ARTHRITIS & RHEUMATISMVol. 44, No. 5, May 2001, pp 1013–1021© 2001, American College of RheumatologyPublished by Wiley-Liss, Inc.

Potential Role of HOXD9 in Synoviocyte Proliferation

Nguyen Dinh Khoa,1 Minako Nakazawa,1 Tomoko Hasunuma,1 Toshihiro Nakajima,1

Hiroshi Nakamura,1 Tetsuji Kobata,2 and Kusuki Nishioka1

Objective. To investigate the role of HOXD9 in theproliferation activity of cultured synoviocytes as well asthe mechanisms that regulate HOXD9 transcription.

Methods. Synoviocytes from patients with rheu-matoid arthritis (RA) and osteoarthritis (OA) weretransfected with HOXD9 complementary DNA to estab-lish stable transformants that overexpressed HOXD9.HOXD9 expression was detected by Western blottingwith anti-HOXD9 antibody. The growth properties ofthe transformants were investigated by proliferationand colony formation assays. The expression of basicfibroblast growth factor (bFGF), tumor necrosis factora (TNFa), interleukin-1b, c-Fos, and c-Myc was exam-ined by Western blotting. Transcriptional regulation ofHOXD9 was examined by transient cotransfection.

Results. HOXD9 protein was highly expressed inRA synoviocytes, but there was no expression in OAsynoviocytes. HOXD9 transfection induced stableHOXD9 protein expression in synoviocytes and showedan increased proliferation rate under both normal andserum-starved conditions, as well as an enhanced ca-pacity to proliferate anchorage independently to formcolonies in soft agar cultures, compared with controltransfectants. Higher levels of bFGF and c-Fos weredetected in HOXD9 transformants than in controls.

Transient cotransfection assays of NIH3T3 fibroblastsand synoviocytes showed that HOXD9 activated theluciferase reporter construct containing the highly con-served region (HCR), an autoregulatory element ofHOXD9 promoter. This activation was significantly in-creased by bFGF, suppressed by TNFa, and unchangedby transforming growth factor b in synoviocytes. Hu-man T lymphotropic virus type I tax also activated theluciferase reporter construct containing the HCR andhad a synergistic effect with HOXD9 on HCR promoteractivation.

Conclusion. Our data suggest that HOXD9 plays apotential role in synovial proliferation. In addition, theysuggest that the involvement of HOXD9 in the regulationof cellular growth might be mediated, at least in part, byup-regulation of growth-related factors such as bFGFand c-Fos and/or might result from increased transcrip-tion activity by its regulators.

Persistent hyperplasia of synovial lining cells inrheumatoid arthritis (RA) is one of the major pathologicchanges that contributes to the progression of synovitisand destruction of bone and cartilage (1). Synoviocyteproliferation in RA is induced and regulated by variouscytokines (2–4). In addition, other studies have sug-gested that the synovial lining cells of RA patients canautonomously and locally proliferate, as evidenced bythe expression and activation of proliferating cell nu-clear antigen/cyclin, c-Myc, nucleolar organizer regions,transcription factors, and certain other proliferation-associated genes in these cells (5–7). There is alsosufficient evidence to suggest that proliferativefibroblast-like synoviocytes in RA exhibit certain fea-tures of transformed cells (8–10), and a number offactors, including certain protooncoproteins, have beenfound to be associated with this phenomenon (11,12).Despite extensive and phenotypic transformation ofsynovial cells in RA, the pathogenesis remains incom-pletely understood.

Homeobox (HOX) genes encode homeoproteins,

Supported by grants from the Ministry of Education, Science,Sports and Culture of Japan, the Ministry of Welfare of Japan, theJapanese Rheumatism Foundation, and the Drug Organization ofJapan, and by a Grant-in-Aid for Scientific Research from the Ministryof Education, Science, Sports and Culture.

1Nguyen Dinh Khoa, MD, Minako Nakazawa, MS, TomokoHasunuma, MD, PhD, Toshihiro Nakajima, MD, PhD, Hiroshi Naka-mura, MD, PhD, Kusuki Nishioka, MD, PhD: St. Marianna UniversitySchool of Medicine, Kawasaki, Japan; 2Tetsuji Kobata, MD, PhD: St.Marianna University School of Medicine, Kawasaki, Japan, andDokkyo University School of Medicine, Kobayashi, Japan.

Address correspondence and reprint requests to ProfessorKusuki Nishioka, MD, PhD, Director, Rheumatology, Immunologyand Genetics Program, Institute of Medical Science, St. MariannaUniversity School of Medicine, 2-16-1 Sugao, Miyamae-ku, Kawasaki216-8512, Japan.

Submitted for publication December 22, 1999; accepted inrevised form January 4, 2001.

1013

which are localized in the nucleus and form one of themajor classes of transcription factors. The homeodo-main, a DNA-binding motif, and its flanking sequencesprovide either activation or repression functions fortarget gene transcription (13,14). In addition to theirroles in axial patterning during embryonic development(15), HOX genes are also involved in the control ofnormal proliferation and differentiation of several adulttissues as well as proliferation and oncogenic transfor-mation in some neoplasms and cancers (16–21).

