**5. Fibroblast-like synoviocytes and mesenchymal stem cells**

FLSs are characterized mainly by *in vitro* analyses*.* The synovial membranes are easily obtained by joint surgery. The cells liberated from the synovial tissues by treatment with collagenase can be cultured under the appropriate conditions. Although primary FLSs are useful, experiments must be carefully designed, because the composition of the cells in the culture changes after 4 passages, when contaminated hematopoietic cells disappear (Zimmermann *et al.*, 2001).

The source of pathogenic FLSs proliferating in the synovium of RA is an intriguing issue, and several possibilities can be considered. Growth of FLSs can be stimulated by adjacent hematopoietic cells in the microenvironment, or initiated by the acquisition of cell-intrinsic properties for unregulated growth. Alternatively, growing FLSs can be derived from resident FLSs in the normal synovial membrane, or migrated from other organs. The latter possibility was demonstrated by experiments to inject FLSs from human TNF- transgenic mice into the knee joint, and to transplant human synovial fibroblasts into SCID mice (Aidinis *et al.*, 2003; Lefevre *et al.*, 2009).

Fibroblast-like cells that initiate growth during the very early stages of RA can originate from mature FLSs or from other mesenchymal cells at the primitive stage, such as mesenchymal stem cells (MSCs). The presence of MSCs in the synovium has been reported by several researchers. MSCs with the potential to differentiate into 3 lineages, osteogenic, adipogenic, and chondrogenic cells, were obtained from the synovial membrane following digestion with collagenase and more than 3 passages (De Bari *et al.*, 2001). Fibroblast-like MSCs expressing fibroblast marker D7-FIB, but not CD45, were detected in the synovial fluid (Jones *et al.*, 2004). The number of these MSCs was lower in the synovial fluid of RA than in that of OA. In addition to the synovium, MSCs have been derived from blood (Zvaifler *et al.*, 2000), adipose tissue (Zuk *et al.*, 2002), and the periosteal region (De Bari *et al.*, 2006). Although not genuine MSCs, circulating CD14+ monocytes may contain progenitors with the potential to differentiate into mesenchymal cells (Kuwana *et al.*, 2003).

In terms of the underlying mechanism for the transformation-like phenotype of FLSs, and the involvement of MSCs, Li & Makarov (2006) reported intriguing data from animal models of RA. Arthritic FLSs contained a substantial fraction of bone marrow-derived precursors, with the ability to differentiate *in vitro* into various mesenchymal cell types. However, inflammation prevented multilineage differentiation. The transcription factor NF- B played a key role in repressing osteogenic and adipogenic differentiation of arthritic FLSs. On the other hand, specific activation of NF-B profoundly enhanced proliferation, motility, and matrix-degrading activity of FLSs by the induction of MMPs. These data suggest an intriguing mechanism, namely, that arthritic FLSs are bone marrow-derived MSCs, which are arrested during the early stages of differentiation, by the activation of NF- B induced by inflammatory cytokines (Li & Makarov, 2006).

#### **6. Search for RA-related genes through the classification of fibroblast-like synoviocytes**

#### **6.1 Subtypes of fibroblast-like synoviocytes in RA**

Kasperkovitz *et al.* (2005) reported that subtypes of FLSs in RA differ in their gene expression. Complementary DNA microarrays of the synovial tissues and cultured FLSs obtained from RA patients revealed that the gene expression profiles of high- and low-grade

FLSs are characterized mainly by *in vitro* analyses*.* The synovial membranes are easily obtained by joint surgery. The cells liberated from the synovial tissues by treatment with collagenase can be cultured under the appropriate conditions. Although primary FLSs are useful, experiments must be carefully designed, because the composition of the cells in the culture changes after 4 passages, when contaminated hematopoietic cells disappear

The source of pathogenic FLSs proliferating in the synovium of RA is an intriguing issue, and several possibilities can be considered. Growth of FLSs can be stimulated by adjacent hematopoietic cells in the microenvironment, or initiated by the acquisition of cell-intrinsic properties for unregulated growth. Alternatively, growing FLSs can be derived from resident FLSs in the normal synovial membrane, or migrated from other organs. The latter possibility was demonstrated by experiments to inject FLSs from human TNF- transgenic mice into the knee joint, and to transplant human synovial fibroblasts into SCID mice

