Col V siRNA Engineered Tenocytes for Tendon Tissue Engineering

The presence of uniformly small collagen fibrils in tendon repair is believed to play a major role in suboptimal tendon healing. Collagen V is significantly elevated in healing tendons and plays an important role in fibrillogenesis. The objective of this study was to investigate the effect of a particular chain of collagen V on the fibrillogenesis of Sprague-Dawley rat tenocytes, as well as the efficacy of Col V siRNA engineered tenocytes for tendon tissue engineering. RNA interference gene therapy and a scaffold free tissue engineered tendon model were employed. The results showed that scaffold free tissue engineered tendon had tissue-specific tendon structure. Down regulation of collagen V α1 or α2 chains by siRNAs (Col5α1 siRNA, Col5α2 siRNA) had different effects on collagen I and decorin gene expressions. Col5α1 siRNA treated tenocytes had smaller collagen fibrils with abnormal morphology; while those Col5α2 siRNA treated tenocytes had the same morphology as normal tenocytes. Furthermore, it was found that tendons formed by coculture of Col5α1 siRNA treated tenocytes with normal tenocytes at a proper ratio had larger collagen fibrils and relative normal contour. Conclusively, it was demonstrated that Col V siRNA engineered tenocytes improved tendon tissue regeneration. And an optimal level of collagen V is vital in regulating collagen fibrillogenesis. This may provide a basis for future development of novel cellular- and molecular biology-based therapeutics for tendon diseases.


Introduction
Tendons and ligaments are frequently targets of injury from trauma in sports and aging [1,2]. It is well known that tendons and ligaments do not heal through a regenerative process; Instead, their healing occurs via the formation of a fibrotic scar [3][4][5]. In the healing tendon, a uniform distribution of small diameter collagen fibrils has been found with poorer mechanical properties than native tissue and shows no improvement of mechanical properties with time [6]. Previous studies have shown that the diameter of collagen fibrils in soft tissues has a positive correlation with collagen mechanical strength [7,8]. So, the generation of larger diameter fibrils is believed to promote biomechanical properties of healing tendons.
Fibril-forming collagens in most connective tissues include collagen types I, III, and V. Collagen V is a quantitatively minor component of the tissues such as dermis, tendon/ligament and bone. Unlike cornea, where collagen V represents 20 to 25% of the total collagen and such high percentage contributes to their transparence [9], in most above tissues the relative concentration of collagen V is significantly lower, only 1 to 3%. However, collagen V has a key role in the regulation of initial fibril assembly. Type V and Type I collagen are co-assembled into heterotypic fibrils. The entire triple-helical domain of the type V collagen molecules is buried within the fibril and type I collagen molecules are present along the fibril surface [10,11], and the retained NH2terminal domains of the type V collagen are exposed at the surface, extending outward through the gap zones. At the early stage of assembly, immature fibril segments is regulated by the NH2-terminal domain of type V collagen [11]. These NH2-terminal domains alter accretion of collagen molecules onto fibrils and therefore lateral growth. A critical density of type V collagen would favor the initiation of new fibrils rather than the continued growth of existing fibrils. Type V collagen molecules may contain a1 (V), a2(V) and a3(V) chains [12,13]. There are several collagen V isoforms that differ in chain composition, such as a1(V) 2 a2(V) heterotrimer, a1(V)a2(V)a3(V) heterotrimer, a1(V) 3 homotrimer [14][15][16]. However, collagen fibrils in tendon and ligament are heterotypic type I/V fibrils and a1(V) 2 a2(V) is the favored and functional heterotrimer [16,17]. Ultrastructural immunolocalization demonstrated that the homotrimer a1(V) 3 was localized at the surface of wide collagen I fibrils as thin filamentous structures and did not regulate fibril assembly, whereas the heterotrimer a1(V) 2 a2(V) was buried in the fibril interior and regulated collagen fibrillogenesis [18]. Studies have indicated that the occurrence of a uniform distribution of small diameter collagen fibrils [6], along with a persistently elevated level of collagen V, for up to 52 weeks after injury [19]. Several investigations have shown that type V collagen plays a role in regulating the diameter of the type I collagen fibril during fibrillogenesis [10,11,20,21]. The collagen fibril diameter was inversely proportional to type V/type I collagen ratios [23], i.e. the higher concentration of type V, the smaller the fibril diameter [19]. Although the involvement of collagen V in tendon matrix organization is well established, the functional significance of the particular chains of collagen V in tendon tissue engineering is remained unknown.
At present, the models for investigating tendon biology include cell and animal models [21,22]. Cell models are simple, however, are not convenient for the subsequent investigation of collagen fibril synthesis [22,23]. Animal models are powerful for detecting collagen fibril synthesis, but contain more influencing factors than in vitro study and ethical difficulties also present. Previous studies by our group successfully utilized mesenchymal stem cell sheets assembled with frozen tendon graft for tendon tissue engineering [24]. In this study, a scaffold-free tissue engineered tendon was developed with tenocyte sheets. This provides a simple and controllable 3-D tissue engineering model, which mimics tissue in vivo and makes the tendon biological study in vitro feasible.
The objective of this study was to investigate the effect of particular chain of collagen V on the fibrillogenesis of tenocytes, as well as the efficacy of Col V siRNA engineered tenocytes for tendon tissue engineering. RNA interference technology is an efficient tool for silencing endogenous or exogenous genes by complementary binding to the target mRNA [25]. Since there are two collagen  V chains, therefore, we use siRNA oligonucleotides to decrease the two chains of type V collagen synthesis separately in tissue engineered tendon model as a logical approach to understanding the role of type V collagen during the tendon healing process.

