Collagen Remodeling along Cancer Progression Providing a Novel Opportunity for Cancer Diagnosis and Treatment

The extracellular matrix (ECM) is a significant factor in cancer progression. Collagens, as the main component of the ECM, are greatly remodeled alongside cancer development. More and more studies have confirmed that collagens changed from a barrier to providing assistance in cancer development. In this course, collagens cause remodeling alongside cancer progression, which in turn, promotes cancer development. The interaction between collagens and tumor cells is complex with biochemical and mechanical signals intervention through activating diverse signal pathways. As the mechanism gradually clears, it becomes a new target to find opportunities to diagnose and treat cancer. In this review, we investigated the process of collagen remodeling in cancer progression and discussed the interaction between collagens and cancer cells. Several typical effects associated with collagens were highlighted in the review, such as fibrillation in precancerous lesions, enhancing ECM stiffness, promoting angiogenesis, and guiding invasion. Then, the values of cancer diagnosis and prognosis were focused on. It is worth noting that several generated fragments in serum were reported to be able to be biomarkers for cancer diagnosis and prognosis, which is beneficial for clinic detection. At a glance, a variety of reported biomarkers were summarized. Many collagen-associated targets and drugs have been reported for cancer treatment in recent years. The new targets and related drugs were discussed in the review. The mass data were collected and classified by mechanism. Overall, the interaction of collagens and tumor cells is complicated, in which the mechanisms are not completely clear. A lot of collagen-associated biomarkers are excavated for cancer diagnosis. However, new therapeutic targets and related drugs are almost in clinical trials, with merely a few in clinical applications. So, more efforts are needed in collagens-associated studies and drug development for cancer research and treatment.


Introduction
Cancer is a serious disease for humans with high morbidity and death. The extracellular matrix (ECM) is a non-ignored factor in cancer progression due to the fact that it is the major component of tumor stroma, playing the roles of the physical scaffold and regulator of cell and tissue function. ECM could not only act as a medium to conduct signals but also elicits biochemical and biophysical signaling to excite cells [1,2]. The interaction between tumor cells and ECM is bidirectional and dynamical, which could reshape the morphology of perimalignant tissue continuously. Recent studies have shown that tumors could directly leverage ECM remodeling to create a microenvironment that promotes tumorigenesis and metastasis. Conversely, many cell behaviors are inspired by transformed ECM, such as adhesion, migration, angiogenesis, and canceration [3,4].

