Gout. Mechanisms of inflammation in gout

An acute attack of gout is a paradigm of acute sterile inflammation, as opposed to pyogenic inflammation. Recent studies suggest that the triggering of IL-1β release from leucocytes lies at the heart of a cascade of processes that involves multiple cytokines and mediators. The NLRP3 inflammasome appears to have a specific role in this regard, but the biochemical events leading to its activation are still not well understood. We review the known mechanisms that underlie the inflammatory process triggered by urate crystals and suggest areas that require further research.


Introduction
Acute gout, also called podagra when it aff ects the fi rst metatarsophalangeal joint, is due to the infl ammatory reaction of joint tissues when monosodium urate (MSU) crystals form within the joint. Th e link between crystal deposition and joint infl ammation was made by Garrod, who developed the thread-test assay of serum urate and stated that 'the deposited urate of soda may be looked upon as the cause, and not the eff ect, of the gouty infl ammation' . Older clinical descriptions of acute gout, from Hippocrates to Charcot, detailed the very sudden and infl ammatory character of an attack.
Th e initial symptoms of a gouty attack are sudden and violent, taking place nearly always at night. Th e patient feels a sudden onset of pain … it often aff ects the big toe 1st MT joint, which becomes rapidly red and swollen; the veins of the leg can become dilated, and the leg can become purple and sometimes accompanied by bruising.
Charcot, Maladies des Vieillards, 1890 Th is account brings to life all of the cardinal manifestations of acute infl ammation. After McCarty and Hollander published their observations that the crystals found in the synovial fl uid were composed of MSU [1], there remained no doubt that urate crystals are the cause of this intense infl ammation. Th ese observations have led to detailed analyses of the infl ammatory pathways triggered by urate crystals and the mechanisms that link crystal deposition to tissue infl ammation. During the past decade, the role of innate immune responses in the initiation of infl ammatory and immune responses to injury and infection has gained prominence. In contrast to the acquired immune system, innate responses orchestrate the immediate and early phases of host defence to microbes as well as to injury, initiating the infl ammatory reaction and recruiting cells of the acquired immune system to the site of infl ammation. Th ese advances bring our attention back to the interplay between cells of the innate immune system and the products of tissue damage and cell death -the so-called danger signals or danger-associated molecular patterns. MSU represents one of these signals that activate cellular defences. In the present review, we cover the recent develop ments in our under standing of MSU's eff ects, integrating this information with previous knowledge regarding the infl ammatory response provoked by urate crystals.

Crystal factors
MSU crystallizes when its plasma concentration exceeds its solubility (around 7 mg/dl, or 420 μmol/l). Concentration may not be the only determinant, however, and other factors have been reported to aff ect urate solubility. We know from clinical observation that patients with hyperuricaemia can be asymptomatic, sometimes for long periods before the fi rst declaration of gout, so there must be modifying factors present in biological fl uids that aff ect urate solubility and their capacity to form crystals. Furthermore, patients can have known MSU deposits that are clinically quiescent -indicating that there is further regulation at the level of the tissue response to urate (see later).
Th e factor or factors predisposing certain hyperuricaemic persons and not others to develop gout are as yet unknown. In vitro, several factors have been described to infl uence urate solubility, including pH, temperature, ionic strength and the binding of urate to plasma macromolecules [2]. Katz and Schubert suggested that proteoglycans of connective tissue origin can increase urate solubility. Th eir experiments showed that increasing the concentration of a cartilage protein polysaccharide extract increased urate solubility threefold, from 6 to 17 mg/100 ml. How this may explain the site of deposition of urate crystals in vivo, however, is not yet clear [3]. Not only is the solubility of urate infl uenced by its binding to plasma proteins, but also its infl ammatory potential. Crystals coated with IgG fragments were more infl ammatory than naked crystals [4,5]. As infl ammation subsides, apolipoprotein B displaces the IgG by competitively coating sites on crystals and could contri bute in part to the resolution of acute gouty arthritis [6].
Th ese fi ndings indicate that crystal-bound proteins modify the cellular response, but the underlying mechanisms of these eff ects remain to be elucidated. As crystals are essentially foreign bodies, they have toxic eff ects on cells (see below). Surprisingly, the size of MSU crystals does not seem to be a major determinant of its infl ammatory potential, as varying the mean crystal size between 2.0 and 11.8 μm did not aff ect TNF production by monocyte/macrophage cells [7].

