Targeting the Extra-Cellular Matrix—Tumor Cell Crosstalk for Anti-Cancer Therapy: Emerging Alternatives to Integrin Inhibitors

The extracellular matrix (ECM) is a complex network composed of a multitude of different macromolecules. ECM components typically provide a supportive structure to the tissue and engender positional information and crosstalk with neighboring cells in a dynamic reciprocal manner, thereby regulating tissue development and homeostasis. During tumor progression, tumor cells commonly modify and hijack the surrounding ECM to sustain anchorage-dependent growth and survival, guide migration, store pro-tumorigenic cell-derived molecules and present them to enhance receptor activation. Thereby, ECM potentially supports tumor progression at various steps from initiation, to local growth, invasion, and systemic dissemination and ECM-tumor cells interactions have long been considered promising targets for cancer therapy. Integrins represent key surface receptors for the tumor cell to sense and interact with the ECM. Yet, attempts to therapeutically impinge on these interactions using integrin inhibitors have failed to deliver anticipated results, and integrin inhibitors are still missing in the emerging arsenal of drugs for targeted therapies. This paradox situation should urge the field to reconsider the role of integrins in cancer and their targeting, but also to envisage alternative strategies. Here, we review the therapeutic targets implicated in tumor cell adhesion to the ECM, whose inhibitors are currently in clinical trials and may offer alternatives to integrin inhibition.

The extra-cellular matrix (ECM) is a dynamic niche continuously undergoing quantitative and qualitative remodeling by renewed synthesis and proteolytic modifications. During ECM remodeling, changes to its physical structure and organization occur, leading to a dysregulation in fiber composition, tissue architecture, and stiffness contributing to cancer progression and fibrosis (1). The cell can sense the surrounding ECM fibers by transmembrane surface molecules, such as integrins or other glycoproteins, acting as cellular mechano-chemical sensors. The relevance of the finely tuned integration and crosstalk between the ECM molecules, the cellular cytoskeleton, and the downstream signaling pathways, has been widely recognized and studied (2,3). Their complex dynamic bi-directional interactions and mechano-transduction control have been associated to fundamental physiological processes such as branching tissues morphogenesis and angiogenesis during development and homeostasis. These interactions are also relevant to pathological conditions including cancer, from initial malignant transformation to the disruption of tissue polarity and promotion of invasiveness toward dissemination and metastasis development (4,5). Integrins represent the key cell surface receptors for the cell to sense the ECM, triggering signaling pathways that determine cell fate and evolution toward a malignant phenotype and resistance to therapy (6,7). Numerous experimental and preclinical studies conducted over the past decades highlighted the central role of integrins in affecting different steps of tumorigenesis, by controlling tumor cell adhesion, proliferation, migration, invasion, and survival (6). This made integrins appealing therapeutic targets leading to the development of integrin inhibitors and their clinical testing in cancer trials. Unfortunately and unexpectedly, integrin inhibitors failed to deliver any tangible therapeutic benefits for cancer patients (8)(9)(10). This failure may be due to the intrinsic complexity of integrin signaling that we still do not fully understand. But they also question the pharmacokinetic/pharmacodynamics properties of the integrin inhibitors developed, the integrin subunit and the associated biological process targeted, the preclinical models used as well as the design of the clinical trials performed (7,8). Addressing those yet unanswered questions is likely to pave the road toward successful introduction of a novel generation of integrin inhibitors in clinical practice. In the meantime, long-ago discovered non-integrin ECM receptors as well as intra-cellular downstream effectors of the ECMtumor cell crosstalk (signaling molecules) taking part in several key aspects of tumor progression, were largely neglected. Considering the clinical failure of integrin inhibitors, these ECMtumor crosstalk targets are potential candidates that may be therapeutically exploited in alternative to integrin inhibitors.
Here we review those currently tested in anti-cancer clinical trials, and portray their biology and activity in promoting tumor evolution.
Altogether, CD44 is involved at multiple steps of tumor progression and its inhibition appears as a promising alternative for tumor-ECM targeting therapies. Low molecular mass HA, soluble CD44, CD44 blocking antibodies, CD44 blocking peptides/aptamers, CD44-targeting sh/siRNA or silibinin (a plant-derived inhibitor of CD44 expression) have all been used successfully to interfere with CD44 function in preclinical FIGURE 1 | Extracellular matrix-tumor cell interactions. In addition to integrins, DDR, CD44, LAMRs, FAK, and SFK represent emerging therapeutic targets currently tested in clinical trials for solid tumors. Downstream effectors interactions were simplified for clarity reasons. DDR, discoidin domain receptor; LAMR, 36/67 kDa laminin receptors; FAK, focal adhesion kinase; PI3K, phosphoinositide-3-kinase; MAPK, mitogen-activated protein kinases; Casp3, caspase 3; NF-κB1, nuclear factor-kappa B1. models of solid tumor progression ( Table 1). The CD44-blocking antibody RO5429083 was tested in a phase I, dose-escalation clinical study in metastatic or locally advanced, CD44-positive malignant solid tumors (NCT01358903) as well as in a phase I clinical study, alone or in combination with cytarabine, for acute myelogenous leukemia (NCT01641250). Alternatively, CD44 targeting may serve to specifically deliver cytotoxic drugs or radioisotopes to tumor cells. Bivatuzumab-mertansine, a CD44v6-specific targeting antibody linked to the cytotoxic drug mertansine, was tested in phase I dose-escalation clinical studies for CD44v6-positive recurrent or metastatic breast cancers (NCT02254031, NCT02254005) and advanced squamous cell carcinoma of the head and neck (NCT02254044, NCT02254018). The 186 Re-labeled bivatuzumab was tested in phase I biodistribution studies for non-small cell lung cancers (NCT02204059) and adenocarcinoma of the breast (NCT02204046). Although preliminary, these results encourage further clinical assessment of CD44-targeting therapies, either alone or in combination.

