Cytokine-like 1 (CYTL1) Regulates the Chondrogenesis of Mesenchymal Cells*

To identify novel molecules regulating chondrogenesis and cartilage development, we screened a cartilage-specific expressed sequence tag data base. Cytokine-like 1 (Cytl1), a possible cytokine candidate with unknown function that was originally identified in bone marrow-derived CD34-positive cells, was selected for functional characterization. In view of the initial observation that Cytl1 is predominantly expressed in chondrocytes and cartilage, we investigated its possible role in chondrogenesis and hypertrophic maturation of chondrocytes. Cytl1 expression was very low in mesenchymal cells, dramatically increased during chondrogenesis, and decreased during hypertrophic maturation, both in vivo and in vitro. The role of Cytl1 in chondrogenesis and hypertrophic maturation was examined by treating chondrifying mesenchymal cells with exogenous Cytl1 or ectopic expression of Cytl1. Notably, exogenous Cytl1 caused chondrogenic differentiation of mouse limb bud mesenchymal cells during micromass culture. Lentivirus-mediated overexpression of Cytl1 additionally induced chondrogenic differentiation of mesenchymal cells. However, Cytl1 did not affect the hypertrophic maturation of chondrocytes. Cytl1 exerted its chondrogenic effect via stimulation of Sox9 transcriptional activity. In addition, Cytl1 caused expression of insulin-like growth factor 1, which has a capacity to induce chondrogenesis. Thus, our results collectively suggest that chondrocyte-specific Cytl1 regulates chondrogenesis as a novel autocrine factor, but not hypertrophic maturation of chondrocytes during cartilage development.

To identify novel molecules regulating chondrogenesis and cartilage development, we screened a cartilage-specific expressed sequence tag data base. Cytokine-like 1 (Cytl1), a possible cytokine candidate with unknown function that was originally identified in bone marrow-derived CD34-positive cells, was selected for functional characterization. In view of the initial observation that Cytl1 is predominantly expressed in chondrocytes and cartilage, we investigated its possible role in chondrogenesis and hypertrophic maturation of chondrocytes. Cytl1 expression was very low in mesenchymal cells, dramatically increased during chondrogenesis, and decreased during hypertrophic maturation, both in vivo and in vitro. The role of Cytl1 in chondrogenesis and hypertrophic maturation was examined by treating chondrifying mesenchymal cells with exogenous Cytl1 or ectopic expression of Cytl1. Notably, exogenous Cytl1 caused chondrogenic differentiation of mouse limb bud mesenchymal cells during micromass culture. Lentivirus-mediated overexpression of Cytl1 additionally induced chondrogenic differentiation of mesenchymal cells. However, Cytl1 did not affect the hypertrophic maturation of chondrocytes. Cytl1 exerted its chondrogenic effect via stimulation of Sox9 transcriptional activity. In addition, Cytl1 caused expression of insulin-like growth factor 1, which has a capacity to induce chondrogenesis. Thus, our results collectively suggest that chondrocyte-specific Cytl1 regulates chondrogenesis as a novel autocrine factor, but not hypertrophic maturation of chondrocytes during cartilage development.
Cartilage formation during embryonic development begins with the aggregation of mesenchymal cells, which ultimately differentiate into chondrocytes. Differentiated chondrocytes proliferate rapidly and secrete a cartilage-specific extracellular matrix such as type II collagen and sulfated proteoglycan to form cartilage. The cartilage serves as a template for endochon-dral ossification, which requires the maturation of hypertrophic chondrocytes (1)(2)(3)(4). These sequential events during chondrogenesis and cartilage formation are precisely regulated by various growth factors released from cartilage elements and perichondrium. Secreted growth factors exert their effects by modulating intracellular signaling (1,2). Although several regulatory growth factors have been identified, including bone morphogenetic proteins, fibroblast growth factors, insulin-like growth factor-1 (IGF-1), 2 transforming growth factor-␤, and parathyroid hormone-related peptide, the precise mechanisms of regulation of chondrogenesis and cartilage development remain to be elucidated. In this study, we analyze a cartilagespecific expressed sequence tag (EST) data base in an attempt to identify novel molecules that modulate chondrogenesis and cartilage development.
