Minor Type IV Collagen α5 Chain Promotes Cancer Progression through Discoidin Domain Receptor-1

Type IV collagens (Col IV), components of basement membrane, are essential in the maintenance of tissue integrity and proper function. Alteration of Col IV is related to developmental defects and diseases, including cancer. Col IV α chains form α1α1α2, α3α4α5 and α5α5α6 protomers that further form collagen networks. Despite knowledge on the functions of major Col IV (α1α1α2), little is known whether minor Col IV (α3α4α5 and α5α5α6) plays a role in cancer. It also remains to be elucidated whether major and minor Col IV are functionally redundant. We show that minor Col IV α5 chain is indispensable in cancer development by using α5(IV)-deficient mouse model. Ablation of α5(IV) significantly impeded the development of KrasG12D-driven lung cancer without affecting major Col IV expression. Epithelial α5(IV) supports cancer cell proliferation, while endothelial α5(IV) is essential for efficient tumor angiogenesis. α5(IV), but not α1(IV), ablation impaired expression of non-integrin collagen receptor discoidin domain receptor-1 (DDR1) and downstream ERK activation in lung cancer cells and endothelial cells. Knockdown of DDR1 in lung cancer cells and endothelial cells phenocopied the cells deficient of α5(IV). Constitutively active DDR1 or MEK1 rescued the defects of α5(IV)-ablated cells. Thus, minor Col IV α5(IV) chain supports lung cancer progression via DDR1-mediated cancer cell autonomous and non-autonomous mechanisms. Minor Col IV can not be functionally compensated by abundant major Col IV.


Introduction
Basement membranes (BMs), specialized extracellular matrices separating epithelial and endothelial cells from underlying mesenchyme, provide cells with structural support, as well as morphogenic and functional cues [1][2][3]. Type IV collagens (Col IV) are major components of BMs [1,3]. Three triple helical protomers, α1α1α2, α3α4α5 and α5α5α6, are formed by the Col IV α chains that further form collagen networks [4,5]. α1α1α2, the major Col IV, is widely expressed as a component of all BMs. α3α4α5 and α5α5α6, known as minor Col IV, have much restricted tissue distribution [4,5].
Col IV-initiated signals are essential survival and growth cues for liver metastasis in diverse tumor types [6]. BM proteins produced by mouse Engelbrecht Holm-Swarm sarcoma, known as Matrigel, enhanced the tumorigenicity of human cancer cells [7]. BM proteins, including α1(IV), protect small cell lung cancer cells from chemotherapy-induced apoptosis [8]. Angiogenesis, required by tumors to supply nutrients and oxygen, and to evacuate metabolic wastes, is dependent on correct interaction between endothelial cells and the vascular BMs [1,9,10]. Col IV plays crucial roles in supporting endothelial cell proliferation and migration. Blood vessel formation and survival are connected with proper collagen synthesis and deposition in BMs. Col IV, by binding to cell surface receptors, activates intracellular signaling events to promote cell survival, proliferation and tumorigenesis [5]. Loss of integrin α1β1 ameliorates Kras G12D -induced lung cancer [11,12]. β1 integrin and its downstream effecter focal adhesion kinase (FAK) are critical in mediating resistance to anoikis, chemotherapy-induced cell death and metastasis [6,8,11].
Despite Col IV is extensively studied, majority of the works focused on the functions of major Col IV, or unfortunately did not distinguish the roles of major and minor Col IV. It is largely unknown whether minor Col IV plays a role in cancer development. It also remains to be elucidated whether major and minor Col IV signal through the same cell surface receptors and intracellular signaling pathways and whether they can functionally compensate for each other.
In the present study, we demonstrate that minor Col IV α5 chain is indispensable in lung cancer development by using α5(IV)-deficient mouse model. α5(IV) supports lung cancer progression via cancer cell autonomous and non-autonomous mechanisms. α5(IV), but not α1 (IV), promotes lung cancer cell proliferation and tumor angiogenesis through non-integrin collagen receptor DDR1-mediated ERK activation. The functions of minor Col IV can not be compensated by abundant major Col IV.