Discoidin Domain Receptors (DDR)
DDR1 and DDR2 belong to the family of the transmembrane receptor tyrosine kinase (RTK) with an extracellular discoidin domain binding to collagen in its native triple-helical conformation (227,228) (Figure 1). DDR1 and DDR2 bind to various collagen isoforms with different affinities. DDR1 typically binds to collagens I-VI and VIII, while DDR2 preferentially binds to collagens I-III and X (228)(229)(230)(231). Upon collagen binding, DDRs cluster and get activated through auto-phosphorylation at multiple tyrosine residues within the cytosolic part of the protein (232,233), leading to the recruitment of adaptor or signaling proteins like ShcA, SHP-2, SFKs, the proline-rich tyrosine kinase 2 (Pyk2), and the non-muscle myosin heavy chain (NMHC) IIA (234,235). In cancer cells, DDR activation was reported to induce Ras/MAPK (236), PI3K/AKT (236), Notch (237), NF-κB (238), PKCα/JAK/Stat (239), and p130CS/JNK pathways (234), thereby participating in various steps of tumor progression (Figure 1). Both DDR1 and DDR2 were shown to promote tumor cell proliferation, survival (236,238,240,241), and migration   (190) Bevacizumab Glioma Tumor growth (191) Rapamycin Liver Tumor growth (192) Paclitaxel Breast, ovary Tumor growth (193,194) Cisplatin Bladder Tumor growth (195) Oxaliplatin Colon Tumor growth (194,196) Gemcitabine Urothelial Tumor growth (197) (242)(243)(244)(245). Interestingly, EMT was reported to rely on the switch from DDR1 (epithelial) to DDR2 (mesenchymal) expression (246), although various reports implicate both DDR1 and DDR2 in EMT-mediated tumor cell invasion (234,247). More recently, DDRs were implicated in the late stages of metastatic tumor progression (244,248). Typically, DDR1 drives site-specific metastasis of lung cancer cells to bone (248). Additionally, the collagen-dependent interaction between Transmembrane 4 L6 Family Member 1 (TM4SF1) and DDR1 regulates dormancy vs. growth at the metastatic site (239). Finally, both DDR1 and DDR2 promote resistance to radio-and chemo-therapy in various preclinical models (94, [236][237][238]249). However, despite these converging evidences implicating DDRs in tumor progression, one should consider that DDR-mediated effects are highly versatile and cell-dependent. For example, DDR1 was shown to either support or prevent integrin α2β1-mediated cell migration in different experimental models (234,250,251). Moreover, the dynamic regulation of DDR expression during tumor progression will determine the consequences of DDR inhibition (231). Thus, the complex regulation of DDR activity in tumor cells may stand for the controversy concerning their contribution to cancer progression (243,248,(252)(253)(254) and affect the potential efficacy of DDR targeting in cancer. Still, the recent identification of activating mutations in the cytoplasmic signaling portions of DDR affecting intracellular signaling (240,(255)(256)(257) opens new perspectives in the identification of patients who might benefit the most from DDR inhibition. DDR1 and DDR2 kinases are efficiently inhibited by multikinase inhibitors like ponatinib, imatinib, dasatinib, and nilotinib (258). Dasatinib, nilotinib, a DDR1 blocking antibody, the selective DDR1 inhibitors 7rh and DDR1-IN-1 and the selective allosteric DDR2 inhibitor WRG-28 were shown to efficiently prevent DDR-mediated tumor progression in preclinical models ( Table 1). Driven by these encouraging results, dasatinib was tested in a phase II clinical trial for patients with advanced non-small cell lung cancers harboring a DDR2 mutation (NCT01514864). Unfortunately, it was abandoned because of lack of efficacy and slow enrollment. Currently, nilotinib is being assessed in a phase II clinical trial for malignant locally advanced or metastatic solid neoplasms presenting DDR1 or DDR2 mutations (NCT02029001). Importantly, noncanonical activation of DDR1 was shown to promote metastasis through tyrosine kinase-independent signaling in preclinical models (239), warranting cautious assessment of RTK inhibitors to target DDR. Further efforts should aim at the development of specific DDR1 and DDR2 inhibitors targeting canonical and non-canonical activation routes, the identification of the patients who may benefit the most from DDR inhibition and their use in combination therapies.
Despite various emerging strategies aimed to target LAMR (300), in vivo preclinical studies assessing the feasibility and efficiency of targeting LAMR are still scant. Both a LAMR 37 blocking antibody and a small molecule inhibitor preventing laminin-LAMR interaction were shown to impede metastatic progression ( Table 1). The green tea-derived epigallocatechin-3-gallate (EGCG) is a small molecule affecting a large number of cellular targets, including LAMR 67 (301) and LAMR 37 (302). EGCG is currently assessed in a phase I study for chemopreventive effect in patients with curative-intent resections of colorectal cancer (NCT02891538). Interestingly, the immunogenic LAMR tumor-associated antigen, referred as oncofoetal antigen immature laminin receptor protein (OFA-iLRP), has been successfully used as a tumor antigen for vaccine-based therapies in preclinical studies ( Table 1). Cellular immunotherapy using autologous dendritic cell loaded with OFA-iLRP was tested in a phase I-II clinical study for metastatic breast cancers (NCT00879489). Altogether, LAMR targeting appears promising for cancer therapy, although major efforts should aim at the development of specific inhibitors and acquisition of stronger preclinical data prior to further clinical trial.