Fibroblast-like cells that initiate growth during the very early stages of RA can originate from mature FLSs or from other mesenchymal cells at the primitive stage, such as mesenchymal stem cells (MSCs). The presence of MSCs in the synovium has been reported by several researchers. MSCs with the potential to differentiate into 3 lineages, osteogenic, adipogenic, and chondrogenic cells, were obtained from the synovial membrane following digestion with collagenase and more than 3 passages (De Bari *et al.*, 2001). Fibroblast-like MSCs expressing fibroblast marker D7-FIB, but not CD45, were detected in the synovial fluid (Jones *et al.*, 2004). The number of these MSCs was lower in the synovial fluid of RA than in that of OA. In addition to the synovium, MSCs have been derived from blood (Zvaifler *et al.*, 2000), adipose tissue (Zuk *et al.*, 2002), and the periosteal region (De Bari *et al.*, 2006). Although not genuine MSCs, circulating CD14+ monocytes may contain progenitors

In terms of the underlying mechanism for the transformation-like phenotype of FLSs, and the involvement of MSCs, Li & Makarov (2006) reported intriguing data from animal models of RA. Arthritic FLSs contained a substantial fraction of bone marrow-derived precursors, with the ability to differentiate *in vitro* into various mesenchymal cell types. However, inflammation prevented multilineage differentiation. The transcription factor NF- B played a key role in repressing osteogenic and adipogenic differentiation of arthritic FLSs. On the other hand, specific activation of NF-B profoundly enhanced proliferation, motility, and matrix-degrading activity of FLSs by the induction of MMPs. These data suggest an intriguing mechanism, namely, that arthritic FLSs are bone marrow-derived MSCs, which are arrested during the early stages of differentiation, by the activation of NF-

**6. Search for RA-related genes through the classification of fibroblast-like** 

Kasperkovitz *et al.* (2005) reported that subtypes of FLSs in RA differ in their gene expression. Complementary DNA microarrays of the synovial tissues and cultured FLSs obtained from RA patients revealed that the gene expression profiles of high- and low-grade

with the potential to differentiate into mesenchymal cells (Kuwana *et al.*, 2003).

B induced by inflammatory cytokines (Li & Makarov, 2006).

**6.1 Subtypes of fibroblast-like synoviocytes in RA** 

**synoviocytes** 

**5. Fibroblast-like synoviocytes and mesenchymal stem cells** 

(Zimmermann *et al.*, 2001).

(Aidinis *et al.*, 2003; Lefevre *et al.*, 2009).

inflammation synovial tissues were characterized by high and low expression of genes of immune-competent cells (T cells, B cells, and APCs), respectively. Furthermore, hierarchical clustering identified 2 groups of FLSs, characterized by distinctive gene expression profiles and correlation with the inflammatory profiles of the synovial tissues. The first group correlated with the high-grade inflammation tissue, and exhibited increased expression of a TGF-/activin A-inducible gene profile, which is characteristic of myofibroblasts, a cell type involved in wound healing. The second group correlated with the low-grade inflammation tissue, and showed increased expression of the genes involved in autocrine growth regulation, cell transformation, complement activation, and oxidative stress. Reflecting the gene expression profile, an increased proportion of myofibroblast-like cells in the heterogeneous population of FLSs were immunohistochemically detected in the high-grade inflammation tissue. These data suggest that the inflammatory state of the synovium is determined by the composition of heterogeneous FLSs.

#### **6.2 Transformed fibroblast-like synoviocyte lines reveal heterogeneity irrespective of arthritis types**

The data of Kasperkovitz *et al.* (2005), Galligan *et al*. (2007) and others indicate that combining gene expression profiling with other parameters, such as clinical data or characteristics of FLS lines, constitutes a powerful tool for identifying novel disease-related genes. To identify the cell-intrinsic abnormalities of RA-FLSs, we established transformed cell lines from the synovium of RA or OA cases, by immortalization with SV40 large T Ag (unpublished data of Ishihara *et al.*). Characterization of FLSs from 2 types of arthritis revealed no significant differences in surface molecules, growth rates, patterns of tyrosinephosphorylated proteins, or expression of the genes related to inflammation (*IL-1, IL-6, MMP-1, MMP-3*, etc.). Since the expression levels of these genes vary (ranges exceeding 1,000-fold) among FLS lines from each type of arthritis, we tentatively categorized them into 2 subtypes reflecting resting (r) and active (a) stages, based on the expression levels of *IL-1* and *MMP-1*. Next, we performed a micro DNA array to obtain the gene expression profiles for 4 representative cell lines, r-OA-FLS, a-OA-FLS, r-RA-FLS, and a-RA-FLS, and obtained 10 gene clusters. Although no disease-specific clusters were obtained, 2 reciprocal, stagespecific clusters were detected, suggesting the validity of our hypothesis for the presence of subtypes in FLSs. Using these data we are presently searching for 2 types of candidate genes; master genes that determine the states of FLSs, and genes that could play a role in the pathophysiology of RA by inference based on our current understanding of FLSs. In the following sections, we will review the potential roles of activation-induced cytidine deaminase (*AID*) (Igarashi *et al.*, 2010) and the *A20/ABIN* family.