Gene expression of injured and normal tendon tissue
To verify the expression of different chains of collagen V in the injured tendon, real time PCR was used to detect their mRNA expression 1 week after surgery and compare with that of normal tendon tissue. The results showed that injured tendon had significant higher expression levels of Col1a1, Col3a1, Col5a1 and Col5a2 than those in normal tendon tissue ( Fig. 1, P ,0.05 or P,0.01), whereas those of Col5a1 and Col5a2 were much higher, so we choose these two for further study.
Morphology and biochemical properties of the in vitro tissue engineered tendon Rat Achilles tenocytes proliferated rapidly and formed coherent cellular sheets within 2 weeks after attaining confluence in the presence of ascorbic acid. After the sheets formed, they could be detached from the culture substratum, so a tissue engineered tendon with living cells and collagen matrix which had a relatively compact structure was obtained ( Fig.2A). Hematoxylin-eosin and Masson trichrome staining of tissue engineered tendon revealed a tissue-specific tendon structure: organized bundles of highly crimped fibrils and cells oriented in parallel (Fig. 2C, D). And the real time PCR results showed that the cultured cell sheet had high level of collagen V expression (Fig. 2B).
Transfection of siRNA against collagen V efficiently suppressed collagen V expression in rat Achilles tenocytes We transfected two silencer pre-designed siRNA sequences against different chains of collagen V into rat achilles tenocytes by Lipofectamine 2000. As demonstrated by fluorescent microscopy, uptake of the siRNAs by rat Achilles tendon was highly efficient. The green fluorescent signal was observed as early as 1 h after transfection and reached a maximum at 6 h and could still be seen 48 h after the application of oligonucleotides (Fig. 3A).
Forty-eight hours after transfection, real time PCR analysis revealed siRNA196227 (Col5a1 siRNA) and siRNA s136862 (Col5a2 siRNA) led to remarkable suppression of Col5a1 and Col5a2 expression respectively compared to the negative control group (NC: scrambled siRNA). There was at least ,70% knockdown of collagen V mRNA ( Fig. 3B, P,0.05). The positive control (PC: GAPDH siRNA) also had evident suppression of GAPDH (data not shown). The positive control was used to verify the methods of siRNA transfection. Taken together, we chose siRNA 196227 and siRNA s136862 in the subsequent experiments.
As shown in the collagen V immunofluorescence staining results ( fig. 3C, D), tendon cells had decreased fluorescence in response to particular siRNA interference, which indicates down-regulation of Col5a1 and Col5a2 at the protein level. The control group (cells treated with transfection agent only) showed high level of fluorescence indicating constitutively high levels of type V procollagen a1 and a2 expression, while Col5a1 siRNA decreased the expression level of Col5a1 by ,36.6%, and Col5a2 siRNA down-regulated the level of Col5a2 by ,33.7%. Such findings were consistent with the mRNA expression results. These data collectively indicate that collagen V was efficiently down-regulated at the mRNA and protein levels by siRNA interference methods.