The EST data base provides important information on novel genes displaying tissue-specific expression profiles (5)(6)(7). We analyzed the human cartilage UniGene library (8), and selected Cytl1 as a possible candidate regulator of chondrogenesis. Cytl1 is a functionally unknown cytokine candidate originally cloned from bone marrow and cord blood mononuclear cells that bear the CD34 surface marker (9). Previous studies report Cytl1 mRNA and protein expression in cartilaginous tissues, including mouse inner ear and human articular cartilage, respectively (10,11). Alignment of the predicted primary amino acid sequences of Cytl1 proteins from multiple species (human, mouse, chicken, and puffer) reveals an N-terminal secretory signal peptide and four ␣-helices, a common characteristic of cytokines. Accordingly, Cytl1 is categorized as a novel cytokine, despite no functional evidence (9 -11). In this study, we investigated the function of Cytl1 in chondrogenesis and cartilage development. Our findings clearly indicate that Cytl1 regulates chondrogenesis, but not hypertrophic maturation of chondrocytes during cartilage development.

EXPERIMENTAL PROCEDURES
Tissue Preparation and Cell Culture-Mouse tissues and rib cartilage were obtained from 4-week-old male ICR mice and 3-day-old newborn mice, respectively. Mouse limb buds were prepared from forelimb buds of mouse embryos at the indicated embryonic day. Mouse rib chondrocytes were prepared from 3-day-old newborn ICR mice, as described previously (12). Briefly, cartilaginous rib cages were preincubated for 45 min in 0.2% type II collagenase (381 units/mg solid; Sigma) and rinsed with PBS. Tissues were further dissociated enzymatically for 4 h in 0.2% type II collagenase. Isolated chondrocytes were maintained in Dulbecco's modified Eagle's medium (DMEM) containing 10% (v/v) fetal bovine serum, 50 g/ml streptomycin, and 50 units/ml penicillin. Mesenchymal cells were isolated from the limb buds of 11.5 dpc ICR mouse embryos and maintained as micromass culture to induce chondrogenesis and hypertrophic maturation (13,14). Briefly, 4.0 ϫ 10 7 cells/ml in Dulbecco's modified Eagle's medium/F-12 medium (2:3) containing 10% (v/v) fetal bovine serum, 50 g/ml streptomycin, and 50 units/ml penicillin were spotted as 15-l drops on culture dishes to induce chondrogenesis for 6 days. The culture medium was altered to DMEM/F-12 (2:3) supplemented with 50 g/ml ascorbic acid, 5 mM ␤-glycerol phosphate, to induce hypertrophic maturation up to 15 days. Chondrogenesis was determined by examining the expression of collagen IIB and aggrecan by RT-PCR, as described below, or Alcian blue staining to detect accumulation of sulfated proteoglycans (15). Hypertrophic maturation of chondrocytes was determined by examining the expression of markers, such as collagen X and matrix metalloproteinase (MMP)-13, by reverse transcription-PCR (RT-PCR) or Alizarin Red staining to detect mineralization (13,14).
Expression Vectors and Reporter Gene Assay-Cytl1 cDNA (GenBank TM accession number EF108311) was subjected to RT-PCR amplification from mouse rib chondrocytes using specific primers (Table 1) designed to introduce EcoRI and XhoI restriction sequences at the 5Ј and 3Ј ends of the amplified fragment, respectively. The resulting cDNA was cloned into the corresponding restriction sites of pcDNA3.1(ϩ) Myc/His vector (Invitrogen). A reporter gene containing the 48-bp Sox9binding site in the first intron of human COL2A1 was constructed by inserting the chemically synthesized sequence in the pGL3-promoter luciferase reporter vector (Promega, Madison, WI), as described previously (16). The Sox9 reporter gene was transfected into a suspension of isolated mesenchymal cells using the Lipofectamine 2000 reagent (Invitrogen). After transfection with pGL3-control or pGL3-Sox9 reporter gene for 2 h, mesenchymal cells were spotted and maintained as micromass culture in complete medium for 12 h, and treated with IGF-1 or Cytl1-conditioned medium in serum-free conditions, as described for each experiment. For the reporter gene assay, the Sox-9 reporter gene (1 g) was co-transfected with pCMV-␤galactosidase expression vector (0.3 g) as an internal control to ensure transfection efficiency. Luciferase activity was normalized against ␤-galactosidase activity, as described previously (16).
