**1.2 Role of synovial fibroblasts in synovitis**

#### **1.2.1 Invasive features of synovial fibroblasts**

There is not a uniform theory to explain how synovitis develops. However, one of the major features of synovitis is the acquisition of invasiveness of synovial fibroblasts. It could be said that rheumatoid synovial fibroblasts exhibit features of transformed cells, but unlike these, they do not show genetic aberrations. Rather, it seems that different activation processes are not correctly balanced by regulatory mechanisms in these cells. In this regard, it is typical of rheumatoid synovial fibroblasts to constitutively express growth factors, adhesion

<sup>\*</sup> Antonio Gabucio1, Astrid Jüngel2, Michel Neidhart2, María Comazzi2, Gabriel Herrero-Beaumont1, Renate E. Gay2 and Steffen Gay2

*<sup>2</sup>Zürich University Hospital, Zürich, Switzerland*

Innate Mechanisms of Synovitis – Fibrin Deposits Contribute to Invasion 113

In order to overcome these limitations, elegant strategies have been employed for the study of invasiveness of rheumatoid synovial fibroblasts. An interesting approach is to specifically look at areas where synovium invades cartilage and bone, the so-called cartilage-pannus junction (CPJ), looking for selective expression of molecules conferring invasiveness (Benito et al., 2004). From this type of study we have learned that cells located in the invasive fronts are mostly macrophages and fibroblasts. Both the up-regulation of anti-apoptotic factors, such as sentrin 1 and the Fas associated death domain-like interleukin 1 converting enzyme inhibitory protein (FLIP), and the expression of protooncogenes, have been found in these areas (Franz et al., 2000; Schedel et al., 2002). On the other hand, studies looking into CPJs are necessarily carried out in specimens obtained at the time of joint replacement surgery. Therefore, we need to be cautious at drawing conclusions, because these samples could

The invasive process has been studied in vitro in a transwell system with Matrigel, a method that was designed for the study of metastasis (Tolboom et al., 2002). More recently, a matrix-associated transepithelial resistance invasion (MATRIN) assay was developed to measure the rate of invasiveness from the breakdown of the electrical resistance generated by an epithelial monolayer (Wunrau et al., 2009). This system provides a means of directly assessing the participation of a particular factor in the ability of cells to scatter through the matrix. Also interesting are several three-dimensional co-culture systems in which minced, artificially generated, or native cartilage is put into contact with different subsets of cells that

The experimental severe cellular immunodeficient (SCID) mouse co-implantation model allows the study of mechanisms of invasion in vivo. In this model, human cartilage and rheumatoid synovial fibroblasts are engrafted in nude mice, that is, in an immune independent environment. Several works using this model have shown the ability of rheumatoid synovial fibroblasts to migrate and to destroy cartilage in the absence of immune cell concurrence (Müller-Ladner et al., 1996). On the other hand, some mechanisms of destruction taking place in rheumatoid arthritis might not show up in the SCID mouse model, as the latter does not provide an inflammatory microenvironment comparable to

There are probably several mechanisms accounting for cartilage and bone destruction in rheumatoid arthritis, and different cells and mediators can be involved. Activation of osteoclasts through the RANK ligand system stands as the principal pathway of bone erosion. In this sense, synovial fibroblasts, among other cell types, have an undeniable role as inductors of osteoclast differentiation and maturation in arthritic joints (Kim et al., 2007). But there is also much evidence of the direct ability of synovial fibroblasts to penetrate the

Metalloproteinases (MMPs) are a family of Zn2+ binding endoproteinases able to degrade the connective tissue. They are synthesized as precursors and cleaved at the N-terminus to their active forms. There is a considerable overlap of functions between them, and some of them are known to trigger the activation of others, as well as interact with additional

Of all MMPs, the interstitial collagenase, MMP-1, and stromelysin 1, or MMP-3, are the best characterized in the setting of synovitis. MMP-3 is able to degrade the most abundant extracellular components of the synovial tissue, including fibronectin and laminin, as well as

**1.2.3 The family of metalloproteinases and related molecules** 

proteases to generate proteolytic cascades in certain systems.

adjacent joint structures as a result of their production of various proteases.

reflect longstanding instead of active disease.

are present at the CPJ.

synovitis (Jüngel et al, 2010).

molecules and proteases, which participate in inflammation and in the destruction of joint tissues (Pap et al., 2005).

One particularly altered regulatory mechanism in rheumatoid synovial fibroblasts is cell growth. In healthy individuals, the synovial membrane is formed by a limiting row of cells, the synovial intima, which overlays connective tissue in an epithelium-like fashion. In rheumatoid arthritis, the intimal layer is characteristically disarranged and hyperplastic (Tarner et al., 2005). Along with the infiltration by leukocytes, there is an increase in the density of synovial fibroblasts. However, the latter does not derive from a high proliferation rate, but from cell longevity. These cells are able to survive in adverse conditions, such as hypoxia and loss of matrix anchorage. Defective apoptosis might therefore be a critical process in acquisition of invasiveness, and different works have shed light on this, describing specific alterations in the regulation of death mechanisms in rheumatoid arthritis (Takami et al., 2003; Jüngel et al., 2006). But on the whole it could be said that the apoptotic machinery of rheumatoid synovial fibroblasts is not impaired, and most of its deregulation could be due to the influence of signals from the inflamed environment (Kammouni et al., 2007).