Effect of different chain of collagen V on cell proliferation and apoptosis
The MTT assay for tendon cell viability revealed that both of the siRNA treatments caused little difference between the groups at day 1 after transfection. The reduction in OD 570 was statistically significant for cells treated with Col5a1 siRNA and Col5a2 siRNA ( Fig. 4A, P,0.05) at day 3. The mean 6 SD OD 570 was 0.22960.0208 and 0.22160.018 for cultures treated with Col5a1 siRNA and Col5a2 siRNA, respectively. While the OD 570 value in control cells was 0.27160.0216. These results clearly indicated that down regulation of both chains of collagen V had a modest negative influence on the viability of tendon cells.
In DAPI staining, the labeled DNA of viable cells was scattered throughout the nuclei, and bright condensation of chromatin revealed apoptotic cells. A few more apoptotic cells were found in the Col5 siRNA group (Fig. 4B). These results indicate that the lower growth rate of cells treated with COL5 siRNA may have been caused by the increased apoptosis when the collagen V was down-regulated (26).

Different chains of collagen V have different effects on fibrillogenesis and collagen fibril diameter
The amount of collagen deposited in the tissue engineered tendon was shown in Fig. 5A. The mean 6 SD of collagen deposition was (35.167.3) % and (38.966.8) % of the value in control cells for cultures treated with Col5a1 siRNA and Col5a2 siRNA, respectively (P,0.05). There was less of a difference between the two experimental groups. The results indicated that down-regulated collagen V impaired collagen expression in Achilles tenocytes.
Transmission electron microscopy (TEM) results showed that the in vitro tissue engineered tendon had organized collagen fibrils. However, tissue engineered tendon treated with Col5a1 siRNA had abnormal fibril ultrastructure, while the fibrils from tendon treated with Col5a2 siRNA was similar to that of the control (data not shown). Comparison of mean fibril diameters in different treatment are shown in Fig. 5B. The relative mean diameters of a1 siRNA and a2 siRNA treatments were about (80.6610.06)% and (87.3610.11)% of the control.

Different chains of collagen V have different effects on matrix gene expression.
To clarify the molecular events after blocking collagen V expression, we also assessed the expression of some important collagen and decorin genes by real time PCR and compared their proportions to collagen type I. The expression of Col1a1 was down regulated when either siRNA was added, especially reduced by Col5a2 siRNA to ,16.5% (Fig. 5C, P,0.01). Quantitive PCR analysis also showed both collagen type III and decorin mRNA levels were decreased after dealing with Col5a1 siRNA or Col5a2 siRNA ( Fig. 5D and F). These finding indicate that the block of Col V may affected the synthesis of these extracellular matrix.
In the Col5a1 siRNA treated group, the ratios of Col5a1/ Col1a1, Col3a1/ Col1a1 and decorin/Col1a1 were sharply decreased by 70.5%, 20.6% and 48.7%, respectively. In the Col5a2 Transmission electron microscopy analysis of tissue engineered tendons formed by coculture of Col5a1 siRNA treated tenocytes and normal tenocytes in different ratios.
Since tissue engineered tendons treated with Col5a1siRNA had abnormal collagen fibrils, a possible explanation is that the corresponding chain of collagen type V is over-suppressed. So we cocultured Col5a1 siRNA treated tenocytes and normal tenocytes with ratios of 1:0, 1:0.5, 1:1, 1: 2, 1:3 and 0:1 (NC group). As shown in Fig.6 (A-F) are their transmission electron microscopic analysis and histogram of collagen fiber diameters of tissue engineered tendons. Compared with collagen fibrils of the control group, collagen fibrils of cocultured groups with a ratio of 1:0.5 or 1:1 had similar morphology and relative larger diameters with the mean diameter (118.867.52)% and (109.865.78)% (Fig.6G) of the control, respectively. While collagen fibrils of cocultured groups with a ratio of 1:2 or 1:3 were (9466.56)% and (84.965.83)% (Fig.6G) of the control group. Those results indicate that regulating the quantity of collagen type V a1 to a proper level is beneficial for the generation of larger diameters.