Preparation of Conditioned Medium for Cytl1-Stable control and mouse fibroblast L929 cell lines expressing mouse Cytl1 were generated by transfection with empty vector and Cytl1 cDNA, respectively. One day after transfection, cells were subcultured and selected by adding 0.8 mg/ml G418. Cytl1 expression was confirmed by RT-PCR and Western blotting. Following growth to 90% confluency, control and Cytl1-expressing L929 cells were washed and maintained in serum-free DMEM for 48 h. Conditioned media were clarified by centrifugation at 10,000 ϫ g for 5 min, followed by filtration (0.2 m pore size), and concentrated 30 times by ultrafiltration in Amicon stirred cells (Millipore) using a YM membrane with a 10-kDa molecular mass cutoff.
Construction and Infection of Lentivirus-Lentivirus bearing mouse Cytl1 was constructed by the Macrogen LentiVector Institute (Seoul, Korea). Briefly, the Cytl1 gene fragment was amplified from pcDNA3.1-Cytl1-myc/his vector by PCR. A Cytl1-specific primer set (Table 1) was designed to tag Myc sequences at the C terminus of the gene. The amplified product was digested with EcoRI-ClaI and inserted into a lentiviral vector (Lenti-mCMV-IRES-puro) containing the mouse cytomegalovirus promoter and a puromycin-resistant gene, using the internal ribosome entry site system. The constructed transfer vector, a vesicular stomatitis virus-G expression vector, and a gag-pol expression vector were co-transfected into 293T cells at a 1:1:1 molar ratio using Lipofectamine Plus (Invitrogen). The culture supernatant containing viral particles was harvested at 48 h after transfection, clarified with a 0.45-m membrane filter (Nalgene, New York, NY), and stored at Ϫ70°C. The titer obtained (ϳ1-5 ϫ 10 7 transduction units) was used without further concentration. Titers were determined using human immunodeficiency virus, type 1, p24 enzyme-linked immunosorbent assay. Mesenchymal cells were maintained as micromass culture for 24 h and infected with mock lentivirus or Cytl1 lentivirus for 18 h. Infected cells were cultured for up to 5 days in serum-free DMEM/F-12 medium. siRNA and shRNA-Five siRNA and two shRNA were used to knock down Cytl1 expression in chondrifying mesenchymal cells during micromass culture. Briefly, siRNAs were designed based on the coding sequence of mouse Cytl1 (supplemental Fig. 3), and siRNA oligonucleotides were synthesized by Samchully Pharm (Daejeon, Korea) as follows: siRNA1, 5Ј-aua cca uca uga acu ccu utt-3Ј (sense) and 5Ј-aag gag uuc aug aug gua utt-3Ј (antisense); siRNA2, 5Ј-ggc uuu acc ugg aca ucc att-3Ј (sense) and 5Ј-ugg aug ucc agg uaa agc ctt-3Ј (antisense); siRNA3, 5Ј-uga cug cag ugc cuu aga att-3Ј (sense) and 5Ј-uuc uaa ggc acu gca guc att-3Ј (antisense); siRNA4, 5Ј-gga cuu ggu auu ccu cuc att-3Ј (sense) and 5Ј-uga gag gaa uac caa guc ctt-3Ј (antisense); and siRNA5, 5Ј-gcc gug aga uca ugg cag att-3Ј (sense) and 5Ј-ucu gcc aug auc uca cgg ctt-3Ј (antisense). After heating the siRNA oligonucleotides for 2 min at 90°C to denature secondary structures within single strands, the sense and antisense strands were allowed to anneal at 30°C for 1 h. Two shRNA constructs for Cytl1 were obtained from Openbiosystems Inc. (Huntsville, AL). The target sequences are 5Ј-gga agc tgt ata cca tca tg-3Ј (shRNA1) and 5Ј-gct tta cct gga cat cca taa-3Ј (shRNA2). Mesenchymal cells were maintained as micromass culture for 12 h and transfected for 6 h with siRNA or shRNA by using Lipofectamine 2000 (Invitrogen) and Arrest-In TM transfection reagent (Openbiosystems), respectively. Transfected cells were cultured for up to 4 days in serum-free medium. Nonsilencing siRNA (Bioneer, Daejeon, Korea) and shRNA (Openbiosystems) were used as a negative control.