The Observation Methods of Collagens in Research and Clinic
The paramorphia of collagens is a significant signal of many diseases, especially fibrosis and precancerosis. Several imaging modalities were developed for quantitative or qualitative analysis collagens. The coarse collagen fiber could be observed directly under the light field of optical microscopy, the visual scenes of which are shown in Figure 1A. However, the tiny collagen fibrils or soft networks are hard to be identified under optical microscopy. Pathological staining technology offers the assistance to catch high-recognition images under light fields, especially for clinicopathological sections. Of note, immunohistochemistry (IHC) could recognize the types of collagens with the assistance of optical microscopy. Fluorescence immunostaining is a method to mark the collagens with a fluorescence dye, then the marked collagens are highlighted under the excitation light of a fluorescence microscope from the background. This method is suitable for all subtypes of collagens, whether fibrous or not; however, this method is hard to apply in practical applications due to the fact that the fluorescence dye is hard to wash completely from the dense networks, causing significant interference to collagen identification. Collagen fiber is a non-centrosymmetry and high second-order nonlinear coefficient structure that could produce the signal of second-harmonic generation (SHG) under two-photon excitation. This provides an excellent opportunity to image collagens, and SHG imaging has been promoted to be a popular and effective approach to investigate collagens in the laboratory. As an example, SHG imaging is developed to investigate collagen fiber organization, which has a high resolution capable of recognizing faint signals [10]. In fact, SHG imaging of collagens is always used combining with confocal microscopy, which could catch the interior signals of the samples to reconstruct 3D images [11,12]. We took an SHG image of type I collagen fibers in an in vitro experiment using confocal microscopy shown in Figure 1B. Due to the fibrous structure differs from other matrices, the pure reflective mode of the confocal microscope is available to take the fiber signal, although the clarity is worse in contrast to SHG imaging [13][14][15]. Figure 1C exhibited the collagen network structure in an in vitro experiment using the reflective mode of the confocal microscope. Electron microscopy is an instrument manufactured by the principle of electron photons, which replaces light beams and optical lenses with electron beams and electron lenses to take the subtle structure imaged with a very high magnification. Fortunately, collagen fibers could be imaged by the two commonly types of electron microscopy, i.e., scanning electron microscopy (SEM) and transmission electron microscopy (TEM) [16][17][18]. SEM could see the morphology of collagens limited to surface and longitudinal. We took the porous structure by SEM, shown in Figure 1D. TEM could evaluate the cross sections through 3D imaging. Figure 1E is the structure of collagen type I under TEM. Cryo-TEM advanced the TEM technique allowing the sample examined to maintain the frozen-hydrated state and removing the step of heavy-metal staining. The early banding analysis of reconstituted collagen fibrils was performed from cryo-TEM images [19][20][21]. Atomic force microscopy (AFM) is another morphology detection instrument based on a completely different principle, i.e., van der Waals force. AFM could be used to confirm the inner assembly of collagen fibrils. As an example, Figure 1F exhibited the microstructure of a single collagen fibril in a high magnification AFM image. However, AFM could only scan the 3D topographic feature, but not the section structure inside the matrix [22][23][24]. subtypes of collagens, whether fibrous or not; however, this method is hard to apply in practical applications due to the fact that the fluorescence dye is hard to wash completely from the dense networks, causing significant interference to collagen identification. Collagen fiber is a non-centrosymmetry and high second-order nonlinear coefficient structure that could produce the signal of second-harmonic generation (SHG) under two-photon excitation. This provides an excellent opportunity to image collagens, and SHG imaging has been promoted to be a popular and effective approach to investigate collagens in the laboratory. As an example, SHG imaging is developed to investigate collagen fiber organization, which has a high resolution capable of recognizing faint signals [10]. In fact, SHG imaging of collagens is always used combining with confocal microscopy, which could catch the interior signals of the samples to reconstruct 3D images [11,12]. We took an SHG image of type I collagen fibers in an in vitro experiment using confocal microscopy shown in Figure 1B. Due to the fibrous structure differs from other matrices, the pure reflective mode of the confocal microscope is available to take the fiber signal, although the clarity is worse in contrast to SHG imaging [13][14][15]. Figure 1C exhibited the collagen network structure in an in vitro experiment using the reflective mode of the confocal microscope. Electron microscopy is an instrument manufactured by the principle of electron photons, which replaces light beams and optical lenses with electron beams and electron lenses to take the subtle structure imaged with a very high magnification. Fortunately, collagen fibers could be imaged by the two commonly types of electron microscopy, i.e., scanning electron microscopy (SEM) and transmission electron microscopy (TEM) [16][17][18]. SEM could see the morphology of collagens limited to surface and longitudinal. We took the porous structure by SEM, shown in Figure 1D. TEM could evaluate the cross sections through 3D imaging. Figure 1E is the structure of collagen type I under TEM. Cryo-TEM advanced the TEM technique allowing the sample examined to maintain the frozen-hydrated state and removing the step of heavy-metal staining. The early banding analysis of reconstituted collagen fibrils was performed from cryo-TEM images [19][20][21]. Atomic force microscopy (AFM) is another morphology detection instrument based on a completely different principle, i.e., van der Waals force. AFM could be used to confirm the inner assembly of collagen fibrils. As an example, Figure 1F exhibited the microstructure of a single collagen fibril in a high magnification AFM image. However, AFM could only scan the 3D topographic feature, but not the section structure inside the matrix [22][23][24].  Clinically, medical imaging technology is an important contribution to disease diagnosis and surgical guidance, which is popular due to its non-invasive, including ultrasound, X-ray, computed tomography (CT), and nuclear magnetic resonance imaging (MRI). The medical imaging technology allows the sample to be examined in a close to physiological hydration state without chemical fixation of sectioning. Diffraction pattern signals are inspired by high-intensity X-rays scattering from the arranged collagen molecules and fibrils in the bulk matrix. So, X-ray is available to analysis the average diameter, lateral arrangement, and alignment of collagen fibrils; as an example, corneal ultrastructure is obtained by X-ray with a powerful synchrotron source [25,26]. CT and MRI are able to analyze the collagen fibers quantitatively. Karjalainen et al. used micro-computed tomography to analyze the three-dimensional collagen orientation of the human meniscus posterior horn in health and osteoarthritis [27]. Eder et al. used MRI to evaluate the regional collagen fiber network in the human temporomandibular joint disc [28]. Ultrasound is developed to assess collagen microstructure based on the integrated backscatter coefficient (IBC). Mercado, Karla P. et al. employed IBC as a quantitative ultrasound parameter to detect the quantify spatial variations of collagen fiber density and diameter [29]. Kenton et al. prospectively characterize the collagen organization in the Achilles and patellar tendon [30]. However, IHC still plays an irreplaceable role in the clinic due to its ability to recognize the types of collagens. Pathological detection based on IHC is the gold standard for tumor diagnosis in the clinic.