Cell types
Th e cells most studied in gouty infl ammation are leucocytes -in particular, neutrophils and macrophages. Addition of MSU crystals to neutrophils results in cell death and release of lysosomal and cytoplasmic enzymes.
In vitro studies showed that when native crystals were added to neutrophil preparations, ingested crystals induced the release of lysosomal contents by rupturing lysosomal membranes [8]. Th is pathway of cell death appears to be abrogated when crystals are coated with immunoglobulin, but release of lysosomal enzymes still took place [9]. Th e importance of neutrophils in the acute response is further supported by the eff ectiveness of colchicine, a potent inhibitor of neutrophil lysosomes as well as cell migration [10].
Neutrophils are attracted to the site of infl ammation by chemotactic factors as well as cytokines and chemokines (see later) that act upon endothelial adhesion and transmigration. Among the factors known to be upregulated during gouty infl ammation are IL-1β, IL-8, CXCL1 and granulocyte colony-stimulating factor [11].
Th e interactions between monocyte/macrophages with MSU do not appear to be so deadly, eliciting principally the secretion of proinfl ammatory cytokines. Both IL-1β and TNFα were secreted when monocytes were cultured in the presence of crystals, and their eff ects on cell proliferation were inhibited by specifi c blocking antibodies [12,13]. Macrophages do not appear to respond as well to MSU as to monocytes. When macrophages were diff erentiated in vitro, they became less proinfl am matory in their response to MSU crystals, even though they retained their capacity for phagocytosis [7]. In a subsequent study by the same group, a potential role of transforming growth factor beta was suggested to account for this phenomenon [14]. Th e autoregulation of infl ammation by phagocytes may explain the self-limiting nature of acute gout, as well as the fact that the presence of MSU crystals in a joint is not always accompanied by infl ammatory signs and symptoms. Th e diff erence in the cellular response seems to be linked to the state of the macrophage, and as our understanding of the phenotypic variations of macrophages evolves, including the concept of M1 and M2 macrophages, we may gain a better understanding on how this impacts on infl ammation. Of particular interest is the role of resident macrophages, as they have been reported to play a primordial part in initiating the tissue response to MSU in a mouse peritonitis model [15].
Among the other cell types that may participate in gouty infl ammation, mast cells deserve a mention. In the air pouch model of gout, the density of mast cells increased in the lining layer rapidly after crystals were injected, and was correlated with the histamine content of the pouch liquid [16]. Using a microarray approach to analyse gene expression by lining cells of the air pouch, it was interesting to observe that the gene encoding histidine decarboxylase, the enzyme that synthesizes histamine, is rapidly upregulated after injection of MSU crystals [17]. In the peritonitis model of gout, depletion of peritoneal mast cells attenuated the neutrophil infl ammatory response [18]. Finally, mast cells are able to release IL-1β on activation of the NALP3 infl ammasome through lipopolysaccharide [4], but we have no information on whether MSU can also perform this function. Th ese results suggest that mast cells participate in the acute response of tissue to MSU crystals and contribute to neutrophil recruitment.

IL-1 as a crucial mediator of gouty infl ammation
Proinfl ammatory cytokines undoubtedly have a critical role in orchestrating the infl ammatory reaction to MSU crystals. Recent attention has focused particularly on the role of IL-1. In murine models of crystal-induced infl ammation, the injection of MSU into a preformed air pouch or within the peritoneal cavity elicited an infl ux of neutrophils that was accompanied by the local accumulation of chemoattractants and cytokines. Although care needs to be taken when extrapolating fi ndings from these models to joint infl ammation in gout, both the air pouch and the peritoneum have a lining that is morphologically similar to the synovium, and both models reconstitute the complex interplay that occurs between resident and infi ltrating immune cells.
Experimental data from crystal-induced peritonitis, crystal-induced air pouch infl ammation or urate crystals injected intra-articularly provide compelling evidence for the role of IL-1 in the pain and infl ammation associated with gout. Two lines of study have been employed -a murine IL-1R1 knockout model, and the eff ects of IL-1 blockade (by antibodies against IL-1β, IL-1RΑ or by mIL-1 Trap). Th ese inhibitors of IL-1 were as eff ective as genetic deletion of IL-1R1 in reducing infl ammation and hyperalgesia in these murine models [11,19,20]. Furthermore, the rapid clinical response of patients with acute gouty arthritis to IL-1 inhibition by a variety of agents (IL-1RA, IL-1Trap or an anti-IL-1β monoclonal antibody) validates the concept that this cytokine plays a key role in gouty infl ammation [20][21][22].
Th e cellular target of IL-1β released by MSU crystals in the peritonitis model does not appear to be leucocytes. Using bone-marrow chimeras, Chen and colleagues demonstrated that MSU-induced infl ammation was not modifi ed when normal wild-type mice received IL-1RI -/or MyD88 -/bone marrow, but infl ammation was attenuated when IL-1RI-defi cient or MyD88-defi cient mice received normal bone marrow [19]. As both IL-1RI and MyD88 are essential for signal transduction on IL-1 binding, these results showed that nonhematopoieitic cells, most probably endothelial cells, are necessary to develop fully the infl ammatory response to IL-1β.

