Type IV collagen α6 chain is a regulator of keratin 10 in keratinization of oral mucosal epithelium

Keratinized mucosa is of fundamental importance to maintain healthy gingival tissue, and understanding the mechanisms of oral mucosa keratinization is crucial to successfully manage healthy gingiva. Previous studies have shown a strong involvement of the basement membrane in the proliferation and differentiation of epithelial cells. Therefore, first, to identify the keratinized mucosa-specific basement membrane components, immunohistochemical analysis for the six alpha chains of type IV collagen was performed in 8-week-old mice. No difference in the expression pattern of type IV collagen α1(IV) and α2(IV) chains was observed in the keratinized and non-keratinized mucosa. Interestingly, however, type IV collagen α5(IV) and α6(IV) chains specifically were strongly detected in the keratinized mucosa. To analyze the functional roles of the type IV collagen isoform α6(IV) in oral mucosa keratinization, we analyzed Col4a6-knockout mice. Epithelial developmental delay and low levels of KRT10 were observed in new-born Col4a6-knockout mice. Additionally, in vitro experiments with loss-of function analysis using human gingival epithelial cells confirmed the important role of α6(IV) chain in epithelial keratinization. These findings indicate that α112:α556 (IV) network, which is the only network that includes the α6(IV) chain, is one regulator of KRT10 expression in keratinization of oral mucosal epithelium.

essential for the development of novel techniques and materials for keratinized gingival tissue reconstruction and regeneration. However, the mechanisms regulating keratinization of gingiva still remain unclear.
Basement membrane separates the epithelial and mesenchymal tissues and it plays important roles in the determination of cell polarity, proliferation and differentiation [10][11][12][13] . In this study, we hypothesized that the basement membrane is a critical regulator of keratinization of the oral mucosal epithelium. The basement membrane, consisting of the lamina lucida and lamina densa, is composed of four major components (i.e., type IV collagen, laminin, nidogen, and perlecan), and separates the epithelium, mesothelium and endothelium from the connective tissue 10 . Type IV collagen is identified primarily in the skin within the basement membrane, and comprises six collagen chains (i.e., α1 to α6). These helical polypeptide α-chains form triple-helical protomers [α1α1α2 (α112), α3α4α5 (α345) and α5α5α6 (α556)], which further assemble into three major networks (α112:α112, α112:α556, α345:α345), interconnected by NC1 domain 14 . Type IV collagen α1 and α2 chains are expressed ubiquitously in basement membranes, although type IV collagen α3-α6 chains have a tissue specific distribution. Indeed, type IV collagen α1 to α5 chains, except α6 chain, are expressed in the kidney, whereas type IV collagen α1, α2, α5 and α6 chains, excluding α3 and α4 chains, are expressed in the skin 15 . It is well known that mutations in type IV collagen α3 to α5 chains causes Alport's syndrome associated with glomerulonephritis, sensorineural deafness and eye abnormalities.
From these reports, we hypothesized that the difference in keratinization of palatal and buccal mucosa could be associated with the composition of the basement membrane. We herein focused especially on type IV collagen, and found that α5(IV) and α6(IV) chains were highly expressed in keratinized oral mucosa. Loss-of-function analysis using Col4a6 knock out mice (Col4a6-KO mice) and gene expression knockdown with siRNA indicated that α112:α556 (IV) network, which is the only network that includes the α6(IV) chain, plays an important role in the keratinization of oral mucosa epithelial cells.

Results
Histological and morphological differences between palatal and buccal mucosa. First, to demonstrate that palatal and buccal mucosa in mice are distinct types of mucosa, histological and morphological analysis were performed. Histological findings showed the presence of parakeratosis in the external layer of both palatal and buccal mucosa (Fig. 1A). Keratin 1 (KRT1) and keratin 10 (KRT10), which are the major markers for keratinized mucosa, were highly expressed in the palatal mucosa, as shown by real time RT-PCR analysis (Fig. 1D). Further immunohistochemical (IHC) analysis confirmed the presence of KRT10 only in the palatal mucosa (Fig. 1B). On the other hand, KRT14 was detected in both keratinized and non-keratinized mucosa (Fig. 1C). Transmission electron microscopy (TEM) was also used to observe the details of palatal and buccal mucosa morphology. TEM images showed that palatal mucosa presented a stratum corneum and a granular layer, where keratohyalin granules were observed. On the other hand, in the buccal mucosa, these two layers were absent (Fig. 1E).