How rheumatoid synovial cells become invasive brings back the chicken or the egg dilemma. Are they mere effectors activated by immune cells or is it the activation of these cells by innate mechanisms what helps to trespass the barrier of peripheral tolerance, at the same time conferring them with aggressive features? Increasing evidence is rising supporting that local factors associated to inflammation can shape the phenotype of rheumatoid synovial fibroblasts. An intriguing feature of rheumatoid synovitis is local hypoxia (Stevens et al., 1991). The formation of new vessels is characteristic of synovitis, but still the density of blood vessels is insufficient to supply the overgrown tissue, so consequently there are focal areas of ischemia inside the inflamed joint. It has been shown that reduced oxygen leads to an activation of hypoxia sensitive elements, orchestrated by hypoxia induced transcription factor-1 (HIF-1). Apparently, there is a low threshold for the induction of the hypoxia program in rheumatoid arthritis, probably resulting from the stabilisation of HIF-1 by inflammatory cytokines. The vascular endothelium growth factor (VEGF) is one of the molecules induced by HIF-1 thought to play a prominent role in the acquisition of invasiveness by rheumatoid synovial fibroblasts (Distler et al., 2004).

#### **1.2.2 Hurdles in the study of rheumatoid synovial fibroblasts' features**

Synovitis is a non-conventional lesion, and it is difficult to put it into experimental coordinates in order to dissect its pathogenic mechanisms. Nonetheless, in primary cultures, rheumatoid synovial fibroblasts but not synovial fibroblasts from osteoarthritic or healthy joints, are able to maintain an activated phenotype after several passages. This rare feature suggests that cells carry with them a stable imprinting of the in vivo circumstances. However, in vitro studies have frequently failed to identify or consistently replicate mechanisms associated to acquisition of invasiveness by rheumatoid synovial fibroblasts. A possible explanation is that cells from the same joint can exhibit a heterogeneous spectrum of phenotypes (Kasperkovitz et al., 2005). In large synovial specimens, an alternation can be observed between overgrown sprouts of tissue (macroscopic villi) and normal-appearing areas, a finding that suggests that the aggressive transformation of synovial fibroblasts is focal. Studies based on primary cultures, as well as on high throughput techniques, can miss the features of small but critical subpopulations of cells.

molecules and proteases, which participate in inflammation and in the destruction of joint

One particularly altered regulatory mechanism in rheumatoid synovial fibroblasts is cell growth. In healthy individuals, the synovial membrane is formed by a limiting row of cells, the synovial intima, which overlays connective tissue in an epithelium-like fashion. In rheumatoid arthritis, the intimal layer is characteristically disarranged and hyperplastic (Tarner et al., 2005). Along with the infiltration by leukocytes, there is an increase in the density of synovial fibroblasts. However, the latter does not derive from a high proliferation rate, but from cell longevity. These cells are able to survive in adverse conditions, such as hypoxia and loss of matrix anchorage. Defective apoptosis might therefore be a critical process in acquisition of invasiveness, and different works have shed light on this, describing specific alterations in the regulation of death mechanisms in rheumatoid arthritis (Takami et al., 2003; Jüngel et al., 2006). But on the whole it could be said that the apoptotic machinery of rheumatoid synovial fibroblasts is not impaired, and most of its deregulation could be due to the influence of signals from the inflamed environment (Kammouni et al.,

How rheumatoid synovial cells become invasive brings back the chicken or the egg dilemma. Are they mere effectors activated by immune cells or is it the activation of these cells by innate mechanisms what helps to trespass the barrier of peripheral tolerance, at the same time conferring them with aggressive features? Increasing evidence is rising supporting that local factors associated to inflammation can shape the phenotype of rheumatoid synovial fibroblasts. An intriguing feature of rheumatoid synovitis is local hypoxia (Stevens et al., 1991). The formation of new vessels is characteristic of synovitis, but still the density of blood vessels is insufficient to supply the overgrown tissue, so consequently there are focal areas of ischemia inside the inflamed joint. It has been shown that reduced oxygen leads to an activation of hypoxia sensitive elements, orchestrated by hypoxia induced transcription factor-1 (HIF-1). Apparently, there is a low threshold for the induction of the hypoxia program in rheumatoid arthritis, probably resulting from the stabilisation of HIF-1 by inflammatory cytokines. The vascular endothelium growth factor (VEGF) is one of the molecules induced by HIF-1 thought to play a prominent role in the

acquisition of invasiveness by rheumatoid synovial fibroblasts (Distler et al., 2004).

Synovitis is a non-conventional lesion, and it is difficult to put it into experimental coordinates in order to dissect its pathogenic mechanisms. Nonetheless, in primary cultures, rheumatoid synovial fibroblasts but not synovial fibroblasts from osteoarthritic or healthy joints, are able to maintain an activated phenotype after several passages. This rare feature suggests that cells carry with them a stable imprinting of the in vivo circumstances. However, in vitro studies have frequently failed to identify or consistently replicate mechanisms associated to acquisition of invasiveness by rheumatoid synovial fibroblasts. A possible explanation is that cells from the same joint can exhibit a heterogeneous spectrum of phenotypes (Kasperkovitz et al., 2005). In large synovial specimens, an alternation can be observed between overgrown sprouts of tissue (macroscopic villi) and normal-appearing areas, a finding that suggests that the aggressive transformation of synovial fibroblasts is focal. Studies based on primary cultures, as well as on high throughput techniques, can miss

**1.2.2 Hurdles in the study of rheumatoid synovial fibroblasts' features** 

the features of small but critical subpopulations of cells.

tissues (Pap et al., 2005).

2007).