HOXD9 (formerly HOX4C in humans, HOX5.2and HOX4.4 in mice), the most 39-located member ofAbdB-related HOXD genes, is the earliest HOXD geneactivated during the initiation and outgrowth of limbbuds (22–24). It is also known to be a potential tran-scription factor that not only autoregulates itself, butalso transactivates other HOX genes as well as certainmembers of other families (25,26). We have recentlydemonstrated that HOXD9 is preferentially expressed inthe proliferative synovium of RA patients (27). Inanother study in mice, we also found that the same HOXgene was expressed in embryonic joint tissues duringearly stages of joint formation, and that it was activatedin arthritic joints of human T lymphotropic virus type I(HTLV-I) tax transgenic mice with RA-resemblingarthropathy (28). Thus, the results of our studies, to-gether with those of other investigators described above,suggest that activation of HOXD9 in the rheumatoidsynovium may play a role in the abnormal growth andphenotypic “transformation” of synoviocytes.

In the present study, we first transfected synovio-cytes with HOXD9 complementary DNA (cDNA) andestablished a stable transformant synovial cell line thatoverexpressed HOXD9. We then investigated the growthproperties of, and cytokine production by, these trans-formant synovial cells. In addition, we also examined theregulation of HOXD9 transcription.

PATIENTS AND METHODS

Synoviocyte preparation and cell culture. Synoviocyteswere obtained by collagenase digestion of synovial tissues frompatients with RA and osteoarthritis (OA), as described previ-ously (29). Adherent cells were cultured and passaged inHam’s F-12 medium (Gibco BRL, Grand Island, NY) supple-mented with 10% heat-inactivated fetal calf serum (FCS;Bioserum, Melbourne, Australia). After 2–3 passages, synovio-cytes appeared homogeneous as fibroblast-like cells. Cellsfrom passages 4–8 were used. NIH3T3 fibroblasts (AmericanType Culture Collection, Rockville, MD) were maintained inDulbecco’s modified Eagle’s medium (DMEM; Gibco BRL)supplemented with 10% FCS.

Plasmids. The pSGH4C plasmid is an expression con-struct that contains the complete open reading frame of the

HOXD9 cDNA cloned into the pSG5 vector. The pTHCRplasmid is a reporter construct that contains a single copy ofthe ;100-bp highly conserved region (HCR), an autoregula-tory element of HOXD9 promoter, cloned into the pT81lucluciferase reporter vector. Both constructs were kindly pro-vided by Dr. Zappavigna (Milan) (25). The pH2Rneo plasmid,which contains the bacterial neomycin-resistant gene under thecontrol of the SV40 early promoter, and the pH2R40Mplasmid, which encodes the HTLV-I tax gene cloned into thepH2Rneo plasmid, were generously provided by Dr. Hatanaka(Kyoto University) (30).

Antibodies. Rabbit anti-HOXD9 antibodies wereraised against full-length recombinant HOXD9 fused withglutathione S transferase (GST). The fusion construct wasdescribed previously (25). Antibodies were purified by passingsera over a GST affinity column (Bio-Rad, Hercules, CA).Polyclonal antibodies for human basic fibroblast growth factor(bFGF), tumor necrosis factor a (TNFa), interleukin-1b (IL-1b), c-Myc, and c-Fos were obtained from Santa Cruz Biotech-nology (Santa Cruz, CA).

Stable transfection and selection. In a 9 cm–diameterdish, synoviocytes were cotransfected with 10 mg of HOXD9expression plasmid (pSGH4C) or control plasmid without aninsert (pSG5) together with 2 mg of the selective markerplasmid (pH2Rneo). After 40–48 hours, transfected cells werecultured and maintained in a selective medium containing 400mg/ml of G418. Between 2 and 3 weeks later, G418-resistantcells were treated with trypsin and expanded for furtheranalysis. We established 3 OA cell lines and 2 RA cell lines.The expression of HOXD9 in stable transformants was evalu-ated by Western blotting.

Transient transfection and luciferase assay. NIH3T3fibroblasts, OA synoviocytes, or RA synoviocytes were seededinto 6-well plates 24 hours before transfection, and the trans-fection was performed using the calcium phosphate precipita-tion method with the aid of the CellPhect transfection kit(Pharmacia Biotech, Uppsala, Sweden). Cells were cotrans-fected with 3 mg of the reporter plasmid, 3 mg of the expressionconstruct or the control plasmid without an insert, and 1 mg ofpSV2Apap encoding human placental alkaline phosphatase asan internal control for normalizing transfection efficiently (27).Twenty-four hours after transfection, the cells were treatedwith recombinant human bFGF (Genzyme, Cambridge, MA),TNFa (Genzyme), or transforming growth factor b (TGFb)(Sigma, St. Louis, MO). Cells were harvested 48–60 hoursafter transfection, lysed, and assayed for luciferase expressionusing the Luciferase Assay System (Promega, Madison, WI).Luciferase activity was measured by a luminometer (LB9501;Berthold Australia, Bundoora, VIC, Australia). Data werecalculated as the mean 6 SEM of 3 independent assays.

Cell proliferation assay. Nonradioactive cell prolifera-tion assay was performed using the CellTiter 96 Non-Radioactive Cell Proliferation Assay kit (Promega). Briefly,HOXD9 or mock stable transformants were seeded in 96-wellplates in Ham’s F-12 medium at a density of 5 3 103/well.After 24 hours, cultures were replaced with a medium contain-ing 0.5%, 5%, or 10% FCS. The proliferation assay was carriedout at days 0, 3, and 6 of cultures using the protocol providedby the manufacturer. Absorption values were determined at anoptical density of 490 nm using an automatic microtiter plate

1014 KHOA ET AL

reader (Bio-Rad). Data were calculated as the mean 6 SEM of3 independent assays.