Analysis of type IV collagen expression in basement membrane of keratinized and non-keratinized mucosa. Previous reports have demonstrated that type IV collagen α1 and α2 chains express ubiquitously in basement membranes, although α3-α6 chains have a tissue specific distribution 16 . To characterize the differences in the composition of the basement membrane between keratinized and non-keratinized mucosa, we performed IHC analysis for the six α-chains of type IV collagen, and found no differences in α1(IV) and α2(IV) expression levels in both keratinized and non-keratinized mucosa ( Fig. 2A-D). The α3(IV) and α4(IV) chains could not be detected in either mucosa ( Fig. 2E-H), but could be detected in the kidney (Supplemental Fig. 1). Interestingly, we found that the protein levels of α5(IV) and α6(IV) chains in keratinized mucosa were markedly high, compared to non-keratinized mucosa in 8-week-old mice ( Fig. 2I-L). Based on these results and on the fact that type IV collagen consists of the three distinct triple-helical networks (i.e., α112:α112, α112:α556, α345:α345), we hypothesized that the α112:α556 network of type IV collagen could play an important role in keratinization of oral mucosa. Since type IV collagen α6 chain is included only in the α112:α556 network, we performed a deeper analysis on the role of type IV collagen α6 chain in epithelial keratinization.
Expression pattern of type IV collagen α6 chain and keratin 10 during embryonic development. Secondary palatal development begins at embryonic day 12.5 (E12.5), and the palatal shelves, which have been vertically oriented on either side of the tongue, are elevated to the horizontal position at E14.5. Fusion between the two palatal shelves occurs by E15.5 17 . Therefore, in order to understand the chronological order of α6(IV) and KRT10 expression during palatal development, HE staining and IHC staining were performed using E12.5 to E18.5 mice including the developmental stages before, during and after palate formation and palatal mucosa keratinization ( Fig. 3A-D). As shown in Fig. 3E-H, the protein expression levels of α6(IV) could be detected in palatal mucosa onward E14.5. Interestingly, however, KRT10 could only be observed at E18.5 ( Fig. 3I-P). Therefore, these results suggested that the α556(IV) protomer contained in α112-α556 network could be an important factor involved in keratin formation in keratinized mucosa.
Functional analysis of type IV collagen using Col4a6-knockout mice. Next, we used type IV collagen knockout (Col4a6-KO) mice to investigate the role of type IV collagen α6 chain on keratinization of oral mucosa. Immunohistological analysis showed a low level of KRT10 (Fig. 4A,B) and KRT1 (Supplemental Fig. 2A) in new-born Col4a6-KO mice. Of note, there was no notable difference in the expression levels of perlecan between Col4a6-KO and WT mice (Supplemental Fig. 3). Since aging has been related to a decrease in type IV collagen expression 18 , we analyzed the keratinized mucosa of 28-week-old aged mice. As shown in Fig. 4C,D, the expression levels of KRT10 were decreased in Col4a6-KO mice compared with WT mice. It is well known that KRT5 and KRT14 are expressed in the basal cells of oral mucosa and involved in the formation of strong networks that help attach keratinocytes together. As these cells enter the terminal differentiation program, KRT5 and KRT14 are substituted by KRT1 and KRT10. Therefore, we also checked the expression level of KRT14, and found that it was decreased in new-born Col4a6-KO mice compared with WT mice (Fig. 4E), but the expression was identical in aged (28-week-old) WT and Col4a6-KO mice (Fig. 4F). Additional IHC analysis for α5(IV) in Col4a6-KO mice showed the loss of α5(IV) in keratinized mucosa (Supplemental Fig. 2B,C). From these results, we concluded that α556(IV) protomer is absent, and that α112:α556 network is not synthesized in keratinized mucosa of Col4a6-KO mice.

Functional analysis of type IV collagen α6 in vitro.