#### **6.3 Ectopic expression of** *AID* **and acquisition of a tumor-like phenotype by fibroblastlike synoviocytes**

#### **6.3.1** *P53* **mutation in fibroblast-like synoviocytes of RA and** *AID* **expression in inflammation**

In addition to the properties described above, the expression of the tumor-suppressor gene *p53* with somatic mutations (Firestein *et al.*, 1997; Inazuka *et al.*, 2000; Kullmann *et al.*, 1999; Reme *et al.*, 1998; Yamanishi *et al.*, 2002), and the down-regulation of the tumor suppressor *PTEN*, a protein phosphatase gene, have been demonstrated in RA-FLSs (Pap *et al.*, 2000b).

Molecular Mechanisms of Rheumatoid Arthritis

inflammatory cytokines *IL-6* and *IL-1*

compared it with the intact *p53* gene sequence.

phenotypes of FLSs in RA.

**the B-cell follicles** 

3 levels, or medication.

Revealed by Categorizing Subtypes of Fibroblast-Like Synoviocytes 83

*AID* transcription is not induced by autocrine cytokines. No clear relationship was observed between aberrant expression of *AID* and other clinical parameters, such as age, serum MMP-

The mutations of the *p53* tumor-suppressor gene frequently found in RA-FLSs could contribute to the tumor-like, and also the pro-inflammatory properties of RA-FLSs, such as aggressive growth, invasion, and destruction of cartilage and bone (Firestein *et al.*, 1997; Inazuka *et al.*, 2000; Kullmann *et al.*, 1999; Reme *et al.*, 1998; Sun *et al.*, 2004; Yamanishi *et al.*, 2002). Although genotoxic and oxidative stresses have been speculated to be causative candidates for the somatic mutation in the *p53* gene in RA-FLSs, the molecular mechanism has not yet been elucidated. As mentioned in 6.3.1, a clear relationship between *AID* expression and the frequency of *p53* somatic mutations has been demonstrated in some non-B lymphocytes, such as hepatocytes and colon epithelial cells (Chan-On *et al.*, 2009; Endo *et al.*, 2008; Komori *et al.*, 2008; Kou *et al.*, 2007; Morisawa *et al.*, 2008). Thus, we speculated that aberrant expression of *AID* might be involved in the introduction of the *p53* gene mutation. We amplified the coding region of *p53* from 3 *AID*+ RA-FLS cell lines with high-fidelity polymerase. We then determined the nucleotide sequence corresponding to that region and

*AID*+ RA-FLSs harbored approximately 2- to 3.5-fold more mutations than the control RA-FLS subsets, which expressed *AID* at a lower level. In addition, the frequency of non-silent mutations was 3 times more than that of silent mutations. Notably, the base substitution pattern in *p53* was biased toward the transition type, which is typical for AID-mediated mutations at the variable region of the immunoglobulin gene (Di Noia & Neuberger, 2007). The mutations were distributed intensively at the DNA-binding domain of the *p53* gene, where the hotspot of somatic mutations is found in some malignant tumors. The Arg248 mutation, one of the cancer hotspot mutations (Ko & Prives, 1996), was found in *p53* from our *AID*+ RA-FLSs. In addition, among the amino acid mutations that we identified, 17% were identical to those previously reported. A further 33% were distinct amino acid mutations; however, the positions of base change were located in the same codons as previously reported. The apparent correlation between ectopic expression of *AID* and increased frequency of somatic mutations of *p53* strongly suggests that AID may be involved in the introduction of mutations to *p53*. Such mutations could lead to reductions or increases in the function of p53, which in turn may result in the tumor-like or anti-apoptotic

**6.3.4 AID is produced by non-transformed RA-FLSs and in the RA synovium outside** 

The aberrant expression of *AID* in some RA-FLS transformed cell lines is not caused by the effects of transformation with SV40 large T Ag. Indeed, 3 to 8 times higher transcription levels of *AID* were observed, even in non-transformed primary FLS cell lines (4 out of 11 RA-FLSs, but none of the 6 OA-FLSs). In addition, cyto-staining with anti-AID antibody revealed a positive signal in *AID*-expressing primary RA-FLSs. Furthermore, dual-color immunohistostaining of the synovial sections from *AID*+ RA patients clearly demonstrated

did not correlate with that of *AID*, suggesting that

, or the pro-

(Endo *et al.*, 2007, 2008; Pauklin *et al.*, 2009). The transcription levels of *TNF-*

