Overexpression of SPACIA1/SAAL1, a newly identified gene that is involved in synoviocyte proliferation, accelerates the progression of synovitis in mice and humans




To identify novel genes associated with dysregulated proliferation of activated synovial fibroblasts, which are involved in arthritic joint destruction.


We performed transcriptome analysis to identify genes that were up-regulated in the foot joints of mice with collagen-induced arthritis (CIA). The effect of candidate genes on proliferation of synovial fibroblasts was screened using antisense oligodeoxynucleotides and small interfering RNAs (siRNAs). We characterized the expression and function of a novel gene, synoviocyte proliferation–associated in collagen-induced arthritis 1 (SPACIA1)/serum amyloid A–like 1 (SAAL1) using antibodies and siRNA and established transgenic mice to examine the effect of SPACIA1/SAAL1 overexpression in CIA.


Human and mouse SPACIA1/SAAL1 encoded 474 amino acid proteins that shared 80% homology. SPACIA1/SAAL1 was primarily expressed in the nucleus of rheumatoid arthritis (RA) synovial fibroblasts and was highly expressed in the hyperplastic lining of inflamed synovium. In addition, its expression level in RA- or osteoarthritis (OA)–affected synovial tissue was positively correlated with the thickness of the synovial lining. Furthermore, SPACIA1/SAAL1 siRNA inhibited the proliferation of synovial fibroblasts, especially tumor necrosis factor α–induced synovial fibroblasts, by blocking entry into the S phase without inducing apoptosis. Finally, transgenic mice overexpressing SPACIA1/SAAL1 exhibited early onset and rapid progression of CIA.


These results suggest that SPACIA1/SAAL1 is necessary for abnormal proliferation of synovial fibroblasts and its overexpression is associated with the progression of synovitis in mice and humans. Thus, therapy targeting SPACIA1/SAAL1 might have potential as an inhibitor of synovial proliferation in RA and/or OA.

Synovitis is a common characteristic of rheumatoid arthritis (RA) and knee osteoarthritis (OA). The major pathologic features of synovitis are hyperplasia of the synovial lining, inflammatory cell infiltration, and high stromal cell density (1). Activated synovial fibroblasts, which are a major component of synovial lining hyperplasia, are important in the pathogenesis of synovitis because they secrete cytokines and chemokines, leading to the exacerbation of inflammation. They also produce matrix metalloproteinases and cathepsins, which destroy bone and cartilage (2). As a result, the activated synovial fibroblasts form pannus, a type of granulation tissue that erodes the joint (3). Therefore, reducing the number of activated synovial fibroblasts is a promising therapeutic strategy for arthritis. For example, adenoviral gene transfer of cyclin-dependent kinase inhibitors, such as p16 or p21, inhibits synovial cell proliferation and has demonstrated high therapeutic efficacy in animal models of RA (4, 5). Similarly, intraarticular injection of anti-Fas IgM monoclonal antibodies induces apoptosis in synoviocytes and infiltrating lymphocytes, which leads to improvement of RA (6).

Although these findings suggest that synoviocyte proliferation is important in the pathogenesis of RA, the exact molecular mechanism of the disease is not known. To clarify the mechanism of the abnormal proliferation of synovial fibroblasts in RA, we used gene expression and functional analyses to seek novel genes that are involved in this process. Herein we describe a novel gene, synoviocyte proliferation–associated in collagen-induced arthritis 1 (SPACIA1)/serum amyloid A–like 1 (SAAL1), which was found to be up-regulated in the foot joints of mice with collagen-induced arthritis (CIA). Knockdown of this gene inhibited the proliferation of human RA synovial fibroblasts (RASFs) in vitro. Finally, we demonstrated that overexpression of SPACIA1/SAAL1 is associated with the progression of synovitis in mice and humans.


Ethical considerations.

All human and animal experimental protocols in this study (nos. 443 and 31M0912T2, respectively) were approved by the Ethics Review Committee of St. Marianna University School of Medicine. Written informed consent was obtained from all patients prior to collection of joint tissue samples.


Male DBA/1J mice were purchased from Japan SLC. DBA/1J mice and transgenic mice were raised under conventional conditions at our facilities. Tap water and food were provided ad libitum.

Induction and assessment of CIA.

For the microarray experiments, 10 DBA/1J mice (7–8 weeks old) were immunized twice with 100 μg bovine type II collagen (CII), as described previously (7). Seven days after the second injection of CII, collagen-injected mice and control mice were killed and foot joints were removed from their forepaws.