A previous study reported that only epithelial cells produce Col4a6 in colon 19 ; thus, to obtain a deeper insight into the effect of type IV collagen α6 chain on epithelial keratinization, we cultured hGECs using 3D culture methods to induce epithelial keratinization, and analyzed the gene expression pattern of COL4A6 and keratinization markers by real time RT-PCR. Gene expression levels of COL4A6 increased by 2.4 times after 1 day and 4.7 times after 7 days (Fig. 5A). Strikingly, the gene expression levels of KRT1, KRT10 and involucrin (INV), one of keratinization-related genes, were markedly low in the initial days of culture, but increased dramatically at day 7 ( Fig. 5B-D). Accordingly, the gene expression levels of KRT5, KRT13, KRT14, KRT15 and KRT16, which have been identified to be expressed in oral mucosa, increased markedly after 7 days of culture (Supplemental Fig. 4). Of note, when hGESs were cultured in normal culture dish (2D culture model), KRT1 mRNA expression could not be detected and KRT10 expression level was low and did not increase even after 28 days in 2D culture (Supplemental Fig. 5); indicating that the 3D culture system is required for proper mucosa keratinization.
Next, we investigated the effect of type IV collagen α6 chain knockdown by inhibition of COL4A6 expression with siRNA specific for COL4A6 on epithelial keratinization using the 3D culture model. Confirmation experiments showed that the expression of COL4A6 mRNA (Fig. 6A), but not that of COL4A1, COL4A2 or COL4A5 (Supplemental Fig. 6), was significantly decreased by COL4A6 siRNA in hGECs. As shown in (Fig. 6B-D,I), mRNA levels of KRT1, KRT10, INV and KRT16 (marker of keratinocyte hyper-proliferation) decreased upon inhibition of COL4A6 expression. The decrease in KRT10 protein levels was also confirmed by western blot (Fig. 6J). On the other hand, blockade of COL4A6 in hGECs induced an increase in gene expression levels of KRT13, a marker of non-keratinized epithelia, but induced no significant changes in the gene expression levels of KRT5, KRT14 and KRT15, which are expressed in basal cells in oral mucosa (Fig. 6E-H). Collectively, these results demonstrate that α556(IV) protomer contained in α112-α556 network is one of regulator of oral mucosa keratinization.

Discussion
In the oral cavity, keratinized and non-keratinized mucosa are consecutive tissues, despite the differences in their morphology and function. Keratinization of oral mucosa is important for the maintenance of healthy periodontal tissue and long-term prognosis of natural tooth and dental implants. However, the molecular mechanism behind keratinization of oral mucosal epithelium is still unclear. Mutations in type IV collagen have been reported to cause Alport's syndrome associated with glomerulonephritis, sensorineural deafness 20,21 . Therefore, we hypothesized that the basement membrane components could be one important factor regulating keratinization in palatal mucosa. The basement membrane, which composition is tissue-specific and undergoes dynamic changes during development and aging, plays important biological function in the development and maintenance of tissues and organs 22 . In this study, we demonstrated that the expression of α6(IV) chain preceded the expression of KRT10 during palatal development (Fig. 3), as well as in in vitro 3D culture of hGECs (Fig. 6). These results indicate a novel role of α6(IV) as one of the regulators of keratinization of oral mucosal epithelium. Since, type IV collagen α1(IV) and α2(IV) chains were expressed equally in keratinized and non-keratinized mucosa, and α3(IV) and α4(IV) chains were not detected in either of the tissues, the results highlight the importance of type IV collagen α6 chain, especially the α122:α556 network, in mucosa keratinization.
The importance of other type IV collagen chains have already been demonstrated in embryogenesis and diseases. Deletion of both Col4a1/Col4a2 locus in mice is known to cause growth retardation and embryonic lethality 12 . Mice deficient of α3(IV) or α5(IV) are viable, but develop renal failures resembling those of Alport's syndrome [23][24][25] . However, ablation of α6(IV) does not disrupt basement membrane assembly. Additionally, no notable phenotype in the kidney was observed. Therefore, this study is the first to demonstrate a phenotype of Col4a6 deletion in the keratinization of oral mucosa in mice.