In Situ Hybridization-Forelimb buds of 14.5 dpc mouse embryos and spots of micromass culture were fixed in 4% paraformaldehyde for 18 h at 4°C, dehydrated with graded ethanol, embedded in paraffin wax, and cut into 4-m sections, as described previously (17,18). For hybridization, dewaxed paraffin sections of prepared samples were pretreated with HCl and proteinase K, followed by acetylation. Sections were incubated with hybridization buffer containing denatured sense or antisense digoxigenin-labeled riboprobes. Next, sections were treated with RNase A and processed with an anti-digoxigenin detection assay kit (Roche Diagnostics). Hybridization signals were visualized with a solution of 4-nitro blue tetrazolium chloride and 5-bromo-4-chloro-3-iodolyl phosphate (Roche Diagnostics). Digoxigenin-conjugated riboprobes for collagen IIB, collagen X, and Cytl1 were synthesized using the digoxigenin RNA labeling mix (Roche Diagnostics). Briefly, the cDNA fragments of mouse collagen IIB and collagen X were amplified by PCR using specific primers designed to introduce HindIII (sense) and EcoRI (antisense) restriction sequences at the 5Ј end ( Table 1). The cDNA fragments were digested with HindIII and EcoRI. The cDNA sequences of mouse Cytl1 inserted in the pcDNA3.0 vector (Invitrogen) were digested with PstI and SacI. Digested collagen IIB and collagen X and Cytl1 cDNA sequences were cloned into the pSPT-18 vector (Roche Diagnostics), linearized with EcoRI or HindIII digestion, and transcribed with SP6 and T7 RNA polymerase to generate sense and antisense probes, respectively.
RT-PCR Analysis and Quantitative Real Time PCR (qRT-PCR)-Total RNA was isolated using RNA STAT-60 (Tel-Test, Woodlands, TX) and reverse-transcribed with ImProm-II TM reverse transcriptase (Promega). The cDNA generated was amplified by PCR with Taq polymerase (Takara Bio, Shiga, Japan). Primers and conditions for amplification are summarized in Table 1, and qRT-PCR was performed using a chromo 4 cycler (Bio-Rad) and SYBR Premix Ex Taq TM (Takara Bio). qRT-PCRs were performed in duplicate, and the amplification signal from the target gene was normalized against that of glyceraldehyde-3-phophate dehydrogenase (GAPDH) in the same reaction.
Northern and Western Blot Analyses-Total RNA was isolated with a single-step guanidinium thiocyanate-phenol-chloroform method using RNA STAT-60 (Tel-Test), according to the manufacturer's protocol. Next, total RNA (3 g) was fractionated on formaldehyde/agarose gels. The probe (417 bp) for Cytl1 transcripts was generated by RT-PCR using a sense primer corresponding to nucleotides ϩ1 to ϩ21 and an antisense primer corresponding to nucleotides ϩ399 to ϩ417 of Cytl1. For Western blotting, whole cell lysates prepared as described previously (16,18) were fractionated by SDS-PAGE and transferred to a nitrocellulose membrane. Anti-type II collagen (Chemicon, Temecula, CA), anti-extracellular signal-regulated protein kinase (ERK)-1 (BD Biosciences), anti-Sox9 (Santa Cruz Biotechnology Inc., Santa Cruz, CA), and anti-Myc-tagged (Cell Signaling Technology, Beverly, MA) antibodies were employed to detect proteins.
Micro Liquid Chromatography-Tandem Mass Spectrometry Analysis of Cytl1-Mouse rib chondrocytes were cultured in DMEM. After growth to 90% confluency, chondrocytes were washed and maintained in serum-free DMEM for 48 h. Supernatant fractions prepared from chondrocytes were precipitated using the trichloroacetic acid/acetone method. Proteins from the precipitated supernatant were reduced, alkylated, and trypsin-digested, as described previously (19). Briefly, digested proteins were loaded onto fused silica capillary columns, and placed in line with an Agilent HP1100 quaternary LC pump. Separated peptides were directly electrosprayed into an LCQ Deca XP Plus mass spectrometer (ThermoFinnigan, Palo Alto, CA). A data-dependent scan consisting of one full MS scan and three data-dependent MS/MS scans was used to generate MS/MS spectra of eluted peptides. MS/MS spectra were searched against an in-house protein data base containing mouse Cytl1 sequences using TurboSequest. Search results were filtered with Bioworks version 3.1.