To induce CIA in SPACIA1-overexpressing mice, CII was administered as described above, except that the amount of CII administered in the second injection was reduced to 50 μg. Subsequently, the mice were assessed 3 times per week, and an arthritis score was assigned based on the grading system described by Hughes et al (8). In addition, blood samples were collected from the tail vein once per week to measure anti-CII antibody titers. Forty-two days after the first injection, the mice were weighed and killed, and samples of their blood, spleen, and knee joints were collected.

DNA microarray.

Total RNA was extracted and pooled from the foot joint samples from the mice with CIA and from the control mice. Messenger RNA (mRNA) was isolated using the PolyATtract mRNA Isolation System (Promega). Double-stranded complementary DNA (cDNA) was synthesized from 1 μg mRNA, followed by preparation of biotinylated complementary RNA (cRNA) with the BioArray HighYield RNA Transcript Labeling Kit (Affymetrix). Biotin-labeled cRNA, which was fragmented according to the Affymetrix procedure, was hybridized to Murine Genome U74v2 microarrays. Subsequently, these microarrays were stained with streptavidin–phycoerythrin and scanned with a GeneArray scanner. The intensity data from each array were normalized using Microarray Suite version 5.0 software. To determine the log2 ratio of signal intensities, we compared the normalized data from mice with CIA to the data from the control group, using Data Mining Tool version 3.0 software. The microarrays, scanner, and software sets were all obtained from Affymetrix, and all of the annotation data for the identified probe sets were obtained from the NetAffx Analysis Center (http://www.affymetrix.com/analysis/index.affx).

Design and use of antisense oligodeoxynucleotides (ODNs) and small interfering RNAs (siRNAs) in cell proliferation assays.

To screen candidate genes, we designed antisense phosphorothioated ODNs near the start codon, using Oligo software (Molecular Biology Insights). We also designed siRNAs, using the design program on the Takara Bio web site (http://www.takara-bio.co.jp/rnai/intro.htm). The ODNs and siRNAs were chemically synthesized at Hokkaido System Science. Their sequences are shown in Supplementary Tables 1 and 2 (available on the Arthritis & Rheumatism web site at http://onlinelibrary.wiley.com/journal/10.1002/(ISSN)1529-0131).

RASFs were isolated and cultured as described previously (9). Twenty-four hours before transfection, the cells were trypsinized, seeded on a 24-well plate, and cultured overnight. To determine the effect of antisense ODNs and siRNAs on the proliferation of RASFs, we transfected 100 nM antisense ODN or 100 nM siRNA into the cells with Oligofectamine (Invitrogen) and then cultured the cells for 66 hours or 96 hours, respectively. Subsequently, we used the Cell Counting Kit-8 assay (Dojindo) to determine the relative number of viable cells.

Similarly, to determine the effect of SPACIA1 siRNA on RASF proliferation induced by serum or tumor necrosis factor α (TNFα), we used SAAL1 (also known as SPACIA1) siRNA, c-fos siRNA, and negative control siRNA (On-Target plus SMARTpool; Dharmacon). RASFs were then trypsinized and seeded in 48-well plates. After 18 hours, 33 nM siRNA was transfected into the cells with Lipofectamine 2000 (Invitrogen) in Opti-MEM medium. Four hours later, the culture medium was changed to Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum (FBS) or 1% FBS and 5 ng/ml TNFα. Finally, 96 hours after transfection, we determined the relative number of viable RASFs, using the Cell Counting Kit-8 assay.

Cloning and expression of human SPACIA1 and preparation of anti-SPACIA1 antibodies.

We designed the following polymerase chain reaction (PCR) primers on the basis of the sequence of human SAAL1, which is identical to SPACIA1 (GenBank Refseq NM_138421; http://www.ncbi.nlm.nih.gov/genbank/): for the first PCR, hSPACIA1+ 5′-AAAG- TCATGGACCGCAAC-3′, hSPACIA1− 5′-CCAATTCAG- GTTTTAAGTCTGAAC-3′; for the second PCR, hSPACIA1+/Not I 5′-ATATGCGGCCGCCGATGGA- CCGCAACCCCTCG-3′, hSPACIA1−/Xho I 5′-TCACTCG- AGGTTTTAAGTCTGAACCTTC-3′. Human RASF cDNA was used as a template for PCR. The product was cloned into a pcDNA3-FLAG vector, which was constructed by inserting the FLAG sequence into pcDNA3 (Invitrogen). To create a GST-SPACIA1 fusion protein, the SPACIA1 gene was subcloned into a pGEX-6P vector (GE Healthcare) and then transformed into Escherichia coli BL21 cells. The transfected cells were cultured for 4 hours at 30°C in the presence of 0.1 mM isopropyl β-D-1-thiogalactopyranoside. Subsequently, the cells were sonicated, and the recombinant human SPACIA1 protein was purified from the lysate with a glutathione Sepharose 4B column (GE Healthcare) and PreScission protease cleavage. Finally, the purified human SPACIA1 protein was used as an antigen to generate polyclonal antibodies and monoclonal antibodies (pAb and mAb, respectively), using standard methods as previously described (10, 11).