Colony formation assay. Colony formation assay wasperformed in a 2-layer soft agar system (31). Briefly, growingsynoviocytes were trypsinized and completely resuspended inDMEM with 10% FCS and containing 0.33% agar (Seaplaque; FMC Bioproducts, Rockland, ME). The cells wereplated in triplicate in 6-well plates containing a solidified layerof 0.6% agar in DMEM with 10% FCS at a density of 1 3104/well. The plates were incubated at 37°C for 4 weeks in ahumidified atmosphere containing 5% CO2. Colonies with atleast 10 cells were then counted and photographed under aninverted microscope.

Western blot analysis. Synoviocytes (;1 3 106) har-vested from cultures were lysed in a lysis buffer containing 50mM Tris, 150 mM NaCl, 1% Nonidet P40, 1 mM phenylmeth-ylsulfonyl fluoride, 10 mg/ml leupeptin, and 10 mg/ml aprotinin.Protein concentration in lysates was quantified by protein assay(Bio-Rad). Lysates were separated by 10–15% sodium dodecylsulfate–polyacrylamide gel electrophoresis and transferredonto nitrocellulose membranes. Nonspecific protein interac-tions were blocked with 2.5% bovine serum albumin and 2.5%skim milk in phosphate buffered saline (PBS). Membraneswere incubated for 60 minutes with a primary antibody andthen incubated for 45–60 minutes with a horseradishperoxidase–labeled anti-rabbit IgG (1:2,000; Santa Cruz Bio-technology). Washing between incubations was performedwith 0.1% PBS–Tween. Bands were visualized using enhancedchemiluminescence (ECL) reagent and exposed to HyperfilmECL (Amersham Life Science, Buckinghamshire, UK). Dilu-tions of the specific antibodies were 1:200 for anti-HOXD9and 1:500 for anti-bFGF, anti-TNFa, anti–IL-1b, anti–c-Fos,and anti–c-Myc.

Statistical analysis. Data were expressed as themean 6 SEM. Differences between groups were examined forstatistical significance using Student’s t-test. P values less than0.05 were considered significant.

RESULTS

HOXD9 protein expression in synoviocytes fromRA and OA patients. First, we examined HOXD9 ex-pression in cultured synoviocytes obtained from 3 RA

and 3 OA patients. HOXD9 expression was detected inRA synoviocytes (Figure 1), while there was no HOXD9expression in OA synoviocytes. HOXD9 was expressedpotently in synoviocytes from RA patients 1 and 3, butonly slightly in those from RA patient 2, suggesting thatthe HOXD9 expression rate differed among RA pa-tients (Figure 1). These results were consistent with ourprevious finding that HOXD9 messenger RNA(mRNA) expression was significantly higher in RA thanin OA synovium.

Establishment of transformant synoviocytesoverexpressing HOXD9. To study the role of HOXD9 insynovial proliferation, we used the method of DNAtransfection by introducing a full cDNA clone ofHOXD9 into OA and RA cultured synoviocytes toestablish transformant synovial cell lines. After 4–5weeks in selective culture media, G418-resistant cellpopulations, HOXD9-transfected synoviocytes, andmock (control) transfectants were established from eachof 3 OA patients and 2 RA patients. HOXD9 proteinexpression of the transformants was evaluated by West-ern blotting with anti-HOXD9 antibody. As shown inFigure 2, HOXD9 protein expression was detected inHOXD9-transfected OA and RA synoviocytes. Littleprotein expression was found in mock-transfected OAsynoviocytes, and HOXD9 protein caused by basal ex-pression was detected in mock-transfected RA synovio-cytes (Figure 2).

Figure 1. Expression of HOXD9 in cultured synoviocytes from pa-tients with rheumatoid arthritis (RA) and osteoarthritis (OA).HOXD9 expression was detected by Western blotting with anti-HOXD9 antibody. Twenty-five micrograms per lane of whole cellextracts from 3 OA patients (extracts OA1, OA2, and OA3) and 3 RApatients (extracts RA1, RA2, and RA3) were loaded and resolved bysodium dodecyl sulfate–polyacrylamide gel electrophoresis followed byWestern blotting with anti-HOXD9 antibody as described in Patientsand Methods. An ;36-kd band was detected as HOXD9 protein.

Figure 2. Detection of HOXD9 protein expression in transformantsynoviocytes by Western blotting. Whole cell lysates (25 mg) fromHOXD9-transfected OA synoviocytes (Clones 1–3 in lanes 1–3, respec-tively) and RA synoviocytes (Clones 1 and 2 in lanes 4 and 5,respectively) were analyzed by Western blotting with anti-HOXD9antibody as described in Patients and Methods. Whole cell lysates (25mg) from mock-transfected OA synoviocytes (Clones 1–3 in lanes 6–8,respectively) and RA synoviocytes (Clones 1 and 2 in lanes 9 and 10,respectively) were also examined. See Figure 1 for definitions.

