Oral Fibroblast Expression of wound-inducible transcript 3.0 (wit3.0) Accelerates the Collagen Gel Contraction in Vitro*

Wounds of the oral mucosa show faster closure with less scar formation than skin wounds in other areas. A differentially expressed cDNA, wound-inducible transcript 3.0 (wit3.0), was isolated from oral mucosal wound in rats (Sukotjo, C., Abanmy, A. A., Ogawa, T., and Nishimura, I. (2002) J. Dent. Res. 81, 229–235). The purpose of this study was to characterize the wit3.0 gene structure and the function of its deduced peptide. Human and rat genome databases revealed that the gene for wit3.0 was located in human chromosome 12p11.23 and rat chromosome 4q44. Its human and rat gene structures were well conserved, composed of 7 exons spread over 20 kb. Exon 5 was alternatively spliced generating two transcripts encoding deduced peptides of 215 and 253 amino acids (wit3.0α and wit3.0β, respectively). The protein families data base of alignments (Pfam) analysis suggested the wit3.0 peptide sequence shared similarity with a portion of the myosin II coiled-coil domain consensus sequence. Fibroblasts isolated from the rat oral wound up-regulated wit3.0 expression and exhibited greater ability to contract collagen gel in vitro than fibroblasts isolated from untreated oral mucosa/gingiva. NIH3T3 and rat oral fibroblasts transfected with expression vector containing the coding sequences of wit3.0α or wit3.0β increased in vitro collagen gel contraction. When treated with TGFβ-1, NIH3T3 fibroblast expression of wit3.0 showed no significant change, whereas alpha smooth muscle actin was increased in a dose-dependent manner. These data suggest that there may be a novel wound healing pathway involving wit3.0 underlying the favorable early wound closure characteristics of oral mucosa.

Tooth extraction is one of the most frequently prescribed oral surgeries (1). The surgical removal of a tooth creates a large open wound, which heals generally without complication (2). Characteristically, the oral mucosa open wound closes rapidly primarily due to the spontaneous approximation of gingival wound margins, often without the aid of surgical suturing (see Fig. 1). Oral mucosa and skin wounds undergo the similar healing sequence, including hemostatis, inflammation, granu-lation tissue formation, and remodeling of the connective tissue matrix. However, clinical observations and experimental animal studies consistently indicate that the extent of granulation and scar formation in oral mucosa is generally small, and wound healing in the oral mucosa demonstrate the better outcome as compared with the equivalent wound in skin (3)(4)(5). Oral mucosa and gingiva are composed of a thin keratinocyte layer with an underlying highly vascularized connective tissue (6). Although a number of investigations had attempted to explain the healing differences between oral mucosa and skin (7)(8)(9), the distinctive mechanism involved in oral mucosal wound healing remains to be elucidated. Isolated oral fibroblasts have been shown to contract the collagen gel in vitro at a faster rate than skin fibroblasts (5, 10 -12). These data suggest that oral fibroblasts possess distinctive characteristics promoting accelerated wound closure. We recently isolated the overlapping full-length cDNAs encoding a differentially expressed gene, wound-inducible transcript 3.0 (wit3.0) from a cDNA library of rat gingiva tissue undergoing wound healing after tooth extraction. We also found, appearing to generate from the wit3.0 gene, two alternative transcripts were presented; these encode 215 (wit3.0␣)-and 253 (wit3.0␤)-amino acid long peptides. In situ hybridization has revealed that the fibroblasts localized in the oral wound were the cellular source of wit3.0 (13). A basic local alignment search tool (BLAST) 1 search matched wit3.0 with human expressed sequence tag FLJ10672; however, the identification and potential function of wit3.0 are still unknown. This study aimed to characterize the wit3.0 gene structure and to identify the potential function of its deduced peptide. The results suggest that there may be a novel wound healing pathway involving wit3.0 facilitating fibroblast-derived oral wound contraction that may contribute to the favorable early wound-closure characteristics of oral mucosa.