Western blotting.

Synovial fibroblasts from patients with RA and synovial tissue from patients with OA and RA were lysed in lysis buffer (20 mM Tris [pH 7.5], 150 mM NaCl, 0.5% Nonidet P40, 1 mM EDTA, 1 mM dithiothreitol, 5 mM NaF, 0.2 mM Na3VO4, and protease inhibitors). Then whole cell lysates (10 μg) or tissue lysates (20 μg) were electrophoresed on 10% sodium dodecyl sulfate polyacrylamide gels and transferred onto activated Immobilon-P polyvinylidene fluoride membranes (Millipore) using a wet transfer method. After blocking with 5% nonfat dry milk in phosphate buffered saline (PBS), the membranes were incubated with anti-human SPACIA1 mAb (clone 1Ac; 1:500 dilution), followed by incubation with horseradish peroxidase (HRP)–labeled anti-mouse IgG antibody (1:20,000 dilution; The Binding Site) for 1 hour at room temperature, and detection with Immobilon Western Chemiluminescent HRP Substrate (Millipore). The signal intensities of the bands specific for SPACIA1 were quantified using ImageJ software (National Institutes of Health; http://rsbweb.nih.gov/ij/). To investigate the expression of SPACIA1 in various human tissues, we used a human multiple tissue blot (Calbiochem). The detection method was the same as described above, except the primary antibody was anti-human SPACIA1 pAb (1:8,000 dilution) and the secondary antibody was HRP-labeled anti-rabbit IgG antibody (1:20,000 dilution).


To determine the intracellular localization of SPACIA1 in RASFs, we performed immunofluorescence staining as previously described (12). Anti-human SPACIA1 mAb (clone 1Ac) and normal mouse IgG were used as the primary antibody and negative control, respectively. The secondary antibody was a rhodamine-conjugated anti-mouse IgG antibody (Millipore). Nuclei were stained with DAPI.


Human synovial tissue specimens and mouse knee joints were fixed in 10% formalin and then embedded in paraffin. To retrieve antigens, deparaffinized tissue sections (5 μm thick) were microwaved for 5 minutes and soaked in 1 mM EDTA solution (pH 8.0) for 40 minutes at 90°C. After blocking of endogenous peroxidase activity for 10 minutes in 3% methanol, the sections were incubated with anti-human SPACIA1 pAb (1.5 μg/ml), normal rabbit IgG (1.5 μg/ml), anti-CD14 mAb (1:150 dilution; Leica), anti-CD163 mAb (1:150 dilution; Leica), or normal mouse IgG (1 μg/ml) for 30 minutes at room temperature. Subsequently, the sections were rinsed and visualized by immunoperoxidase staining with the Vectastain ABC-PO kit (Vector) and 3,3′-diaminobenzidine tetrahydrochloride substrate. Mayer's hematoxylin was used as a counterstain. Normal rabbit IgG and normal mouse IgG were used as negative controls.

Cell cycle phase analysis.

To analyze the effect of SPACIA1 siRNA on the cell cycle of RASFs, we used SPACIA1 siRNA and mock siRNA. The target sequences were as follows: human SPACIA1 3′-untranslated region siRNA 5′-GAAUUACUUCUGUACAAGAA-3′, mock siRNA (negative control) 5′-UAAGGCUAUGAAGAGAUAC-3′. Ninety-six hours after transfection of siRNA into RASFs, the cells were harvested, washed, and suspended in PBS with 0.1% Triton X-100. The suspension was filtered through nylon mesh to remove aggregates. Subsequently, RNase and propidium iodide were added to the suspension, which was analyzed on a FACSCalibur flow cytometer (BD Biosciences). The results were analyzed using FlowJo software (Tree Star).

Annexin V–based apoptosis assay.

To evaluate apoptotic cells, we used the same siRNAs that were used for the cell cycle analysis and analyzed RASFs 96 hours after transfection. RASFs that were treated with 10 μM staurosporine for 3 hours and cultured for 12 hours were used as a positive control. Staurosporine-treated or siRNA-transfected RASFs were stained with fluorescein isothiocyanate (FITC)–labeled annexin V (MBL) using a previously described immunofluorescence staining procedure (12), and the percentage of annexin V–positive RASFs in 3 microscopic fields (200×) was calculated.

Bromodeoxyuridine (BrdU) incorporation assay.