HOXD9 AND SYNOVIOCYTE PROLIFERATION 1015

Proliferation activity of synoviocytes transfectedwith HOXD9. To evaluate the effect of HOXD9 on thegrowth properties of synoviocytes in monolayer cultures,

we examined the proliferative activity of transfectedsynoviocytes from OA patients under different cultureconditions. Control transfectant cells proliferated poorlyin cultures containing very low concentrations of FCS(0.5%). In contrast, HOXD9-transfected synoviocyteswere able to grow at a significantly higher rate under thiscondition at both time points examined (days 3 and 6)(Figure 3A). At higher concentrations of FCS (5% and10%), the proliferation rate of HOXD9-transfected cellswas still higher than that of control transfectants, al-though this was only significant at day 6 of culture(Figures 3B and C). We confirmed the proliferationactivity of RA synoviocytes under conditions of lowserum concentration and found that cell numbers weresignificantly increased, but there was no significantdifference between HOXD9-transfected and nontrans-fected RA synoviocytes (data not shown).

Anchorage-independent growth of synoviocytesenhanced by HOXD9. To further characterize the growthbehavior of synoviocytes transfected with HOXD9, weevaluated the ability of HOXD9 transformants to formcolonies in soft agar. Unlike RA control transfectants,few colonies or none were formed in cultures of OAcontrol transfectants. However, both OA and RA syno-viocytes transfected with HOXD9 formed numerouscolonies, 3–6-fold more than the respective controls(Table 1 and Figure 4). In addition, microscopic exam-ination showed that many colonies of HOXD9 transfor-mants were larger than those of control transformants(Figure 4).

Increased production of bFGF and c-Fos inHOXD9-transfected synoviocytes. To identify some ofthe factors that might be associated with the overgrowthof synoviocytes following HOXD9 transfection, we ex-amined the levels of several cytokines and protoonco-

Figure 3. Increased proliferation activity of HOXD9-transfected syno-viocytes. HOXD9-transfected synoviocytes or control transfectantswere seeded in 96-well titer plates and serum starved for 24 hours, thenstimulated with A, 0.5%, B, 5%, or C, 10% fetal calf serum (FCS).Nonradioactive proliferation assay was performed at the indicatedtime intervals. Values are the mean 6 SEM of the data obtained from3 HOXD9 transfectants or mock transfectants established from osteo-arthritis synoviocytes. p 5 P , 0.05 versus day 0; pp 5 P , 0.01 versusday 0. OD 5 optical density.

Table 1. Effect of HOXD9 on anchorage-independent growth ofsynoviocytes in colony formation assay*

Synovial cellline

Colonies per 1 3 104 cells

Control transfectant HOXD9 transfectant

OA1 None 58 6 11OA2 12 6 4 74 6 8OA3 9 6 3 47 6 5RA1 45 6 7 132 6 13RA2 39 6 6 156 6 14

* Colony formation assay was performed using 2-layer, semi-solidculture system. HOXD9-transfected or control (mock-transfected)synoviocytes (1 3 104 in each case) derived from 3 osteoarthritis (OA)and 2 rheumatoid arthritis (RA) patients were suspended in 0.3% agarmedium and plated in 6-well plates containing solidified 0.6% agar.Colonies were counted after 4 weeks of culture. Values are the mean 6SEM of 2 independent experiments performed in triplicate for eachcell line.

1016 KHOA ET AL

proteins (bFGF, TNFa, IL-1b, c-Fos, and c-Myc) instably transfected synoviocytes by Western blotting.While the levels of TNFa, IL-1b, and c-Myc did notdiffer between HOXD9 transformants and controls, thelevels of bFGF and c-Fos were considerably increased inHOXD9-transfected cells (Figure 5). This difference wasparticularly evident in control transfectants of OA syno-viocytes, in which the expression of bFGF and c-Fos wasvery low or undetectable. We performed several exper-iments with 3 transformants from OA synoviocytes and 2transformants from RA synoviocytes, and although basalexpression levels were slightly different, elevations ofbFGF and c-Fos expression were reproducibly detected.We showed results typical of the expression of bFGF,

TNFa, IL-1b, c-Fos, and c-Myc obtained from clone 2 ofHOXD9 transformant from OA synoviocytes and clone 2of that from RA synoviocytes. Note that these cloneshad strong HOXD9 expression (Figure 2) and potentcolony-forming ability (Table 1) compared with otherclones.

Different effects of bFGF and TNFa on HOXD9transcriptional activity. Basic FGF, TNFa, and TGFbare abundantly expressed in rheumatoid synovium, andare important cytokines involved in the regulation ofcellular growth (4,5,32,33). Furthermore, bFGF andTGFb also activate the expression of several HOX genes(27,34–36). Accordingly, we compared the effects ofthese cytokines on HOXD9 transcription in cotransfec-

Figure 4. HOXD9 enhancement of capacity of synoviocytes to form colonies in soft agar cultures. Colony formation assay was performed as notedin Table 1. Photographs were taken using an inverted microscope after 4 weeks of culture and represent the results of assays performed with mock(A and C) or HOXD9 (B and D) transfectants from 1 OA patient (A and B) and 1 RA patient (C and D). Note that HOXD9 transfectants formednumerous colonies (arrows), some of which were larger than those formed by mock transfectants. (Original magnification 3 200.) See Figure 1 fordefinitions.

HOXD9 AND SYNOVIOCYTE PROLIFERATION 1017

tion assays in NIH3T3 fibroblasts and synoviocytes usinga luciferase reporter construct carrying HCR, an auto-regulatory element of HOXD9 promoter. To avoid thesecondary effect of constitutively activated bFGF pro-duction by HOXD9 stable transformants, we used atransient transfection system for the reporter assay. Ingeneral, the peak levels of luciferase activity were higherin NIH3T3 cells than in RA and OA synoviocytes,probably due to the higher efficiency of transfection inNIH3T3 cells. However, similar tendencies were ob-served in both cell lines.