Cytl1 Is Predominantly Expressed in Chondrocytes and
Cartilage-To identify cartilage-specific regulatory genes, we analyzed the human normal cartilage library (Library 8940) deposited in the UniGene data base at the NCBI (www.wcbi.nlm.nih.gov). Briefly, functionally unknown and novel cartilage-specific genes in the library were classified on the basis of reference criteria (6,7). A gene was categorized as "cartilage-specific" if the number of cartilage ESTs was significantly higher than that of non-cartilage ESTs analyzed using monochromatic serial analysis of gene expression (SAGE)/cDNA virtual Northern (cgap.nci.nih.gov/SAGE). Among these, Cytl1 was selected for functional characterization during chondrogenesis and hypertrophic maturation of chondrocytes. Alignment of Cytl1 amino acid sequences from multiple species (human, mouse, chicken, and puffer) revealed four ␣-helices and six conserved cysteine residues, which are common characteristics of cytokines and interleukins (supplemental Fig. 1).
We initially examined the expression patterns of Cytl1 in various mouse tissues. RT-PCR, and quantitative RT-PCR analyses disclosed that Cytl1 is predominantly expressed in cartilage (Fig. 1, A and B). Among the cell types examined, Cytl1 was detected in primary culture chondrocytes but not HTB-94 chondrosarcoma or ATDC5 chondroprogenitor cells (Fig. 1C). Upon ectopic expression in chondrocytes, Cytl1 was detected as an ϳ1.0-kb single transcript in primary culture mouse rib chondrocytes ( Fig. 2A), and protein was identified in both cell lysates and culture supernatant (Fig. 2B).
The secreted form of Cytl1 displayed increased molecular weight (approximately ϳ6.4 kDa), indicative of post-translational modifications during secretion. Secretion of endogenous Cytl1 was further confirmed by micro-liquid chromatography-tandem mass spectrometry analysis of the culture supernatant of chondrocytes. Specifically, trypsin-digested protein from the culture supernatant fractions was reduced, alkylated, and analyzed by micro liquid chromatography-tandem mass spectrometry. As shown in Fig. 2C, two peptides corresponding to mouse  Cytl1 ( 77 LRDFVASPQCWK 88 and 89 MAWVDTLKDR 98 ) were detected, confirming its role as a secretory protein in chondrocytes.
Expression of Cytl1 during Cartilage Development-To elucidate the in vivo expression pattern of Cytl1 during cartilage development, we performed RT-PCR analysis using mouse forelimb buds isolated from embryos at 11.5-15.5 dpc displaying limb development and cartilage formation (4,20). Collagen IIB was detected at 14.5 and 15.5 dpc limb buds in which individual fingers are visible and longitudinal growth of cartilage occurs (Fig. 3A). Cytl1 expression occurs prior to collagen IIB expression with a peak level at 14.5 dpc limb buds (Fig. 3A). In situ hybridization in developing limb buds at 14.5 dpc led to the detection of significant Cytl1 transcript levels in the cartilage region, which was determined by collagen IIB expression and accumulation of sulfated proteoglycans (Fig. 3, B and C). In developing ulna, Cytl1 transcript was detected in the region where collagen IIB is expressed, but it was not detected in the region where collagen X is expressed (Fig. 3D), indicating that Cytl1 is expressed in differentiated and proliferating chondrocytes but not in hypertrophic chondrocytes.
Next, we examined Cytl1 expression during in vitro chondrogenesis and hypertrophic maturation of chondrocytes caused by micromass culture of mesenchymal cells isolated from 11.5 dpc mouse limb buds (13,14). For this purpose, mesenchymal cells were maintained as micromass culture to induce chondrogenesis up to day 6, and hypertrophic maturation was induced by switching chondrogenic medium to hypertrophic medium up to day 15. As expected, the chondrocytespecific markers collagen IIB and aggrecan (1,21) were expressed at day 3, reached peak levels at day 6, and decreased during hypertrophic maturation (Fig. 4, A and B). Markers of hypertrophic chondrocytes, collagen X and MMP-13, were detected at day 9 (Fig. 4, A and B). Cytl1 expression was very low in undifferentiated mesenchymal cells (day 1), dramatically increased during chondrogenesis, and decreased during hypertrophic maturation (Fig. 4, A and B), similar to collagen IIB. Cells expressing Cytl1 and collagen IIB were identified from sections of day 6 micromass culture spots by in situ hybridization. Collagen IIB expression analyses supported the presence of significant amounts of the Cytl1 transcript in cells located in cartilage nodules composed of differentiated chondrocytes (Fig. 4C). Analogous to the expression pattern of Cytl1 during chondrogenesis and hypertrophic maturation of mesenchymal cells, Cytl1 expression level in undifferentiated ATDC5 cells (a chondroprogenitor cell line) was very low, elevated during chondrogenesis, and decreased during hypertrophic maturation (supplemental Fig. 2). The above results clearly demonstrate chondrocyte-specific expression of Cytl1 during cartilage development, both in vivo and in vitro.