EXPERIMENTAL PROCEDURES
Rat Tooth Extraction Wound Healing Model-An experimental wound in oral mucosa was created in rats by unilateral maxillary molar extraction (14,15). Forty-day-old Sprague-Dawley rats were anesthetized with 2% isoflurane inhalation. The gingival tissue firmly attached to the maxillary molars was carefully dissociated by a dental explorer, and the maxillary first, second, and third molars were extracted from the left jaw. The wound healing tissue and the contralateral untreated * This work was supported by the UCLA Academic Senate Faculty Grant Program (to I. N.) and a Procter & Gamble ACP Fellowship (to C. S.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
RT-PCR Studies on Steady-state mRNA Levels of the Extracellularmatrix molecules, Wound-related Molecules, and wit3.0 during Wound Healing-The oral mucosa wound specimen, as well as untreated contralateral gingival tissue, were harvested at postsurgery days 4 and 7. Each tissue specimen was homogenized separately, and total RNA was extracted by the guanidium isothiocyanade method (TRIzol, Invitrogen, Grand Island, NY). After the DNase treatment, 1 g of total RNA sample was used to synthesize cDNA using random hexamer primer (Clontech, Palo Alto, CA). Steady-state mRNA levels of the following extracellular matrix (ECM) molecules and wound-related peptides were examined using RT-PCR: type I collagen alpha 1 chain (col1a1), type III collagen alpha 1 chain (col3a1), type XII collagen alpha 1 chain isoforms A, B, 1, and 2 (col12a1A, col12a1B, col12a1-1, and col12a1-2, respectively), TGF-␤1, ␣-smooth muscle actin (␣-SM actin) as well as wit3.0␣ and wit3.0␤. Primer sequences and PCR conditions achieving the exponential amplification for each target mRNA molecule are summarized in Table I. During the preliminary experiments, PCR conditions were determined to represent the exponential amplification cycle for each target molecule. Throughout the experiment glyceraldehyde-3-phosphate dehydrogenase was used as the normalization control.
In Vitro Gel Contraction Assay of Fibroblasts from Wounded and Untreated Rat Oral Mucosa/Gingiva-Fibroblasts from healing tooth extraction wounds and untreated gingiva were harvested at postsurgery day 7. The tissue was minced into ϳ2 mm 3 under phosphatebuffered saline, pH 7.4 (PBS), then rinsed twice with PBS. Minced pieces of the tissue were planted onto the surface of 10-mm cell culture dishes. After 1 h of incubation at 37°C, 10 ml of growth medium: Dulbecco's modified Eagle's medium (DMEM) and 10% fetal bovine serum (FBS), with 0.25 g/ml amphotericin B, 100 units/ml penicillin, and 100 g/ml streptomycin antibiotics, was added. The fibroblasts were obtained by trypsinization of the primary outgrowth of cells. Isolated primary cells were cultured under 37°C, 5% CO 2 , and 80% humidity condition. Growth medium was continually replenished every 4 days. The cells were routinely passaged using 0.05% trypsin in PBS containing 0.53 mM EDTA. Cells used for the experiments were collected from between the 4th and 10th passages. The steady-state level of wit3.0 mRNA was examined by RT-PCR as described above.
Collagen gels were cast in 6-well plates from type I collagen/DMEM solution composed of 5 parts of bovine skin collagen type I (Vitrogen 100, Cohesion Corp., Palo Alto, CA), 2 parts of 5ϫ DMEM containing Hepes and gentamicin, 1 part of NaOH (0.142 M), 5 parts of FBS, and 1.5 parts of PBS (16). The gels were in liquid form at 4°C and solidified at 37°C. Primary fibroblasts were seeded into the collagen gel (1.2 ϫ 10 5 cells/well) and incubated at 37°C, humidity of 80%, and CO 2 level of 5%. Fibroblasts-gel complex contraction was monitored by standardized photography every 10 h for 120 h.
The area of fibroblast-gel complex in digitized photographs was measured using Image Pro Plus software (Media Cybernetics, Silver Spring, MD). The ratio of collagen gel area against the culture well area was calculated at each measurement point. Multiple data sets of different groups and different time points were analyzed by repeated measures analysis of variants (ANOVA), or two-way ANOVA at a 5% level.