In order to overcome these limitations, elegant strategies have been employed for the study of invasiveness of rheumatoid synovial fibroblasts. An interesting approach is to specifically look at areas where synovium invades cartilage and bone, the so-called cartilage-pannus junction (CPJ), looking for selective expression of molecules conferring invasiveness (Benito et al., 2004). From this type of study we have learned that cells located in the invasive fronts are mostly macrophages and fibroblasts. Both the up-regulation of anti-apoptotic factors, such as sentrin 1 and the Fas associated death domain-like interleukin 1 converting enzyme inhibitory protein (FLIP), and the expression of protooncogenes, have been found in these areas (Franz et al., 2000; Schedel et al., 2002). On the other hand, studies looking into CPJs are necessarily carried out in specimens obtained at the time of joint replacement surgery. Therefore, we need to be cautious at drawing conclusions, because these samples could reflect longstanding instead of active disease.

The invasive process has been studied in vitro in a transwell system with Matrigel, a method that was designed for the study of metastasis (Tolboom et al., 2002). More recently, a matrix-associated transepithelial resistance invasion (MATRIN) assay was developed to measure the rate of invasiveness from the breakdown of the electrical resistance generated by an epithelial monolayer (Wunrau et al., 2009). This system provides a means of directly assessing the participation of a particular factor in the ability of cells to scatter through the matrix. Also interesting are several three-dimensional co-culture systems in which minced, artificially generated, or native cartilage is put into contact with different subsets of cells that are present at the CPJ.

The experimental severe cellular immunodeficient (SCID) mouse co-implantation model allows the study of mechanisms of invasion in vivo. In this model, human cartilage and rheumatoid synovial fibroblasts are engrafted in nude mice, that is, in an immune independent environment. Several works using this model have shown the ability of rheumatoid synovial fibroblasts to migrate and to destroy cartilage in the absence of immune cell concurrence (Müller-Ladner et al., 1996). On the other hand, some mechanisms of destruction taking place in rheumatoid arthritis might not show up in the SCID mouse model, as the latter does not provide an inflammatory microenvironment comparable to synovitis (Jüngel et al, 2010).

#### **1.2.3 The family of metalloproteinases and related molecules**

There are probably several mechanisms accounting for cartilage and bone destruction in rheumatoid arthritis, and different cells and mediators can be involved. Activation of osteoclasts through the RANK ligand system stands as the principal pathway of bone erosion. In this sense, synovial fibroblasts, among other cell types, have an undeniable role as inductors of osteoclast differentiation and maturation in arthritic joints (Kim et al., 2007). But there is also much evidence of the direct ability of synovial fibroblasts to penetrate the adjacent joint structures as a result of their production of various proteases.

Metalloproteinases (MMPs) are a family of Zn2+ binding endoproteinases able to degrade the connective tissue. They are synthesized as precursors and cleaved at the N-terminus to their active forms. There is a considerable overlap of functions between them, and some of them are known to trigger the activation of others, as well as interact with additional proteases to generate proteolytic cascades in certain systems.

Of all MMPs, the interstitial collagenase, MMP-1, and stromelysin 1, or MMP-3, are the best characterized in the setting of synovitis. MMP-3 is able to degrade the most abundant extracellular components of the synovial tissue, including fibronectin and laminin, as well as

Innate Mechanisms of Synovitis – Fibrin Deposits Contribute to Invasion 115

2004). While uPA is able to degrade the extracellular matrix, it also activates MMPs and proteoglycanases through the cleavage of their precursors. Several studies have shown that uPA is over-expressed in joints from patients with rheumatoid arthritis, in correlation with disease severity (van der Laan et al., 2000). In spite of this role, uPA could be a doubleedged therapeutic sword in rheumatoid arthritis, due to its activity in extra-vascular fibrinolysis, as we discuss in the next section. Interesting evidence was drawn in mice with antigen-induced arthritis, since uPA-deficient animals depicted a more severe phenotype as compared to wild type littermates (Busso et al., 1998). From subsequent studies, it can be concluded that aggressive features mediated by uPA are linked to its cell attachment activity, through the binding of its high affinity receptor, uPAR. New released work has found that the uPA-uPAR pair is a mediator of invasiveness in the SCID mouse coimplantation model (Serratì et al., 2011). In turn, uPAR is part of a larger complex, the urokinase plasminogen activating system (uPAS), formed by its assembly with 4 serin protease inhibitors at the cell surface. Triggering of uPAS is associated to proliferation, adhesion, migration and neoangiogenesis in tumours. These findings point to the complex

as a better therapeutic target than the protease itself (Ulisse et al., 2009).

a constitutive pathway (Weinberg et al., 1991; Ronday, et al., 1996).

**1.3.1 Haemostasis activation overflows the fibrinolytic capacity in the joints with** 

Since the extra-vascular activation of haemostasis is a characteristic feature of inflammation, fibrin deposition in the inflamed synovial tissue is considered a non-specific event. During inflammation, the exudation of plasma into joints can result in coagulation factors achieving high concentrations at the synovial effusion. In fact, joints affected with osteoarthritis, infections, and trauma, often show fibrin deposits, albeit not as widespread as found in rheumatoid arthritis (Clemmensen et al., 1983). The striking abundance of fibrin in rheumatoid synovial tissues has been attributed to both an increased formation and a low clearance of the clots. As pointed out in different studies, rheumatoid arthritis flares provoke a status of extra-vascular thrombophilia, so that the influx of fibrinogen and its immobilization are high (Carmassi et al., 1996). Fibrin networks are thicker in patients with rheumatoid arthritis than in controls, and presumably more resistant to proteolysis as well (Kwasny-Krochin et al., 2010). This feature along with a reduced fibrinolytic activity can explain the accumulation of fibrin inside rheumatoid joints. Of the two regulatory systems that activate plasmin to degrade fibrin, the tissue plasminogen activator (tPA) is reduced in rheumatoid synovial tissues. Similarly, there is an increased production of the inhibitors of plasminogen activator, PAI-1 and PAI-2, which act by preventing fibrin dissolution through

