**7. Conclusion**

222 Rheumatoid Arthritis – Treatment

residues on the external surface of the TLR3 homodimer (Liu et al., 2008). More recent modelling of TLRs 7 and 9 also suggests direct ligand binding to the TLR molecule (Kubarenko et al., 2010). In contrast, TLR1:TLR2 hetero-dimerisation results in the formation of a hydrophobic pocket into which the lipopeptide PAM3Cys fits (Jin et al., 2007). Lastly, the structure of the TLR4:MD2:LPS complex reveals that LPS does not initially make direct contact with TLR4, but rather first binds to MD2, altering its conformation and allowing it to bind to and cause homodimerisation of TLR4 (Park et al., 2009). In this case, TLR4 residues important for LPS activity include those required for MD2 binding and those required for receptor homodimerisation. TLR4 responses to LPS also require the presence of CD14 which

Our knowledge of the receptor complexes used by DAMPs is far from complete, but it is already clear that these molecules have a further level of complexity in their receptor organisation. Thus, of those that require TLR4, some, such as HSP70, biglycan and s100 also require both CD14 and MD2 for activity. Others, such as Gp96, HMGB1 and fibronectin EDA require only MD2, whilst another group that includes surfactant protein A and lactoferrin require only CD14 (Piccinini et al., 2010). The last, and probably the most diverse group of DAMPs use co-receptors or accessory molecules that are distinct from CD14 and MD2. Immune complexes of citrullinated fibrinogen for example have recently been shown to use a combination of Fcgamma receptors and TLR4 (Sokolove et al., 2011), and may also use CD11b/ CD18 (Barrera et al., 2011). A-SAA, which has been associated with RA and uses the TLR2 receptor has been shown to also use both CD36 and FPRL-1 as co-receptors, while low molecular weight hyaluronan forms complexes with TLR2 and CD44, and biglycan, which may use both TLR2 and TLR4, uses a variety of molecules including CD14, MD2, P2x4 and NLRP3 (Babelova et al., 2009). Other DAMPs such as tenascin-C do not use CD14 or MD2 (Midwood et al., 2009). Whether they bind directly to TLR4 or use an as yet

Because of their alternative use of co-receptor molecules, DAMPs are therefore likely to use different residues on the TLR4 molecule than those used by LPS, so it may not be surprising that SNPs that affect LPS binding and antibodies designed to prevent LPS activation of TLR4 are inactive in RA where DAMP mediated activation of TLR4 may be critical to disease pathology. This hypothesis is confirmed by studies of the D299G and T399I mutations in TLR4. These have been shown to prevent LPS activation, but to enhance the

The signalling mechanisms used by DAMPs are also not well defined, and there is very little data available at present to suggest therapeutic targets for DAMPs. However, the use of TLR molecules suggests that many of the same pathways activated by PAMPs may be relevant. This has been confirmed by recent studies of oxidised LDL signalling (a TLR4 DAMP), which shows activation of many familiar pathways such as those involving IKK and the MAP kinases. Studies with tenascin-C also show that MyD88 signalling is important in response to this molecule (Midwood et al., 2009). Any differences between DAMP and LPS mediated signalling pathways are likely to come from the DAMP use of alternative co-receptors. Molecules such as CD36, CD44 and integrins already have defined signalling pathways. Whether DAMP signalling will prove to be simply a combination of TLR:MyD88 driven pathways and those emanating from any co-receptor molecules remains to be defined. It is more likely however, that the signal transduction mechanism of DAMPs will be a result of a

combination of both pathways, where each is able to modulate / modify the other.

facilitates the transfer of LPS to MD2.

undefined co-receptor molecule(s) is unclear at present.

ability of TLR4 to respond to fibrinogen. (Hodgkinson et al, 2008)

In conclusion, it is clear that whilst the advent of specific biological therapies for the treatment of RA has significantly improved treatment of this disease in the last 10 years, there is still a significant un-met clinical need for therapies that target the cause of disease chronicity rather than its consequences.

 Increasing evidence from *in vitro* studies, murine models of disease, and human studies, suggests that the TLRs play a significant role in RA. In particular, TLRs 2 and 4 and the intracellular TLRs 3 and 8 are increasingly regarded as key to the pathogenesis of this disease. However, PAMPs derived from infectious agents are not found in RA joints and are unlikely to be the causative TLR ligands in RA. Rather, there is now increasing evidence that endogenously derived DAMPs, either in their native form or in a citrullinated state, are able to drive the chronicity of RA. DAMPs comprise an enormously diverse subset of molecules and targeting them as a form of RA therapy could be achieved in a number of different ways. Many of these approaches have already found some success in other fields, but have yet to be tested in RA.