Cytl1 Induces Chondrogenesis of Mesenchymal Cells without Effects on Hypertrophic Maturation-
Because the chondrocyte-specific expression of Cytl1 suggests a possible function in chondrogenesis and/or hypertrophic maturation, we examined its role in chondrogenic differentiation of mesenchymal cells and hypertrophic maturation of chondrocytes. For this purpose, recombinant Cytl1 was prepared from serum-free conditioned medium of L929 cells expressing ectopic Cytl1 and applied to mesenchymal cells maintained as micromass culture in serum-free medium. IGF-1 (100 ng/ml) was used as the positive control for induction of chondrogenic differentiation (22). Exogenous Cytl1 did not modulate precartilage condensation, as determined by peanut agglutinin staining (Fig. 5A), but caused chondrogenic differentiation of mouse limb bud mesenchymal cells, as evident from the increased syn-  . Chondrogenic effects of exogenous Cytl1. A, mesenchymal cells were maintained as micromass culture for 1.5 or 4 days in the presence of control conditioned medium (60 l), IGF-1 (100 ng/ml), or Cytl1 conditioned medium (60 l) under serum-free conditions. Cells were stained with Alcian blue to determine chondrogenesis at day 4 or peanut agglutinin (PNA) to determine precartilage condensation at day 1.5. B, Alcian blue-stained cells were extracted and quantified by measuring absorbance. C, mesenchymal cells were maintained as micromass culture for 4 days in the presence of control (Con) conditioned medium (40 l) or IGF-1 (100 ng/ml) or the indicated amounts (l) of Cytl1 conditioned medium. Expression levels of collagen IIB mRNA and collagen II protein were determined by RT-PCR and Western blotting (WB), respectively. GAPDH and ERK were employed as the loading controls. Cytl1 secreted into conditioned medium from L929 cells was detected by Western blotting using the anti-Myc antibody (D). The data represent a typical result from five independent experiments. thesis of sulfated proteoglycan (Fig. 5, A and B) and expression of collagen IIB (Fig. 5C). The role of Cytl1 in chondrogenesis was further confirmed by overexpression in chondrifying mesenchymal cells using lentivirus prepared in serum-free medium. To increase cell viability, mesenchymal cells were maintained as micromass culture in complete medium for 24 h, followed by exposure to lentivirus in serum-free conditions. Under these conditions, mesenchymal cells underwent slight chondrogenic differentiation in control culture (Fig. 6A). Lentivirus-mediated overexpression of Cytl1 led to increased synthesis of sulfated proteoglycan (Fig. 6, A and B) and expression of collagen IIB (Fig.  6C), confirming enhancement of chondrogenesis.
We next examined a role of Cytl1 in hypertrophic maturation of chondrocytes. Exogenous Cytl1 was added to hypertrophic medium from days 5 to 13 (the time period of hypertrophic maturation). Contrasting the effects on chondrogenesis, exogenous Cytl1 did not affect hypertrophic maturation of chondrocytes (Fig. 7A). Additionally, lentivirus-mediated ectopic expression of Cytl1 did not affect hypertrophic maturation of chondrocytes (Fig. 7B). The results provide conclusive evidence that Cytl1 induces chondrogenesis of mesenchymal cells without significant effects on hypertrophic maturation of differentiated chondrocytes.