Local activation of complement is an inflammation-dependent mechanism that can help to stabilize fibrin clots thereby decreasing the fibrinolytic potential of the joint. In particular, we explored some years ago the local production of the regulatory factor C4b-binding protein (C4BP). The protein C-S anticoagulatory system is a principal mechanism for preventing the uncontrolled activation of haemostasis. The beta chain of C4BP binds protein S with high affinity in an equimolecular fashion (Dahlbäck, 1989). Only free protein S is active and the free fraction depends on the availability of C4BP beta. Reduced levels of free protein S are associated with an increased risk of thrombotic events. Interestingly, we showed local production of C4BP beta by rheumatoid synovial fibroblasts, as well as its colocalization with fibrin-rich areas at the synovial tissue (Sánchez-Pernaute et al., 2006). The

**1.3 The role of fibrin in rheumatoid arthritis** 

**rheumatoid arthritis** 

collagens I and III. In addition, it activates MMP-1, which not only shares some of MMP-3 cleavage targets, but is also able to degrade collagen II, the principal component of articular cartilage. Plasma levels of both proteases have been found to be increased in patients with rheumatoid arthritis (Manicourt et al., 1995). Moreover, they correlated with disease activity, while intra-joint concentrations of the enzymes increased in parallel with the degree of joint inflammation (Ishiguro et al., 1996). Plasma levels of MMP-3 are currently regarded as a surrogate marker of severity, a fact that reflects its relevance as a mediator of joint destruction in rheumatoid arthritis. With immune-detection techniques, MMP-1 and -3 show a patchy distribution throughout the inflamed synovial tissue. Both molecules are consistently found at CPJs, where a diffuse immune-reactivity has been described (Tetlow & Woolley, 1995). Their pattern of distribution has confirmed synovial fibroblasts as the major source of these molecules in the joint.

Also of interest are the group of membrane-anchored (MT-) MMPs, which are bound to integrin chains, and, upon activation, digest pericellular matrix. Of this family, MT1-MMP (also MMP-14), which is over-expressed in rheumatoid synovial fibroblasts, is thought to confer to these cells some of their invasive potential (Yamanaka et al., 2000). In this regard, experiments carried out in the SCID mouse co-implantation model have shown the participation of MMP-1 and MT1-MMP in the degradation of cartilage by rheumatoid synovial fibroblasts (Rutkauskaite et al., 2004; Rutkauskaite et al., 2005). Based on this evidence, the pathway of MMPs has been for a while a promising area of research for therapeutics, not only in rheumatoid arthritis but also in metastatic tumours. The members of the family of tissue inhibitor of metalloproteinases (TIMPs), which act as natural regulators of MMPs, appeared as ideal candidates to develop anti-invasive compounds. In the SCID mouse model, TIMP-1 and -3 over-expressing mutants were able to slow the invasive process (van der Laan et al., 2003). However, less convincing results have been drawn so far in therapeutic experimental approaches and clinical trials.

In summary, it appears that synovial fibroblasts are the main effectors of destruction, a fact that could be considered natural. Fibroblasts are in charge of connective tissue remodelling, both under physiologic conditions and in disease. Production of proteases allows them to migrate through the matrix and restore the injured site in wound healing processes (Woessner, 1991). The same mechanisms take place during invasive processes. In this regard, rheumatoid synovial fibroblasts have been compared to tumour-associated stromal cells, which are non-neoplastic fibroblasts that contribute to metastatic growth by the production of MMPs (Hotary et al., 2003). Interestingly, the presence of MMPs at the synovial tissue is not related to the stage of the disease, and in fact the proteases can be abundant in early synovitis (Katrib et al., 2001). Therefore, fibroblast activation is not necessarily a consequence of longstanding disease, but could be one of the distinguishing processes between non-progressive disease and rheumatoid arthritis.

To help understand why rheumatoid and not other synovial fibroblasts turn invasive, a revealing study put in relationship the mRNA expression levels of MMPs with local hypoxia. Not only hypoxic cultures resulted in an increase in MMP-1 and MMP-3 transcripts, but also HIF-1 siRNA transfects yielded 50% lower mRNA levels of MMP-3 (Ahn et al., 2008).

Pulling the thread of research coming from invasive neoplasms and stromal cells, additional synovial fibroblast-dependent proteases were discovered at CPJs, showing potent in vitro capacity to destroy bone and cartilage. One of these molecules, that heralds the aggressive behaviour of tumours, is the urokinase type plasmin activator (uPA) (Duffy & Duggan,

collagens I and III. In addition, it activates MMP-1, which not only shares some of MMP-3 cleavage targets, but is also able to degrade collagen II, the principal component of articular cartilage. Plasma levels of both proteases have been found to be increased in patients with rheumatoid arthritis (Manicourt et al., 1995). Moreover, they correlated with disease activity, while intra-joint concentrations of the enzymes increased in parallel with the degree of joint inflammation (Ishiguro et al., 1996). Plasma levels of MMP-3 are currently regarded as a surrogate marker of severity, a fact that reflects its relevance as a mediator of joint destruction in rheumatoid arthritis. With immune-detection techniques, MMP-1 and -3 show a patchy distribution throughout the inflamed synovial tissue. Both molecules are consistently found at CPJs, where a diffuse immune-reactivity has been described (Tetlow & Woolley, 1995). Their pattern of distribution has confirmed synovial fibroblasts as the major