**6.3.3 Accumulation of** *p53* **gene mutations in** *AID***-expressing RA-FLSs** 

In particular, the somatic mutation of the *p53* gene appears consistent, not only in terms of increased resistance to apoptosis, but also with respect to pro-inflammatory responses such as production of IL-6 and MMP-1 (Han *et al.*, 1999; Sun *et al.*, 2004; Yamanishi *et al.*, 2005). However, little is known about the mechanism by which the somatic mutations are introduced into the *p53* gene in RA-FLSs.

AID is a member of the APOBEC family, which is a cellular cytidine deaminase involved in protection from retroviral infection or regulation of cholesterol metabolism (Goila-Gaur & Strebel, 2008). AID was originally identified as an indispensable molecule for somatic hypermutation at the immunoglobulin variable region, and also for class-switch recombination in germinal center B lymphocytes (Di Noia & Neuberger, 2007; Honjo *et al.*, 2004). Recently, several investigators have demonstrated up-regulation of *AID* in nonlymphoid tumor cells such as breast cancer, cholangiocarcinoma, hepatoma, and colorectal cancer cells (Babbage *et al.*, 2006; Chan-On *et al.*, 2009; Endo *et al.*, 2007, 2008; Komori *et al.*, 2008; Kou *et al.*, 2007; Morisawa *et al.*, 2008). During the process of oncogenesis, NF-B activation in inflammation is thought to be important for aberrant expression of *AID*. For example, the infection of gastric mucosal cells with *Helicobacter pylori*, or of hepatocytes with hepatitis C virus, activates NF-B and successfully induces local production of pro-inflammatory cytokines such as TNF- and IL-1. Together, these secreted cytokines also activate NF-B, and lead to the induction of AID. In fact, stimulation with TNF- or IL-1 induces *AID* expression even in non-tumor hepatocyte or colon epithelial cells. Moreover, the somatic mutations of *p53* found in these cancer cells appeared to be a direct target of AID (Endo *et al.*, 2008; Kou *et al.*, 2007; Takai *et al.*, 2009). RA is characterized by an environment rich in pro-inflammatory cytokines and the existence of mutations in the *p53* gene. Thus, under chronic inflammatory circumstances, it is possible that aberrant expression of *AID* could introduce mutations into the *p53* gene of FLSs.

#### **6.3.2 Aberrant expression of** *AID* **in RA-FLSs**

First, we assessed the expression of the *AID* gene in the transformed FLS cell lines described in 6.2, by real-time reverse transcription polymerase chain reaction (RT-PCR). *AID* was transcribed in more than half of the RA-FLS cell lines (5 out of 9) and in none of the OA-FLS cell lines. Quantitative assay by RT-PCR showed 7- to 18-fold higher *AID* transcription in the RA-FLS lines compared to the OA-FLS lines that expressed a low but detectable level of *AID* transcription. The possibility of contaminated signals from *AID*-expressing B cells was excluded by the absence of pan B cell marker transcription. The translation of AID was further confirmed by the detection of protein in the cell lysate from RA-FLSs, with western blot analysis.

Patients who provided *AID*-expressing FLSs showed a tendency toward higher levels (approximately 2.7 times) of CRP in the serum. Regarding gender, the number of female patients with *AID*+ FLSs was approximately 1.9 times higher than the number of male patients. Although our data are not statistically significant because of the small sample numbers used, it appears that *AID* expression in FLSs is facilitated under conditions of inflammation in female patients. Indeed, we observed that estrogen, a representative female hormone, or TNF-, a representative pro-inflammatory cytokine, augmented the transcription of *AID* in *AID*+ RA-FLSs to more than 20-fold higher levels compared with the basal levels in OA-FLSs. These results are similar to those previously reported for other cells

In particular, the somatic mutation of the *p53* gene appears consistent, not only in terms of increased resistance to apoptosis, but also with respect to pro-inflammatory responses such as production of IL-6 and MMP-1 (Han *et al.*, 1999; Sun *et al.*, 2004; Yamanishi *et al.*, 2005). However, little is known about the mechanism by which the somatic mutations are