In Silico Search for Human, Rat, and Mouse Genome Databases and the Protein Families Data Base of Alignments-The nucleotide se-quence of rat wit3.0 was used to search chromosomal assignment and gene structure in silico using the public human, rat, and mouse genome databases. The genomic sequences were further compared with the cDNA sequences, and the exon structure was determined. The deduced peptide sequences of wit3.0␣ and wit3.0␤ were submitted to the online protein families data base of alignment (Pfam) search (16,17), for a search of the available consensus functional domains. The wit3.0 peptide sequence and the candidate sequences were compared by the protein BLAST search.
Transfected NIH3T3 fibroblastic cells were centrifuged at 500 ϫ g for 2-3 min; nuclear and cytoplasmic extracts were separately collected using an NE-PER Nuclear and Cytoplasmic Extraction Reagents kit (Pierce, Rockford, IL) using the manufacturer's protocol. The extracts in Laemmli buffer containing ␤-mercaptoethanol were subjected to 4 -20% SDS-gel electrophoresis. The transferred Western blot was examined with M2 anti-FLAG monoclonal antibody using HSP89 monoclonal antibody as a positive cytoplasmic protein control.
For fluorescent cytology experiments, NIH3T3 fibroblastic cells were cultured on a Permanox Chamber Slide (Lab Tek, Nalgen Nunc Int., Rochester, NY). Transfection of NIH3T3 fibroblastic cells using recombinant expression vector containing coding sequences of wit3.0␣, wit3.0␤, and bacterial alkaline phosphatase (BAP: positive control) was performed as previously described. After 24 -48 h, cells were washed with Trisbuffered saline/CA (50 mM Tris, 150 mM NaCl, pH 7.4, containing 1 mM calcium chloride) and fixed with freshly prepared 1:1 mix of acetone/ methanol for 1 min. Ten micrograms/ml of M5 anti-FLAG monoclonal antibody was used as a primary antibody to recognize the FLAG epitope. After 1 h of primary antibody incubation, the chambers were washed five times with Tris-buffered saline/CA. Texas Red, fluorescein-conjugated antibody (Molecular Probes, Eugene, OR) was used as a secondary antibody. Sytox green (Molecular Probes) was used to stain the cell's nucleus, and a confocal laser scanning microscope was used to determine the terminal localization of the wit3.0-translated peptides.
In Vitro Gel Contraction Assay for NIH 3T3 Fibroblastic Cells and Rat Gingival Fibroblasts Transfected with wit3.0 Expression Vectors-To test the hypothesis that wit3.0 participates in facilitating fibroblast-derived wound contraction, NIH3T3 fibroblastic cells transfected with the expression vectors containing wit3.0␣-FLAG, wit3.0␤-FLAG, and bacterial alkaline phosphatase-FLAG fusion peptides were subjected to the in vitro collagen gel contraction assay. The areas of collagen gel of the following four groups were compared: NIH3T3 plus wit3.0␣-FLAG, NIH3T3 plus wit3.0␤-FLAG, NIH3T3 plus BAP-FLAG (transfection control), and NIH3T3 cells alone (no transfection control).
Antisense oligonucleotide treatment was performed to validate the effect of wit3.0 on gel contraction. Antisense oligonucleotide: 5Ј-CT-GAATGGTGCAGCTCAT-3Ј and Sense oligonucleotide: 5Ј-ATGAGCT- GCACCATTCAG-3Ј were dissolved in DMEM culture medium into 100 M stock solution and sterilized by filtration through 0.2 M cellulose acetate filter. Serum free medium containing 2 M oligonucleotide was mixed with LipofectAMINE and incubated for 20 min at room temperature. The mixture was later added to the NIH3T3 fibroblastic cells that had previously received one of each expression vectors containing wit3.0␣-FLAG, wit3.0␤-FLAG, or BAP-FLAG fusion peptide. Cells were incubated for 4 h without serum prior to addition of FBS. Subsequently, the cells were divided into two groups. The first group was further incubated for 24 h at 37°C and subjected for Western blot assay using M2 anti-FLAG monoclonal antibody. The second group was mixed with bovine type I collagen gel solution (Vitrogen 100, Cohesion Corp, Palo Alto, CA) as described above to a concentration of 1 ϫ 10 5 cells/ml/well. Non-transfected cells served as an untreated control in both groups. A consistent volume of collagen gel was used throughout the experiment. Solidified gel was released from the well after 2 h of incubation. Standardized photography was used to monitor the longitudinal gel contraction.