For measurement of BrdU incorporation, we used SPACIA1 siRNA, negative control siRNA, and PCNA siRNA (proliferating cell nuclear antigen) (On-Target plus SMARTpool). RASFs were trypsinized and then seeded on a 6-well plate (1.4 × 105 cells/well). After 18 hours, 33 nM siRNA was transfected into the cells with Lipofectamine 2000. Twenty-four hours later, 10 μM aphidicolin was added to the culture media for 24 hours to synchronize the cells at the G1 phase. Then the aphidicolin was removed and DNA synthesis was determined by measurement of BrdU incorporation for 8 hours. Subsequently, the cells were fixed, permeabilized, treated with DNase I, and stained for 1 hour with an FITC-labeled anti-BrdU antibody (BD Biosciences). Fluorescence intensity was measured with a Cellomics ArrayScan VTI HCS reader (Thermo Fisher Scientific).

Generation of SPACIA1-overexpressing mice.

On the basis of the sequence of mouse SAAL1 (GenBank Refseq NM_030233), we designed the following primers: for the first PCR, mSPACIA1+ 5′-CATCGGCATGGATCGAAAC–3′, mSPACIA1− 5′-ATGCCAACTTCGAACCATAG-3′; for the second PCR, mSPACIA1+/Mun I 5′-ATACAATTGA- TGGATCGAAACCCGTCTC-3′, mSPACIA1−/Xho I 5′-TAACTCGAGTTAAGTCTGTGCCTTCAC-3′. PCR was performed with a mouse embryo (17th day of gestation) cDNA as a template. The PCR product was cloned into the pcDNA3-FLAG vector, and the FLAG-tagged mouse SPACIA1 gene was subcloned into the mammalian expression vector pJC13-1 (a kind gift from Dr. Suming Huang, University of Florida, Gainesville) with a CAG promoter and β-globin insulators (13). The linearized FLAG-tagged mouse SPACIA1 pJC13-1 expression vector was injected into the pronucleus of fertilized eggs from C57BL/6 mice (Macrogen). To confirm the overexpression of mouse SPACIA1, genomic DNA and protein were extracted from tails and used for PCR and Western blotting analyses, respectively. Finally, the SPACIA1-overexpressing mice were backcrossed onto the DBA/1J strain for 7 generations.

Histopathologic examination.

Human synovial tissue specimens were obtained from patients with RA (n = 9) or OA (n = 9) and processed as described previously (14). Synovitis was scored on a 0–3 scale using the grading system of Krenn et al (1).

Mouse knee joints were removed 42 days after CIA induction, fixed in 10% formalin, and embedded in paraffin. Deparaffinized tissue sections (5 μm thick) were stained with hematoxylin and eosin. The sections were scored on a scale of 0–4 under blinded conditions, according to the degree of hyperplasia in the synovial lining, mononuclear cell infiltration, and pannus formation, as described previously (15).

Statistical analysis.

Fisher's exact test was used to analyze the correlation between expression levels of SPACIA1 and pathologic features of synovitis. The significance of RASF proliferation was analyzed by one-way analysis of variance and Tukey's post hoc test. The chi-square test (or Fisher's exact test) was used to compare the incidence of arthritis between transgenic and wild-type mice. Arthritis scores and histopathologic scores in transgenic and wild-type mice were compared by Mann-Whitney U test. All other statistical analyses were performed using Student's unpaired t-test. P values less than 0.05 were considered significant.


Identification of novel genes associated with synoviocyte proliferation.

In the transcriptome of the foot joints of mice with CIA, we detected 80 genes (among ∼36,000 mRNA probes) that were up-regulated >8-fold compared with those of healthy control mice (Figure 1A). By transfecting human RASFs with antisense ODNs specific for the human homologs of these 80 genes (see Supplementary Table 1, available at http://onlinelibrary.wiley.com/journal/10.1002/(ISSN)1529-0131), we identified the top 15 genes ranked by their proliferative inhibition rate. Subsequently, by using siRNA against these genes (Supplementary Table 2 http://onlinelibrary.wiley.com/journal/10.1002/(ISSN)1529-0131), we reduced the number of genes associated with synoviocyte proliferation to 8 candidates (Figure 1). Among them, there were 2 genes with unknown functions. We named these genes synoviocyte proliferation–associated in collagen-induced arthritis 1 (SPACIA1) (GenBank accession no. AB489136 [human] and AB489137 [mouse]) and SPACIA2 (GenBank accession no. AB541014 [human isoform 1], AB541015 [human isoform 2], and AB541013 [mouse]). The proliferation of RASFs was most strongly inhibited by SPACIA1 siRNA, followed by lactotransferrin (LTF), proteinase 3 (PRTN3), glycine amidinotransferase (GATM), SPACIA2, placenta-specific gene 8 (PLAC8), peptidoglycan recognition protein 1 (PGLYRP1), and interleukin 6 (IL6). Therefore, we focused upon the role of SPACIA1.

