumor core following contact with the area of increased stiffness [66,136]. Continuous ECM remodeling and overexpression of certain matrix components also pro-moted the recruitment of bone marrow cells, which are ultimately polarized to support ECM remodeling, CTL inhibition, tumor proliferation, and invasion [137,138]. Analysis of T-cell trajectories in tumors revealed that both CD4+ and CD8+ T-cells were trapped in fibronectin-and collagen-rich stromal regions, suggesting that the ECM may facilitate immune evasion by hindering proper infiltration of T-cells into tumors [66]. In addition, studies have speculated that laminin can induce a similar mechanism of dendritic cell tolerance and promote tumor immune escape [139]. Proteoglycan and Hyaluronic acid from ECM components can bind Toll-like receptors-2 and -4 to induce inflammatory gene expression, thereby exacerbating tumor-site inflammation [140,141]. Subsequently, cytokines and other inflammatory mediators that are secreted by tumor-associated immunosuppressive cells contribute to the stemness, tumorigeniccity, and metastatic potential of CSCs [142]. It has been confirmed that CSCs create a microenvironment to promote immunosuppression and tumor metastasis by secreting signals of tumor-specific CTL function and overexpression of these matrix components [143].
The process of CSC migration involves ECM degradation, remodeling of cell-cell and cell-ECM interactions, formation of adhesive plaques and invasiveness, and EMT transformation, which requires a lot of cellular energy [144]. The glycolysis dependence of CSCs can drive the degradation of the ECM, the formation of invasive structures, and cellular protrusions, leading to the migration and invasion of cancer cells [145]. Interestingly, the release of lactate, the final product of this process, alters the extracellular pH and promotes ECM degradation, leading to cancer cell migration [146].
In addition, massive ECM deposition, mainly collagen and hyaluronic acid, resulted in increased tumor stiffness in PDAC. A stiffer ECM induces the production of fibronectin, which binds extracellular collagen, fibrin, and heparan sulfate proteoglycans on one side and integrins on the other. An ECM with increased stiffness can enhance cell adhesion to the ECM, connect the ECM to the cytoskeleton through local adhesion proteins, and increase cytoskeletal tension through Rho/ROCK signaling activation [147]. Next, integrin aggregation initiates the recruitment of focal adhesion signaling molecules, such as FAK, Src, paxillin, and Rac, Rho, and Ras, which promote tumor progression [148]. Moreover, ECM stiffening can directly enhance PI3K activity and tumor invasiveness [149]. Surprisingly, increased matrix stiffness leads to elevated ROCK activity in PDAC with mutant SMAD4. These responses in turn stimulate ECM production, assembly of focal adhesions and signaling, and activators of transcription-3 (STAT-3) signaling that drive tumor progression [150]. Matrix stiffening can also induce EMT, which, in addition to leading to the acquisition of a more aggressive phenotype, also facilitates the transformation of cancer cells into stem cells that are more favorable for migration and invasion [151,152].
The YAP/TAZ transcriptional coactivator is a core component of the Hippo pathway and a sensor of the structural and mechanical characteristics of the cellular microenvironment [153]. Increased matrix stiffness in PDAC leads to YAP/TAZ activation, which in turn promotes the production of profibrotic mediators and ECM proteins. Activated YAP/TAZ can enhance tumor proliferation and survival by transactivating target genes that are related to cell cycle progression and anti-apoptosis and can also induce the expression of the proto-oncogene c-Myc to promote cell cycle progression [154,155]. More importantly, YAP/TAZ is closely related to ZEB1/2 and Twist, both of which can be used to regulate EMT, thereby inducing malignant features of cancer cells and CSC-related properties, such as tumor initiation, drug resistance, and metastasis [156]. Furthermore, YAP-mediated activation of myosin light chain 2 is critical for the generation of CAFs, contributing to matrix remodeling and tumor invasion [157]. Table 1 lists the above-mentioned mechanisms that are related to ECM and PCSC-induced metastasis.