Discussion
The present study for the first time demonstrated the use of a scaffold-free tissue engineered tendon model for investigating the biological function of collagen V in tendon fibrillogenesis. We found that the type V procollagen a1 and a2 chains had different effects on regulating tendon matrix gene expression to some extent and an optimal level of collagen type Va1 is vital in regulating collagen fibrillogenesis. Col V siRNA engineered tenocytes improved tendon tissue engineering. These findings present a good example of in vitro tissue engineering model for tendon biology investigation and may provide basis for future development of cell or gene therapy for tendon repair.
The basic function of tendon is to transmit force from muscle to bone, which makes limb and joint movement possible. Therefore tendons must be capable of resisting high tensile forces with limited elongation. Parry and his colleagues reported that the mechanical properties of tendons are related to the fibril diameter distribution, large fibrils could withstand higher tensile forces [8]. Moreover, investigations have also showed that a greater fibril diameter generates greater tensile strength and the relative smaller fibril diameter provides greater elastic properties, which is of consistence to the region-specific function of patellar tendon. So for the tendon to function properly it seems that there is need for both small and large diameter fibrils [27,28]. However, during the process of repair, collagen fiber diameter is notably smaller compared to the normal health one and the larger fiber in the health tendon occupy the most volume, so now for tendon repair, regeneration of larger diameter is one of the major challenge. Previous studies of tendon biology used either in vitro cell cultures or animal models [21,22], but both of them have several inherent disadvantages. In our study, a relative well-organized collagen fibrils resembling tendon ultrastructure is formed. Meanwhile, the tissue engineered tendon is scaffold-free, so there are fewer exogenous factors. It can be easily and efficiently used to evaluate the role of gene intervention (e.g. RNAi and antisense techniques) in tendon research. In addition, our tissue engineered tendon is an excellent setup to simulate injured tendon. It has a uniform distribution of small diameter collagen fibrils along with a high level of type V collagen expression, which is characteristic of injured tendon [6,19]. Although the tissue engineered tendon lacks a fair bit in the mechanical perspective and has less efficient fibrillogenesis, it still illustrates a novel model and strategy for future tendon biology research and studies.

Effects of different chains of collagen V on tendon fibrillogenesis
This study assessed the function of two chains of collagen V. The amount of collagen deposited in the tissue engineered tendon indicated that down-regulated collagen V impaired collagen formation in Achilles tenocytes. The down-regulation of Col I expression and consequently possible affected Col I turnover may be one of the reasons that contributed to the collagen deposition decrease. Moreover, it seems that the influence of Col5a1 on fibril formation was more apparent than that of Col5a2. Tissue engineered tendon treated with Col5a1 siRNA had abnormal collagen fibril morphology while tendon treated with Col5a2 siRNA had collagen fibrils similar to controls. Published data indicate that cells deficient in Col5a1 would be null for type V collagen since all three isoforms contain this chain, while cell deficient in Col5a2 would be expected to assemble the a1(V) 3 homotrimer. However, unlike the heterotrimer a1(V) 2 a2(V), the homotrimer a1(V) 3 was shown to not assemble into heterotypic type I/V fibrils and hence did not participate in collagen fibrils formation [18]. So, the difference of collagen fibrillogenesis between Col5a1and Col5a2 siRNA treatments may be caused by the over-depression of Col5a1. Real time PCR results indicated that Col5a1 siRNA and Col5a2 siRNA had different effects on the expression of Col1a1 and decorin at the mRNA level. Col5a1 siRNA decreased the ratios of Col5a1, Col3a1 and decorin to Col1a1, whereas Col5a2 siRNA increased the ratios of Col5a1, Col3a1 and decorin to Col1a1. Collagen type I is the major quantitative component in collagen fibrils, whereas Col5a1 and Col3a1 are two minor quantitative contents. The ratios between these two genes and Col1a1 are reversely related to the fibril diameter [29][30][31][32][33]. Thus, it is a paradox since all the ratio values were decreased while the fibrils diameter is also smaller in the Col5a1 siRNA group. Over silencing of Col5a1 may be the answer for such contradiction. The sharp reduction of collagen V and decorin by Col5a1 siRNA may could not assemble enough a1(V) 2 a2(V) functional isoform and proteoglycan to form V/I heterotypic fibrils and collagen-proteoglycan heteropolymers to fulfill the minimal requirement of collagen I fibril formation [23]. In the Col5a2 siRNA-treated group, those ratios are consisted with the diameters. Decorin belongs to small leucine-rich proteoglycans/ proteins (SLRPs) and the formation of collagen-proteoglycan heteropolymers is important in regulation of fibrillogenesis [34][35][36][37]. Whether the ratios of decorin to collagen type I may also be inversely proportional to collagen fibril diameter is needed to be determined.
Collagen formation processes are driven by the loss of solvent molecules in cell-surface crypts [38] and result in assemblies with a circular cross-section. Type V and type I collagen co-assemble into heterotypic fibrils and regulate the initial fibril assembly. This assembly is completely processed outside the cell [39]. Overdepression of collagen V in tissue engineered tendon using Col5a1 siRNA maybe caused small and abnormal collagen fibrils, so we cocultured Col5a1 siRNA treated tenocytes and normal tenocytes with different ratios. The coculture experimental results revealed that collagen fibrils of cocultured groups with a ratio of 1:0.5 or 1:1 had similar morphology and larger diameters, while collagen fibrils of cocultured groups with a ratio of 1:0, 1:2 or 1:3 were smaller. These results indicated that there was an optimal level of collagen type V and the level was vital in regulating collagen fibrillogenesis. The undetermined precise level of collagen type V is the limitation of the present study and warrants further investigation to obtain. Although the mechanisms involved need further investigation, and the effects of both chains blocked should be studied, these findings provided useful information for the development of molecular biology-and cell-based therapeutics for improving injured tendon repair in future.