As shown in Figure 6, transfection of HCRreporter with or without pSG5 plasmid yielded a lowlevel of luciferase activity, while cotransfection of HCRreporter with HOXD9 expression construct resulted inan activating reporter gene with an average activity;4.5-fold (RA synoviocytes), ;2.5-fold (OA synovio-cytes), or ;8-fold (NIH3T3 cells) higher than the basal

Figure 5. Western blot analysis of cytokine and protooncoproteinexpression in stable transfectant RA and OA synoviocytes (RASC andOASC, respectively). Transfected and selected synoviocytes were lysedin a lysis buffer. After normalization of protein concentration, equalamounts of lysates were separated by sodium dodecyl sulfate–polyacrylamide gel electrophoresis and transferred onto nitrocellulosemembranes. Membranes were incubated with primary antibodies toHOXD9, basic fibroblast growth factor (bFGF), tumor necrosis factora (TNFa), interleukin-1b (IL-1b), anti–c-Fos, and anti–c-Myc, andthen with a horseradish peroxidase–labeled secondary antibody. Bandswere visualized using enhanced chemiluminescence reagent. Note thehigh levels of bFGF and c-Fos in HOXD9-transfected cells (D9)compared with those in control transfectants (C). Results are repre-sentative of experiments performed with transfectants from OA andRA synoviocytes. See Figure 1 for other definitions.

Figure 6. Different effects of bFGF and TNFa on transcriptionalactivity of HOXD9 in transient cotransfection assays. Highly conservedregion (HCR) promoter activity is shown as luciferase reporteractivity, in light units, assayed from extracts of cotransfected RAsynoviocytes (A), OA synoviocytes (B), and NIH3T3 cells (C). In 6-wellplates, cells were transfected with 3 mg of pTHCR luciferase reporterplasmid (HCR1) or pT81luc control plasmid (HCR2) together with 3mg of HOXD9 expression construct (HOXD91) or pSG5 controlplasmid (HOXD92) and 1 mg of pSV2Apap as an internal standard.Basic FGF (10 ng/ml or 20 ng/ml [shaded bars]), TNFa (5 ng/ml or 10ng/ml), or transforming growth factor b (TFGb; 10 ng/ml or 50 ng/ml)were added to the cultures 24 hours after transfection, and the cellswere cultured for an additional 24 hours before harvesting for theluciferase assay. Values are the mean and SEM of 3 independentexperiments. See Figures 1 and 5 for other definitions.

1018 KHOA ET AL

level. The results from NIH3T3 cells were consistentwith the results of Zappavigna et al (25). The addition ofbFGF (10 ng/ml or 20 ng/ml) significantly increasedluciferase activity in HCR/HOXD9 cotransfection (1.5–2.5-fold the level obtained in cotransfection withoutcytokine treatment). In contrast, treatment of cotrans-fected cells with TNFa (5 ng/ml or 10 ng/ml) led to an;2-fold reduction in luciferase activity compared withno cytokine treatment in OA and RA synoviocytes. Theaddition of 10 ng/ml or 50 ng/ml of TGFb to cotrans-fected cells did not have a clear effect on the reporteractivity. None of the cytokines tested in this studychanged the basal level of luciferase activity in cotrans-fection of HCR reporter plasmid with the control plas-mid pSG5 and without the HOXD9 insert (data notshown). These data suggested that bFGF and TNFaregulated HOXD9 transcription in a HOXD9-dependentmanner.

HOXD9 transcription activated by HTLV-Ithrough the HCR sequence. We have recently shownthat HTLV-I infection and tax gene transfection intocultured synoviocytes induce HOXD9 expression (28). In

the present study, we performed cotransfection assaysusing synoviocytes from RA patients to further under-stand the effect of the tax gene on HOXD9 transcription.Interestingly, cotransfection of HCR and tax (pH2R40M)resulted in increased activity of the pTHCR reportereither in the absence or presence of HOXD9 expressionplasmid, although the presence of HOXD9 provided afurther enhancement. Cotransfection of pTHCR orpH2Rneo control plasmid did not affect the basal activ-ity (Figure 7). This finding suggests that the tax geneactivates the reporter gene via the HCR sequence.

DISCUSSION

Like many other HOX genes, HOXD9 is wellknown for its involvement in embryogenesis, especiallyin limb patterning (22,23,37). However, very little isknown about its function in normal and pathologic adulttissues. In a series of recent studies, we have demon-strated preferential expression and activation of HOXgenes in the rheumatoid synovium of adult patients,indicating that these genes are involved in the pathogen-esis of arthritis (27,28). We previously reported thatHOXD9 mRNA was characteristically expressed in RAsynovium (27), and in this study we provide additionalevidence that HOXD9 protein was detected in culturedrheumatoid synoviocytes. Since there was no HOXD9expression in OA synoviocytes, consistent with our pre-vious data from an in vivo study (27), we attempted touse HOXD9 transformants as a useful tool for providingdirect evidence of the implication of HOXD9 in syno-viocyte activation. We established stable transfectants ofHOXD9 in OA synoviocytes and revealed that experi-mentally induced overexpression of HOXD9 not only ledto increased proliferative activity of these cells underboth normal and serum-starvation conditions, but alsoenhanced their capacity to form multiple large coloniesin soft agar cultures. These results support our in vivofindings implicating HOXD9 in synoviocyte activation(27) and clearly indicate that HOXD9 induces the strongproliferation capacity of synoviocytes in vitro.