Precancerous Lesions
Stromal alterations are the reference precursors to predict the progression of carcinoma, especially collagens, which is the main component of the ECM. The occurrence of fibrosis and the base membrane abnormality, which are mainly involved by collagens, are the important cues orienting to deterioration, having a significant clinical meaning. In precancerous lesions, it is likely to occur due to the abnormal ratio of collagen types, new collagen types secreted, and abnormal molecular structure, causing the abnormal ECM.
The fibrosis of organ tissue is an important cue to draw attention to, which is likely to be a stage of deterioration, such as liver fibrosis, lung fibrosis, oral submucosa fibrosis, and so on. Most hepatocellular carcinoma develops through the progression of chronic liver injury, hepatic inflammation, and fibrosis, so liver fibrosis is a precursor of cancerization. It has been confirmed that high fibrosis index is positively correlated with the risk of hepatocellular carcinoma [31,32]. The configuration of collagen types is changed greatly in the progression of fibrillation. In normal liver, the collagens in ECM are type IV and VI, which are non-fibrillar; however, a great accumulation of fibrillar collagens occurs in the fibrotic liver, such as collagen type I and III [33]. Similarly, idiopathic pulmonary fibrosis is considered to have a high risk of concomitant lung cancer in the clinic. What is worse, patients with idiopathic pulmonary fibrosis have a poor prognosis with a 2-5 year survival time, which is worse than liver fibrosis [34]. Enhancing nodules in post-radiation fibrosis in CT imaging could be an early detecting method of recurrent lung cancer [35]. The deposition of collagens in the interstitium is the direct reason for fibrosis. In early pulmonary fibrosis, collagen type III predominates in the matrix; however, the proportion is gradually replaced by collagen type I along the process of pulmonary fibrosis to the late stage [36]. Oral submucous fibrosis is a precancerous disorder and has a 1.5-15% chance of transforming into a malignant tumor. The characteristics of oral submucous fibrosis are abnormal collagen deposition. In oral submucous fibrosis cells, the collagen synthesis is increased and the ratio of the α1(I) to α2(I) chains of type I collagen is~3:1 whereas~2:1 in normal cells [37,38].
The analysis of collagen fibers is significant in predicting cancer. Despotovic et al. caught the SEM images of the perimalignant tissue shown in Figure 2A. They found that the altered organization of collagen fibers was observed at 10 cm and 20 cm away from the malignant tumor. The alignment of collagen fibers is step increased as proximity to the tumor [39]. Wu et al. focused on the base membrane in intraductal carcinoma, a precancerous lesion of invasive ductal carcinoma. They found that the base membrane was distorted and elongated compared with the normal cases ( Figure 2B). Several types of gynecological cancers are reported differently in terms of precancerous lesions, such as breast cancer, ovarian cancer, and vulvar cancer. Castor et al. characterized the collagen fibers in preneoplastic lesions compared with normal tissue and squamous carcinoma in vulvar cancer using SHG microscopy. They found that the collagen fibers showed better organization in the normal tissue than in the other two stages. The devise parameters of collagen fibers showed reducing in squamous carcinoma and preneoplastic lesions compared with normal tissue, i.e., quantity, organization, and uniformity; however, no obvious difference was observed between squamous carcinoma and preneoplastic lesions [40]. In cervical precancers, several collagen-associated indicators directly affect the quantitative classification of precancerous stages, including the density and degree of linear arrangement, collagen degradation, and the breakage of collagen cross-links. Zaffar et al. focused on this valuable information by developing a series of studies of the spatial frequencies of collagens for cervical precancer detection [41,42]. The expression of collage IV seriously affects the integrity of the base membrane because collagen IV is the main complement of the base membrane. As a precancerous lesion of squamous cell carcinoma in malignant skin tumors, actinic keratosis has shown the premonition of collagen IV low expression. Hirakawa et al. compared the expression of collagen IV using immunohistochemical in actinic keratosis tissue. The result showed that collagen IV in dysplastic areas of actinic keratosis samples was lower than peri-lesional tissue and no longer continuous [43].
distorted and elongated compared with the normal cases ( Figure 2B). Several types of gynecological cancers are reported differently in terms of precancerous lesions, such as breast cancer, ovarian cancer, and vulvar cancer. Castor et al. characterized the collagen fibers in preneoplastic lesions compared with normal tissue and squamous carcinoma in vulvar cancer using SHG microscopy. They found that the collagen fibers showed better organization in the normal tissue than in the other two stages. The devise parameters of collagen fibers showed reducing in squamous carcinoma and preneoplastic lesions compared with normal tissue, i.e., quantity, organization, and uniformity; however, no obvious difference was observed between squamous carcinoma and preneoplastic lesions [40]. In cervical precancers, several collagen-associated indicators directly affect the quantitative classification of precancerous stages, including the density and degree of linear arrangement, collagen degradation, and the breakage of collagen cross-links. Zaffar et al. focused on this valuable information by developing a series of studies of the spatial frequencies of collagens for cervical precancer detection [41,42]. The expression of collage IV seriously affects the integrity of the base membrane because collagen IV is the main complement of the base membrane. As a precancerous lesion of squamous cell carcinoma in malignant skin tumors, actinic keratosis has shown the premonition of collagen IV low expression. Hirakawa et al. compared the expression of collagen IV using immunohistochemical in actinic keratosis tissue. The result showed that collagen IV in dysplastic areas of actinic keratosis samples was lower than peri-lesional tissue and no longer continuous [43].

Breaching Base Membrane
In the development of cancer progression, collagens become from a passive barrier resisting cancer cells to an accomplice in promoting the progression. The base membrane is a baffle between tumor cells and normal tissue originally; however, it would be breached at the early stage of carcinogenesis. The main reason is collagen IV, which occupies the major complement of the base membrane, is degraded directly or indirectly by tumor cells. In the degrading progress, matrix metalloproteinase plays an indispensable role, which is secreted by tumor cells or stimulated epithelial or stromal cells. In the subsequent progression of cancer development, degrading collagens remains an effective strategy to create roads to invasion or migration. Yan et al. reported a period that collagens exhibited a significant loss in invasive ductal carcinoma compared to the normal case and precursor lesion [44]. Recently, another cue is revealed that cancer cells could break the base membrane just facilitated by physical forces, which is a completely different manner independent of protease. As evidence, the collagen IV meshwork exhibits a densified structure at the adjacent disruption; on the contrary, the collagen IV scaffold should be decreased under the degradation theory. Piercing filopodia is captured in further studies, which is proved to have a pivotal role in the mechanical response model. The force of push and pull by the contractility of piercing filopodia is considered an explanation of the base membrane non-protease disruption [45][46][47].