Role of NLRP3 (NALP3) infl ammasome in gout
IL-1β is the prototypic infl ammatory cytokine and has multiple actions on cells and tissues (reviewed in [23]). Th e cytokine is produced as an inactive pro-molecule by immune cells such as macrophages, monocytes and dendritic cells, and is then cleaved into the active p17 form of IL-1β to be secreted out of the cell. Cleavage of pro-IL-1β is catalysed by caspase-1 (also known as IL-1converting enzyme). Caspase-1 is a member of the family of infl ammatory caspases that include caspase-4, caspase-5, caspase-11 and caspase-12; in the context of IL-1β processing, caspase-1 accounts for the major activity that requires the formation of a molecular platform known as the infl ammasome (reviewed in [24]). Other pathways of IL-1β processing independent of caspase-1 have been described, implicating neutrophilderived and mast cell-derived proteases [25,26].
Th e infl ammasome is a cytoplasmic protein complex composed of a protein of the NLRP (or NALP) family, an adapter ASC protein as well as an infl ammatory caspase. Fourteen NLRPs have been identifi ed in humans, and they all have multiple domains that share common features. Th e ASC adapter contains a PYD domain that mediates interaction with a homologous domain on NLRP, as well as a CARD domain that interacts with caspase-1. For the NLRP3 infl ammasome, a second adapter -the Cardinal protein -is needed to recruit a second molecule of caspase-1, so as to form a dimer with the caspase-1 recruited by ASC (Figure 1). Numerous danger-asso cia ted molecular patterns and pathogen associated molecular patterns have been reported to signal via the NLRP3 infl ammasome, including bacterial cell wall components, bacterial RNA or bacteria themselves, such as Listeria monocytogenes or Staphylococcus aureus. In addition, NLRP3 can be activated by signals that induce potassium effl ux, such as ATP, via the P2X7 receptor, or by toxins such as nigericin.
Th e group of Tschopp discovered that many inorganic particles, including MSU and calcium pyrophosphate dihydrate crystals, are capable of activating the NLRP3 infl ammasome to process and secrete active IL-1β as well as IL-18 [27,28]. Th e list of NRLP3 triggers is currently ever expanding, and includes alum, hemozoin and DNA [29][30][31]. Macrophages defi cient for components of the NLRP3 infl ammasome were unable to secrete active IL-1β following stimulation with MSU and calcium pyrophosphate dihydrate crystals. Moreover, MSUinduced peritonitis was decreased in ASC-defi cient or caspase-1-defi cient mice. Colchicine, a drug commonly used in the treatment of acute gout, was found to block IL-1β maturation, probably by infl uencing crystal endocytosis and/or presentation of crystals to the infl ammasome. Th e results show that MSU crystals initiate an infl ammatory cascade, the starting point being the release of active IL-1β from monocytes and macrophages. Th ese fi ndings, however, also raise questions about the steps that connect cellular contact with crystals with infl ammasome activa tion, processes that are still not completely understood. It may involve general mechanisms shared by other infl ammasome activators, such as potassium effl ux that is regulated by K + channels like P2X7, or by the sensing of reactive oxygen species that are released during cell stress.