Figure 1.

Identification of SPACIA1 by whole-genome microarray analysis and functional screening. A, Flow chart of the procedure that was used to select 8 genes from >36,000 mouse transcripts. Eighty genes were up-regulated >8-fold in the foot joints of mice with collagen-induced arthritis (CIA). Among these genes, 15 genes that inhibited the proliferation of human rheumatoid arthritis synovial fibroblasts (RASFs) were identified with antisense oligodeoxynucleotides (ODNs). Finally, 8 candidate genes were isolated using small interfering RNAs (siRNAs) that were targeted against these 15 genes. B, Rate of proliferation of RASFs that were transfected with siRNA targeted against the candidate genes. The results are expressed as the percentage of the proliferation rate of RASFs that were transfected with green fluorescence protein (GFP) siRNA (negative control). Small interfering RNA against synoviolin (SYVN), an E3 ubiquitin ligase that we previously identified as a pathogenic factor in arthropathy (7), was used as a positive control. All experiments were performed in triplicate. Values are the mean ± SD.

Characterization of human and mouse SPACIA1.

According to the NCBI nucleotide database (http://www.ncbi.nlm.nih.gov/nuccore), SPACIA1 is identical to serum amyloid A–like 1 (SAAL1). The human and mouse SPACIA1/SAAL1 genes are localized to chromosomes 11p15.1 and 7B4, respectively. These loci contain a gene cluster of the serum amyloid A (SAA) superfamily, which includes SAA1, SAA2, SAA3, and SAA4 (16). The human and mouse SPACIA1/SAAL1 proteins are 82% and 80% homologous at the nucleotide and amino acid levels, respectively (Figure 2). Furthermore, the NCBI HomoloGene database (http://www.ncbi.nlm.nih.gov/homologene) indicates that the sequence of the SPACIA1/SAAL1 gene (unique identifier no. 34706) is conserved from zebrafish to humans. Although SPACIA/SAAL1 does not have any signal peptides, transmembrane domains, classic nuclear localization signals, or nuclear export signals, the InterPro database (http://www.ebi.ac.uk/interpro/) indicates that it has an armadillo-type motif (InterPro accession no. IPR016024) between amino acids 82 and 211 (Figure 2). In addition, previous proteomic studies showed that human and mouse SPACIA1/SAAL1 are phosphoproteins (Figure 2).

Figure 2.

Amino acid sequence alignment of mouse and human SPACIA1. The alignment was generated using the ClustalW2 multiple sequence alignment program (http://www.ebi.ac.uk/Tools/clustalw2/). Identical residues are marked with asterisks. Colons and periods indicate conserved substitutions and semiconserved substitutions, respectively. Shaded regions indicate armadillo repeats (InterPro accession no. IPR016024). Boxed amino acids indicate phosphorylation sites.

The SAA protein family includes acute-phase proteins that are secreted in response to inflammation (19). Two members of this family, SAA1 and SAA2, are precursors of amyloid A, which causes amyloidosis secondary to RA (20). Thus, SAA plays an important role in inflammatory conditions, such as RA. However, the amino acid sequence of SPACIA1 was not very similar to that of SAA; SPACIA1 and SAA1 are only 27% homologous. Moreover, unlike SAA proteins, which have a signal peptide and a conserved SAA domain, human and mouse SPACIA1/SAAL1 do not have either of these characteristics. As a result, we refer to SPACIA1/SAAL1 below as SPACIA1.

Expression analysis of human SPACIA1.

Human SPACIA1 is a 55-kd protein that is expressed in RASFs (Figure 3A) and is predominantly localized to the nuclei (Figure 3B). In addition, SPACIA1 was strongly expressed in the testis and ovary, but only weakly expressed in the lung, spleen, and heart (Figure 3C). A 40-kd protein in the heart and a 30-kd protein in the pancreas were also detected with the anti-SPACIA1 pAb, but these results may have been due to nonspecific binding. Although SPACIA1 was also expressed abundantly in the synovial tissue of patients with RA or OA (Figure 3D), we did not observe any specific pattern in the expression levels of SPACIA1 in RA or OA synovial tissue. However, the expression level of SPACIA1 was significantly correlated with the thickness of the synovial lining, but not with the density of stromal cells or inflammatory cell infiltrates (Table 1). As expected from these results, SPACIA1 was strongly expressed in the hyperplastic synovial lining and moderately expressed in synovial stromal cells, endothelial cells, and plasma cells (Figure 3E). Interestingly, the localization of SPACIA1-positive cells was the opposite of that of CD14- or CD163-positive cells, which are considered to be synovial macrophages (Figure 3F).