Enhancing ECM Stiffness
Enhancing the stiffness of ECM is another strategy to promote tumor cell migration and invasion through the pathway of activating integrins to increase the adhesion between cells and substrate. Castor et al. found that the parameters of collagen fibers present higher in metastatic vulvar cancer patients than in that without metastases [40]. High ECM stiffness is mainly realized by increasing the secretion of fibrous collagens and the deposition of non-fibrous collagens, most notably collagen type I and type IV [48,49]. In medical statistics, ECM stiffness is considered a reason for tumor rise incidence with aging because it is a fact that the aged tissues are stiffer due to containing more aberrant crosslinked collagens [50]. High collagen density in tumors is often closely correlated with poor prognosis; however, the association between collagen density and cancer progression is not completely clear. Recent reports studied it from various perspectives. The most accepted view is that ECM stiffness is closely related to cancer-associated fibroblasts (CAFs) [51][52][53][54]. This viewpoint is supported because CAFs are the main producer of abnormal collagen fibers. Shibata et al. reported that CAFs promote ECM stiffness in response to the signals from yes-associated protein 1 (YAP1) [55]. However, Farhat et al. found that abnormal activation and expression of the Lox family of proteins, a group of extracellular enzymes catalyzing the cross-linking of collagens, would lead to the ECM toward increased rigidity and fibrosis [56].

Orienting the Collagen Fibers
Orienting the collagen fibers is a significant manner of remodeling the ECM by cancer cells. The stress of tumor growth remodels the collagen fibers toward the tumor circumference at the tumor periphery. Those oriented collagen fibers provide a highway to cancer cell invasion directly [57][58][59][60]. Meanwhile, the tension of aligned collagen fiber bunches contributes to ECM stiffness [61]. Many studies reported the high orientation of collagen fibers in the tumor location [13,57,59,62]. The mechanism of collagen fibers orientation in the malignant tumor is still unclear. Many scientists dedicated themselves to this study, and they revealed it is a complex process, maybe participated in by tumor cells, CAFs, mesenchymal cells, and interstitial fluid. Tumor cells remodel the collagen fibers through various manners, including the stress of tumor growth, the pseudopod with contractility, the protease, and so on, which is a complex process with multiple pathways [61,[63][64][65][66][67]. The examples are shown below. Ray et al. reported that the traction forces from the di-rected migration of cancer cell clusters are a mechanism of collagen fiber alignment [68]. Drifka et al., using a pancreatic ductal adenocarcinoma model, found that human pancreatic stellate cells could orchestrate the alignment of collagen fibers, and they further found that the aligned collagen fibers, in turn, enhanced cancer cell migration. Bayer et al. showed that the collagen receptor DDR2 in CAFs reorganizes collagen fibers at the tumor-stromal boundary [69]. Hanley et al. reported that CAFs could induce the formation of elongated collagen fibers [70]. Del Amo et al. constructed a collagen-based osteoblasts model in 3D microfluidic devices, and the data suggested that a high rate of interstitial fluid flow could modify the orientation of collagen fibers [71].

Promoting Angiogenesis
In cancer progression, limited oxygen and nutrients are always insufficient for the rapid growth of cancer cells. Angiogenesis is induced under a poor supply of oxygen and nutrient. The angiogenesis process is always accompanied by the upregulation of collagens and ECM-modifying enzymes, which has been confirmed in different human tumor types. That is because the collagen network provides the scaffold for recruited endothelial cells migrating during angiogenesis. Even the remote collagen fiber network is induced by tumor cells into orientation steers for angiogenesis. We illustrated the process of angiogenesis induced by collagen fibers in Figure 3A. Piotrowski-Daspit et al. took a confocal image of a representative tissue comprised of breast cancer cell MDA-MB-231. They found that the collagen fibers showed high alignment obviously ( Figure 3B). The collagen network provides the intact physical connection with tumor tissue for vascular endothelial cells' mechanical sensing, which is considered essential during angiogenesis [72]. Nathaniel et al. observed the fibril organization around angiogenic sprouts and growing neo-vessels in real time. They found that a strong association of fibrillar collagens occurred during vessel reconstitution and a substantial collagen fibril reorganization at the sites of sprout and neo-vessel tips [73]. Luthria et al. investigated the vasculature around the tumor in the dense collagen micro-environment-shown in Figure 3C [74]. Niels et al. reported that type VIII collagen was elevated in diseases associated with angiogenesis and vascular remodeling, such as pulmonary fibrosis and cancer [75]. The examples are shown below. Ray et al. reported that the traction forces from the directed migration of cancer cell clusters are a mechanism of collagen fiber alignment [68]. Drifka et al., using a pancreatic ductal adenocarcinoma model, found that human pancreatic stellate cells could orchestrate the alignment of collagen fibers, and they further found that the aligned collagen fibers, in turn, enhanced cancer cell migration. Bayer et al. showed that the collagen receptor DDR2 in CAFs reorganizes collagen fibers at the tumorstromal boundary [69]. Hanley et al. reported that CAFs could induce the formation of elongated collagen fibers [70]. Del Amo et al. constructed a collagen-based osteoblasts model in 3D microfluidic devices, and the data suggested that a high rate of interstitial fluid flow could modify the orientation of collagen fibers [71].