Other cytokines
Besides IL-1β, both IL-6 and TNFα are also upregulated when monocytes are in contact with MSU in vitro and with gouty tissues in vivo [11]. Although there is clear evidence that TNF is released at the site of infl ammation in animal models, there has not been any systematic study into the eff ects of interfering with the TNF signalling pathway in gout. Unpublished data obtained by the authors showed that TNF blockade with an anti-TNFα antibody did not block neutrophil infl ux in the peritonitis model, arguing against a major role for TNF; but there are anecdotal data of chronic gout that responded to anti-TNF therapy in patients [32][33][34]. Th e role of IL-6 is even less clear, although its known proinfl ammatory properties on the acute phase response and on leucocytes and endothelium would suggest a role in amplifying infl ammation.
Th ere is a growing body of data showing that chemokines also play a central role in acute gouty infl ammation, particularly in neutrophil recruitment. IL-8 (CXCL8) and closely related chemokines, such as GROα (CXCL1), bind to the receptor CXCR2 to promote neutrophil chemo taxis and can also aff ect angiogenesis. In this context, acute MSU crystal-induced neutrophilic infl ammation (4 hours after crystal injection into a subcutaneous air pouch) was completely dependent on chemokine ligands of CXCR2 in mice, as demonstrated in experiments using CXCR2 knockout mice [35]. A role for IL-8 was also confi rmed by neutralizing IL-8 with an antibody in a rabbit model of MSU-induced arthritis, resulting in inhibition of the delayed phase of neutrophil infl ux at 24 hours [36].
Th e evidence of a role for these and other cytokines is summarized in Table 1.

Toll-like receptors in gout: are they important?
Toll-like receptors (TLRs) are pivotal sensors of infectious and danger signals on leucocytes, and are an integral part of the innate immune system [38] -and their role in the pathogenesis of infl ammatory diseases has become increasingly clear [39]. MSU crystals may be a danger signal released from injured cells that trigger innate immune responses via TLRs, but the experimental evidence available to date provides no clear answer.
Using knockout mice in models of crystal-induced infl ammation, macrophages from TLR2 and TLR4 knockout mice showed decreased expression of infl ammatory cytokines such as IL-1β, TNFα and transforming growth factor beta-1 on exposure to MSU [40]. Furthermore, CD14 -an adaptor molecule that is shared by TLR2 and TLR4 to promote infl ammatory signalling -appears to be an important component of the infl ammatory response both in vitro and in vivo. Bone-marrow-derived macrophages from CD14 knockout mice showed reduced secretion of IL-1β in response to MSU, and leucocyte infl ux was reduced in the air pouch model in these same mice [41]. Th ese experiments also suggest that the phagocytosis phase of the infl ammatory response can be dissociated from the responder phase, when proinfl ammatory cytokines are produced and secreted. Th e absence of TLR2 or TLR4 reduced phagocytotic uptake as well as infl ammation, but defi ciency of CD14 reduced the production of IL-1β without any eff ect of crystal phagocytosis.

Figure 1. Composition of the NALP3 infl ammasome and its activation by monosodium urate.
Phagocytosis of monosodium urate (MSU) crystals leads to the generation of reactive oxygen species (ROS) through activation of NADPH oxidases. This event activates the NLRP3 infl ammasome. MSU crystals may also induce the secretion of ATP, which in turn activates P2X7R. On activation of the P2X7 receptor, there is rapid exit of intracellular potassium that triggers the NLRP3 infl ammasome. A rise in intracellular calcium is also required for the secretion of processed IL-1β. The macromolecular complex (infl ammasome) consists of NLRP3, ASC and procaspase-1, and CARDINAL. Assembly leads to activation of caspase-1, which in turn cleaves pro-IL-1β to produce biologically active IL-1β. ASC, apoptosis-associated speck-like protein containing a caspase recruitment domain (CARD); FIIND, domain with function to fi nd; LRR, leucine-rich repeat; MDP, muramyl dipeptide; NACHT, domain conserved in NAIP, CIITA, HET-E and TP1; NALP3, NACHT-containing, LRR-containing and PYD-containing protein; PYD, pyrin death domain.
Th e fi ndings in favour of a role for TLRs, however, are put into question by results from another group. Using the peritonitis model, Chen and colleagues found that none of the nine TLR-defi cient mouse strains available showed an impairment in neutrophil infl ux. To exclude the role of TLR5 or TLR8 (mice defi cient in these TLRs are not available), transfectants were generated with all the known TLRs and the NF-κB response to MSU was measured. None of the 11 TLRs tested responded to MSU stimulation. Although these fi ndings do not exclude a role for an unidentifi ed co-receptor that may cooperate with TLRs, they strongly suggest that TLRs do not recognize MSU or calcium pyrophosphate dihydrate crystals and that TLR2 and TLR4 may not be crucial for MSU-induced infl ammation as previously suggested [19].
Why these two sets of results are so diff erent is still not understood, but one can invoke possibly diff erent types of resident cells in the two models that have diff erential requirements for activation, or perhaps local modifying factors (such as proteins).
Events infl uencing the early phase of innate immune response in gouty infl ammation include expression of the triggering receptor expressed on myeloid cells-1 (TREM-1). Th is molecule is a member of a family of surface receptors that act to amplify innate responses in infectious and infl ammatory states [42]. TREM-1 expression was by murine resident peritoneal macrophage s, and infi ltrating leucocytes were upregulated in the murine air pouch model of crystal-induced infl ammation [17]. Interestingly, co-stimulation of resident peritoneal macrophages with MSU crystals and an anti-TREM-1 agonist antibody synergistically increased the production of both IL-1β and monocyte chemoattractant protein-1 compared with stimulation with crystals alone [43]. Little is known to date about the ligands that activate TREM-1, but TREM-1 blockade appears to attenuate infl ammation in animal models.