AID is a member of the APOBEC family, which is a cellular cytidine deaminase involved in protection from retroviral infection or regulation of cholesterol metabolism (Goila-Gaur & Strebel, 2008). AID was originally identified as an indispensable molecule for somatic hypermutation at the immunoglobulin variable region, and also for class-switch recombination in germinal center B lymphocytes (Di Noia & Neuberger, 2007; Honjo *et al.*, 2004). Recently, several investigators have demonstrated up-regulation of *AID* in nonlymphoid tumor cells such as breast cancer, cholangiocarcinoma, hepatoma, and colorectal cancer cells (Babbage *et al.*, 2006; Chan-On *et al.*, 2009; Endo *et al.*, 2007, 2008; Komori *et al.*, 2008; Kou *et al.*, 2007; Morisawa *et al.*, 2008). During the process of oncogenesis, NF-B activation in inflammation is thought to be important for aberrant expression of *AID*. For example, the infection of gastric mucosal cells with *Helicobacter pylori*, or of hepatocytes with hepatitis C virus, activates NF-B and successfully induces local production of pro-inflammatory cytokines such as TNF- and IL-1. Together, these secreted cytokines also activate NF-B, and lead to the induction of AID. In fact, stimulation with TNF- or IL-1 induces *AID* expression even in non-tumor hepatocyte or colon epithelial cells. Moreover, the somatic mutations of *p53* found in these cancer cells appeared to be a direct target of AID (Endo *et al.*, 2008; Kou *et al.*, 2007; Takai *et al.*, 2009). RA is characterized by an environment rich in pro-inflammatory cytokines and the existence of mutations in the *p53* gene. Thus, under chronic inflammatory circumstances, it is possible

that aberrant expression of *AID* could introduce mutations into the *p53* gene of FLSs.

First, we assessed the expression of the *AID* gene in the transformed FLS cell lines described in 6.2, by real-time reverse transcription polymerase chain reaction (RT-PCR). *AID* was transcribed in more than half of the RA-FLS cell lines (5 out of 9) and in none of the OA-FLS cell lines. Quantitative assay by RT-PCR showed 7- to 18-fold higher *AID* transcription in the RA-FLS lines compared to the OA-FLS lines that expressed a low but detectable level of *AID* transcription. The possibility of contaminated signals from *AID*-expressing B cells was excluded by the absence of pan B cell marker transcription. The translation of AID was further confirmed by the detection of protein in the cell lysate from RA-FLSs, with western

Patients who provided *AID*-expressing FLSs showed a tendency toward higher levels (approximately 2.7 times) of CRP in the serum. Regarding gender, the number of female patients with *AID*+ FLSs was approximately 1.9 times higher than the number of male patients. Although our data are not statistically significant because of the small sample numbers used, it appears that *AID* expression in FLSs is facilitated under conditions of inflammation in female patients. Indeed, we observed that estrogen, a representative female hormone, or TNF-, a representative pro-inflammatory cytokine, augmented the transcription of *AID* in *AID*+ RA-FLSs to more than 20-fold higher levels compared with the basal levels in OA-FLSs. These results are similar to those previously reported for other cells

introduced into the *p53* gene in RA-FLSs.

**6.3.2 Aberrant expression of** *AID* **in RA-FLSs** 

blot analysis.

(Endo *et al.*, 2007, 2008; Pauklin *et al.*, 2009). The transcription levels of *TNF-*, or the proinflammatory cytokines *IL-6* and *IL-1* did not correlate with that of *AID*, suggesting that *AID* transcription is not induced by autocrine cytokines. No clear relationship was observed between aberrant expression of *AID* and other clinical parameters, such as age, serum MMP-3 levels, or medication.

#### **6.3.3 Accumulation of** *p53* **gene mutations in** *AID***-expressing RA-FLSs**

The mutations of the *p53* tumor-suppressor gene frequently found in RA-FLSs could contribute to the tumor-like, and also the pro-inflammatory properties of RA-FLSs, such as aggressive growth, invasion, and destruction of cartilage and bone (Firestein *et al.*, 1997; Inazuka *et al.*, 2000; Kullmann *et al.*, 1999; Reme *et al.*, 1998; Sun *et al.*, 2004; Yamanishi *et al.*, 2002). Although genotoxic and oxidative stresses have been speculated to be causative candidates for the somatic mutation in the *p53* gene in RA-FLSs, the molecular mechanism has not yet been elucidated. As mentioned in 6.3.1, a clear relationship between *AID* expression and the frequency of *p53* somatic mutations has been demonstrated in some non-B lymphocytes, such as hepatocytes and colon epithelial cells (Chan-On *et al.*, 2009; Endo *et al.*, 2008; Komori *et al.*, 2008; Kou *et al.*, 2007; Morisawa *et al.*, 2008). Thus, we speculated that aberrant expression of *AID* might be involved in the introduction of the *p53* gene mutation. We amplified the coding region of *p53* from 3 *AID*+ RA-FLS cell lines with high-fidelity polymerase. We then determined the nucleotide sequence corresponding to that region and compared it with the intact *p53* gene sequence.