ECM, PCSCs, and Chemoresistance
To date, the mechanisms of ECM-related chemoresistance have been identified, which can be roughly divided into two major categories, namely physical barriers (abnormal vascularization and matrix stiffness) and cell adhesion-related means (ECM tissue, mechanical signaling pathways, and pro-survival signaling pathways) [158] (Figure 1). Aberrant ECM remodeling results in increased matrix stiffness, vessel collapse, and reduced blood flow, which greatly reduces the ability of drugs to enter the tumor [159]. As previously described, dense fibrosis and abnormal vascularization in PDAC contribute to the formation of a hypoxic and pH-abnormal tumor microenvironment. Hypoxia also affects drug movement from the bloodstream to the tumor microenvironment, specifically affecting the activity of drug transporters and the expression and activity of phase I drug metabolizing enzymes [160][161][162]. Glycolysis under hypoxia results in the production of large amounts of lactic acid, which reduces the extracellular pH. The ability of the drug to cross the hydrophobic membrane is greatly reduced when the drug is electrically charged in an acidic environment [160]. In addition to the dense fibrotic ECM that impairs the ability of drugs to spread from blood vessels to cancer cells, most ECM proteins contribute to chemoresistance by activating EMT and oncogenic signaling pathways, including MAPK, PI3K, and YAP [163][164][165][166]. CSCs are also an important factor leading to chemotherapy resistance, but there are only few studies that are available on the relationship between the ECM and PCSC in chemoresistance. Therefore, this part focuses on the research progress of ECM-PCSC interactions and chemoresistance. Table 2 lists the mechanisms of action of the ECM in relation to chemotherapy resistance in this paragraph.

Figure 1.
Signaling pathways that are associated with matrix stiffness. Matrix stiffness activates a large number of mechano-responsive signaling pathways in different cells through integrins, ion channels, and other transmembrane proteins. Pathways such as PI3K, RhoA-ROCK, and YAP/TAZ play major roles in this conduction. The central players in these signaling pathways can be connected to other molecules to eventually transform the changes in the ECM into related biological changes and phenotypes.

Pharmacological Targeting of ECM
CSCs rely on the favorable environment that is created by the ECM to obtain the necessary support for its development and survival. Thus, targeted therapy for the ECM may inhibit the growth of CSCs by disrupting its niche, and ultimately improving patient survival [177]. Since Hyaluronic acid leads to both increased intra-tumoral pressure that affects drug delivery and the acquisition of PCSC characteristics, it was hypothesized that targeting Hyaluronic acid would improve the efficacy of chemotherapy and patient prognosis [169,178]. The effect of PEGylated human recombinant PH20-Hyaluronidase (PEGPH20) was evaluated in combination with mFOLFIRINOX and albumin-bound paclitaxel plus gemcitabine in SWOG S1313 and HALO trials, respectively [179,180].
Collagen is one of the most basic components in the ECM, and it is also one of the ideal therapeutic directions. TGF-β plays a key role in collagen synthesis, so TGF-β signaling is the most promising target for inhibiting collagen synthesis. In animal models of PDAC, an anticoccidial named halofuginone has been shown to reduce collagen synthesis by inhibiting TGF-β signaling [181]. TGF-β is usually overexpressed in PDAC and blocking TGF-β-mediated signaling may enhance antitumor effects [173,174]. A clinical trial aiming to target TGF-β in PDAC is investigating the antitumor activity of SAR439459, a pan-TGF-β neutralizing antibody [182]. Vitamin D may also block collagen secretion by disrupting the TGF-β signaling pathway, helping to prevent tumor metastasis and enhance drug responses [183,184]. The inhibition of collagen cross-linking is also a therapeutic strategy to target ECM stiffness in cancer. In 2017, a randomized Phase II study of Simtuzumab, a humanized IgG4 monoclonal antibody that inhibits extracellular LOXL2, was performed in combination with gemcitabine in patients that were suffering from metastatic PDAC [132]. However, the results showed that a combination of gemcitabine and Simtuzumab did not improve clinical outcomes. Therefore, the complex tumor-stromal interaction of LOXL2 in PDAC tumorigenesis requires further investigation and re-eval- Figure 1. Signaling pathways that are associated with matrix stiffness. Matrix stiffness activates a large number of mechano-responsive signaling pathways in different cells through integrins, ion channels, and other transmembrane proteins. Pathways such as PI3K, RhoA-ROCK, and YAP/TAZ play major roles in this conduction. The central players in these signaling pathways can be connected to other molecules to eventually transform the changes in the ECM into related biological changes and phenotypes. Recently, the physical properties of ECM and the role of its direct or indirect signaling pathways in the survival and maintenance of CSCs have become increasingly apparent [167]. In addition, to provide anchoring, the ECM receptor of CSCs mediates paracrine signaling that is involved in self-renewal and differentiation processes [168]. As previously described, hyaluronic acid-CD44 interactions increase the stemness (including NANOG and SOX2) and expression of drug resistance factors (MDR1) of PCSCs. Hyaluronic acid synthase 1-3 (HAS1-3) is a key enzyme in Hyaluronic acid synthesis, and its expression level is closely related to the prognosis of patients. 4-Methylumbelliferone inhibits Hyaluronic acid synthesis and is itself an approved drug for bile treatment [136]. Strikingly, Hyaluronic acid accumulation was significantly reduced in mouse models of pancreatic cancer that were treated with 4-MU [169]. Furthermore, other in vivo studies using a murine PDAC model demonstrated that treatment with pegylated human recombinant PH20-Hyaluronidase reduced Hyaluronic acid levels and improved gemcitabine therapy [15].