Complement and soluble mediators of infl ammation
MSU crystals also promote infl ammation by indirect eff ects on activation of infl ammatory cells, as exemplifi ed by MSU crystal-induced activation of the classical pathway of complement in vitro [44]. Th is classical complement activation process does not require immunoglobulin, but is amplifi ed by both C-reactive protein and IgG [45]. MSU crystals also activate the alternative pathway in vitro [46] and, in this process, direct cleavage of C5 to C5a and C5b is triggered by the formation of a stable C5 convertase on the MSU crystal surface [47].
Th e formation of leucotrienes and arachidonic acid metabolites induced by MSU has been extensively studied in the past, and is largely accounted for by the eff ects of MSU crystals on neutrophils and platelets recruited to the infl ammatory site [48,49]. Pretreatment of animals with a nonsteroidal anti-infl ammatory drug blocked the generation of infl ammatory prosta glandins and exudate formation, but did not inhibit local neutrophil accumulation [50].

Unresolved questions
Although our understanding of how MSU crystals trigger infl ammation has advanced considerably in the past 20 years, a number of fundamental and clinically relevant questions remain unanswered.
Th e fi rst and foremost question concerns the exact mecha nisms that lead to spontaneous resolution of infl am mation in gout. We have already discussed the possible role of transforming growth factor beta secretion by mature macrophages, but is it possible that the infl am matory response of phagocytes to MSU may require multiple signals beyond those that are already men tioned? In in vitro experiments, triggering of the infl amma some to process IL-1β is a multistep process. In the absence of a fi rst signal that induces production of pro-IL-1β, such as phorbol myristate acetate or ultrapure lipopolysaccharide, monocytes/macrophages do not spontaneously secrete mature IL-1β when cultured with MSU crystals (personal observation). Is it possible that such a fi rst signal is also required in man? If this is the case, what may it be in the clinical situation? Could remission be related to the removal of such a fi rst signal? A second question that has vexed researchers in this fi eld concerns the explanation for why crystals are not always infl ammatory. In part, this could be linked to the mechanisms evoked in the paragraph above, but another possible explanation is the surface modifi cation of crystals by proteins and other agents that may modify their phlogistic potential. A body of research from the 1990s showed that protein coating modifi ed crystalinduced infl ammation -but how does this work, and how do we integrate this with the more recent data on NLRP3 activation?
Finally, a mechanism to explain why urate-lowering therapies provoke acute fl ares needs to be found. Th is is a common clinical problem that may lead patients to abandon urate-lowering therapies. Although the strategy for prevention of gout fl ares is straightforward, a better understanding of its cause(s) will lead to more eff ective prevention.

Conclusion
Gout is a paradigm for acute sterile infl ammation that is triggered by interactions between MSU microcrystals and the local tissue environment. Our current understanding suggests that resident cells with such tissues, such as resident macrophages or monocytes, react to crystal deposition by uptake of crystals through phagocytosis. Th rough a series of steps that are still not well understood, these phagocytes activate the NLRP3 infl ammasome, resulting in processing and secretion of IL-1β. Th e subsequent recruitment of infl ammatory leucocytes to the site, mediated most probably by endothelial activation, can account for the subsequent release of infl ammatory mediators and the recognized infl ammatory manifestations of acute gout (Figure 2). Although multiple cytokines and chemokines are involved in the infl ammatory process, there is a suggestion that these are produced in a hierarchical manner, secondary to IL-1β release. Th e understanding of these mechanisms off ers us a novel opportunity to interfere with this infl ammatory pathway, and may one day develop more eff ective therapies for gout.