Effect of different chains of collagen V on cell proliferation and apoptosis
Our results in siRNA treated tendon cells suggest collagen V involvement in biological processes besides fibrillogenesis. A few more Col5a2 siRNA-treated cells were DAPI positive and exhibited morphological signs of apoptosis, as judged by fluorescence microscopy, while Col5a1 siRNA had a modest negative influence on cell growth and apoptosis. The apoptosis of Col5a2 siRNAtreated cells may be a consequence of the loss of important cellmatrix interactions. This finding is in line with the established role of the extracellular matrix in sustaining cell survival. Changes in matrix composition may affect the adhesion signals of cells [26] and the storage and activation of growth factors [40]. Previous studies using mice harboring a targeted deletion of the Col5a2 gene have shown increased apoptosis of stromal fibroblasts in skin [26].

RNAi for tendon matrix gene regulation
This study showed that siRNA interference successfully inhibited procollagen a1 and a2 (V) synthesis in rat Achilles tendon fibroblasts. RNAi is the process of sequence specific, post-transcriptional gene silencing directed by short interfering 21-23 nt double-stranded RNA (siRNA) [41]. siRNAs act as guides to activate the RNA-induced silencing complex, which cleaves homologous mRNA molecules [25]. Many studies have demonstrated that the introduction of siRNAs into mammalian and human cells causes specific and effective suppression of the corresponding mRNA molecules [25,[41][42][43]. Several lines of evidence that siRNAs inhibit the in vivo expression of endogenous genes provides further support for the notion that specific siRNAs may be a new alternative to gene-specific therapeutics of human diseases [43]. For the therapeutic application of siRNA technology, it is critical to employ an efficient gene delivery system for transduction of siRNA into target cells.
In conclusion, analysis of the different chains of collagen type V shed light on the role of this minor collagen type in fibrillogenesis and tendon regeneration. Our experiments document the critical contribution of the a1 (V) and a2 (V) chains to fibrillogenesis, fibril organization, and cell viability. In addition, this work successfully used scaffold-free tendon tissue engineered model to mimic tissue in vivo for investigating the biological process of collagen fibrillogenesis. This provides a simple and controllable model for future tendon biology studies. Meanwhile, the optimal level of collagen V in injured tendon repair is vital and its details need further investigation.