For example, for DAMPs whose expression is specifically up-regulated during inflammation it may be possible to manipulate this induction of expression at the genetic level. This approach has been taken in a murine lung carcinoma model where knockdown of versican expression in Lewis lung carcinoma cell lines (LLC) ablated their tumorigenic capability, promoting mouse survival and reduced metastasis. Conversely, over expression of versican in LLC lines with low innate metastatic potential increased lung metastasis (Kim et al., 2009).

The use of micro RNAs (miRNAs) to manipulate gene expression is also attracting considerable attention. MicroRNAs are endogenous RNAs that post-transcriptionally modulate gene expression (reviewed in (Guo et al., 2010). Not surprisingly, regulation of gene expression by microRNAs has also been extended to the TLR signalling paradigm (reviewed in (O'Neill et al., 2011)) where they impose several levels of regulation on the TLR signalling axis. For example, miR-155, miR-21 and miR-147 regulate the expression of TLRs 2-4, downstream signalling mediators such as MyD88 and TRIF, as well as transcription factors NF-B and IRF3 (reviewed in (O'Neill et al., 2011)). Recent studies have reported that miR-155, miR-146a and miR-203 are upregulated in RA synovial fibroblasts, resulting in altered cytokine and MMP synthesis (Stanczyk et al., 2008; Li et al., 2010; Stanczyk et al., 2011). These insights may create a novel approach to limiting excessive TLR activation during inflammation.

Targeting DAMP Activation of

Toll-Like Receptors: Novel Pathways to Treat Rheumatoid Arthritis? 225

blocking elastase could reduce the production of endogenous TLR4 activators. Indeed, pretreatment with neutrophil elastase inhibitor before induction of hepatic ischemiareperfusion injury has already been shown to ameliorate liver damage (Uchida et al., 2009). Thirdly, as we have discussed here, targeting the TLRs that are critical for DAMP recognition is increasingly considered as a viable therapeutic option in RA. This can take the form of agents that antagonise DAMP:TLR association as has been tested with the antagonistic TLR2 and TLR4 antibody studies and the DNA based TLR7/9 antagonist IMO-3100 and with the studies on *Bartonella Quintana* LPS and the ST2 protein. Other approaches tested so far include the use of antagonistic bent oligonucleotides that have a high affinity for HMGB1 and suppress HMGB1-induced proliferation and migration of smooth muscle cells *in vitro* (Musumeci et al., 2007). An engineered mutant fragment of HMGB1 (HMGB1 Mut (102-105)) that carries two glycine substitutions has also been shown to decrease TNF release induced by the full-length HMGB1 protein in human monocyte cultures (Yuan et al., 2008). In addition, the N-terminal domain of thrombomodulin, an endothelial anticoagulant cofactor, has been shown to exert anti-inflammatory effects in a model of lethal endotoxemia partly by binding to and sequestering HMGB1 (Abeyama et al., 2005). Targeting the DAMP

co-receptor molecules may also prove to be a viable therapeutic approach.

TLR signalling pathways, activated during DAMP recognition, would also represent tractable targets in RA and the success in *in vitro* studies and animal models of RA of small molecule inhibitors directed against signalling molecules such as p38, IKK2, PDE4, Syk and Btk may in large part be due to their effect on DAMP mediated signalling pathways. In addition, it will be interesting to see if DAMP-mediated signalling pathways are subject to the same control mechanisms as PAMP-mediated TLR signals. The role of naturally occurring molecules such as SIGIRR and SARM which are reported to modulate TLR signalling, as are the TAM receptors Tyro3, Axl and Mer. Many of these have not been

examined in the context of DAMP:TLR activity and may yield further areas of study.

prove to be an exciting one for many years.

*Rheum* 58(12): 3753-3764.

The authors are supported by funding from Arthritis Research UK.

**8. Acknowledgements** 

**9. References** 

"www.opsona.com."

In summary, it is now clear that the TLR:DAMP axis represents a key point in RA pathology that will be susceptible to therapeutic attack. However, in order to mount such an attack we need to understand a number of key points: which DAMPs are relevant to RA pathology?; how are these DAMPs produced and / or modulated to become pathogenic?; how do the TLRs recognise DAMPs, including the role of co-receptor molecules?; and which signals are generated? The field of DAMP research in RA should

Abdollahi-Roodsaz, S., L. A. Joosten, et al. (2008). "Shift from toll-like receptor 2 (TLR-2)

toward TLR-4 dependency in the erosive stage of chronic streptococcal cell wall arthritis coincident with TLR-4-mediated interleukin-17 production." *Arthritis* 