Figure 3.

Expression analysis of SPACIA1 in human tissue. A, Western blotting of RASFs that were transfected with control or SPACIA1 siRNA; an anti-SPACIA1 monoclonal antibody (mAb) was used for detection. B, Subcellular localization of SPACIA1 in RASFs. Immunofluorescence staining was performed with an anti-SPACIA1 mAb and normal mouse IgG. The images are representative of 3 separate experiments. C, Western blotting of various human tissues with an anti-SPACIA1 polyclonal antibody. D, Comparison of SPACIA1 expression levels in osteoarthritis (OA) and RA synovial tissue, by Western blotting. The membrane was stained with ponceau S to ensure equal protein loading in all lanes. The intensity of the SPACIA1-specific bands was quantified by densitometry. Another set of Western blots also showed no significant differences between RA and OA (data not shown). E, Expression of SPACIA1 in human synovium with or without hyperplasia of the synovial lining. SPACIA1-positive cells are brown. The images are representative of the results obtained in experiments with specimens from 5 different patients with RA or OA. Original magnification × 100. F, Localization of SPACIA1, CD163, and CD14 in serial sections of RA synovial membrane. Arrowheads show SPACIA1-positive, CD163-negative, and CD14-negative cells. Arrows show a capillary blood vessel. Original magnification × 200. See Figure 1 for other definitions.

Table 1. Correlation between expression level of SPACIA1 and pathologic features of synovitis in patients with RA or OA*
FeatureLow SPACIA1 expression, no. of patientsHigh SPACIA1 expression, no. of patientsP
  • *

    Synovial tissue samples from 9 patients with rheumatoid arthritis (RA) and 9 patients with osteoarthritis (OA) were classified as having high or low expression of SPACIA1 based on the intensity of SPACIA1-specific bands shown on Western blotting. Synovial lining thickness, stromal cell density, and inflammatory cell infiltration were scored according to the synovitis grading system described by Krenn et al (1).

  • Versus score of 0–1, by Fisher's exact test.

Synovial lining thickness   
 Score 0–1 (≤4 layers)9 (5 RA, 4 OA)2 (0 RA, 2 OA) 
 Score 2–3 (>4 layers)07 (4 RA, 3 OA)0.0011
Stromal cell density   
 Score 0–1 (normal–mild)8 (4 RA, 4 OA)6 (1 RA, 5 OA) 
 Score 2–3 (moderate–severe)1 (1 RA, 0 OA)3 (3 RA, 0 OA)0.2882
Inflammatory cell infiltration   
 Score 0–1 (normal–mild)5 (2 RA, 3 OA)2 (0 RA, 2 OA) 
 Score 2–3 (moderate–severe)4 (3 RA, 1 OA)7 (4 RA, 3 OA)0.1674

Antiproliferative effect of SPACIA1 siRNA in synovial fibroblasts.

As shown in Figure 4A, SPACIA1 siRNA inhibited the proliferation of RASFs that were stimulated with 10% FBS. However, this effect was significantly weaker than that of the positive control, c-fos siRNA, which strongly inhibits the proliferation of RASFs (21). On the other hand, the inhibitory effect of SPACIA1 siRNA on the proliferation of RASFs that were stimulated with TNFα was comparable to that of c-fos siRNA (Figure 4B). Furthermore, flow cytometric analysis demonstrated that transfection of SPACIA1 siRNA into RASFs increased the number of cells in the G0/G1 phase and reduced the number of cells in the S and G2/M phase compared with mock siRNA transfection (Figure 4C). A BrdU incorporation assay showed that the percentage of BrdU-positive RASFs was 24%, 12%, and 4% after transfection with mock, SPACIA1, or PCNA siRNA, respectively (Figure 4D). Moreover, the percentage of annexin V–positive RASFs was similar after transfection with SPACIA1 and mock siRNA (Figure 4E).

Figure 4.

Antiproliferative effect of SPACIA1 siRNA on RASFs by partial inhibition of entry into the S phase. A and B, SPACIA1 siRNA suppressed the proliferation of RASFs in response to 10% fetal bovine serum (FBS) (A) or 5 ng/ml tumor necrosis factor α (TNFα) (B); c-fos siRNA was used as a positive control. Data are representative of 5 separate experiments in A and 3 separate experiments in B. Values are the mean ± SD. ∗ = P < 0.05; ∗∗ = P < 0.01; ∗∗∗ = P < 0.001. NS = not significant. C, After culture of RASFs in medium supplemented with 10% FBS, SPACIA1 siRNA reduced the number of RASFs in the S and G2/M phase, compared with mock siRNA. Representative histograms from 3 different experiments are shown. D, After culture of RASFs in medium supplemented with 10% FBS, SPACIA1 siRNA significantly inhibited the entry of RASFs into the S phase in a bromodeoxyuridine (BrdU) incorporation assay, compared with mock siRNA. PCNA siRNA was used as a positive control. Values are the mean ± SD (n = 3 experiments). E, After culture of RASFs in medium supplemented with 10% FBS, SPACIA1 siRNA did not induce apoptosis in an annexin V apoptosis assay. RASFs that were treated with staurosporine (STS) for 3 hours and cultured for 12 hours were used as a positive control. Values are the mean ± SD. Experiments in A, B, D, and E were performed in triplicate. See Figure 1 for other definitions.