DOWNSTREAM EFFECTORS OF INTEGRIN-MEDIATED TUMOR CELL ADHESION TO THE ECM Focal Adhesion Kinase (FAK)
Focal adhesion kinase (FAK) is a cytoplasmic non-receptor protein tyrosine kinase. It is an important cell signaling hub highly phosphorylated upon integrin activation, and has long been recognized as promoting cancer cell migration, proliferation, and survival/chemoresistance through downstream activation of Rho-GEF, talin, cortactin, SFKs, PI3K/AKT, Ras/MAPK, or NF-κB pathways (303, 304) (Figure 1). More recent studies have described that besides its classical localization at the plasma membrane of tumor cells, FAK can also translocate to the nucleus and act as a transcription factor driving the expression of cytokines and chemokines favoring tumor immune evasion, independently of integrin signaling (305). In pancreatic cancer, FAK inhibition increases the immune infiltrate within the tumor environment, thereby sensitizing tumors to immune-checkpoint blockade (306). In addition, FAK inhibition also affect stromal cells. By targeting carcinoma-associated endothelial cells, FAK inhibition enhances vascular permeability, drug delivery, and overcomes chemo-resistance to DNA-damaging agents (307). Altogether, these data largely support the potential for therapeutic benefits of FAK inhibitors, used alone or in combination therapies, in the arsenal of anti-cancer strategies, illustrated by their success in various preclinical models ( Table 1). FAK inhibition mostly relies on small molecule inhibitors working through various mechanisms: ATP competitive kinase inhibition (TAE-226, VS-4718, VS-6062, VS-6063, GSK-2256098, PF-573228), FAK scaffold inhibition (compounds 14, Y11, Y15, C4, INT2-31, M13, R2), or more recently ATP competitive non-kinase inhibition (BI853520) ( Table 1). In combination, FAK inhibition was reported to improve the efficacy of chemotherapeutic agents (docetaxel, paclitaxel, temzolomide, 5-FU, gemcitabine, doxorubicin), targeted therapies (EGFR inhibitor, Src inhibitor, sunitinib, BRAF inhibitor, CXCR4 inhibitor, HDAC inhibitor), or immunotherapy (PD1 antagonists, T cell immunotherapy) ( Table 1) (317)]. This unexpected failure may have been prevented by the stratification of the patients based on FAK amplification/activity in order to select for the best responders. VS-6063 is currently tested in multiple clinical trials: (i) a phase II clinical trial in a pre-operative setting for malignant mesothelioma (NCT02004028); (ii) a phase II clinical trial in association with the PD-1 inhibitor pembrolizumab for advanced solid tumors (NCT02758587, NCT03727880); (iii) a phase I clinical trial in association with the RAF/MEK inhibitor RO5126766 for advanced solid tumors (NCT03875820); (iv) a phase I clinical trial in association with the anti-PDL1 antibody avelumab for epithelial ovarian cancer (NCT02943317); (v) a phase I clinical trial in association with pembrolizumab and gemcitabine for advanced solid tumors (NCT02546531). The results of these ongoing clinical trials will be decisive to shape the future development of FAK inhibitors in clinical practice.