Promoting Angiogenesis
In cancer progression, limited oxygen and nutrients are always insufficient for the rapid growth of cancer cells. Angiogenesis is induced under a poor supply of oxygen and nutrient. The angiogenesis process is always accompanied by the upregulation of collagens and ECM-modifying enzymes, which has been confirmed in different human tumor types. That is because the collagen network provides the scaffold for recruited endothelial cells migrating during angiogenesis. Even the remote collagen fiber network is induced by tumor cells into orientation steers for angiogenesis. We illustrated the process of angiogenesis induced by collagen fibers in Figure 3A. Piotrowski-Daspit et al. took a confocal image of a representative tissue comprised of breast cancer cell MDA-MB-231. They found that the collagen fibers showed high alignment obviously ( Figure 3B). The collagen network provides the intact physical connection with tumor tissue for vascular endothelial cells' mechanical sensing, which is considered essential during angiogenesis [72]. Nathaniel et al. observed the fibril organization around angiogenic sprouts and growing neovessels in real time. They found that a strong association of fibrillar collagens occurred during vessel reconstitution and a substantial collagen fibril reorganization at the sites of sprout and neo-vessel tips [73]. Luthria et al. investigated the vasculature around the tumor in the dense collagen micro-environment-shown in Figure 3C [74]. Niels et al. reported that type VIII collagen was elevated in diseases associated with angiogenesis and vascular remodeling, such as pulmonary fibrosis and cancer [75].  Another vasculature-like structure could also support tumors with blood, which is called vasculogenic mimicry constructed by cancer cells without endothelial cells. The primary trigger is dense collagen, inducing cancer cells to form interconnected networks [4]. Velez et al. found that the collagen matrixes with small pores and short fibers induced vasculogenic mimicry. The upregulation of β1-integrin is triggered by the collagen matrix architecture and is considered a significant reason in the follow-up research [77]. The non-fibril collagens contribute to vasculogenic mimicry, such as collagen type IV, VI, and XVI. Bedal et al. found the NC11 domain of human collagen XVI, one non-fibril collagen, induced vasculogenic mimicry in oral squamous cell carcinoma cells. The process is realized by triggering the generation of tubular-like net structures on a laminin-rich matrix [78].

Promoting Invasion
Escaping from the primary site is another strategy employed by cancer cells to obtain supplies; collagens also play a significant role in this process. Collagen fibers induce cancer cell migration by contact guidance [68]. Especially, the aligned collagen fibers are confirmed to provide "a highway" to cancer cell invasion [79][80][81]. The guidance of collagen fibers runs through the major process of cancer progression. In the early stage, the collagen fibers that are perpendicular to the solid tumor provide conduits to escape and penetrate the base membrane [58,82]. In the following process, the interface between the collagen fiber bundle in ECM and the peripheral interstitial components provides the opportunity for breakthrough for tumor cells. We illustrated the intravasation guided by collagen fibers in Figure 4A. The role of the "highway" has been verified by several reports. In an experiment on the orientation of collagen fibers, cancer cells could break through the high density of Matrigel along with the collagen fibers, while they were unable to do that in the absence of oriented collagen fibers-shown in Figure 4B [59]. In the process of intravasation (cancer cell entering blood vessel) and extravasation (cancer cell exuding blood vessel), collagen fibers as important participators are remodeled firstly and, in turn, guide the cancer cell invasion [83][84][85][86]. Moreover, the orientation and the deposition of collagens make the matrix stiff. The stiffness of the ECM could induce the enhancement of transforming growth factorβ (TGF-β) to increase the cell adhesion to the substrate, further promoting the migration and invasion of tumor cells. Masoud et al. reported that ECM stiffness could make a bridge in the base membrane through the related pathway of transforming growth factor-β (TGF-β), which contributed to EMT [87]. Figure 4C illustrates the process of EMT induced by the high stiffness ECM. Han et al. displayed a merged image of the SHG signal and light field of the pathological section ( Figure 4D). They found that the tumor cells invaded along with the direction of fibers [59]; however, the mechanism of the stiffness ECM acting on tumor cells is complex; there are other pathways being investigated as well. For instance, Anne et al. investigated the immune modulatory properties of collagens in cancer, and they revealed that collagens could affect the function and phenotype of various types of tumor-infiltrating immune cells, such as tumor-associated macrophages and T cells [88]. Yu et al., from the perspective of drug transport, revealed that ECM stiffness was a barrier to drug screening at the tumor site [89].
Non-fibril collagens also participate in the invasion process. Fang et al. reported that collagen type IV occurred a series of changes to provide a proper tumor microenvironment for cancer invasion. In the initial stage, collagen IV presents an irregular sheath in the base membrane, and then it is degraded and accompanied by linear redeposition to form invasion fronts, which would become the escape sites of cancer cells [90]. Zhang et al. reported that collagen XIII could promote invasive tumor growth, enhance the stemness of cancer cells, and induce anoikis resistance [91]. Karagiannis et al. investigated collagen type XII by immunohistochemistry and found that collagen XII was highly expressed in the invasion front of cancer cells [92]. Kumagai et al. found that the intercellular expression of type XVII collagen could promote collective invasion by producing intercellular adhesion sites for contact following [93]. Miyake et al. reported that collagen IV and collagen XIII played a pivotal role in tumor invasion by inducing tumor budding [94]. Non-fibril collagens also participate in the invasion process. Fang et al. reported th collagen type IV occurred a series of changes to provide a proper tumor microenviro ment for cancer invasion. In the initial stage, collagen IV presents an irregular sheath the base membrane, and then it is degraded and accompanied by linear redeposition form invasion fronts, which would become the escape sites of cancer cells [90]. Zhang al. reported that collagen XIII could promote invasive tumor growth, enhance the stem ness of cancer cells, and induce anoikis resistance [91]. Karagiannis et al. investigated co lagen type XII by immunohistochemistry and found that collagen XII was highly e pressed in the invasion front of cancer cells [92]. Kumagai et al. found that the intercellul expression of type XVII collagen could promote collective invasion by producing interce lular adhesion sites for contact following [93]. Miyake et al. reported that collagen IV an collagen XIII played a pivotal role in tumor invasion by inducing tumor budding [94].