Progression of CIA in transgenic mice overexpressing SPACIA1.

A schematic representation of the mammalian expression construct that was used to create transgenic mice overexpressing SPACIA1 is shown in Figure 5A. The overexpression of SPACIA1 in the transgenic mice was confirmed by Western blotting (Figure 5B) and PCR (results not shown). These transgenic mice did not spontaneously develop arthritis or cancer. Since findings of a preliminary experiment suggested that SPACIA1-transgenic mice are more sensitive to CIA than wild-type mice, we reduced the amount of CII that was used to induce arthritis, to more easily detect the difference in their sensitivity to CIA. Despite this modification, the incidence of CIA in the transgenic mice increased significantly more rapidly than in the wild-type mice, particularly at 30–37 days after the first injection of collagen (Figure 5C). Furthermore, the arthritis score in the paws of transgenic mice with CIA was significantly higher than that of wild-type mice at 30–39 days after the first injection (Figure 5D), and the histopathologic scores of the knee joints of the transgenic mice were significantly higher than in the wild-type mice (Figure 5E). Finally, immunohistochemistry was performed to examine the expression levels of SPACIA1 protein in the joints of the mice with CIA (Figure 5F). The number of SPACIA1-positive cells was markedly increased in the joints of both wild-type and transgenic mice with CIA.

Figure 5.

Early onset and rapid progression of collagen-induced arthritis (CIA) in transgenic (Tg) mice overexpressing SPACIA1. A, Schematic representation of the transgene for overexpression of mouse SPACIA1. B, Immunoblot (IB) analysis of lysates from tail tissue of wild-type (WT) and Tg mice, using an anti-FLAG antibody. C, Development of arthritis in 23 WT and 23 Tg mice after injection of bovine type II collagen (CII) on days 0 and 21. Arthritis incidence was significantly higher in Tg mice than in WT mice on days 30–37. ∗ = P < 0.05 versus WT mice. D, Arthritis severity in 23 WT and 23 Tg mice after injection of CII on days 0 and 21. Arthritis scores were significantly higher in Tg mice than in WT mice on days 30–39. Values are the mean ± SD. ∗ = P < 0.05; # = P < 0.01, versus WT mice. E, Histopathologic scores in the knee joints of 23 WT and 23 Tg mice after injection of CII on days 0 and 21. F, Expression of SPACIA1 protein in synovial tissue from the knee joints of WT and Tg mice with or without CIA. Original magnification × 100.


In this study of ∼36,000 mRNA transcripts, we identified a novel gene, SPACIA1, that is involved in the dysregulated proliferation of synovial fibroblasts in the foot joints of mice with CIA. However, since the forepaw tissue samples included not only synovium but also bone, cartilage, muscle, and tendon tissue, the transcriptome analysis was not specific for genes that are expressed in the synovium. To overcome this problem, we used human RASFs to screen the candidate genes that were identified from the transcriptome analysis. We found that SPACIA1 was expressed not only in cultured RASFs (Figure 3A and B) but also in the inflamed joints of mice and humans (Figures 3E and 5F). Notably, SPACIA1 was strongly expressed in synovial fibroblasts, and not in synovial macrophages, in the intimal layer of the synovial membrane of human RA patients (Figure 3F). Furthermore, its expression in human synovial tissue was not specific to RA, as was demonstrated by its expression in synovial tissue from patients with OA as well (Figure 3D). Similar to RA, synovitis commonly occurs in end-stage OA (22). In fact, 90% of joints from OA patients undergoing arthroplasty contain pannus-like tissue on the articular surface (23). As a result, SPACIA1 may be expressed at comparable levels in OA and RA synovia.