There are several lines of evidence that HOXgenes promote cellular growth and transformation, ei-ther through perturbation of the cell cycle, which resultsin alterations in cellular behavior (31), or through DNAbinding ability and transcriptional interactions with theirpartners (31,38,39). Given that HOXD9 can activatetranscription factors (25), it is necessary to identify thetarget genes in order to clarify its function.

At present, the putative targets of HOXD9 in-clude other members of the HOX family and adhesion

Figure 7. Human T lymphotropic virus type I (HTLV-I) tax–enhanced transcriptional activity of HOXD9 in transient cotransfectionassays of synoviocytes. RA synoviocytes from 3 patients were trans-fected with 3 mg of pTHCR luciferase reporter plasmid (HCR1) orpT81luc control plasmid (HCR2), with or without 3 mg of pSGH4C(the HOXD9 expression construct; HOXD91) or pSG5 control plas-mid (HOXD92), together with 3 mg of pH2R40M (the HTLV-I taxexpression construct; Tax [shaded bars]) or pH2Rneo control plasmidfor tax (Mock). Cells were harvested for the luciferase assay 48 hoursafter transfection. Values are the mean and SEM of 3 independentexperiments. See Figures 1 and 6 for other definitions.

HOXD9 AND SYNOVIOCYTE PROLIFERATION 1019

molecules such as liver cell adhesion molecule (25,26).In this regard, at least one HOX gene, HOXB7, has beenshown to activate bFGF transcription in melanomasthrough binding to the promoter of bFGF gene (40).Moreover, transduction of the breast carcinoma cell linewith the HOXB7 gene induces bFGF expression andincreases cell proliferation (41). It is also reported thatexpression of HOX1.3 precedes the induction of c-Fosduring mesodermal differentiation of teratocarcinomacells (42). In the case of HOXD9, however, it is still notclear whether it has a direct effect on the regulation ofbFGF and c-Fos, a subject that needs further investiga-tion. Nevertheless, the finding that bFGF and c-Fos,important factors for cell proliferation and transforma-tion, were up-regulated in HOXD9-transfected synovio-cytes is significant in further demonstrating the effect ofHOXD9 on the growth behavior of synoviocytes. Incontrast to bFGF elevation induced by HOXD9 trans-fection, there was little effect on TNFa expression. Weconsider that HOXD9 plays a role in bFGF-inducedproliferation, but not so much so in the TNFa pathway.

When we examined the involvement of HOXD9in synovial proliferation, we also expected to gain abetter insight into the regulatory mechanisms of HOXD9transcription. We recently showed that bFGF up-regulates HOXD9 expression in RA synoviocytes andincreases DNA binding activity of HOXD9 to the C2element in the HOXD9 promoter (27). In the presentstudy, the results of cotransfection assays using a re-porter construct containing the HCR sequence (anotherregulatory element of the HOXD9 promoter [14,25])together with a HOXD9 expression plasmid led us toidentify another regulatory mechanism. These resultsindicate that bFGF might interact with a different regula-tory element involved in HOXD9 transcription, and thatthe effect of bFGF on HCR-mediated transcription ofHOXD9 was HOXD9 dependent. It is already knownthat HOXD9 products can bind to the HCR and auto-regulate its transcription (25). Our data suggests thatbFGF acts as a stimulator for this binding activity. It isinteresting that bFGF seems to be a putative down-stream regulator for this HOX gene. If this autocrineloop really exists in vivo, especially in pathologic condi-tions like arthritis, it would subsequently result in furtherdeterioration of the pathologic condition.

HTLV-I is known as the causative agent ofseveral inflammatory diseases in humans, includingHTLV-I–associated arthropathy, in which the tax gene(an extracellular transcription transactivator not only ofviral, but also of cellular, genes) is considered critical forsynovial proliferation (42,43). Using transient cotrans-

fection assays, we demonstrated in the present study thattax can interact with the HCR sequence and activateHOXD9 transcription, and that tax and HOXD9 havesynergistic effects on activation of HOXD9. These find-ings suggest that tax gene products, unlike bFGF andTNFa, might be able to bind directly to the HCRelement in the HOXD9 promoter and activate HOXD9transcription. Further experiments are necessary to con-firm and clarify such interactions. Studies of transcrip-tional interactions and the DNA binding activity be-tween HOXD9 and tax, as well as studies of some otherpossible cofactors of HOXD9, are currently under way inour laboratory.

In conclusion, although the effect of HOXD9should be clarified by findings from assays of loss offunction of HOXD9, such as those using antisenseHOXD9 oligonucleotides, our present study providesevidence for a potential role of HOXD9 as a novel targetof regulation for synoviocyte activation. We propose thatHOXD9, associated with its regulators and/or targetfactors, is an essential factor in synovial homeostasis.Dysregulation of this HOX gene during pathologic pro-cesses like arthritis may be part of the molecular changesleading to synovial hyperplasia and phenotypic transfor-mation in RA.