Also of interest are the group of membrane-anchored (MT-) MMPs, which are bound to integrin chains, and, upon activation, digest pericellular matrix. Of this family, MT1-MMP (also MMP-14), which is over-expressed in rheumatoid synovial fibroblasts, is thought to confer to these cells some of their invasive potential (Yamanaka et al., 2000). In this regard, experiments carried out in the SCID mouse co-implantation model have shown the participation of MMP-1 and MT1-MMP in the degradation of cartilage by rheumatoid synovial fibroblasts (Rutkauskaite et al., 2004; Rutkauskaite et al., 2005). Based on this evidence, the pathway of MMPs has been for a while a promising area of research for therapeutics, not only in rheumatoid arthritis but also in metastatic tumours. The members of the family of tissue inhibitor of metalloproteinases (TIMPs), which act as natural regulators of MMPs, appeared as ideal candidates to develop anti-invasive compounds. In the SCID mouse model, TIMP-1 and -3 over-expressing mutants were able to slow the invasive process (van der Laan et al., 2003). However, less convincing results have been

In summary, it appears that synovial fibroblasts are the main effectors of destruction, a fact that could be considered natural. Fibroblasts are in charge of connective tissue remodelling, both under physiologic conditions and in disease. Production of proteases allows them to migrate through the matrix and restore the injured site in wound healing processes (Woessner, 1991). The same mechanisms take place during invasive processes. In this regard, rheumatoid synovial fibroblasts have been compared to tumour-associated stromal cells, which are non-neoplastic fibroblasts that contribute to metastatic growth by the production of MMPs (Hotary et al., 2003). Interestingly, the presence of MMPs at the synovial tissue is not related to the stage of the disease, and in fact the proteases can be abundant in early synovitis (Katrib et al., 2001). Therefore, fibroblast activation is not necessarily a consequence of longstanding disease, but could be one of the distinguishing

To help understand why rheumatoid and not other synovial fibroblasts turn invasive, a revealing study put in relationship the mRNA expression levels of MMPs with local hypoxia. Not only hypoxic cultures resulted in an increase in MMP-1 and MMP-3 transcripts, but also HIF-1 siRNA transfects yielded 50% lower mRNA levels of MMP-3

Pulling the thread of research coming from invasive neoplasms and stromal cells, additional synovial fibroblast-dependent proteases were discovered at CPJs, showing potent in vitro capacity to destroy bone and cartilage. One of these molecules, that heralds the aggressive behaviour of tumours, is the urokinase type plasmin activator (uPA) (Duffy & Duggan,

drawn so far in therapeutic experimental approaches and clinical trials.

processes between non-progressive disease and rheumatoid arthritis.

source of these molecules in the joint.

(Ahn et al., 2008).

2004). While uPA is able to degrade the extracellular matrix, it also activates MMPs and proteoglycanases through the cleavage of their precursors. Several studies have shown that uPA is over-expressed in joints from patients with rheumatoid arthritis, in correlation with disease severity (van der Laan et al., 2000). In spite of this role, uPA could be a doubleedged therapeutic sword in rheumatoid arthritis, due to its activity in extra-vascular fibrinolysis, as we discuss in the next section. Interesting evidence was drawn in mice with antigen-induced arthritis, since uPA-deficient animals depicted a more severe phenotype as compared to wild type littermates (Busso et al., 1998). From subsequent studies, it can be concluded that aggressive features mediated by uPA are linked to its cell attachment activity, through the binding of its high affinity receptor, uPAR. New released work has found that the uPA-uPAR pair is a mediator of invasiveness in the SCID mouse coimplantation model (Serratì et al., 2011). In turn, uPAR is part of a larger complex, the urokinase plasminogen activating system (uPAS), formed by its assembly with 4 serin protease inhibitors at the cell surface. Triggering of uPAS is associated to proliferation, adhesion, migration and neoangiogenesis in tumours. These findings point to the complex as a better therapeutic target than the protease itself (Ulisse et al., 2009).

#### **1.3 The role of fibrin in rheumatoid arthritis**

#### **1.3.1 Haemostasis activation overflows the fibrinolytic capacity in the joints with rheumatoid arthritis**

Since the extra-vascular activation of haemostasis is a characteristic feature of inflammation, fibrin deposition in the inflamed synovial tissue is considered a non-specific event. During inflammation, the exudation of plasma into joints can result in coagulation factors achieving high concentrations at the synovial effusion. In fact, joints affected with osteoarthritis, infections, and trauma, often show fibrin deposits, albeit not as widespread as found in rheumatoid arthritis (Clemmensen et al., 1983). The striking abundance of fibrin in rheumatoid synovial tissues has been attributed to both an increased formation and a low clearance of the clots. As pointed out in different studies, rheumatoid arthritis flares provoke a status of extra-vascular thrombophilia, so that the influx of fibrinogen and its immobilization are high (Carmassi et al., 1996). Fibrin networks are thicker in patients with rheumatoid arthritis than in controls, and presumably more resistant to proteolysis as well (Kwasny-Krochin et al., 2010). This feature along with a reduced fibrinolytic activity can explain the accumulation of fibrin inside rheumatoid joints. Of the two regulatory systems that activate plasmin to degrade fibrin, the tissue plasminogen activator (tPA) is reduced in rheumatoid synovial tissues. Similarly, there is an increased production of the inhibitors of plasminogen activator, PAI-1 and PAI-2, which act by preventing fibrin dissolution through a constitutive pathway (Weinberg et al., 1991; Ronday, et al., 1996).