Cytotoxic agents have been the cornerstone of chemotherapy for PDAC, but more studies are needed to understand their effects on ECM dynamics and CSCs-acquired chemotherapy resistance. Recent studies have shown that the abnormal expression and activation of the transcription factor nuclear factor-erythroid 2-related factor 2 (NRF2) and its major negative regulator Kelch-like ECH-associated protein 1 (Keap1) are observed at different stages of PDAC, and are involved in tumor development, metastasis, and drug resistance [170,171]. During PDAC progression and metastasis, Keap1 is frequently mutated and silenced, resulting in abnormal stabilization of NRF2 [171]. NRF2 can be activated by stromal-derived factor 1α (SDF-1α) and IL-6, that are secreted by PSCs [172]. The activation of NRF2 can induce drug resistance of CSCs through the up-regulation of glutathione pathway. The activation of the glutathione pathway involves binding of TGF-β to its membrane receptor, followed by upregulation of P21. CSCs cell cycles that respond to TGF-β are slowed-down or stopped, which contributes to chemotherapy resistance because cytotoxic chemotherapy induces apoptosis by causing DNA damage exclusively in rapidly dividing cells [173,174]. In addition, the Keap1-NRF2 pathway is also involved in the chemotherapy resistance of PDAC by regulating the expression of drug resistancerelated genes and cytoprotective antioxidant genes [175]. The tumor microenvironment promotes chemoresistance by maintaining the phenotype of CSCs, for example, collagen promotes CSC self-renewal through integrin signaling. The JNK signaling has been shown to promote the up-regulation of ECM-related genes. Thus, the JNK signaling may further lead to chemoresistance by establishing and regulating the CSC niche [176].

Pharmacological Targeting of ECM
CSCs rely on the favorable environment that is created by the ECM to obtain the necessary support for its development and survival. Thus, targeted therapy for the ECM may inhibit the growth of CSCs by disrupting its niche, and ultimately improving patient survival [177]. Since Hyaluronic acid leads to both increased intra-tumoral pressure that affects drug delivery and the acquisition of PCSC characteristics, it was hypothesized that targeting Hyaluronic acid would improve the efficacy of chemotherapy and patient prognosis [169,178]. The effect of PEGylated human recombinant PH20-Hyaluronidase (PEGPH20) was evaluated in combination with mFOLFIRINOX and albumin-bound paclitaxel plus gemcitabine in SWOG S1313 and HALO trials, respectively [179,180].
Collagen is one of the most basic components in the ECM, and it is also one of the ideal therapeutic directions. TGF-β plays a key role in collagen synthesis, so TGF-β signaling is the most promising target for inhibiting collagen synthesis. In animal models of PDAC, an anticoccidial named halofuginone has been shown to reduce collagen synthesis by inhibiting TGF-β signaling [181]. TGF-β is usually overexpressed in PDAC and blocking TGF-β-mediated signaling may enhance antitumor effects [173,174]. A clinical trial aiming to target TGF-β in PDAC is investigating the antitumor activity of SAR439459, a pan-TGF-β neutralizing antibody [182]. Vitamin D may also block collagen secretion by disrupting the TGF-β signaling pathway, helping to prevent tumor metastasis and enhance drug responses [183,184]. The inhibition of collagen cross-linking is also a therapeutic strategy to target ECM stiffness in cancer. In 2017, a randomized Phase II study of Simtuzumab, a humanized IgG4 monoclonal antibody that inhibits extracellular LOXL2, was performed in combination with gemcitabine in patients that were suffering from metastatic PDAC [132]. However, the results showed that a combination of gemcitabine and Simtuzumab did not improve clinical outcomes. Therefore, the complex tumor-stromal interaction of LOXL2 in PDAC tumorigenesis requires further investigation and re-evaluation of the efficacy of using anti-LOXL2 antibodies or small molecule LOXL2 inhibitors in the treatment landscape of metastatic PDAC [185].
Considering the importance of CD44-hyaluronic acid interactions in tumor cells, they may be promising therapeutic targets for cancer therapy. Several research groups are working to evaluate the antitumor effects of CD44 antibodies, such as bivacizumab, the first humanized monoclonal antibody against CD44v6 [186]. Now, more CD44 antibodies are entering clinical trials, such as the RO5429083 trial in solid tumors. Research on the application of fibronectin in cancer therapy has mainly focused on its application as a target for precision drug delivery. Several targeted therapies targeting the extra domain B of fibronectin have been developed in tumors such as bowel cancer and skin cancer [187]. But therapeutic trials and data for pancreatic cancer are currently lacking.