Transgenic mice overexpressing SPACIA1 exhibited earlier onset and more rapid progression of CIA than wild-type mice (Figures 5C–E). In addition, the expression level of SPACIA1 was positively correlated with the thickness of the synovial lining in humans (Table 1). These results suggest that overexpression of SPACIA1 accelerates the progression of synovitis by promoting synovial cell proliferation. However, it is unlikely that SPACIA1 overexpression affected the immune system, because we did not observe any significant difference in the ratio of spleen weight to body weight between the wild-type and transgenic mice with CIA (data not shown). In addition, there were no significant differences in anti-CII antibody titers in serum from the wild-type and transgenic mice with CIA (data not shown). With regard to the use of transgenic mice overexpressing FLAG-tagged mouse SPACIA1, there is both a benefit and a limitation that should be considered. The benefit is that FLAG-tagged mouse SPACIA1 is easy to distinguish from endogenous mouse SPACIA1. The limitation is that we were not able to identify any potential side effects of the FLAG tag when we evaluated these transgenic mice. However, to minimize this potential effect, we confirmed that there are no significant domains in the N-terminal region of SPACIA1, e.g., a signal peptide.

Since SPACIA1 knockdown inhibited synoviocyte proliferation, our results indicate that SPACIA1 should be implicated in the abnormal proliferation of synovial fibroblasts in synovitis due to an antiapoptotic effect or stimulation of cell cycle progression. In this regard, although SPACIA1 knockdown inhibited the entry of RASFs into the S phase (Figures 4C and D), it did not induce apoptosis in RASFs (Figure 4E). In addition, our ongoing studies indicate that SPACIA1 knockdown halves the expression of the genes for cyclin E2 and cyclin-dependent kinase 2, which are involved in the progression of the cell cycle from G1 to S (Sato T, et al: unpublished observations). Therefore, it is more likely that SPACIA1 regulates the progression of the cell cycle than that it regulates apoptosis in RASFs.

What is the possible mechanism of action of SPACIA1? When we screened the molecules that interact with SPACIA1 using a yeast 2-hybrid assay, most of the molecules obtained were transcription regulatory factors that are located in the nucleus (data not shown). This suggests that SPACIA1 might act as a transcriptional regulator by forming a complex with certain molecules in the nucleus. We next considered the setting in which SPACIA1 acts. Because serum contains a variety of growth factors and hormones, the addition of serum to the culture medium represents a general stimulation that leads to proliferation of cells, whereas administration of TNFα represents an inflammatory stimulation that induces cell proliferation. Since SPACIA1 knockdown was more efficient at blocking TNFα-induced proliferation than serum-induced proliferation (Figures 4A and B), SPACIA1 might be especially involved in the inflammatory signaling pathways that induce the proliferation of RASFs. Based on our findings, we speculate that SPACIA1 may act downstream of inflammatory signaling and regulate the transcription of cell cycle–implicated genes, which then regulate the expression of cyclin E2 and cyclin-dependent kinase 2. However, further investigation is needed to clarify the precise mechanism of action of SPACIA1, e.g., whether it affects any of the major signaling pathways, including the NF-κB or JNK pathway.

Moreover, there is in vivo evidence that the effect of SPACIA1 on synoviocyte proliferation is limited to inflammatory conditions. The transgenic mice overexpressing SPACIA1 did not spontaneously develop arthritis or cancer, and administration of a collagen emulsion was still required for induction of synoviocyte proliferation. Therefore, unlike typical anticancer drugs, an SPACIA1 inhibitor may be able to specifically suppress abnormal synoviocyte proliferation in inflammatory environments with relatively few side effects.

In conclusion, the overexpression of SPACIA1, a novel protein that is involved in the proliferation of synovial fibroblasts, is associated with the progression of synovitis in mice and humans. Consequently, SPACIA1 might be a potential therapeutic target for inhibiting synovial proliferation in RA and OA.


All authors were involved in drafting the article or revising it critically for important intellectual content, and all authors approved the final version to be published. Dr. Fujii had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Study conception and design. Fujii, Konomi, Nishioka, Nakajima.

Acquisition of data. Sato, Fujii, Konomi, Yagishita, Aratani, Araya.

Analysis and interpretation of data. Sato, Fujii, Konomi, Yagishita, Aratani, Araya, Aono, Yudoh, Suzuki, Beppu, Yamano, Nishioka, Nakajima.


Santen Pharmaceutical collaborated with St. Marianna University School of Medicine in this study. Employees of Santen Pharmaceutical contributed to the study design, data collection, data analysis, and writing of the manuscript; however, Santen Pharmaceutical was not involved in the decision to submit the manuscript for publication or in approval of the content of the submitted manuscript, and publication of the manuscript was not contingent upon the approval of Santen Pharmaceutical.


We would like to thank N. Watanabe-Asakura, K. Takahashi, N. Yamamoto, A. Une, N. Furuya, S. Shinkawa, Y. Nakagawa, K. Suzuki, S. Asada, T. Sato-Mogi, H. Ogasawara, Y. Sato, Y. Urbanczyk, M. Yamanashi, and M. Ishikawa for excellent technical assistance.