A second method of modulating DAMP activity in RA may be to block their production or level of expression. Indeed, ethyl pyruvate, stearoyl lysophosphatidylcholine and nicotine have already been shown to be efficacious in ameliorating experimental sepsis by preventing HMGB1 release (Ulloa et al., 2003; Wang et al., 2004; Chen et al., 2005). However, the mechanism by which they do so is unclear and these compounds are likely also to affect numerous other cell processes. HMGB1 is released from cells by two distinct mechanisms: it is either liberated from cells undergoing necrosis (Scaffidi et al., 2002), or it is hyperacetylated and then actively secreted from stimulated cells. Other DAMPs including the S100 proteins are also secreted in the same way (Foell et al., 2007) and targeting this pathway therefore may potentially offer a means to modulate the release of intracellular DAMPs. Other DAMPs are generated by the degradation of extracellular matrix and inhibition of this process may also be therapeutically beneficial. This has already been tested in the case of immune stimulatory heparin sulphate (HS) fragments which are released from the ECM as a result of elastase activity (Brunn et al., 2005). Injection of elastase into the peritoneal cavity of mice caused the release of HS and induced sepsis, nearly as effectively as direct injection of HS or LPS (Johnson et al., 2004). Thus therapeutic measures aimed at

Red star = Potential site of DAMP manipulation.

Fig. 1. Potential sites at which DAMP activity could be modulated for therapeutic advantage.

A second method of modulating DAMP activity in RA may be to block their production or level of expression. Indeed, ethyl pyruvate, stearoyl lysophosphatidylcholine and nicotine have already been shown to be efficacious in ameliorating experimental sepsis by preventing HMGB1 release (Ulloa et al., 2003; Wang et al., 2004; Chen et al., 2005). However, the mechanism by which they do so is unclear and these compounds are likely also to affect numerous other cell processes. HMGB1 is released from cells by two distinct mechanisms: it is either liberated from cells undergoing necrosis (Scaffidi et al., 2002), or it is hyperacetylated and then actively secreted from stimulated cells. Other DAMPs including the S100 proteins are also secreted in the same way (Foell et al., 2007) and targeting this pathway therefore may potentially offer a means to modulate the release of intracellular DAMPs. Other DAMPs are generated by the degradation of extracellular matrix and inhibition of this process may also be therapeutically beneficial. This has already been tested in the case of immune stimulatory heparin sulphate (HS) fragments which are released from the ECM as a result of elastase activity (Brunn et al., 2005). Injection of elastase into the peritoneal cavity of mice caused the release of HS and induced sepsis, nearly as effectively as direct injection of HS or LPS (Johnson et al., 2004). Thus therapeutic measures aimed at

Red star = Potential site of DAMP manipulation.

advantage.

Fig. 1. Potential sites at which DAMP activity could be modulated for therapeutic

blocking elastase could reduce the production of endogenous TLR4 activators. Indeed, pretreatment with neutrophil elastase inhibitor before induction of hepatic ischemiareperfusion injury has already been shown to ameliorate liver damage (Uchida et al., 2009).

Thirdly, as we have discussed here, targeting the TLRs that are critical for DAMP recognition is increasingly considered as a viable therapeutic option in RA. This can take the form of agents that antagonise DAMP:TLR association as has been tested with the antagonistic TLR2 and TLR4 antibody studies and the DNA based TLR7/9 antagonist IMO-3100 and with the studies on *Bartonella Quintana* LPS and the ST2 protein. Other approaches tested so far include the use of antagonistic bent oligonucleotides that have a high affinity for HMGB1 and suppress HMGB1-induced proliferation and migration of smooth muscle cells *in vitro* (Musumeci et al., 2007). An engineered mutant fragment of HMGB1 (HMGB1 Mut (102-105)) that carries two glycine substitutions has also been shown to decrease TNF release induced by the full-length HMGB1 protein in human monocyte cultures (Yuan et al., 2008). In addition, the N-terminal domain of thrombomodulin, an endothelial anticoagulant cofactor, has been shown to exert anti-inflammatory effects in a model of lethal endotoxemia partly by binding to and sequestering HMGB1 (Abeyama et al., 2005). Targeting the DAMP co-receptor molecules may also prove to be a viable therapeutic approach.

TLR signalling pathways, activated during DAMP recognition, would also represent tractable targets in RA and the success in *in vitro* studies and animal models of RA of small molecule inhibitors directed against signalling molecules such as p38, IKK2, PDE4, Syk and Btk may in large part be due to their effect on DAMP mediated signalling pathways. In addition, it will be interesting to see if DAMP-mediated signalling pathways are subject to the same control mechanisms as PAMP-mediated TLR signals. The role of naturally occurring molecules such as SIGIRR and SARM which are reported to modulate TLR signalling, as are the TAM receptors Tyro3, Axl and Mer. Many of these have not been examined in the context of DAMP:TLR activity and may yield further areas of study.

In summary, it is now clear that the TLR:DAMP axis represents a key point in RA pathology that will be susceptible to therapeutic attack. However, in order to mount such an attack we need to understand a number of key points: which DAMPs are relevant to RA pathology?; how are these DAMPs produced and / or modulated to become pathogenic?; how do the TLRs recognise DAMPs, including the role of co-receptor molecules?; and which signals are generated? The field of DAMP research in RA should prove to be an exciting one for many years.