Src Family Kinases (SFK)
The SFK, composed of c-Src, Fyn, Yes, Lck, Lyn, Hck, Fgr, and Blk, are cytoplasmic non-receptor protein tyrosine kinases. Their prominent functions are mediated by their SH2 and SH3 domains interacting with various RTKs (such as EGF-R, HER2, IGF-R, HGF-R, and PDGF-R), thereby participating in integration and regulation of RTK signaling. But SFK also participate in ECM-mediated signaling. Through phosphorylation of FAK, SFK activation stabilizes focal adhesion complexes enhancing cell adhesion to the ECM (318) (Figure 1). Altogether, SFK are implicated in many steps of tumorigenesis, including proliferation, migration, invasion, survival in the circulation and at distant metastatic sites (319-324), achieved through modulation of various downstream effectors as PI3K/AKT, Ras/MAPK, or Stat3 (325,326). Additionally, SFK activation confers therapeutic resistance to targeted RTK therapies (e.g., Trastuzumab/Herceptin for HER2), to hormonereceptor endocrine therapies (e.g., Tamoxifen for Estrogen Receptor), as well as to traditional chemo-and radiotherapies (327). Given their central role in tumor cell signaling and pleiotropic functions in cancer, SFK represent a promising target for anti-cancer therapies. SFK are currently most efficiently targeted using non-specific ATP-competitive multikinase inhibitors, such as dasatinib, bosutinib, saracatinib, ponatinib, and vandetanib, targeting many different tyrosine kinases (such as BCR-ABL, Kit, PDGFR, EGFR, RET, VEGFR) in addition to SFK members (328). With the exception of vandetanib, approved for the treatment of thyroid medullary carcinoma, dasatinib, ponatinib, and bosetanib have been approved by the FDA for hematological malignancies only, based on their BCR/Abl inhibitory capacity (328). In vivo preclinical data, however, suggest their potential efficacy in solid tumors as well, alone or in combination, although not necessarily through SFK inhibition ( Table 1). Up to date, the results of phase II clinical trials with SKF inhibitors in monotherapy have been disappointing, as they showed only modest or no efficacy (326,329). Such failure may be largely attributed to the current lack of biomarkers for the identification patients with aberrant SFK, the lack of specificity of SFK inhibitors, and the sometimes opposing effects of SFK members at various steps of tumor progression (330,331). The interpretation of the numerous ongoing clinical trials (http://www.clinicaltrials.gov/) as well as the design of future successful clinical trials testing SFK inhibitors for solid tumors will largely depend on our capacity to overcome these important issues.

CONCLUSION
Despite huge expectations based on preclinical studies, integrin inhibitors failed to deliver anticipated results and have not entered the clinical practice yet. Understanding and surmounting the pitfalls of integrin inhibition will be crucial to further sustain the targeting of tumor cell-ECM interactions as an anticancer strategy. Yet, other long-time discovered molecules at the interface between tumor cell and ECM as CD44, DDR, LAMR, FAK, and SFK, are emerging as alternative therapeutic targets in clinical trials. Alike integrin inhibitors, their therapeutic relevance will depend on the specificity and pharmacokinetic/dynamic properties of the inhibitors developed, on the adequacy of the preclinical models used for validation, on the biological process targeted, on the biomarkers used for the identification of best responders and on the combination strategies applied in clinical trials. Importantly, our growing knowledge of the biology of ECMtumor cell interactions will be instrumental in overcoming these important pitfalls and extend the arsenal of clinically valuable inhibitors targeting the ECM-tumor cells crosstalk in the near future.

AUTHOR CONTRIBUTIONS
GL wrote the review and edited the manuscript. CR edited the manuscript. FK planned the outline, wrote the review, and edited the manuscript. All authors read and approved the submitted version of the manuscript.