In a separate experiment, rat gingival fibroblasts were transfected with each of expression vectors containing wit3.0␣-FLAG or wit3.0␤-FLAG, or both simultaneously. The expression levels of wit3.0 mRNAs were confirmed by RT-PCR. The transfected and untreated rat gingival fibroblasts as well as rat oral wound fibroblasts were subjected to in vitro collagen contraction assay as described above.
Transforming Growth Factor-␤1 Treatment of NIH3T3 Fibroblastic Cells-NIH3T3 fibroblastic cells were cultured in a 6-well plate at a cell density of 1 ϫ 10 4 cells/cm 2 . The cells were starved for 12 h in DMEM supplemented with 0.5% FBS and 1ϫ Amphotericin B and penicillin (3 ml/well). After 12 h, 0.1 or 5.0 ng/ml of TGF-␤1 was added to the culture medium. No TGF-␤1 addition was used as a control. The cells were cultured at 37°C, humidity of 80%, and 5% CO 2 level for an additional 18 h until total RNA from the cells was extracted using the TRIzol method. ␣-SM actin and wit3.0 expressions were examined by RT-PCR. The housekeeping gene, ␤-actin, was used to standardize the gene expression level. Primer sequences and PCR conditions representing experiments performed in triplicate are listed in Table I.

RESULTS
Oral Mucosa Wound Healing Profile-Rapid wound healing as shown in the rat oral wound model was reinforced in our studies. Following tooth extraction, by day 4 of wound healing, granulation tissue, through which thin epithelial cell layers migrate (Fig. 1C, arrowheads), was apparent. By day 7, the wound closure progresses by approximation of the wound margins (Fig. 1D). During early wound healing stage, TGF␤-1 and ␣-SM actin mRNAs increased moderately, whereas synthesis of both wit3.0␣ and wit3.0␤ mRNAs was robust. Expression of type I collagen mRNA during the first week remained at the baseline level, whereas synthesis of type III, type XIIB-1, and XIIB-2 collagens increased (Fig. 1E).
Fibroblasts isolated from day 7 wound healing tissue maintained the high level of wit3.0 expression even after 10 passages. Both wound fibroblasts and untreated gingival fibroblast controls exhibited more than 95% contraction during the first 15-20 h of incubation in the in vitro collagen gel contraction assay (Fig. 1F). The wound fibroblasts-gel complex showed significantly greater intensity of contraction than untreated control fibroblasts-gel complex (p Ͻ 0.05). Although the contraction rate of both leveled off after 20 h, fibroblasts from the wounded tissue exhibited greater overall gel contraction (42% Ϯ 7.3 of its original area) compared with that of untreated gingival fibroblasts (76% Ϯ 9.6) as shown in Fig. 1F.
In Silico Evaluation of the wit3.0 Gene and Peptide Structures-Matches in the online genome databases located the nucleotide sequence of wit3.0 within human chromosome 12p11.23 and rat chromosome 4q44 (Fig. 2A). The genomic DNA structures of human and rat wit3.0 were highly conserved and encoded by seven exons spread over 20 kbp (Fig. 2B). The deduced peptide sequence was encoded by a part of exon 2, exons 3-6, and a part of exon 7. Exon 1 encoded the 5Ј-untranslated region; exon 7, the large 3Ј-untranslated region. Exon 5 encoded the in-frame insertion sequence found in wit3.0␤ but not in wit3.0␣ (Fig. 2, B and C). Rat wit3.0-deduced peptide sequence was searched in BLAST for homologous sequences. Results showed matches with myosin heavy chain among Oryctolagus cuniculus (44% positive amino acid match, including 22% identical match), Drosophila melanogaster (42% positive amino acid match, including 21% identical match), and Argopecten irradians (43% positive amino acid match, including 24% identical match). The 228 amino acids that match with skeletal muscle myosin heavy chain belong to a conserved domain: pfam01576.8 (35% positive amino acid match, including 15% identical match). With a consensus sequence of 860 amino acids, pfam 01576.8 myosin tail family consists of the coiled-coil myosin heavy chain tail region (Fig. 3). In addition, wit3.0 matched with pfam COG4372 (499 amino acids long), which is an uncharacterized protein conserved in bacteria with the myosin-like domain (data not shown).