α5(IV) chain is required for lung cancer progression
A LacZ gene trap cassette including En2 splice acceptor/ECMV IRES/LacZ/SV40 polyadenylation site was inserted into intron 35 of mouse Col4a5 gene on chromosome X to generate Col4a5 knockout mice (S1A and S1B Fig) [13]. RT-PCR analyses demonstrated the absence of Col4a5 mRNA in the KO tissues (S1C and S1D Fig). The LacZ reporter reflects endogenous Col4a5 expression. Strong LacZ staining was observed in lung bronchia (S1E Fig). Immunofluorescent staining demonstrated that α5(IV) chain is expressed in lung bronchia at high levels, and in lung alveolar epithelial cells at lower levels in Col4a5 +/Y (hereafter refereed as wild-type, WT) mice (S1F Fig). The α5(IV) signal is absent in Col4a5 LacZ/Y (hereafter refereed as knockout, KO) lungs (S1F Fig), further demonstrating that the mutant Col4a5 allele is indeed null.
BM proteins promote cancer cell proliferation and protect cancer cells from apoptosis. Tumors in Kras/α5 KO mice had significantly reduced tumor cell proliferation (Fig 1E and 1F), compared with those in Kras/α5 WT mice. Few apoptotic signal was evident in both groups ( Fig 1E). Hemorrhage was evident in α5 KO lungs, but not in WT lungs (Fig 1A). Hemorrhage lesions indicate improper organization of capillaries and blood vessels in α5 KO lungs. As tumor angiogenesis provides tumor cells nutrients and oxygen necessary for sustained tumor growth, this promoted us to examine whether neo-angiogenesis was compromised in Kras/α5 KO tumors. Indeed, tumors in Kras/α5 KO mice were significantly less vascularized ( Epithelial α5(IV) supports tumor cell growth and tumorigenicity α5(IV) is expressed in lung bronchia and alveolar epithelial cells (S1 Fig). To study the functions of epithelial α5(IV) in lung cancer development, endogenous α5(IV) was knocked down in A549 lung adenocarcinoma cells (Fig 2A). α5(IV) knockdown significantly reduced A549 cell proliferation, migration and anchorage-independent cell growth ( Fig 2B-2D), compared to cells expressing scramble control shRNA. This is not due to the off-target effect of α5(IV) shRNAs, as expression of mouse α5(IV) could rescue the phenotypes of α5(IV)-knockdown A549 cells (S2 Fig). α5(IV) knockdown in CRL-5810 lung cancer cells similarly resulted in impaired cell proliferation and anchorage-independent cell growth (S3 Fig). Therefore, the endogenous α5(IV)-constituted BMs are essential in supporting lung cancer cell proliferation. To determine whether in vitro phenotypes were reflected in vivo, tumorigenic ability of A549 cells was tested by injecting control or α5(IV)-knockdown cells subcutaneously into nude mice. α5 (IV) knockdown resulted in slower growing A549 xenograft tumors (Fig 2E). Less proliferating cells were detected in α5(IV)-knockdown xenograft tumors (Fig 2F and 2G).

α5(IV) is expressed in endothelial cells and regulates angiogenesis
Kras/α5 KO tumors were significantly less vascularized (Fig 1). However, knockdown of α5 (IV) in A549 cells only mildly affected neo-angiogenesis in the xenograft tumors, which was not statistically significant (Fig 2F and 2H). This suggests that less angiogenesis observed in Kras/α5 KO tumors may be mainly due to ablation of stromal α5(IV). To examine the roles of stromal α5(IV) in tumor progression, murine Lewis lung cancer (LLC) cells were implanted in Col4a5 WT or KO mice. Tumors grew significantly slower in KO than in WT mice ( Fig 3A). Less proliferating cells were detected in the tumors from KO mice, than in that from WT mice (Fig 3B and 3D). Unlike the Kras-driven lung tumors, which were slowly growing and rare apoptosis was evident (Fig 1E), the LLC transplant tumors grew much faster. Apoptosis was evident in the LLC transplant tumors, due to rapid tumor growth ( Fig 3C). More apoptotic cells were detected in the tumors from KO mice, than in that from WT mice (Fig 3C and 3D). These data collectively suggest stromal α5(IV) provides necessary survival and proliferation cues to support rapid LLC tumor growth. Tumors trigger profound angiogenesis to support vast nutrient and oxygen demand during rapid LLC transplant tumor growth in WT mice ( Fig 3B). Fewer blood vessels formed in the LLC transplant tumors in the KO mice, compared to that in the WT mice ( Fig 3B). The impaired tumor angiogenesis in the KO mice was not only reflected by decreased number of CD31-positive endothelial cells (Fig 3E), but also by dramatically decreased number of sinusoid microvessels ( Fig 3F) and average vessel diameter ( Fig 3G). To further test if stromal α5(IV) plays a role in regulating angiogenesis, VEGF containing Matrigel plugs were implanted subcutaneously in Col4a5 WT or KO mice. Abundant blood vessels, visualized by FITC-dextran, formed in the Matrigel plugs implanted in the WT mice, but not in the KO mice ( Fig 3H). CD31 staining on Matrigel plug sections further revealed~12-fold reduction of capillary numbers in the plugs in KO mice (Fig 3H and 3I). α5(IV) partially colocalized with endothelial cell marker CD31 in the lung ( Fig 4A). Knockdown of α5(IV) in human microvascular endothelial cell-1 (HMEC-1) cells ( Fig 4B) significantly reduced endothelial cell proliferation ( Fig 4C) and migration ( Fig 4D). Knockdown of α5(IV) in HMEC-1 cells also significantly impaired the tubule formation capability of endothelial cells ( Fig 4E). Thus, endothelial α5(IV) may be responsible for efficient tumor angiogenesis.