Collagen-Associated Biomarkers for Cancer Diagnosis
Collagens, as the main component of the ECM, would be abnormally expressed deposited in the location or the pericarcinoma of the cancer site. It is for this reason th collagens are proposed as a diagnostic biomarker in cancer types. Type I collagen is r ported to be a good diagnostic marker to detect the metastasis of lung cancer, the expre sion level of which is a significant indicator of distinct bone involvement happening not in lung cancer metastasis [95]. The degradation products of type I collagen in seru are a significant reference for diagnosis and prognosis. Nurmenniemi et al. reported th  the carboxyl terminal peptide beta-special sequence, total type I procollagen amino terminal propeptide, alkaline phosphatase of collagen I in serum bone metastases of lung cancer [100] type I collagen C-terminal telopeptide in serum head and neck squamous cell carcinoma [96] MMP-generated fragments of type I collagen in serum lung cancer; breast cancer [97,101] the expression level of collagen type I alpha 1 hepatocellular carcinogenesis and metastasis [102] the carboxyterminal collagen type I telopeptid breast cancer with bone metastases [103] the expression level of collagen type I alpha 2 gastric cancer [104] Type III MMP-generated type III collagen fragment, procollagen type III N-peptide in serum breast cancer [101,105] the propeptide of type III collagen in serum pancreatic cancer [106] the level of procollagen III aminoterminal propeptide malignant head and neck cancer; cervical carcinoma [107,108] procollagen type III N-peptide in serum gastric cancer; ovarian cancer; lung cancer [109][110][111] type III collagen in serum hepatocellular carcinoma [112] type III collagen in cancer tissue malignant pleural mesothelioma [113] the expression of collagen III and collagen III mRNA Ewing's sarcoma [114] Type IV the type IV collagen, MMP-generated type IV collagen fragment in serum in serum breast cancer [101,105] 7S domain of type IV collagen in serum gastric cancer; pancreatic cancer; lung cancer; extrahepatic cancer; hepatocellular carcinoma [111,[115][116][117][118] the expression level of collagen IV malignancy glioma [119] the differential localization of the type IV collagen alpha5/alpha6 chains colorectal epithelial tumors [120] MMP-degradation of type IV collagen ovarian cancer [121] the expression level of collagen IV prostate cancer [122] the expression of collagen type IV urothelial carcinoma; oral squamous cell carcinoma [123,124] the expression of the type IV collagen alpha1 and alpha2 chains hepatocarcinogenesis [125] the expression of the type IV collagen alpha1 bladder cancer [126] the expression of collagen IV in tumor location invasive adenocarcinoma [127] serum N-terminal pro-peptide of type IV collagen 7S domain hepatocellular carcinoma in patients with liver cirrhosis [128] [133] serum collagen type VI alpha 3 pancreatic ductal adenocarcinoma [134] stromal collagen type VI salivary gland cancer [135] the expression of COL6 pancreatic cancer [136] collagen type VI alpha 1chain esophageal squamous cell carcinoma [137] Type VII the expression level of type VII collagen identifying lung cancer subtypes; skin cancer; gastric cancer [138][139][140] the expression level of collagen type VII a 1 chain clear cell renal cell carcinoma [141] Type VIII the NC1 domain of human type VIII collagen a1 chain in serum colorectal cancer [142] collagen type VIII alpha 1 chain breast cancer; colon Adenocarcinoma [143,144] Type XI the expression level of Collagen Type XI Alpha 1 Chain colorectal cancer; breast carcinoma invasiveness; colon cancer; gastric cancer; pancreatic ductal adenocarcinoma; pancreatic adenocarcinoma; pancreatic cancer [145][146][147][148][149][150][151] the expression of procollagen XI Alpha 1 Chain breast cancer [152] type XIII the expression of the type XIII collagen alpha1 bladder cancer [126] Type XIV the expression of type XIV collagen breast cancer [153] Type XV fragments of collagen XV collagen alpha1 in urine gastrointestinal cancer [154] Type XVIII the expression of type XVIII collagen pancreatic ductal adenocarcinoma [155] serum endostatin (a fragment of collagen XVIII) levels colorectal cancer [156] Type XXIII the expression of collagen type XXIII alpha 1 chain clear cell renal cell carcinoma [157]

The Predictive Value of Collagens
Collagens have high value in monitoring cancer processes, prognosis, and recurrence. The predictive value is exhibited in many types of cancer, including breast cancer, prostate adenocarcinoma, lung cancer, hepatocellular carcinoma, colon cancer, and pancreatic cancer. In breast cancer, the qualitative descriptors of collagens at the boundaries between tumor and stroma are important indicators for tumor staging. In invasive breast cancer, the 5-year disease-free survival in patients with low tumor-stromal ratios is poorer than the patients with high tumor-stroma ratios. In lung cancer, high levels of collagen I in serum and tissue demonstrated a significant decrease in survival. Furthermore, collagen metabolic components and prolyl hydroxylases have been marked as the predictive factors of lung cancer presence, progression, and outcome. Abnormal collagen expression is associated with cancer overall survival (OS). The prognosis value is outstanding in enriched stromal cancers, such as pancreatic cancer and colon cancer. It is confirmed that the alignment of collagens was associated with the patient survival of pancreatic ductal adenocarcinoma (PDAC). Highly aligned collagen fibers report poor prognosis in PDAC, according to clinical statistics. Collagen types I, III, VI, and XI were shown to be associated with the diverse response of pancreatic cancers, such as proliferation, migration, decreasing E-cadherin expression, and cancer-associated fibroblasts [158]. Furthermore, the percent survival with a high level of collagen type I is observably lower than the low level. It was shown that the level of collagen type I had a negative correlation to OS for pancreatic cancer [159]. In colon cancer, the level of collagen expression is the key indicator to predicting the OS and risk, especially the types of COL1A1, COL1A2, COL3A1, COL4A3, and COL4A6. Please refer to reference [158] for more information on the predictive value of collagens.