Cell culture
Primary tenocytes were isolated from the Achilles tendon of Sprague-Dawley rats. The Protocols were approved by the animal care and use committee of Zhejiang University. The ethical grant number is zju2007102001. In brief, Achilles tendons were harvested and adjacent tissues were stripped. After soaking in phosphate-buffered saline (PBS) containing penicillin (100 units/ml) and streptomycin (100 mg/ml) (Gibco, Carlsbad, CA, USA) for 10 min, the tendon was cut into small fragments followed by enzyme digestion with a 0.2% (W/V) collagenase mixture (Gibco, Carlsbad, CA, USA) at 37uC in a humidified atmosphere of 5% CO 2 and 95% air. After 4 h, an equal volume of Dulbecco's modified Eagle medium (DMEM) containing 10% (V/V) fetal bovine serum (FBS; Gibco, Carlsbad, CA, USA) was added and the mixture digested overnight. Next day, the tissue residue was removed and the resulting cell suspension was centrifuged at 1,200 rpm. The harvested cells were maintained in DMEM containing 10% (V/V) FBS with medium changed every 2-3 days. When cultured primary cells reached 80% confluence, they were detached by treatment with 0.25% (W/V) trypsin and 0.1% ethylenediaminetetraacetic acid (Gibco, Carlsbad, CA, USA) and subcultured at a density of 1 6 10 4 /cm 2 . Cultured cells before passage 5 were used for experiments.

siRNA synthesis and transfection
Two target sequences ( Table 1)

Quantitative reverse transcription-polymerase chain reaction (qRT-PCR)
In this study, the gene expressions of the a1 type I collagen gene (Col1a1), the a1 type III collagen gene (Col3a1) , Col5a1 and Col5a2 of injured tendon tissue, normal tendon tissue and tissue engineered tendon were detected. The injured tendon tissue was removed from the Achilles tendon of Sprague-Dawley rats, which had suffered a gap wound model (1 mm in width and 4 mm in length) and was repaired with no treatment after one week [44]. The normal tendon tissue was removed from the Achilles tendon of Sprague-Dawley rats.
In the siRNAs treated experiment, tenocytes were cultured for 48 h after transfection. Real-time PCR was used to detect the mRNAs encoding Col5a1, Col5a2, Col1a1, Col3a1, decorin and GAPDH (n = 3). Total cellular RNA was isolated by lysis in Trizol (Invitrogen, Carlsbad, CA, USA) followed by one-step phenol chloroform-isopropyl alcohol extraction as described in the protocol. For real-time PCR, 2 mg of total RNA from each sample was reversed transcribed using mMLV reverse transcriptase (Invitrogen, Carlsbad, CA, USA). 1/1000 of the reverse transcription reaction was used for subsequent qRT-PCRs, which were set up using TaKaRa Universal PCR Master Mix on an ABI 7900 HT Real-Time PCR System (Applied Biosystems). Target genes were normalized against the housekeeping gene b-actin. Specific primers are shown in Table 2.

Immunofluorescence staining and quantitative analysis
Immunofluorescence staining was used to detect procollagen type V a1 and a2 chains expression in protein level after treated with specific siRNAs. Tenocytes were grown on the bottom of plastic dishes containing growth medium, and were treated with specific siRNAs. After 48 h, cells were fixed in 4% paraformaldehyde in PBS (pH 7.4) for 15 minutes and then immersed in blocking solution containing 1% bovine serum albumin and 0.3% Triton X-100 for 1 h at room temperature. After washing 3 times in PBS, cells were incubated overnight at 4uC with rabbit and goat polyclonal antibodies against Col5a1 and Col5a2 respectively (1:100, Santa Cruz, USA), followed by incubation with fluorescence isothiocyanate-conjugated (green,1:200, Beyotime, China) and cyanine 3-conjugated (red, 1:30, Sigma, USA) secondary antibody for 1 h. Cells treated with transfection agent only were used as control group. Finally, the incubated cells were rinsed in PBS and observed under a microscope (Olympus, Tokyo, Japan) (n = 3). After immunofluorescence experiment, three pictures of separate view fields were chosen randomly for each sample, and then Image-Pro Plus 5.1.2 was used to quantify the expression level of each procollagen by calculating the related color density.