α1(IV) can not functionally compensate for α5(IV)
Major Col IV is known to provide survival and growth cues to cancer cells. α5(IV) may regulate tumor progression through modulating major Col IV expression and basement membrane assembly. Electron microscopy on the lungs from 6-month old KO mice did not reveal overt defect in the basement membranes underneath lung alveolar epithelial cells (

Loss of α5(IV) impaired ERK activation
FAK is one of the major effecters transducing signals from Col IV [14]. FAK further phosphorylates and activates downstream signaling molecules, including Src [14]. Knockdown of α5 (IV), however, did not affect phosphorylation levels of FAK and Src in A549 and CRL-5810 lung cancer cells (Fig 5A and S3A Fig). Instead, significantly lower phosphorylation levels of ERK and Akt, kinases essential in supporting cell survival, proliferation and transformation [15,16], were detected in α5(IV)-knockdown A549 and CRL-5810 cells (Fig 5A and S3A Fig). Ectopic expression of mouse α5(IV) in α5(IV)-knockdown A549 cells restored phosphorylation of ERK and Akt (S2E Fig). Interestingly, knockdown of α1(IV) resulted in impaired phosphorylation of Akt and Src, but not ERK or FAK in A549 cells (Fig 5A), reinforcing the notion that major and minor Col IV may regulate cancer cell behavior through overlapping, but not identical intracellular signaling pathways. Similar to that in lung cancer cells, knockdown of α5 (IV), but not α1(IV), significantly decreased ERK phosphorylation in HMEC-1 cells (S7A Fig). To study if impaired ERK activation is responsible for the defects in cell proliferation and
Knockdown of DDR1 in A549 cells resulted in decreased phosphorylation of ERK and Akt (Fig 7A), unaffected phosphorylation of FAK and Src (Fig 7A), as well as impaired cell proliferation (Fig 7B), migration ( Fig 7C) and anchorage-independent cell growth (Fig 7D) The similar phenotypes observed in the α5(IV)and DDR1-knockdown cells indicate that DDR1 may be the receptor transducing signals from α5(IV). DDR1 is a receptor tyrosine kinase that its phosphorylation is indicative of receptor activation and important in transducing downstream signals. Significantly less phosphorylation of DDR1 was detected in α5(IV)knockdown A549 cells, compared to that in cells expressing scramble shRNA (Fig 8A). DDR1 expression was reduced in α5(IV)-knockdown cells and less amount of DDR1 was immunoprecipitated ( Fig 8A). To more accurately examine DDR1 phosphorylation levels in α5(IV)knockdown cells, DDR1 was expressed back to endogenous levels. Less DDR1 phosphorylation was detected in α5(IV)-knockdown A549 cells expressing exogenous DDR1 than the control cells, despite similar amount of DDR1 was immunoprecipitated (Fig 8A). Overexpression of DDR1 was not able to restore phosphorylation levels of ERK and Akt in α5(IV)-knockdown A549 cells (S10 Fig). These data collectively suggest that α5(IV) not only affects DDR1 stability and expression, but also is required for DDR1 activation.