*AID*+ RA-FLSs harbored approximately 2- to 3.5-fold more mutations than the control RA-FLS subsets, which expressed *AID* at a lower level. In addition, the frequency of non-silent mutations was 3 times more than that of silent mutations. Notably, the base substitution pattern in *p53* was biased toward the transition type, which is typical for AID-mediated mutations at the variable region of the immunoglobulin gene (Di Noia & Neuberger, 2007). The mutations were distributed intensively at the DNA-binding domain of the *p53* gene, where the hotspot of somatic mutations is found in some malignant tumors. The Arg248 mutation, one of the cancer hotspot mutations (Ko & Prives, 1996), was found in *p53* from our *AID*+ RA-FLSs. In addition, among the amino acid mutations that we identified, 17% were identical to those previously reported. A further 33% were distinct amino acid mutations; however, the positions of base change were located in the same codons as previously reported. The apparent correlation between ectopic expression of *AID* and increased frequency of somatic mutations of *p53* strongly suggests that AID may be involved in the introduction of mutations to *p53*. Such mutations could lead to reductions or increases in the function of p53, which in turn may result in the tumor-like or anti-apoptotic phenotypes of FLSs in RA.

#### **6.3.4 AID is produced by non-transformed RA-FLSs and in the RA synovium outside the B-cell follicles**

The aberrant expression of *AID* in some RA-FLS transformed cell lines is not caused by the effects of transformation with SV40 large T Ag. Indeed, 3 to 8 times higher transcription levels of *AID* were observed, even in non-transformed primary FLS cell lines (4 out of 11 RA-FLSs, but none of the 6 OA-FLSs). In addition, cyto-staining with anti-AID antibody revealed a positive signal in *AID*-expressing primary RA-FLSs. Furthermore, dual-color immunohistostaining of the synovial sections from *AID*+ RA patients clearly demonstrated

Molecular Mechanisms of Rheumatoid Arthritis

currently under investigation.

**7. Conclusion** 

**8. Acknowledgment** 

Revealed by Categorizing Subtypes of Fibroblast-Like Synoviocytes 85

These observations indicate that there is heterogeneity of RA-FLSs in the responsiveness to TNF- stimulation and suggest that these "inhibitors" might not play negative regulatory roles in RA-FLS. The precise mechanism, cell-lineage specificity, disease specificity, and significance in cell biology of this unexpected possible positive role for A20/ABINs are

Anti-cytokine therapy for RA is a prominent achievement in the field of autoimmune diseases. Accumulated evidence from clinical and basic medical research indicates pivotal roles for FLS in the pathogenesis and pathophysiology of RA. Data from genome-wide screening, transcriptional profiling, and animal models indicate that RA consists with heterogeneous disease subsets. Together with several other researchers, we have presented evidence for heterogeneity in FLS. Based on this finding, we have successfully searched for disease-related genes by subtyping FLS. We have identified 2 groups of genes, *AID* and *A20/ABINs*. *AID* is involved in the irreversible transformation of FLS, whereas *A20/ABINs* participate in the reversible, but potentially harmful, responsiveness of them. Both groups of genes are constituent elements for distinct levels of heterogeneity in FLS, which may be involved in resistance to anti-cytokine therapies. Subtyping of FLS based on expression of AID did not coincide with that based on responsiveness to signal-utilizing NF-B, which is reasonable because RA is a multi-factorial disease. We believe that our approach to categorizing subsets of FLS based on differential gene expression, or on responsiveness to inflammatory stimuli, will facilitate a comprehensive understanding of the pathogenesis and pathophysiology of RA.