Local activation of complement is an inflammation-dependent mechanism that can help to stabilize fibrin clots thereby decreasing the fibrinolytic potential of the joint. In particular, we explored some years ago the local production of the regulatory factor C4b-binding protein (C4BP). The protein C-S anticoagulatory system is a principal mechanism for preventing the uncontrolled activation of haemostasis. The beta chain of C4BP binds protein S with high affinity in an equimolecular fashion (Dahlbäck, 1989). Only free protein S is active and the free fraction depends on the availability of C4BP beta. Reduced levels of free protein S are associated with an increased risk of thrombotic events. Interestingly, we showed local production of C4BP beta by rheumatoid synovial fibroblasts, as well as its colocalization with fibrin-rich areas at the synovial tissue (Sánchez-Pernaute et al., 2006). The

Innate Mechanisms of Synovitis – Fibrin Deposits Contribute to Invasion 117

cells increases with fibrinogen transformation into fibrin (Shainoff et al., 1990). Considering that blood cells express different alpha beta integrin chains that could bind circulating fibrinogen, the affinity of the soluble molecule needs to be low. In this regard, the shear forces elicited by fibrin networks may have an influence in the avidity towards cell receptors. This mechanism has been recently found to account for activation of colon tumour cells upon engagement of fibrin by the hyaluronate receptor, CD44 (Alves et a.,

The cell binding activity of fibrin networks is further enriched by its cross-linking with different matrix proteins, including collagens, proteoglycans or fibronectin, which also bear

Yet to be understood, is why rheumatoid but not non-rheumatoid hosts develop an invasive response to fibrin clots inside joints. It could be simply a matter of magnitude, as we discussed several years ago (Sanchez-Pernaute et al, 2003b). It could also rely on intrinsic features of rheumatoid synovial fibroblasts, that up to date have not been found. Alternatively, it is attractive to speculate that resistance of the clots to plasmin proteolysis in rheumatoid joints makes local macrophages and fibroblasts secrete additional proteases that degrade fibrin through a non-constitutive pathway (Bini et al., 1999). Between these potential fibrin-degrading proteases stands MMP-3, which is also one of the major mediators of joint destruction. In this way, the active process of digestion of the insoluble macromolecule could be regarded as a favourable environment for the destruction of

The early works focused on the histopathology of rheumatoid arthritis, described a gradient in fibrin deposition from more abundant and solid-like at the surface, to patchy and reticular in inner areas (Andersen & Gormsen, 1970). Working with specimens from patients undergoing joint replacement surgery we described that fibrin-rich areas of the synovium were organized differently than non-fibrinous regions (Sánchez-Pernaute et al., 2006). With immune-detection techniques we found that both cells and extracellular matrix elements had a differential distribution in fibrinous and non-fibrinous areas, and there were transition zones between them (Figure 1). In this regard, matrix deposition and fibroblast-like synovial cells, as well as vessels, increased in density in fibrin-rich areas, up to an interface with solid-like fibrin deposits at superficial areas, where cells were scant and there were no vessels. Macrophages were clustered in the vicinity of fibrin deposits, although they were also abundant in non-fibrinous regions; but lymphoid aggregates clearly stayed apart (unpublished observations). This particular organization has led us to focus on the interaction between fibrin and synovial fibroblasts aiming to find mechanisms of

Studies conducted in synovial specimens are frequently based on small biopsies. These explants constitute a fine way to reflect events taking place in the whole synovial tissue, in particular as regards cell activation and the participation of subtypes of infiltrating leukocytes (Smith et al., 2006). Elegant studies have proved that the pathology of the synovial tissue from biopsies can be employed to measure response to therapy and even unveil specific molecular predictive markers. However, this kind of study rarely describes the features of fibrin-rich areas, probably because in these regions the architecture of the tissue is distorted and difficult to read (as illustrated in figure 1). We believe that these areas are routinely discarded and thus

critical components of the synovial pathology might be underestimated.

**1.3.4 A potential role of fibrin in the architecture of synovitis** 

invasiveness induced by fibrin in these cells.

2009).

a variety of cell binding sites.

structures nearby.

beta chain of C4BP has also been found in omen fibroblasts participating in the invasion and resorption of the corpus luteum, therefore indicating that besides its prothrombotic role, the molecule is important in fibroblast-dependent remodelling processes.

In the light of these experimental data, a potential participation of fibrin in synovitis has been argued by different groups including ourselves (Busso & Hamilton, 2002; So et al., 2003; Sánchez-Pernaute et al., 2003b), and several antithrombotic strategies have been tried, proving useful in attenuating the inflammatory process in experimental models of rheumatoid arthritis (Busso et al., 1998; Varisco et al., 2000).

#### **1.3.2 Fibrin as an autoantigen**

Fibrin is one of the major substrates for peptidyl deiminases (PAD) inside inflamed joints. These enzymes transform arginine residues into citrulline, and subsequently change the physical properties of the protein. This modification could alter binding sites of plasmin, making the polymer resistant to proteolytic degradation. Moreover, it can also turn the molecule antigenic (Schellekens et al., 1998). This fact was confirmed with the characterization of anti-citrullinated peptide antibodies (ACPA), since they were shown to target epitopes from fibrin in a specific association with rheumatoid arthritis (Masson-Bessiere et al., 2001). Thus, fibrin can been considered a key mediator in the loss of immune tolerance in the disease. Since ACPA antibodies reach higher concentrations inside joints than at the periphery, deposition of fibrin in joints and exposure of its citrullinated form to immune-competent cells should be an early pathogenic event. On the other hand, ACPA can be found in pre-clinical stages in at least half of patients. To explain this contradiction, it has been suggested that the immune system is primed for citrullinated epitopes outside joints, for example in the lung or the oral cavity, with arthritis coming on a second wave (Quirke et al, 2011). We propose a different mechanism. According to recent studies, synovitis can remain asymptomatic during the first stages of the disease. A first mild flare of arthritis could, therefore, be the event during which citrullinated fibrinopeptides are generated inside joints and presented to the central immune system (van de Sande et al., 2011).