The inhibition of ECM production by modulating the activation state of PSCs is considered a viable therapeutic strategy. The transcriptional regulation of CAFs and PSCs is regulated by vitamin D receptors (VDR), so the treatment with Paricalcitol can reprogram the matrix, reduce inflammation, and improve the response to gemcitabine [188,189]. Vitamin D receptor agonists restore PSCs to a quiescent state, thereby reducing tumor fibrosis and enhancing chemotherapeutic drug delivery. Likewise, PSCs also express retinoic acid receptors, which interact with the vitamin A metabolite all-trans retinoic acid (ATRA) [190]. By binding to retinoic acid receptor β, ATRA reduces PSC activation, inhibits ECM remodeling, and also attenuates the ability of PSCs to sense external mechanical signals from a stiff ECM [190]. The composition of the ECM can be further altered by inhibiting angiotensin, a profibrotic cytokine. Losartan, an angiotensin receptor blocker, was found to reduce the expression of TGF-β, hyaluronan synthase 1-3, and collagen I in cancer-associated fibroblasts [191,192]. Enalapril, another renin-angiotensin system inhibitor, has been shown to inhibit the progression of PDAC in combination with aspirin [193].
There are other therapeutic ideas aiming to target the MMP to alter ECM stiffness. For example, MMP-9 promotes aggressive and metastatic phenotypes in tumor cells, and its overexpression increases the aggressiveness of cancer cell lines in vitro [194]. Targeting integrin-mediated signaling is another strategy for ECM therapy. In theory, it is possible to disrupt signals from the extracellular or intracellular environment by using integrin inhibitors to disrupt ECM mechanical sensing [177]. Targeted glutamine metabolism and hexosamine biosynthesis pathways has been shown to have profound effects on the ECM and CSC self-renewal of PDAC among metabolic inhibitors [195]. However, due to the specific environmental effects of these factors and the complex interaction between the ECM and tumor cells, the current treatment for these regimens is not beneficial. Thus, a detailed understanding of the interactions between ECM proteins, glycans, cancer-associated fibroblasts, and cancer cells will ultimately lead to more effective treatment strategies of cancer. An overview of ECM targeting inhibitors and monoclonal antibodies that are currently under investigation in clinical trials is shown in Table 3. In addition, targeting related genes and signal transduction that is regulated by the ECM is also one of the therapeutic directions. JNK signaling promotes the up-regulation of ECM-related genes and CSC self-renewal. JNK signaling is involved in regulating the niche that allows CSCs to evade chemotherapy and promote metastasis in mouse models. Treatment with JNK inhibitors reduces ECM protein expression and potentiates the chemotherapy effect [176]. The STAT3 inhibitor BBI-608 (Napabucasin) in combination with albumin-bound paclitaxel and gemcitabine is currently being tested in a Phase III clinical trial in patients with metastatic PDAC [196]. The interaction of CSCs with the tumor microenvironment also regulates the plasticity and function of CSCs, which contributes to intra-tumor heterogeneity [197]. Treatment with Sonic hedgehog (Shh) inhibitors improved tumor microvascular density and survival in mouse models [198]. However, in clinical trials, Shh inhibitors in combination with chemotherapy failed to improve the overall survival and even resulted in worse outcomes [199,200]. Recent studies suggested that this is a consequence of the heterogeneity of CAFs in tumors. Blocking the Shh signaling pathway may lead to a transformation of myofibroblast-CAF into inflammatory-CAF, promoting an immunosuppressive tumor microenvironment [201]. Therefore, a deeper understanding of the biology and function of different CAF subtypes is beneficial to improve the therapeutic effects. Clinical trials involving different inhibitors are being conducted in different clinical settings, primarily for ECM-induced stemness of PCSCs. For example, the inhibitors of Rhoassociated kinase (ROCK) alter the contractility of PDAC cytoskeleton and CAF, may benefit drug delivery and inhibit metastasis. ECM disorder occurrence and PDAC migration and invasion are prevented when ROCK is suppressed. Other mouse models showed that a prechemotherapy administration of ROCK inhibitors increased the chemotherapy response of the primary tumor and helped to prevent the development of liver metastases [202][203][204]. T13148 was the first ROCK inhibitor that was investigated for the treatment of solid tumors, but further development of this compound is advised due to drug side effects [205].

Conclusions
The ECM provides a favorable environment for CSCs to survive and interact with each other, which play an important role in metastasis formation of PDAC and chemotherapy resistance. Preclinical studies suggest that a treatment aiming to target the dense fibroproliferation-promoting ECM proteins may provide promising new therapeutic options for PDAC patients. However, ECM-targeted therapy still needs more in-depth molecular characterization and discussion according to the clinical practice results. We need to dissect the role of abnormal ECM dynamics in re-shaping the tumor microenvironment to develop more feasible therapeutic strategies for deadly cancers such as PDAC.

Conflicts of Interest:
The authors declare no conflict of interest.