ACKNOWLEDGMENTS

The authors thank Dr. Zappavigna and Dr. Hatanakafor kindly providing the pSGH4C, pTHCR, pH2Rneo, andpH2R40M vectors. The authors also thank Dr. F. G. Issa forreviewing the English usage in this manuscript.

REFERENCES

1. Gay S, Gay RE, Koopman WJ. Molecular and cellular mechanismsof joint destruction in rheumatoid arthritis: two cellular mecha-nisms explain joint destruction? Ann Rheum Dis 1993;52 Suppl1:S39–47.

2. Remmers EF, Lafyatis R, Kumkumian GK, Case JP, Roberts AB,Sporn MB, et al. Cytokines and growth regulation of synoviocytesfrom patients with rheumatoid arthritis and rats with streptococcalcell wall arthritis. Growth Factors 1990;2:179–88.

3. Alvaro-Gracia JM, Zvaifler NJ, Firestein GS. Cytokines in chronicinflammatory arthritis. V. Mutual antagonism between interferon-gamma and tumor necrosis factor-alpha on HLA-DR expression,proliferation, collagenase production, and granulocyte macro-phage colony-stimulating factor production by rheumatoid arthri-tis synoviocytes. J Clin Invest 1990;86:1790–8.

4. Goddard DH, Grossman SL, Williams WV, Weiner DB, Gross JL,Eidsvoog K, et al. Regulation of synovial cell growth: coexpressionof transforming growth factor b and basic fibroblast growth factorby cultured synovial cells. Arthritis Rheum 1992;35:1296–303.

5. Qu Z, Garcia CH, O’Rourke LM, Planck SR, Kohli M, Rosen-baum JT. Local proliferation of fibroblast-like synoviocytes con-tributes to synovial hyperplasia: results of proliferating cell nuclear

1020 KHOA ET AL

antigen/cyclin, c-myc, and nucleolar organizer region staining.Arthritis Rheum 1994;37:212–20.

6. Trabandt A, Gay RE, Gay S. Oncogene activation in rheumatoidsynovium. APMIS 1992;100:861–75.

7. Aicher WK, Heer AH, Trabandt A Jr, Bridges SL, Schroeder HWJr, Stransky G, et al. Overexpression of zinc-finger transcriptionfactor Z-225/Erg-1 in synoviocytes from rheumatoid arthritis pa-tients. J Immunol 1994;152:5940–8.

8. Fassbender HG. Histomorphologic basis of articular cartilagedestruction in rheumatoid arthritis. Coll Relat Res 1983;3:58–64.

9. Lafyatis R, Remmers EF, Roberts AB, Yocum DE, Sporn MB,Wilder RL. Anchorage-independent growth of synoviocytes fromarthritic and normal joints: stimulation by exogenous platelet-derived growth factor and inhibition by transforming growthfactor-beta. J Clin Invest 1989;83:1267–76.

10. Firestein GS. Invasive fibroblast-like synoviocytes in rheumatoidarthritis: passive responders or transformed aggressors? ArthritisRheum 1996;39:1781–90.

11. Trabandt A, Aicher WK, Gay RE, Sukhatme VP, Nilson-HamiltonM, Hamilton RT, et al. Expression of the collagenolytic andRas-induced cysteine proteinase cathepsin L and proliferation-associated oncogenes in synovial cells of MRL/I mice and patientswith rheumatoid arthritis. Matrix 1990;10:349–61.

12. Muller-Ladner U, Kriegsmann J, Gay RE, Gay S. Oncogenes inrheumatoid arthritis. Rheum Dis Clin North Am 1995;21:675–90.

13. Zappavigna V, Sartori D, Mavilio F. Specificity of Hox proteinfunction depends on DNA-protein interactions, both mediated bythe homeodomain. Genes Dev 1994;8:732–44.

14. Renucci A, Zappavigna V, Zakany J, Izpisua-Belmonte JC, BurkiK, Duboule D. Comparison of mouse and human HOX-4 com-plexes defines conserved sequences involved in the regulation ofHox-4.4. EMBO J 1992;11:1459–68.

15. Krumlauf R. Hox genes in vertebrate development. Cell 1994;78:191–201.

16. Quaranta MT, Petrini M, Tritarelli E, Samoggia P, Care A,Bottero L, et al. HOXB cluster genes in activated natural killerlymphocytes: expression from 39–59 cluster side and proliferativefunction. J Immunol 1996;157:2462–9.

17. Care A, Testa U, Bassani A, Tritarelli E, Montesoro E, SamoggiaP, et al. Coordinate expression and proliferative role of HOXBgenes in activated adult T lymphocytes. Mol Cell Biol 1994;14:4872–7.

18. Magli MC, Largman C, Lawrence HJ. Effects of HOX homeoboxgenes in blood cell differentiation. J Cell Physiol 1997;173:168–77.

19. Cillo C, Barba P, Freschi G, Bucciarelli G, Magli MC, BoncinelliE. HOX gene expression in normal and neoplastic human kidney.Int J Cancer 1992;51:892–7.

20. De Vita G, Barba P, Odartchenko N, Givel JC, Freschi G,Bucciarelli G, et al. Expression of homeobox-containing genes inprimary and metastatic colorectal cancer. Eur J Cancer 1993;29A:887–93.

21. Thorsteinsdottir U, Sauvageau G, Humphries RK. Hox homeoboxgenes as regulators of normal and leukemic hematopoiesis. He-matol Oncol Clin North Am 1997;11:1221–37.