We are grateful to Dr. Jun Hashimoto, Dr. Tetsuya Tomita, and Professor Hideki Yoshikawa in Department of orthopaedics (Osaka University) for collaboration on our study of FLS in

**6.4.2 Possible positive effect of A20/ABINs on pro-inflammatory cytokine induction**  A20, also termed TNFAIP3 (TNF-induced protein 3), was originally identified as an inducible zinc finger protein in human umbilical vein endothelial cell lines following stimulation with TNF-. A20 has dual enzymatic activities, namely, ubiquitination and deubiquitination (Dixit *et al.*, 1990). The induction of A20 upon stimulation with TNF- is NF-B dependent; moreover, induced A20 reversely suppresses the activation of NF-B through the regulation of ubiquitin-mediated degradation of NF-B activator (Vereecke *et al.*, 2009). This negative feedback loop is thought to be necessary to terminate inflammation and protect tissues from unnecessary damage. Recently, it was reported that the expression level of A20 in RA-FLSs was lower than that in OA-FLSs (Elsby *et al.*, 2010). Although the difference was not significant, this finding could provide *in vitro* evidence of altered *A20* transcription by 6q23 intergenic SNPs associated with RA (Dieguez-Gonzalez *et al.*, 2009; Orozco *et al.*, 2009). Thus, we speculated that the down-regulation of NF-B inhibitors might be a possible mechanism for enhanced activation of NF-B in high-responder FLSs. Contrary to our speculation, the high-responder group with abundant mRNA levels of proinflammatory cytokines also exhibited marked induction of *A20* following stimulation with TNF-. Furthermore, the transcription of the NF-B inhibitory molecules *ABIN* (A20 binding inhibitor of NF-B activation, also called TNIP, TNFAIP3 interacting protein)*-1* and

*ABIN*-*3*, but not of *ABIN-2*, was increased (Igarashi et al., in press).

the production of AID by FLSs in the RA synovial tissues (Figure 1), providing definitive evidence for the occurrence of ectopic and aberrant expression of *AID* in RA.

Fig. 1. Immunofluorescence staining of AID on synovial tissue sections from a representative RA patient. Sections were stained simultaneously with rat mAb for AID and anti-CD20 (B-cell marker) mAb. AID was visualized with alexa 488 fluoro-dye conjugated anti-rat secondary Ab (green); CD20 was visualized with alexa 594 fluoro-dye conjugated anti-mouse secondary Ab (red). The nucleus was stained with 4',6-diamino-2-phenylindole (blue). Scale bar is 100 m.

We concluded that AID is selectively expressed by a proportion of RA-FLSs and that its expression is associated with an increased frequency of somatic mutations in *p53* (Igarashi *et al.*, 2010). Thus, it is possible that the aberrant expression of AID within certain RA-FLSs induces somatic mutations in *p53*, leading to the acquisition of pro-inflammatory or tumorlike phenotypes.

#### **6.4 Heterogeneous responsiveness of fibroblast-like synoviocytes to TNF- 6.4.1 RA-FLS cell lines differentially respond to TNF-**

The chronic inflammation circuit in the joints of RA is initiated by the production of inflammatory cytokines by FLSs, following stimulation with TNF- secreted from the surrounding inflammatory cells. In this context, the TNF-/NF-B pathway plays an essential role in the transcription of pro-inflammatory cytokines. However, the regulation of NF-B activity downstream of TNF- in FLSs is not fully understood. To investigate the heterogeneous responsiveness of RA-FLS cell lines to TNF- stimulation, we examined the panels of primary RA-FLS cell lines for their induction levels of pro-inflammatory cytokines following TNF- stimulation. Interestingly, RA-FLS cell lines can be clearly categorized into 2 types based on the responsiveness to TNF-, namely, whether the transcription levels of pro-inflammatory cytokine gene are high (designated as the high-responder group) or not (designated as the low-responder group). This facilitated production of pro-inflammatory cytokines can be explained by the significant elevation of NF-B activity in the highresponder FLS lines compared with that in the low-responder lines.

#### **6.4.2 Possible positive effect of A20/ABINs on pro-inflammatory cytokine induction**

A20, also termed TNFAIP3 (TNF-induced protein 3), was originally identified as an inducible zinc finger protein in human umbilical vein endothelial cell lines following stimulation with TNF-. A20 has dual enzymatic activities, namely, ubiquitination and deubiquitination (Dixit *et al.*, 1990). The induction of A20 upon stimulation with TNF- is NF-B dependent; moreover, induced A20 reversely suppresses the activation of NF-B through the regulation of ubiquitin-mediated degradation of NF-B activator (Vereecke *et al.*, 2009). This negative feedback loop is thought to be necessary to terminate inflammation and protect tissues from unnecessary damage. Recently, it was reported that the expression level of A20 in RA-FLSs was lower than that in OA-FLSs (Elsby *et al.*, 2010). Although the difference was not significant, this finding could provide *in vitro* evidence of altered *A20* transcription by 6q23 intergenic SNPs associated with RA (Dieguez-Gonzalez *et al.*, 2009; Orozco *et al.*, 2009). Thus, we speculated that the down-regulation of NF-B inhibitors might be a possible mechanism for enhanced activation of NF-B in high-responder FLSs. Contrary to our speculation, the high-responder group with abundant mRNA levels of proinflammatory cytokines also exhibited marked induction of *A20* following stimulation with TNF-. Furthermore, the transcription of the NF-B inhibitory molecules *ABIN* (A20 binding inhibitor of NF-B activation, also called TNIP, TNFAIP3 interacting protein)*-1* and *ABIN*-*3*, but not of *ABIN-2*, was increased (Igarashi et al., in press).