Mechanisms of Cytl1 Regulation of Chondrogenesis-To clarify the regulatory mechanisms of Cytl1 in chondrogenesis, we investigated whether the protein modulates the expression and/or transcriptional activity of Sox9, a master transcription factor for chondrogenic differentiation (1,23). A reporter gene containing the 48-bp Sox9-binding site in the first intron of human COL2A1 (16,24) was employed to examine transcriptional activity. Exogenous Cytl1 did not affect the Sox9 protein level (Fig. 8A) but enhanced its transcriptional activity (Fig. 8B). These results strongly support our theory that Cytl1 is a soluble factor that induces chondrogenic differentiation of mesenchymal cells by stimulating Sox9 transcriptional activity. The regulatory mechanisms of Sox9 transcriptional activity were further elucidated by examining signaling pathways. Because our previous results indicated that chondrogenesis is enhanced by the inhibition of ERK (25) whereas differentiation is blocked by the inhibition of p38 mitogen-activated protein (MAP) kinase (15) or protein kinase C (PKC) ␣ (25), a role for these signaling pathways in Cytl1-induced chondrogenesis was examined by using specific inhibitor. Consistent with our previous reports (15,25), Cytl1-induced collagen II expression and accumulation of sulfated proteoglycan were enhanced by the inhibition of ERK with PD98059, whereas inhibition of p38 MAP kinase with SB203580 or PKC␣ with Go6976 blocked Cytl1-induced collagen IIB expression and proteoglycan accumulation (Fig. 8,  C and E). Similar to the effects on collagen II expression, ERK inhibition enhanced Sox9 activity, whereas Sox9 activity was blocked by the inhibition of p38 MAP kinase or PKC␣ (Fig. 8D). Interestingly, Sox9 expression was not affected by the inhibition of ERK or p38 MAP kinase but blocked by the inhibition of PKC␣ (Fig. 8C).
Finally, we examined whether there is a cross-talk between Cytl1 and IGF-1 in chondrogenesis of mesenchymal cells. As shown in Fig. 9, A and B, Cytl1 conditioned medium or lentivirus-mediated overexpression of Cytl1 caused enhanced expression of Cytl1, indicating a positive feedback regulation of Cytl1 expression. In addition, IGF-1 expression was significantly increased by Cytl1 conditioned medium or lentivirus-mediated overexpression of Cytl1 with a concomitant expression of collagen II. In contrast, IGF-1 did not modulate Cytl1 expression whereas collagen II expression was induced by IGF-1 (Fig. 9C). Taken together, the above results suggest that Cytl1 causes IGF-I expression which induces chondrogenesis of mesenchymal cells during micromass culture.

DISCUSSION
We demonstrate in this study that Cytl1 expression is strongly correlated with chondrocyte differentiation, both in vitro and in vivo. Moreover, Cytl1 is a chondrocyte-specific secreted protein with the capacity to induce chondrogenesis of mesenchymal cells. Cytl1 was selected from a human cartilage UniGene library in an attempt to identify novel regulatory genes involved in cartilage development. The protein contains four ␣-helices and six conserved cysteine residues, which may form intra-disulfide bonds to give a globular structure. These structural characteristics are common in cytokines (26,27), leading to the assumption that Cytl1 is a cytokine-like protein. SDS-PAGE analyses additionally disclose that secreted Cytl1 displays slower mobility, compared with its cellular form. Therefore, it is likely that Cytl1 undergoes post-translational modifications, although the nature of these modifications remains to be established. Similar sequences to Cytl1 have not been detected in invertebrates, such as Drosophila or Caenorhabditis elegans, but are present in several other vertebrates, such as human, chicken, and Fugu rubripes, as revealed by comparative protein sequence analyses with mouse Cytl1 (supplemental Fig. 1).
Thus, Cytl1 appears to originate early in vertebrate evolution (28). Cytl1 is specifically expressed in differentiated chondrocytes and cartilage. In our experiments, Cytl1 expression was low in undifferentiated mesenchymal cells, dramatically elevated in differentiated chondrocytes, and decreased during hypertrophic maturation. In view of this chondrocyte-specific expression pattern, we initially hypothesized that the protein regulates chondrogenesis, maintenance of differentiated phenotype of chondrocytes, and/or hypertrophic maturation of chondrocytes. Experiments with exogenous Cytl1 and lentivirus-mediated overexpression of Cytl1 clearly support a capacity of the protein to induce chondrogenesis without affecting hypertrophic maturation. The role of Cytl1 in chondrogenesis was further confirmed with a loss-of-function study. We prepared five types of siRNA oligonucleotides and two types of vector-based shRNAs (supplemental Fig. 3). Although control experiments clearly indicated efficient transfection of siRNA oligonucleotides and shRNA vectors, the examined siRNAs and shRNAs did not induce knockdown of Cytl1 expression during micromass culture of mesenchymal cells for unknown reasons (supplemental Fig. 3). Nevertheless, based on the facts that Cytl1 is expressed in chondrocytes with a capacity to induce chondrogenesis of mesenchymal cells and that Cytl1 expression is regulated by a positive feedback mechanism, Cytl1 may act as an autocrine factor that regulates chondrogenesis and cartilage development.