Therapeutic Opportunities of Cancer Target to Collagens and the Collagen Associated Molecules
Collagen remodeling plays a significant role in cancer progression, as the content described above. It provides the possibility for therapeutic cancer targeting to collagens undergoing remodeling, including inhibiting the synthesis of collagens and interdicting the reactivator on the signal pathway. Of note, many small molecule inhibitors of collagen synthesis and functioning are available as anti-cancer drugs. They could accurately point to the target, avoiding damage to normal cells, which is an absolute advantage over conventional chemotherapeutic agents.

Inhibiting the Synthesis and Secretion of Collagens
Inhibiting collagen synthesis and secretion is a strategy for cancer therapeutics, which could prevent the series of effects initiated by collagen remodeling. For instance, as a catalyst in a key step of collagen biosynthesis, CP4H was linked with cancer metastasis in recent studies. Undoubtedly, CP4H has confirmed a new target for anti-cancer drugs. Several drugs targeted to CP4H have been developed as CP4H inhibitors, such as Ethyl 3,4-dihydroxybenzoate (EDHB) and 2-(5-carboxythiazol-2-yl) pyridine-5-carboxylic acid (pythiDC) [160]. Lysyl hydroxylation is a key step for collagen cross-link and deposition, which is a potential target to inhibit collagen synthesis. Aiming at the collagen remodeling induced by aberrant lysyl hydroxylation and collagen cross-link, lysyl hydroxylation is developed as a potential target for cancer therapeutic. Minoxidil, as an inhibitor of lysyl hydroxylation, is confirmed to have anti-invasive effects on human breast cancer [161]. Relatively, procollagen-lysine 2-oxoglutarate 5-dioxygenase (PLOD) drew much attention as the catalyzer of the process of lysyl hydroxylation. Increased PLOD expression has been detected in many types of cancer. Targeting PLODs is considered a potential strategy for cancer treatment; however, there are still no reports revealing the related anti-cancer drugs [162]. More available targets and the related drug aiming at collagen synthesis are summarized in Table 2.

Interdicting the Receptors
The collagen receptors are the direct trigger of the interaction between cancer cells and collagens. Cancer cells sense the surrounding microenvironment by responding to the biochemical and mechanical properties of transmembrane receptors, including integrins and discoidin domain receptors (DDRs). Interdicting the receptors of collagens is another effectual strategy. We summarize the related targets and the corresponding drug below.

Integrins
Integrins are the important receptor of collagens on the cell membrane. It is a large family, at least including 24 different functional heterodimeric receptors distinguished by 18α-subunits and 8β-subunits [190,191]. According to the reports by far, the integrins of α1β1, α2β1, α3β1, α4β1, α5β1, α6β1, α9β1, α10β1, α11β1, α5β3, and α5β8 have been found to be involved in tumor growth and metastasis by the regulation of collagen-binding integrin signal [190,[192][193][194][195][196][197][198][199][200]. Further, integrin-mediated pathways are reported many times to connect to drug resistance [201,202]. So, integrins are considered an attractive drug target for cancer therapeutics [203]. This concept is largely encouraged by preclinical studies. Cilengitide, an inhibitor of integrin αvβ3 and αvβv, has been developed as an anti-cancer drug in various tumor types. Several clinical trials were carried out on diverse cancers, such as lung cancer, breast cancer, glioblastoma, prostate cancer, melanoma, and squamous cell cancer [204,205]. Abituzumab (EMD 525797), a monoclonal antibody targeting integrin alpha nu heterodimers, was also demonstrated as an anti-cancer drug. The phase I clinical trial was completed in ovarian cancer patients with liver metastases and the phase II clinical trial was completed in metastatic colorectal cancer; however, successful clinical trials are few in number. For example, cilengitide failed to improve survival for glioblastoma patients in the phase III trial, despite the standard care. At the same time, the trial data of combining abituzumab showed no improvement in the progression-free survival of patients compared with the standard of care alone. So, further verified trials are essential based on stratifying the patient population [206]. Exploiting tumor-specific integrin expression profiles or downstream integrin effectors is also an alternative strategy to target for the development of anti-cancer drugs. Many typical integrin activation factors are noticed, such as FAK, LOX, mucins, and the corresponding inhibitors are developed for cancer therapeutic. Moreover, snake venom disintegrins were confirmed to inhibit integrins and further effected cancer treatment. We summarize the data in Table 3 below.