These observations indicate that there is heterogeneity of RA-FLSs in the responsiveness to TNF- stimulation and suggest that these "inhibitors" might not play negative regulatory roles in RA-FLS. The precise mechanism, cell-lineage specificity, disease specificity, and significance in cell biology of this unexpected possible positive role for A20/ABINs are currently under investigation.

### **7. Conclusion**

84 Rheumatoid Arthritis – Etiology, Consequences and Co-Morbidities

the production of AID by FLSs in the RA synovial tissues (Figure 1), providing definitive

evidence for the occurrence of ectopic and aberrant expression of *AID* in RA.

Fig. 1. Immunofluorescence staining of AID on synovial tissue sections from a

**6.4 Heterogeneous responsiveness of fibroblast-like synoviocytes to TNF-**

**6.4.1 RA-FLS cell lines differentially respond to TNF-**

responder FLS lines compared with that in the low-responder lines.

(blue). Scale bar is 100 m.

like phenotypes.

representative RA patient. Sections were stained simultaneously with rat mAb for AID and anti-CD20 (B-cell marker) mAb. AID was visualized with alexa 488 fluoro-dye conjugated anti-rat secondary Ab (green); CD20 was visualized with alexa 594 fluoro-dye conjugated anti-mouse secondary Ab (red). The nucleus was stained with 4',6-diamino-2-phenylindole

We concluded that AID is selectively expressed by a proportion of RA-FLSs and that its expression is associated with an increased frequency of somatic mutations in *p53* (Igarashi *et al.*, 2010). Thus, it is possible that the aberrant expression of AID within certain RA-FLSs induces somatic mutations in *p53*, leading to the acquisition of pro-inflammatory or tumor-

The chronic inflammation circuit in the joints of RA is initiated by the production of inflammatory cytokines by FLSs, following stimulation with TNF- secreted from the surrounding inflammatory cells. In this context, the TNF-/NF-B pathway plays an essential role in the transcription of pro-inflammatory cytokines. However, the regulation of NF-B activity downstream of TNF- in FLSs is not fully understood. To investigate the heterogeneous responsiveness of RA-FLS cell lines to TNF- stimulation, we examined the panels of primary RA-FLS cell lines for their induction levels of pro-inflammatory cytokines following TNF- stimulation. Interestingly, RA-FLS cell lines can be clearly categorized into 2 types based on the responsiveness to TNF-, namely, whether the transcription levels of pro-inflammatory cytokine gene are high (designated as the high-responder group) or not (designated as the low-responder group). This facilitated production of pro-inflammatory cytokines can be explained by the significant elevation of NF-B activity in the highAnti-cytokine therapy for RA is a prominent achievement in the field of autoimmune diseases. Accumulated evidence from clinical and basic medical research indicates pivotal roles for FLS in the pathogenesis and pathophysiology of RA. Data from genome-wide screening, transcriptional profiling, and animal models indicate that RA consists with heterogeneous disease subsets. Together with several other researchers, we have presented evidence for heterogeneity in FLS. Based on this finding, we have successfully searched for disease-related genes by subtyping FLS. We have identified 2 groups of genes, *AID* and *A20/ABINs*. *AID* is involved in the irreversible transformation of FLS, whereas *A20/ABINs* participate in the reversible, but potentially harmful, responsiveness of them. Both groups of genes are constituent elements for distinct levels of heterogeneity in FLS, which may be involved in resistance to anti-cytokine therapies. Subtyping of FLS based on expression of AID did not coincide with that based on responsiveness to signal-utilizing NF-B, which is reasonable because RA is a multi-factorial disease. We believe that our approach to categorizing subsets of FLS based on differential gene expression, or on responsiveness to inflammatory stimuli, will facilitate a comprehensive understanding of the pathogenesis and pathophysiology of RA.