Because the levels of chondrocyte-specific markers, such as Col2A1, Col9A2, Col11A2, and Agc, are modulated by Sox9, a key transcriptional regulator of chondrogenesis (29), we investigated the effects of Cytl1 on Sox9 expression and/or transcriptional activity. Cytl1 clearly enhanced Sox9 transcriptional activity without affecting its expression. This stimulation of Sox9 transcriptional activity by Cytl1 possibly contributes to its chondrogenic effect. Moreover, Sox9 transcriptional activity is FIGURE 8. Activation of Sox9 transcriptional activity by exogenous Cytl1 in mesenchymal cells. A and B, mesenchymal cells were maintained as micromass culture up to 4 days in the presence of the indicated amounts (l) of control conditioned medium, IGF-1 (100 ng/ml), or Cytl1 conditioned medium (l) under serum-free conditions. Expression levels of collagen (Coll) II and Sox9 were determined by Western blotting. ERK was employed as the loading control (A). Cells were transfected with the Sox9 reporter gene and maintained as micromass culture up to 4 days in the presence of control (Con) conditioned medium (40 l), IGF-1 (100 ng/ml) or the indicated amounts (l) of Cytl1 conditioned medium. Sox9 transcriptional activity was determined with the reporter gene assay (B). C-E, mesenchymal cells were maintained as micromass culture up to 4 days in the absence (None) or presence of IGF-1 (100 ng/ml) or 40 l of Cytl1 conditioned medium. Thirty min prior to Cytl1 treatment, the cells were exposed to 10 M of PD98059 (PD) to inhibit ERK, 10 M of SB203580 (SB) to inhibit p38 MAP kinase, or 1 M Go6976 (Go) to inhibit PKC. Expression levels of collagen II and Sox9 were determined by Western blotting (C). Sox9 transcriptional activity was determined with the reporter gene assay (D). Accumulation of sulfated proteoglycans was detected by Alcian blue staining (E). The data represent a typical result (A, C, and E) and mean values with standard deviation (B and D) from four independent experiments. controlled by protein-protein interactions, for instance, positive modulation by factors such as cAMP-response elementbinding protein-binding protein (CREB/p300) and peroxisome proliferator-activated receptor ␥ co-activator 1␣ (PGC-1␣) (30,31). Protein modification, such as phosphorylation, is another mechanism for regulating Sox9 transcriptional activity. Indeed, our current results indicated that Cytl1-induced activation of Sox9 transcriptional activity is regulated protein kinase signaling pathways, including ERK, p38 MAP kinase, and PKC. However, Western blotting to determine whether Sox9 is phosphorylated at Ser-211 revealed that Cytl1 does not affect the phosphorylation status. Accordingly, we conclude that modulation of Sox9 phosphorylation is not related to Cytl1 regulation of transcriptional activity and that the regulatory role of ERK, p38 MAP kinase, and PKC␣ is not because of direct modification of Sox9 by these protein kinases. It is likely that Cytl1 controls Sox9 transcriptional activity via regulatory binding proteins or modifications, such as phosphorylation at unidentified sites. For example, L-Sox5 and Sox6 cooperatively regulate chondrocyte-specific genes, such as Col2A1 and aggrecan, with Sox9 (32). In addition to the modulation of Sox9 activity, our results indicated a cross-talk between Cytl1 and IGF-1 in chondrogenesis of mesenchymal cells. Because Cytl1 significantly enhances IGF-1 expression and IGF-1 has a capacity to induce chondrogenesis, our results strongly suggest that Cytl1-induced IGF-1 expression might contribute to the induction of chondrogenesis.
Although the chondrogenic effects of Cytl1 are evident, our current results demonstrate that the protein does not regulate hypertrophic maturation of chondrocytes. Because the Cytl1 level in chondrocytes undergoes a dramatic decrease during hypertrophic maturation, we initially assumed that exogenous or sustained Cytl1 expression inhibits or delays hypertrophic maturation of chondrocytes. However, our results suggest that down-regulation of Cytl1 is not a critical factor in the hypertrophic maturation process. Our findings provide collective evidence that Cytl1 is a novel regulatory factor predominantly expressed in chondrocytes and developing cartilage, which induces chondrogenesis of mesenchymal cells as an autocrine factor.