#### **1.3.3 Fibrin as a scaffold**

We focused our studies on "a non-immune" participation of fibrin in the development of synovitis. Since fibrin networks provide binding sites for the migration of cells, in this way facilitating wound healing, we proposed that the synovial tissue might grow through the engulfing of fibrin deposits at the lining surface. In a time-dependent approach, we studied events taking place from the first stages of the disease in antigen induced arthritis, and were able to describe the transition from acute inflammation, to deposition of fibrin clots, and subsequent synovitis-like tissue modifications taking place at fibrin-synovium interfaces (Sánchez Pernaute et al., 2003a). We then proposed that the binding of the free-surface of the lining cells would alter their polarity, thereby changing their resting phenotype into a migrating one, a mechanism that would contribute to invasiveness.

#### **1.3.3.1 Immobilization of fibrin affects cell binding**

Soluble fibrinogen turns into solid fibrin through the release of fibrinopeptides A and B, with the rest of its alpha and beta chains remaining mostly unchanged. Therefore, most cell binding sites are shared by the two macromolecules. But it is in our opinion the insolubility of fibrin which accounts for an invasive response of synovial fibroblasts in rheumatoid arthritis. Experiments carried out in macrophages demonstrated that the binding affinity of

beta chain of C4BP has also been found in omen fibroblasts participating in the invasion and resorption of the corpus luteum, therefore indicating that besides its prothrombotic role, the

In the light of these experimental data, a potential participation of fibrin in synovitis has been argued by different groups including ourselves (Busso & Hamilton, 2002; So et al., 2003; Sánchez-Pernaute et al., 2003b), and several antithrombotic strategies have been tried, proving useful in attenuating the inflammatory process in experimental models of

Fibrin is one of the major substrates for peptidyl deiminases (PAD) inside inflamed joints. These enzymes transform arginine residues into citrulline, and subsequently change the physical properties of the protein. This modification could alter binding sites of plasmin, making the polymer resistant to proteolytic degradation. Moreover, it can also turn the molecule antigenic (Schellekens et al., 1998). This fact was confirmed with the characterization of anti-citrullinated peptide antibodies (ACPA), since they were shown to target epitopes from fibrin in a specific association with rheumatoid arthritis (Masson-Bessiere et al., 2001). Thus, fibrin can been considered a key mediator in the loss of immune tolerance in the disease. Since ACPA antibodies reach higher concentrations inside joints than at the periphery, deposition of fibrin in joints and exposure of its citrullinated form to immune-competent cells should be an early pathogenic event. On the other hand, ACPA can be found in pre-clinical stages in at least half of patients. To explain this contradiction, it has been suggested that the immune system is primed for citrullinated epitopes outside joints, for example in the lung or the oral cavity, with arthritis coming on a second wave (Quirke et al, 2011). We propose a different mechanism. According to recent studies, synovitis can remain asymptomatic during the first stages of the disease. A first mild flare of arthritis could, therefore, be the event during which citrullinated fibrinopeptides are generated

inside joints and presented to the central immune system (van de Sande et al., 2011).

migrating one, a mechanism that would contribute to invasiveness.

**1.3.3.1 Immobilization of fibrin affects cell binding** 

We focused our studies on "a non-immune" participation of fibrin in the development of synovitis. Since fibrin networks provide binding sites for the migration of cells, in this way facilitating wound healing, we proposed that the synovial tissue might grow through the engulfing of fibrin deposits at the lining surface. In a time-dependent approach, we studied events taking place from the first stages of the disease in antigen induced arthritis, and were able to describe the transition from acute inflammation, to deposition of fibrin clots, and subsequent synovitis-like tissue modifications taking place at fibrin-synovium interfaces (Sánchez Pernaute et al., 2003a). We then proposed that the binding of the free-surface of the lining cells would alter their polarity, thereby changing their resting phenotype into a

Soluble fibrinogen turns into solid fibrin through the release of fibrinopeptides A and B, with the rest of its alpha and beta chains remaining mostly unchanged. Therefore, most cell binding sites are shared by the two macromolecules. But it is in our opinion the insolubility of fibrin which accounts for an invasive response of synovial fibroblasts in rheumatoid arthritis. Experiments carried out in macrophages demonstrated that the binding affinity of

molecule is important in fibroblast-dependent remodelling processes.

rheumatoid arthritis (Busso et al., 1998; Varisco et al., 2000).

**1.3.2 Fibrin as an autoantigen** 

**1.3.3 Fibrin as a scaffold**

cells increases with fibrinogen transformation into fibrin (Shainoff et al., 1990). Considering that blood cells express different alpha beta integrin chains that could bind circulating fibrinogen, the affinity of the soluble molecule needs to be low. In this regard, the shear forces elicited by fibrin networks may have an influence in the avidity towards cell receptors. This mechanism has been recently found to account for activation of colon tumour cells upon engagement of fibrin by the hyaluronate receptor, CD44 (Alves et a., 2009).

The cell binding activity of fibrin networks is further enriched by its cross-linking with different matrix proteins, including collagens, proteoglycans or fibronectin, which also bear a variety of cell binding sites.

Yet to be understood, is why rheumatoid but not non-rheumatoid hosts develop an invasive response to fibrin clots inside joints. It could be simply a matter of magnitude, as we discussed several years ago (Sanchez-Pernaute et al, 2003b). It could also rely on intrinsic features of rheumatoid synovial fibroblasts, that up to date have not been found. Alternatively, it is attractive to speculate that resistance of the clots to plasmin proteolysis in rheumatoid joints makes local macrophages and fibroblasts secrete additional proteases that degrade fibrin through a non-constitutive pathway (Bini et al., 1999). Between these potential fibrin-degrading proteases stands MMP-3, which is also one of the major mediators of joint destruction. In this way, the active process of digestion of the insoluble macromolecule could be regarded as a favourable environment for the destruction of structures nearby.