The Effect of wit3.0 on Collagen Gel Contraction-Gel contraction rate in NIH3T3 fibroblastic cells overexpressing wit3.0␣ or wit3.0␤ was significantly higher than the control groups (p Ͻ 0.05) as shown in Table II. The BAP transfection control and non-transfected control groups showed no differences in gel area size at each time point. The antisense oligonucleotide designed to block wit3.0␤ mRNA partially inhibited the wit3.0␤-FLAG peptide synthesis. The accelerated collagen gel contraction in the wit3.0␤ transfection group was partially but significantly blocked by the antisense treatment (p Ͻ 0.05) (Fig. 5A). Treatment with the sense oligonucleotide also decreased the wit3.0␤-FLAG peptide synthesis and proportionately decreased the accelerated collagen gel contraction. The effect of wit3.0 on the collagen contraction lasted up to 15-20 h. Similarly, the rat gingival fibroblasts overexpressing wit3.0␣ or wit3.0␤ peptides exhibited the increased collagen gel contraction as compared with the untransfected gingival fibroblast control (p Ͻ 0.05). The gingival fibroblasts received both wit3.0␣ and wit3.0␤ expression vectors demonstrated more increased collagen gel contraction as compared with the untransfected gingival fibroblast control (p Ͻ 0.01); however, the contraction rates of these transfected gingival fibroblast groups did not reach the collagen gel contraction rate of wound fibroblasts (Fig. 5B).
The Effect of TGF-␤1 on wit3.0 Expression-The expression of ␣-SM actin mRNA in NIH3T3 fibroblastic cells increased 1.6-fold with 0.1 ng/ml and 1.8-fold with 5.0 ng/ml of TGF-␤1, compared with the untreated control as shown in Fig. 5C. NIH3T3 fibroblasts expressed both wit3.0␣ and wit3.0␤ mRNAs at baseline level. When treated with TGF-␤1 (0.1 and 5.0 ng/ml), NIH3T3 fibroblasts showed non-significant increase in wit3.0␣ and wit3.0␤ mRNA levels (Fig. 5D). Factorial ANOVA analysis revealed that only ␣-SM actin increased significantly with TGF-␤1 treatment (p Ͻ 0.05). The steady-state mRNA levels of wit3.0␣ and wit3.0␤ were not significantly affected by the TGF-␤1 treatment. DISCUSSION As shown, rapid wound closure by the approximation of wound margins in the rat oral mucosal wound by tooth extraction (Fig. 1, A-D) was observed, along with accelerated collagen gel contraction during the initial 15-to 20-h period (Fig. 1F). The inflammatory response to the injury in oral mucosa has been shown less in the equivalent dermal wound (9), and the expression of cytokines is relatively minimal in oral wound tissue during the healing period compared with other tissues (5, 10 -12, 18). It has been postulated that saliva contains these cytokines and supplies them to oral wounds (19,20). The RT-PCR analysis in the present study generally support that the transcriptional response to the tooth extraction wound was mild to moderate among the genes examined. Therefore, it was distinctively noted that the steady-state mRNA level of wit3.0 increased the most from an initial minimal state to 40-fold within 4 days and over 50-fold after 1 week of healing.
Utilizing the genomic DNA sequence databases of humans and other animals recently available, the chromosomal location of wit3.0 was determined to be 12p11.23 in the human and 4q44 in the rat. The mouse data base indicated a wit3.0 match, but the chromosomal location has not yet been determined. The Online Mendelian Inheritance in Man (OMIM) data base (21,22) for known inheritable diseases linked to the 12p11.23 allele did not reveal wit3.0 linkage to any recorded diseases.