However, the cellular origin of type IV collagen, from epithelial cells or mesenchymal cells, is still not well understood. It has been reported that the mRNA of COL4A2 is observed in both epithelial and mesenchymal cells in colon, but the mRNA of COL4A6 is localized in the epithelial cells 19 . In the present study, we could not show the localization of Col4a6 mRNA by using in situ hybridization. However, we could observe the increased Col4a6 mRNA level during epithelial keratinization of hGECs in the 3D culture model, indicating, at least, that epithelial cells can synthesize COL4A6 in oral mucosa. From these data, we assume that epithelial cells are able to secrete α6(IV) chain, which in turn could activate intracellular pathways that direct cell differentiation in an autocrine manner. Additionally, the epithelial-mesenchymal interaction could play fundamental roles in determining the differential expression of α6(IV) chain in keratinized mucosa. In fact, we had performed a preliminary study using hGECs, which were co-cultured with mesenchymal cells isolated either from keratinized or non-keratinized mucosa. We observed that keratinization occurred only in hGECs co-cultured with mesenchymal cells from It is conceivable from our IHC data that, collagen IV α112:α112 and α112:α556 networks exist in the basement membrane of both keratinized and non-keratinized gingiva, and a larger amount of α112:α556 network exists in keratinized gingiva. Interestingly, in the Col4a6-KO mice, not only α6(IV) chain, but also α5(IV) chain were absent in basement membrane of gingiva (Supplemental Fig. 2). It has been reported that the mutation of α5 chain in X-linked Alport's syndrome caused the loss of α6 chain, and failure of assembly of α556 protomer and α112:α556 network (Zheng, A. J. Pathology, 1999). Accordingly, there could also have an absence of α112:α556 network due to loss of α6 chain in the keratinized gingiva of Col4a6-KO mice.
A previous study reported that as many as 10-12% of basal layer cells are stem cells in skin and colon 26 , and these stem cells play an important role in the development and maintenance of epithelium by regulating cell proliferation and differentiation 2 . Basement membrane, especially type IV collagen which serves as a ligand for integrin α1β1 and α2β1, is supposed to contribute to the niche of tissue-specific stem cells. It has been reported that integrin β1 positive keratinocytes have high properties of self-renewal, compared with integrin β1 negative cells 27 . Additionally, integrin β1 deletion in skin epithelial cells caused not only a disorder in skin wound healing 28 but also severe defects in basement membrane assembly and organization in mice 29 . In oral mucosa, oral keratinocyte stem cells are located in basement membrane of epithelium 2 , and have high affinity to type IV collagen 30 . Based on the fact that the development of epithelium is delayed in Col4a6-KO new-born mice, and that cell proliferation marker (cyclin D1) was also decreased in Col4a6 silenced hGECs (data not shown), it was estimated that type IV collagen, especially α6(IV) chain, may provide a niche environment for keratinocyte stem cells, and regulate keratinocyte proliferation and differentiation in oral mucosal epithelium. However, the biological mechanism of α6(IV) chain on stem cells has not been fully understood. Further researches, such as immunoassay and cell receptor assay using the recombinant protein α556(IV), are necessary and expected. Other members of the basement membrane, such as laminin, nidogen and perlecan have also been reported to play important roles in determining epithelial cell fate [31][32][33] . Nevertheless, IHC analysis showed no difference in perlecan levels in keratinized gingiva and non-keratinized gingiva between new-born Col4a6-KO and WT mice. Due to the large number of molecules in the basement membrane, and possible compensation effects, further investigations are necessary to allow a deeper insight on the role of other components of the basement membrane in the keratinization of oral mucosal epithelium.
In summary, our study demonstrated that α5(IV) and α6(IV) chains were highly expressed in keratinized mucosa. Keratinization of oral mucosal epithelium of Col4a6-KO mice was decreased in new-born mice and aged mice. Additionally, down-regulation of COL4A6 in hGECs caused suppression of epithelial keratinization. Collectively, these data indicate that type IV collagen α6 chain, and consequently α112:α556 network, is involved in KRT10 expression in keratinization of oral mucosal epithelium.

Materials and Methods
Animal experiments. Pregnant and 8-weeks-old c57BL/6 J mice were purchased from CLEA Japan (Tokyo, Japan). Col4a6-KO mice were generated by replacing parts of exon 2 and intron 2 with a Neomycin cassette in Col4a6 gene as described before 34 . Col4a6-KO mice were then backcrossed with C57BL/6 J (Charles River) for ten generations. The animal experiment protocols used in this study (OKU-2016495, OKU-2017051) were approved by Okayama University Animal Research Committee. All animals were handled according to the guidelines of Okayama University Animal Research Committee.