#### **1.3.4 A potential role of fibrin in the architecture of synovitis**

The early works focused on the histopathology of rheumatoid arthritis, described a gradient in fibrin deposition from more abundant and solid-like at the surface, to patchy and reticular in inner areas (Andersen & Gormsen, 1970). Working with specimens from patients undergoing joint replacement surgery we described that fibrin-rich areas of the synovium were organized differently than non-fibrinous regions (Sánchez-Pernaute et al., 2006). With immune-detection techniques we found that both cells and extracellular matrix elements had a differential distribution in fibrinous and non-fibrinous areas, and there were transition zones between them (Figure 1). In this regard, matrix deposition and fibroblast-like synovial cells, as well as vessels, increased in density in fibrin-rich areas, up to an interface with solid-like fibrin deposits at superficial areas, where cells were scant and there were no vessels. Macrophages were clustered in the vicinity of fibrin deposits, although they were also abundant in non-fibrinous regions; but lymphoid aggregates clearly stayed apart (unpublished observations). This particular organization has led us to focus on the interaction between fibrin and synovial fibroblasts aiming to find mechanisms of invasiveness induced by fibrin in these cells.

Studies conducted in synovial specimens are frequently based on small biopsies. These explants constitute a fine way to reflect events taking place in the whole synovial tissue, in particular as regards cell activation and the participation of subtypes of infiltrating leukocytes (Smith et al., 2006). Elegant studies have proved that the pathology of the synovial tissue from biopsies can be employed to measure response to therapy and even unveil specific molecular predictive markers. However, this kind of study rarely describes the features of fibrin-rich areas, probably because in these regions the architecture of the tissue is distorted and difficult to read (as illustrated in figure 1). We believe that these areas are routinely discarded and thus critical components of the synovial pathology might be underestimated.

Innate Mechanisms of Synovitis – Fibrin Deposits Contribute to Invasion 119

in vitro system that attempts to reproduce the interaction between fibrin and fibroblasts that takes place inside the joint. This model of cell stimulation had originally been developed to study leukocyte migration through vessel walls and is known as in situ fibrin polymerization (Qi & Kreutzer, 1995). In contrast to other types of cultures involving cells and matrix proteins, this approach conserves the shear forces of freshly clotted fibrin

Synovial tissues were obtained during joint replacement surgery from 8 patients with rheumatoid arthritis according to the American College of Rheumatology criteria (Arnett et al., 1988). For histologic studies, the synovial membrane was immediately fixed in formaldehyde, dehydrated in ethanol and embedded in paraffin. When tissues included bone edges, they were decalcified by a 48 hour incubation in formic acid. For in vitro studies, cells were isolated by disruption with 1.5 mg/ml dispase II at 37ºC for 1 hour in agitation, and cultured in 10% fetal calf serum (FCS) enriched Dulbecco's modified Eagle's medium (DMEM) supplemented with 2 mM L-glutamine, 50 UI/ml penicillin-streptomycin,

We studied the distribution of fibrin, MMP-1 and MMP-3 with double labelling immunedetection methods. Tissues were rehydrated, blocked with 6% bovine albumin and 3% serum of second antibody hosts, and incubated with the specific antibodies at 10 g/ml overnight, at 4ºC. Secondary antibodies were applied for 1 hour, at room temperature. As control, tissues were incubated with an isotype IgG from the species of primary antibodies. Development of fibrinogen immune-reactivity was done with peroxidase and Histogreen, using nuclear fast red for counterstaining. Diaminobenzidine was applied after a byotinilated secondary antibody to develop MMP-3, and counterstaining was done with hematoxylin. An alkaline phosphatase labelled antibody was employed to detect MMP-1, and nitroblue tetrazolium with 5-bromo 4-chloro 3-indolyl phosphate was used as substrate for development, plus nuclear fast red as counterstaining. Inhibition of endogenous peroxidase was done with 1% H2O2 methanol. Alkaline phosphatase activity was blocked

Between passages 4th and 7th cells were grown to confluence at 37ºC in 5% CO2, starved from serum during 48 hours and exposed to in situ clotted fibrin. Chilled fibrinogen was mixed in 0.5% foetal calf serum enriched DMEM at 1 mg/ml on ice, and 0.75 UI/ml thrombin was added. The mixture was immediately spread on top of the cell monolayers and the cultures

Four different cell cultures were employed. Cells were incubated with fibrin or medium alone for 12 hours. At the end of the incubation period, the clots and supernatants were removed, cells were washed and total RNA was isolated. Following retrotranscription, gene expression studies were done with quantitative PCR (qPCR) using cDNA as templates and

were transferred to the incubator to allow formation of fibrin clots.

networks and exposes cells to the deposits by their apical surface.

**2.1 Experimental methods**

**2.1.1 Obtention and handling of samples** 

0.2% amphotericin B, and 10 mM HEPES.

**2.1.2 Immune-detection techniques** 

with 5 mM levamisole.

**2.1.3 Fibrin-cell cultures** 

**2.1.4 Gene expression studies** 

Fig. 1. Stratification of synovitis in relationship to fibrin deposits in a representative rheumatoid synovial tissue

Low-magnification microphotographs of a representative rheumatoid synovial tissue with an overlying fibrin deposit. Left column shows staining of the different tissue regions with haematoxylin-eosin. Middle column shows cell distribution with anti actin antibodies. On the right side, immune-localization of macrophages with anti CD68 antibodies is shown.