Discussion
Col IV, the major BM component, is essential in maintenance of tissue integrity and proper function. In addition to broadly expressed and extensively studied major Col IV α1α1α2, minor Col IV α3α4α5 and α5α5α6 are less abundantly expressed with restricted tissue distribution [4]. Physiological and pathological functions of minor Col IV, however, are less well understood. In this report, we present evidences that minor Col IV α5(IV) is essential in supporting lung cancer development via cancer cell autonomous and non-autonomous mechanisms. Minor but not major Col IV signals through non-integrin receptor DDR1.
Delayed tumor progression in α5(IV)-deficient mice suggests proper signal from α5(IV) is important in supporting cancer cell survival and proliferation. Col IV transduces signals through cell surface receptors. Cell surface integrin expression is unaffected in α5(IV)-knockdown cells. However, expression of DDR1, the non-integrin collagen receptor functioning independent of integrins [20][21][22], is decreased in α5(IV)-knockdown cells. DDR1 is highly phosphorylated in non-small cell lung cancer (NSCLC) [23], and DDR1 overexpression is associated with poor prognosis in NSCLC [24]. Inhibition of DDR1 reduces cell survival, homing and colonization in lung cancer metastasis [25]. Consistently, DDR1 expression is elevated in lung tumors with Kras activation, compared to normal lung tissues (compare Fig 6B and 6C). Ablation of α5(IV) results in decreased DDR1 expression in both normal lung tissues and Kras lung tumors. DDR1-knockdown cells phenocopied α5(IV)-knockdown cells. More importantly, expression of constitutively active DDR1 in α5(IV)-knockdown cells can rescue the proliferation and migration defects, suggesting DDR1 is functionally downstream of α5(IV). α5(IV) knockdown impaired DDR1 phosphorylation. Overexpression of exogenous wild-type DDR1 can not restore ERK phosphorylation in α5(IV)-knockdown cells. These data indicate that the function of DDR1 requires the presence of α5(IV) and DDR1 may directly mediate the functions of α5(IV).
Despite α5(IV) knockdown does not affect integrin cell surface expression, the possibility exists that integrins are functional receptors for α5(IV). Col IV was reported to bind integrins through sites within the triple-helical cyanogen bromide-derived fragments and noncollagenous domains [5]. Such studies were largely based on purified Col IV or Col IV fragments. It should be noted that proper collagen network assembly and geometry are critical in the biological functions of Col IV. Ablation of endogenous Col IV using gene knockout or silencing will provide more physiologically relevant insights into receptor binding, signaling and biological functions of Col IV. It remains to be elucidated whether integrins have selectivity and specificity towards major and minor Col IV under different physiological and pathological circumstances. DDR1 and integrins may have cooperative or opposing functions in response to collagens [26,27]. The crosstalk between DDR1 and integrins upon α5(IV) binding may provide the cells more robustness.
Ablation of α5(IV) does not affect major Col IV expression, or disrupt basement membrane assembly. The inability of abundant major α1α1α2(IV) to support efficient tumor growth and progression in α5(IV)-deficient mice indicates that major Col IV can not functionally compensate for the deficiency of minor Col IV. This is supported by the fact that mutations of Col IV α chains cause distinct heritable diseases. Mutations in COL4A1 cause encephaloclastic porencephaly, characterized by degenerative cavities and cerebral lesions in the brain [28]. Deletion of Col4a1/Col4a2 locus in mice results in growth retardation and embryonic lethality [29]. However, mutations in COL4A5 (Alport's syndrome) or auto-antibody recognizing α3(IV) (Goodpasture's syndrome) result in progressive renal failure [4,5]. Mice deficient of α3(IV) [30,31] or α5(IV) [32] are viable, but develop renal phenotypes reminiscent of that in Alport's syndrome. Knockdown of major Col IV α1(IV) does not affect DDR1 expression. The overlapping, but not identical spectrum of altered signaling events in α5(IV)and α1(IV)-knockdown cells suggests that major and minor Col IV may exert their biological functions via different cell surface receptors and intracellular signaling pathways.
Major and minor Col IV share same domain structure and high sequence similarity. It is yet unclear how highly similar major and minor Col IV recognize different cell surface receptor and activate different intracellular signaling pathways. α3α4α5(IV) is highly cross-linked due to its larger degree intra-and inter-chain disulfide bonds, relative to α1α1α2(IV) [33]. As a result, α3α4α5(IV) has different biochemical properties from α1α1α2(IV) that α3α4α5(IV) is more resistant to proteolytic degradation [33]. Different biomechanical force from major and minor Col IV may be responsible for the receptor specificity. It should be noted that Col IV protomers further form α1α1α2(IV)-α1α1α2(IV), α3α4α5(IV)-α3α4α5(IV) and α1α1α2(IV)-α5α5α6(IV) networks [4,5]. These networks may differentially recognize cell surface receptors and activate intracellular signaling pathways, thus provide signal specificity and redundancy.
α5(IV) regulates cancer progression via cancer cell autonomous and non-autonomous mechanisms. The DDR1-ERK signaling cascade is required for the functions of both cancer cells and endothelial cells. Stromal components, including blood vessels, constitute proper microenvironment to support tumor progression. It is reported that stable microvasculature sustains cancer cells at dormancy, whereas sprouting neovasculature rescues cancer cells from cell cycle arrest and promotes cancer cell proliferation [34]. Col IV assembly is critical for vascular BM integrity and structural organization. Small-molecule inhibitors that interfere Col IV biosynthesis were shown to prevent angiogenesis and tumor growth [35]. α5(IV) is expressed in the endothelium. Deficiency of α5(IV) delayed in vitro and in vivo angiogenesis. It warrants further study if the cancer cells in α5(IV) KO mice remain dormant due to impaired neoangiogenesis.
In summary, we provides evidences in this study that α5(IV) deficiency significantly delays tumor progression. α5(IV) signals through non-integrin collagen receptor DDR1 in lung cancer cells and endothelial cells. α5(IV) promotes tumor growth via both cancer cell autonomous and non-autonomous mechanisms. Abundant major Col IV is not able to compensate for α5 (IV) deficiency.