The overlapping rat wit3.0 cDNAs of 2746 bp suggest a small open reading frame of 645 bp (13). The online genome data base search showed that the open reading frame was encoded from exon 2 through exon 7 in both humans and rats. The disproportionately large 3Ј-untranslated region was encoded in the last exon 7. Based on the complete agreement between cDNA and genomic DNA exon sequences, as well as the highly conserved gene structure, we concluded that the coding sequence of wit3.0 might translate to a functional protein. Furthermore, our new data clearly show that alternative splicing of exon 5 accounts for the previously reported two different transcripts, wit3.0␣ and wit3.0␤ (13).
The wit3.0 deduced peptide was correlated with the Pfam consensus sequence motif of myosin heavy chain II coiled-coil domain (Fig. 3). Myosin heavy chain II is a multidomain protein important for both cellular structure and contraction (23,24). Myosin heavy chain II molecules are dimerized through the coiled-coil ␣-helix and provide the phosphorylation site. Thus, the coiled-coil domain is thought to regulate the ability of the myosin to produce force (25)(26)(27). The peptide sequence of wit3.0␤ suggests a potential ␣-helix structure (Fig. 2C). The wit3.0-FLAG peptide examined on the SDS-PAGE Western blot appeared as a single peptide (Fig. 4, A and B); however, because the myosin dimer is formed through ionic bonding, the possible molecular interaction by wit3.0 remains to be examined.

wit3.0 Facilitates Wound Contraction
Our studies localized both wit3.0␣ and wit3.0␤ to cytoplasmic sites in the fibroblasts we studied. Together with the possible structural relationship with the myosin heavy chain II, which plays a regulatory role in the myofilament-cytoskeleton complex, we postulated that wit3.0 may participate in the fibroblast-derived wound contraction. Tested in our in vitro collagen gel contraction assay, fibroblasts overexpressing wit3.0␣ or wit3.0␤ peptides contracted the collagen gel at the faster rate than controls. The accelerated in vitro gel contraction was temporally limited to the first 15-to 20-h period. Because the presence of wit3.0-FLAG fusion peptides was confirmed by Western blot after the collagen contraction assay (data not shown), the change in the collagen contraction rate only during the initial period is unlikely due to depletion of wit3.0 expression vectors. There is a considerable similarity in the gel contraction profile between transfected fibroblasts overexpressing wit3.0 and oral wound fibroblasts, which continued to up-regulate the endogenous wit3.0 expression. Thus, we speculated that the contribution of wit3.0 to wound contraction might require unidentified cofactors or partner molecules that are not simulated in the in vitro experiments.
Early wound contraction, typically characterized by actively proliferating, migrating epithelial cells and loose connective tissue ECM, is an essential initial healing process that establishes the epithelial integrity in the dermal and mucosal open wound. Clinical observations suggest that migration of fibroblasts into and through the ECM during the initial phase of wound healing, prior to the expression of ␣-SM actin, appears to be a fundamental component of wound contraction (28). Due to the greater compliance of immature connective tissue, early wound contraction does not result in generating tensile stress. The wound repair process continues to precipitate ECM, and the wound connective tissue increasingly becomes rigid. The inflexible ECM combined with a group of cytokines stimulates the differentiation of myofibroblasts expressing ␣-SM actin (29,30), which are more prevalent in the late wound repair tissue. The findings in our studies supported that TGF␤ stimulated the ␣-SM actin expression. Other studies showed TGF␤ has a wit3.0␣ and ␤ mRNA increased 1.12-and 1.04-fold with 0.1 ng/ml and 1.23-and 1.13-fold with 0.5 ng/ml TGF-␤1, respectively. capacity to accelerate the contraction rate of NIH3T3, BHK-21 cell lines, and human foreskin fibroblast cultured in collagen gel (31), which has been believed to be caused by the increased synthesis of ␣-SM actin.
Although the myofibroblast-mediated wound contraction mechanism has been well investigated, the mechanism responsible for the initial wound contraction is poorly understood. The involvement of wit3.0 in the initial wound contraction may shed the light to new investigations on this important woundhealing process.