DDRs
Another important receptor is DDRs family, including DDR1 and DDR2, which have been proved to regulate various cellular signaling pathways, including cell proliferation, adhesion, migration, and matrix remodeling. DDRs are a subfamily of receptor tyrosine kinases, activated by the triple-helical structure of collagens in the interaction. It is revealed that DDRs possess a special activation mechanism, which initiates the pathways leading to autophosphorylation through collagen binding. In the fibrillar collagens, DDR1 and DDR2 could respond to collagen type I, II, III, and V, while in the non-fibrillar collagens, type IV, VI, VIII, and X are also the activator. However, the mechanism of extracellular collagen binding and activation of the cytosolic kinase domain of the receptors is not clear so far. A recognized theory is that DDRs occur dimerization before the ligand binding with collagens, which is different from the other receptor tyrosine kinases undergoing dimerization after ligand binding [210]; then, the amino acid produces residues of collagens as the sites bind with DDRs dimerization [211]. In turn, the activated DDRs could trigger the signal transduction pathways of cell behaviors, such as proliferation, migration, and invasion [212]. In many cancers, overexpression of DDRs is associated with a poor prognosis. Deng et al. demonstrated that collagen-induced DDR1 activation in cancer cells could recruit tumorassociated neutrophils to form extracellular traps, enhancing the subsequent cancer cell invasion and metastasis [213]. The imbalance expression of DDRs has been demonstrated to be associated with most cancers. Huo et al. revealed that high expression of DDR1 was associated with poor prognosis in pancreatic ductal adenocarcinoma [214]. Xie et al. found that overexpression of DDR1 promoted the aggressive growth, migration, and invasion of bladder cancer cells, in which process collagen IV was a signal axis [215]. As DDRs play a significant role in cancer progression, DDRs would be new promising targets for cancer treatment, such as the design of DDR inhibitors for use in clinical settings. Some drugs were developed to inhibit DDR expression, such as dasatinib, imatinib, nilotinib, and ponatinib. Dasatinib was confirmed enabling to inhibit gastric cancer cell migration and invasion in the assays. Several clinical trials were completed in phases I and II, such as lung cancer, breast cancer, and prostate cancer. Nilotinib was reported to reduce metastatic colorectal cancer invasion by inhibiting DDR1 kinase activation. Many clinical trials are currently underway in several types of cancer, such as breast cancer, gastrointestinal stromal tumors, and so on. The antibody-drug conjugate targeting DDRs is utilized in anti-cancer drug development. As an example, T4H11-DM4 is demonstrated to be effective for colon cancer. Many other drugs were developed to target DDRs in recent years-shown in Table 4.

Targeting to Collagen-Induced Chemoresistance
More and more data implicated the desmoplastic reaction is substantially related to chemoresistance in chemotherapeutics. It is reported that the pancreatic cancer cells grown in collagens demonstrated low sensitivity to gemcitabine chemotherapy. A further study showed that three-dimensional collagens enabled an increase in ERK1/2 signaling, which is known to promote chemotherapy resistance in several cancers. In other malignancies, it is repeatedly reported that collagens protect cancer cells against chemotherapy. For instance, in lung cancer models, collagens are shown to provide survival signals to attenuate the effects of chemotherapy. In this regard, MT1-MMP plays a critical physiological role in modulating growth factors and integrin signaling to enhance ERK1/2 phosphorylation in the collagen microenvironment. Moreover, the increasing density of the ECM initiated by collagen fibers and collagen deposition attenuates the permeability of drug delivery in chemotherapy. Targeting collagen-induced chemoresistance is an effective strategy to promote the chemotherapy effect. Some drugs are developed targeting the intermediary in the signal pathway. Moreover, enhancing the collagen penetration of anticancer drugs is an available strategy for cancer therapeutics, which targets collagen-associated stiffness ECM and high dense collagen fibers, increasing the drug efficacy [252]. We summarize the related drugs in Table 5.

Conclusions and Future Perspectives
Collagens are the main component of the ECM, and their remodeling occurs along with all processes of cancer progression. Collagens occur in abnormal morphology and distribution in precancerous lesions, which seriously affect the topography of the ECM. This provides a chance for clinicians to discover the pathology through medical imaging or pathological for cancer early diagnosis. In cancer development, collagens cause different stages of ECM topography, which contributes to cancer staging. Moreover, collagenassociated biochemical indicators are the significant biomarkers of cancer diagnosis and prognosis. Collagen participation in cancer progression is not only reflected in the remodeling under the influence of tumor cells or tumor-associated cells but is further revealed as the role of guider or inducer for cancer cell invasion. The "highway" for cancer cell invasion and the high stiffness of ECM are all enhanced by collagens, corresponding with the aligned collagen fiber bundle and collagen deposition, respectively. Based on the mechanism of the interaction between collagens and cancer cells, many opportunities for cancer therapeutics are revealed by disturbing or blocking the requirement in the interaction. The targets are diversely located across the pathway of collagen synthesis, binding to receptors, degradation, and drug transport. Overall, collagens provide many opportunities, whether for cancer diagnosis or cancer treatment; however, there are still many challenges from the exploration of therapeutic targets to drug development because there are only a few drugs allowed to enter the clinical application, while most of them are just in clinical trials or waiting for further clinical studies. Several questions and propositions are provided for further research. Firstly, the mechanism of collagen remodeling and its interaction with cancer cells is not completely clear. So further studies are essential on the mechanisms of collagen remodeling and the interaction between collagens and cancer cells. In our opinion, small molecule inhibitors are worth developing as anti-cancer drugs due to their excellent location and potential to cause no damage to normal cells. It would be a potential strategy to block the collagen remodeling; however, it is hard to implement for the reconstructed collagens. So, how to reduce the effectiveness of the reconstructed collagens is a valuable issue. In addition to the targets in the action pathway, we propose that ablating the reconstructed collagens is another strategy for cancer therapeutics. Moreover, to enhance the efficacy, we highlight the significance of finding a suitable nanocarrier to increase the drug transport capacity. Overall, further efforts are urgent in the collagen-associated mechanism and the therapeutic strategy for cancer.
Author Contributions: K.S. and Z.H., designed the study; Z.Y. and G.L. drew the schematic diagrams; K.S., X.Z. and Y.X. wrote and modified the paper. All authors have read and agreed to the published version of the manuscript.

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