Ethics statement
All mice were housed in specific pathogen-free environment at the Shanghai Institute of Biochemistry and Cell Biology and treated in strict accordance with protocols approved by the Institutional Animal Care and Use Committee of Shanghai Institute of Biochemistry and Cell Biology (Approval number: SIBCB-NAF-15-003-S325-006).

In vitro and in vivo angiogenesis assay
In vitro angiogenesis assay was performed as described [42] by seeding HMEC-1 cells in the rat tail type I collagen sandwich gel in the presence of VEGF. Cells were photographed after 24 hours. In vivo Matrigel plug assay was performed as described [43] by subcutaneously injecting growth factor reduced Matrigel containing 50 ng recombinant human vascular endothelial growth factor into 8-week-old WT or Col4a5 LacZ/Y mice in C57/Bl background. On day 14, Dextran-FITC was injected through the tail vein 30 min before the mice were sacrificed. Matrigel plugs were fixed and sectioned for CD31 staining. Histological vascular parameters, including microvascular density (MVD), sinusoid microvessel number, and vascular diameter, were measured [44].

Mouse treatment
All mice were housed in specific pathogen-free environment at the Shanghai Institute of Biochemistry and Cell Biology and treated in strict accordance with protocols approved by the Institutional Animal Care and Use Committee. Col4a5 LacZ/Y mice were generated and maintained in C57/Bl background by the European Conditional Mouse Mutagenesis Program [13]. Kras G12D mice were back crossed to C57/Bl background 3 generations before cross with Col4a5 LacZ/Y mice. LLC cells were transplanted at the armpit of lower limb of 8-week old WT or Col4a5 LacZ/Y mice in C57/Bl background. To minimize the possible effects of mouse genetic background on tumor behavior, wild-type littermates were used as control for Col4a5 LacZ/Y mice in all experiments. A549 cells were subcutaneously injected into Balb/c nude mice.

Transmission electron microscopy (TEM)
Lung tissues isolated from 6-month old Col4a5 +/Y and Col4a5 LacZ/Y mice were fixed in 3% glutaraldehyde in 0.1M PBS (pH7.4) for 4 hours at room temperature and then in 1% osmium tetroxide overnight at 4°C. The fixed lung tissue were dehydrated through an alcohol series and embedded in Epon812 Resin at 60°C for 48 hours. Ultrathin sections (70 nm) were collected on copper grids. The grids were stained in 2% uranyl acetate for 40 minutes and in 0.5% lead citrate for 8 minutes orderly. The samples were examined under FEI Tecnai G2 Spirit TEM.

Statistical analysis
Data were analyzed using the two-sided Student t test, and considered statistically significant when the P value was less than 0.05.