Histological analysis. For preparation of frozen sections from non-fixed and un-decalcified hard tissues, Kawamoto's film methods were used 35,36 . Briefly, samples were freeze-embedded with super cryoembedding medium (SECTION-LAB Co. Ltd., Hiroshima, Japan) and cut in thickness of 5 μm with tungsten carbide blade after mounting the adhesive film onto the sample surface. Samples were then immediately fixed with 4% paraformaldehyde (PFA) for 20 min and stained with hematoxylin and eosin. IHC analysis for type IV collagen was performed according to the previous reports 37, 38 . Briefly, the specimens were fixed with acetone for 20 minutes, followed by incubation with primary antibodies at 4 °C overnight, after blocking with 5% normal goat serum (Life [B66]) were used. These antibodies were kind gift from Dr. Sado (Okayama Univ., Japan). All antibodies for type IV collagen were diluted to 1:100 except for H22, which was diluted to 1:50. H11-, H22-, H31-, and b14-stained sections were treated with 6 M urea in 0.1 M glycine/HCl (pH 3.5) for 10 minutes and B66 for 1 min before blocking. Primary antibodies for KRT10 (ab76318), KRT1 (ab185629) and KRT14 (ab181595) were purchased from Abcam (Cambridge, UK). After washing, the specimens were incubated with secondary antibody Alexa Fluor 488 donkey anti-rabbit IgG (Life technologies, Tokyo, Japan), Alexa Fluor 488 donkey anti-rat IgG (Life technologies) or Alexa Fluor 647 anti-rabbit for 60 minutes at room temperature. All images were taken by fluorescence microscope. Quantification of KRT10 positive area in epithelium was based on fluorescent IHC images taken with BZ-700 microscope and analyzed with a BZ analyzer (Keyence, Osaka, Japan).
Real time RT-PCR analysis. The collected palatal and buccal mucosa were homogenized using PowerMasher II (Nippi, Tokyo., Japan) and BioMasher II (Nippi, Tokyo., Japan). After homogenization, total RNA was extracted using TRIzol ® Reagent (Life technologies) according to a conventional method and purified using a PureLink ® RNA Mini Kit (Life technologies). The total RNA from cultured cells was extracted and purified using PureLink ® RNA Mini Kit.
RNA samples were reverse-transcribed by using iScript cDNA synthesis kit (Bio-Rad, Hercules, CA, USA). Real time RT-PCR was performed to quantify the expression of the target gene by using KAPA SYBR FAST qPCR  Table 1.
Morphological analysis. The palatal and buccal mucosa were harvested from eight-week-old female c57BL/6 J mice. The collected samples were fixed with 2% glutaraldehyde and 2% PFA overnight. After post-fixation with 1% osmium tetroxide and dehydration with ethanol, the samples were embedded in spurr low-viscosity embedding media (Polysciences, Warrington, PA, US). Ultrathin sections were then cut using a diamond knife and microtome (LEICA EM UC7, Leica Mikrosysteme GmbH, Vienna, Austria). The sections were double-stained with uranyl acetate and lead citrate and observed using a transmission electron microscope (TEM: H-7650, HITACHI, Tokyo, Japan).
Cells and culture medium. Human gingival epithelial cells (hGECs) were purchased from CELLnTEC advanced cell systems AG (Stauffacherstrasse, Bern, Switzerland). hGECs were cultured in CnT-prime epithelial culture medium (CELLnTEC advanced cell systems AG, Stauffacherstrasse) at 37 °C in 5% CO 2 . To induce differentiation of hGECs, the cells were seeded onto cell culture inserts in a 12-well plate (ThinCert TM , Greiner Japan, Tokyo, Japan) using CnT-Prime 3D barrier medium (3D-medium, CELLnTEC advanced cell systems AG, Stauffacherstrasse). Briefly, the cells were seeded at 5.0 × 10 5 cells in the upper chamber with 0.5 mL of 3D-medium. Additional 5 mL of 3D-medium was added to the lower chamber. For the down-regulation of the Col4a6 gene, 5 nM of siRNA targeting the Col4a6 gene (StelthTM SiCol4a6; Life Technologies) was transfected into hGECs using Lipofectamine RNAiMAX (Life Technologies), according to the manufacturer's instructions. StelthTM RNAi Negative Control High GC Duplex (Life Technologies) was used as the negative control. Transfected hGECs were seeded at 2.0 × 10 6 cells in the upper chamber with 0.5 mL of 3D-medium, and 5 mL of 3D-medium was added to the lower chamber. After 24 h, 3D-media of both chambers were aspirated, and 4 mL of 3D-medium was added only to the lower chamber, and the cells were cultured for additional 7 days.  Statistical analysis. The results obtained from quantitative experiments were reported as the mean values ± SD. Statistical analyses were performed with one-way factorial ANOVA followed by Tukey's multiple comparison tests or Student's unpaired t-tests.