22. Dolle P, Izpisua-Belmonte JC, Falkenstein H, Renucci A,Duboule D. Coordinate expression of the murine Hox-5 complexhomeobox-containing genes during limb pattern formation. Na-ture 1989;342:767–72.

23. Dolle P, Duboule D. Two gene members of the murine HOX-5complex show regional and cell-type specific expression in devel-oping limbs and gonads. EMBO J 1989;8:1507–15.

24. Morgan BA, Tabin C. Hox genes and growth: early and late rolesin limb bud morphogenesis. Dev Suppl 1994:181–6.

25. Zappavigna V, Renucci A, Izpisua-Belmonte JC, Urier G, PeschleC, Duboule D. HOX4 genes encode transcription factors withpotential auto- and cross-regulatory capacities. EMBO J 1991;10:4177–87.

26. Goomer RS, Holst BD, Wood IL, Jones FS, Edelman GM.Regulation in vitro of an L-CAM enhancer by homeobox genesHoxd9 and HNF-1. Proc Natl Acad Sci U S A 1994;91:7985–9.

27. Xue C, Hasunuma T, Asahara H, Yin W, Maeda T, Fujisawa K, etal. Transcriptional regulation of the HOX4C gene by basic fibro-blast growth factor on rheumatoid synovial fibroblasts. ArthritisRheum 1997;40:1628–35.

28. Khoa ND, Hasunuma T, Kobata T, Kato T, Nishioka K. Expres-sion of murine HOXD9 during embryonic joint patterning and inhuman T lymphotropic virus type I tax transgenic mice witharthropathy resembling rheumatoid arthritis. Arthritis Rheum1999;42:686–96.

29. Goto M, Sasano M, Yamanaka H, Miyasaka N, Kamatani N,Inoue K, et al. Spontaneous production of an interleukin 1-likefactor by cloned rheumatoid synovial cells in long-term culture.J Clin Invest 1987;80:786–96.

30. Tanaka A, Takahashi C, Yamaoka S, Nosaka T, Maki M, Ha-tanaka M. Oncogenic transformation by the tax gene of humanT-cell leukemia virus type I in vitro. Proc Natl Acad Sci U S A1990;87:1071–5.

31. Krosl J, Baban S, Krosl G, Rozenfeld S, Largman C, Sauvageau G.Cellular proliferation and transformation induced by HOXB4 andHOXB3 proteins involves cooperation with PBX1. Oncogene1998;16:3403–12.

32. Melnyk VO, Shipley GD, Sternfeld MD, Sherman L, RosenbaumJT. Synoviocytes synthesize, bind, and respond to basic fibroblastgrowth factor. Arthritis Rheum 1990;33:493–500.

33. Qu Z, Huang XZ, Ahmadi P, Andresevic J, Plank SR, Hart CE, etal. Expression of basic fibroblast growth factor in synovial tissuesfrom patients with rheumatoid arthritis and degenerative jointdisease. Lab Invest 1995;73:339–46.

34. King MW, Moore MJ. Novel HOX, POU and FKH genesexpressed during bFGF-induced mesodermal differentiation inXenopus. Nucleic Acids Res 1994;22:3990–6.

35. Kloen P, Visker MH, Olijve W, van Zoelen EJ, Boersma CJ. Celltype-specific modulation of Hox gene expression by members ofthe TGF-beta superfamily: a comparison between human osteo-sarcoma and neuroblastoma cell line. Biochem Biophys ResCommun 1997;233:365–9.

36. Boudreau N, Andrews C, Srebrow A, Ravanpay A, Cheresh DA.Induction of the angiogenic phenotype by Hox D3. J Cell Biol1997;139:257–64.

37. Nelson CE, Morgan BA, Burke AC, DiMambro E, Murtaugh LC,Gonzales E, et al. Analysis of Hox gene expression in the chicklimb bud. Development 1996;122:1449–66.

38. Kroon E, Krosl J, Thorsteinsdottir U, Baban S, Buchberg AM,Sauvageau G. Hoxa9 transforms primary bone marrow cells throughspecific collaboration with Meis1a but not Pbx1b. EMBO J 1998;17:3714–25.

39. Qin XF, Luo Y, Suh H, Wayne J, Misulovin Z, Roeder RG, et al.Transformation by homeobox genes can be mediated by selectivetranscriptional repression. EMBO J 1994;13:5967–76.

40. Care A, Silvani A, Meccia E, Mattia G, Stoppacciaro A, ParmianiG, et al. HOXB7 constitutively activates basic fibroblast growthfactor in melanomas. Mol Cell Biol 1996;16:4842–51.

41. Care A, Silvani A, Meccia E, Mattia G, Peschle C, Colombo MP.Transduction of the SkBr3 breast carcinoma cell line with theHOXB7 gene induces bFGF expression, increases cell prolifera-tion and reduces growth factor dependence. Oncogene 1998;16:3285–9.

42. Nakajima T, Aono H, Hasunuma T, Yamamoto K, Maruyama I,Nosaka T, et al. Overgrowth of human synovial cells driven by thehuman T cell leukemia virus type I tax gene. J Clin Invest1993;92:186–93.

43. Nishioka K. HTLV-I arthropathy and Sjogren syndrome. J AcquirImmune Defic Syndr Hum Retroviral 1996;13 Suppl 1:S57–62.

HOXD9 AND SYNOVIOCYTE PROLIFERATION 1021