MTT assay
Cell survival was determined by MTT assay 1 day and 3 days after treatment of Col5a1 and Col5a2 siRNAs (n = 3). This measure was based on the cleavage of MTT into a blue-colored product (formazan) by the mitochondrial enzyme succinate dehydrogenase [45][46][47]. MTT (50 mg/ml) was added and incubated at 37uC for 4 h. Then, the MTT solution was discarded and 150 mL of dimethyl sulfoxide was added to dissolve the formazan crystals. The plates were then fully shaken to dissolve the formazan crystals formed. The optical density was measured at

Cell morphological assessment and DAPI staining
To detect morphological evidence of apoptosis, cell nuclei were visualized after DNA staining with the fluorescent dye DAPI. Briefly, cells were seeded on coverslips in 24-well tissue culture plates and treated with specific siRNA oligonucleotides. After 3 days, the morphology of cells was monitored under an inverted light microscope. Cells were fixed with 4% paraformaldehyde for 20 min, washed with PBS, and then incubated with DAPI (1 mg/ mL) for 10 min. After washing with PBS, cells were observed using fluorescent microscopy (Olympus, Tokyo, Japan) at a peak excitation wavelength of 340 nm (n = 3).

Fabrication of the tissue engineered tendon
To conveniently detect the collagen fibril synthesis in in vitro study, a scaffold-free tissue engineered tendon was developed with tendon cell sheets. Tenocytes were cultured in high glucose DMEM with 10% (v/v) FBS. To prevent cell senescence and promote extracellular matrix synthesis 50 mg/mL of ascorbic acid was added to the medium. Cells proliferated rapidly and formed coherent cellular sheets within 2 weeks. The cell sheets could be detached from the substratum by applying a small roll-up force [24]. The cell sheets were called tissue engineered tendon and used for gene expression andcollagen content detection , as well as transmission electron microscopy analysis in subsequent experiments.

Histological examination and Masson trichrome staining
To validate whether the tissue engineered tendon mimics tissue in vivo, the morphology property of tissue engineered tendon was detected. The engineered tendon without siRNA treatment was fixed in 10% neutral buffered formalin, dehydrated through an alcohol gradient, cleared, and embedded in paraffin. Histological sections (10 mm) were prepared using a microtome and subsequently stained with hematoxylin-eosin and Masson trichrome according to standard procedures to examine the general appearance of the collagen fibrils [48,49] (n = 3).

Determination of collagen content of cell sheet
For collagen synthesis studies, cells were transfected with specific siRNA oligonucleotide every four days until a cell sheet was formed. The cells were washed with PBS three times. The amount of deposited collagen in the cell sheet was quantified using a collagen assay kit and following the manufacturer's protocol, which was a colorimetric-based hydroxyproline assay used to estimate total collagen content and concentration(n = 3, Jiancheng Ltd., Nanjing, China) [48]. The content was measured using the Multiskan Spectrum ELIASA at 550 nm.

Transmission electron microscopic analysis
To verify that extracellular matrix deposited in the tissue engineered tendon, transmission electron microscopy (TEM) was used to assess the diameter of the collagen fibrils and fibril alignment. Briefly, samples were prefixed with 2.5% glutaraldehyde for 2 h at 4uC and washed twice with PBS at 4uC followed by postfixation with 1% osmic acid for 2 h at 4uC. After two washes in PBS, the samples were dehydrated in an ethanol gradient and dried to a critical point. Then the samples were mounted and sputter-coated with gold for viewing in a TEM (Quanta 10 FEI) [50]. About 500 collagen fibrils were measured in each sample to obtain a true representation of the fibril diameter distribution (n = 3).

Ratio coculture experiments
To find an optimal level of collagen type V in regulating collagen fibrillogenesis, we duplicated the experiment of coculturing Col5a1 siRNA treated tenocytes with normal tenocytes in different ratios. The tenocytes were treated with Col5a1 siRNA three times every 4 days and then detached using 0.25% trypsin and 0.1% ethylenediaminetetraacetic acid (EDTA). Finally, those treated tenocytes were cocultured with untreated tenocytes in a ratio of 1:0, 1:0.5, 1:1, 1: 2 or 1:3. Also, we cultured a control group of normal tenocytes. After two weeks, cell sheets were formed and the TEM analysis was detected (n = 3).

Statistical analysis
The results are presented as mean 6 standard deviation of the 3 samples from each group. The significance of differences was analyzed by Q-test of one-way ANOVA and p ,0.05 was considered as statistically significant.