In vivo approaches to investigate ANCA-associated vasculitis: lessons and limitations

Anti-neutrophil cytoplasm autoantibody (ANCA)-associated diseases are autoimmune conditions characterized by necrotizing inflammation of small blood vessels. The immunogenesis and etiology of these conditions are unknown, but our knowledge of the immunopathogenesis has increased considerably in recent years. In this review, we discuss the animal models currently used to investigate the mechanisms of vascular injury and to test novel therapies. We outline their advantages and limitations and propose potential directions for future research.


Introduction
Anti-neutrophil cytoplasm autoantibody (ANCA)-associated systemic small vessel vasculitis (SVV) (comprising Wegener granulomatosis and microscopic polyangiitis) is a group of related autoimmune disorders characterized by infl ammatory necrosis of small blood vessels [1]. Th is results in dysfunction of supplied organs, and the principal clinical consequences are renal failure and lung hemorrhage. If the condition is untreated, the mortality at 1 year is almost 90% [2]. Despite modern treatment protocols, there are still mortality rates of 15% and 36% at 1 and 5 years, respectively [3], signifi cantly higher than other autoimmune diseases and certain malignancies. Th e immunosuppressive therapies used carry a heavy burden of adverse events; one recent study found that death in the fi rst year is three times more likely to be due to an adverse event than to the vasculitis itself [4]. In those patients not recovering renal function, renal replacement therapy carries an additional average annual cost of €31,000 to €40,000 ($42,240 to $54,500) per patient.
ANCAs are directed against enzymes stored in the azurophilic granules of neutrophils and the lysosomes of monocytes [5]. Several antigenic targets for ANCAs have been identifi ed, but ANCAs directed against myelo peroxidase (MPO) [6] and proteinase 3 (Pr3) [7,8] are most common. However, vasculitic lesions contain only scant immune deposits ('pauci-immune') and do not contain ANCAs. Th erefore, it has been argued that these antibodies are unrelated to the actual vasculitic injury and that they are epi-phenomena and not part of the disease pathogenesis. Th e last two to three decades have witnessed the gradual emergence of an empirically supported paradigm that seeks to explain how these antibodies, which are so tightly associated with clinical disease, could exert a pathogenic eff ect by direct action on neutrophils.

ANCA-SVV pathogenesis
Evidence for a pathogenic role for ANCAs comes from numerous in vitro observations that support the contention that ANCA-mediated eff ector mechanisms contribute to endothelial injury (reviewed in [9]). Th e concept that has emerged from these observations is that ANCAs and proinfl ammatory stimuli (most likely of infectious origin) synergize to cause a destructive infl ammatory process. Th e primary event in this process is that ANCAmediated activation of neutrophils causes the generation of reactive oxygen species, release of proteases, and cytokine production. Full-blown ANCA-mediated neutrophil activation requires priming with minor proinfl ammatory stimuli that induce translocation of the ANCA antigens to the cell surface, facilitating interaction with ANCAs [10]. Moreover, a recent study suggests that neutrophils from ANCA-SVV patients also have increased transcription of the ANCA antigens because of epigenetic modifi cations associated with gene silencing and thus increased autoantigen availability [11]. Following engagement of the F(ab') 2 portion of ANCAs with ANCA antigens on the cell surface, and Fc receptormediated interactions, neutrophil activation is triggered [12]. Importantly, ANCAs also increase neutrophil adher ence to endothelial monolayers, and co-incubation of ANCA-activated neutrophils and endothelial cells results in endothelial cell lysis [13]. Th ere is a large body of in vitro experimental evidence to support this paradigm. However, to study the interplay between ANCAs, neutrophils, and infectious stimuli in the complex multicellular three-dimensional environment of renal and other tissues patrolled by elements of the innate and active immune system, animal models are required. Here, we will review the animal models of ANCA SVV that have been developed and address their advantages and limitations. In addition, we will discuss how these models have contributed to dissecting the pathogenic mechanisms involved in ANCA-mediated vasculitis and how they have provided us with a test bed for novel therapies.

Insights from animal models into the pathogenesis of pauci-immune SVV in the presence of ANCAs: pathogenicity of anti-MPO antibodies
Development of animal models of MPO-ANCAmediated vasculitis has been an essential step in proving the direct pathogenic potential of anti-MPO antibodies in vivo. During the 1990s, several rodent models were developed in an eff ort to model the eff ect of anti-MPO antibodies [14]. Th ese involved inducing autoimmunity with mercuric chloride (a polyclonal B-cell stimulator) [15,16], planting of ANCA antigens within the kidney by direct infusion into the renal artery [17], or focusing the immune response to MPO on the kidney by administration of subnephritogenic doses of anti-glomerular basement membrane (anti-GBM) antibody [18]. Th ese models provided evidence for the ability of anti-MPO antibodies to exacerbate renal injury. However, the development of crescentic nephritis was reliant upon the presence of immune complexes in the kidney, or the anti-MPO response was part of a broad, nonspecifi c, antibody profi le. Th erefore, these approaches, though informative, did not accurately model the pathology of anti-MPOassociated SVV in humans.
In 2002, Xiao and colleagues [19] demonstrated that systemic administration of purifi ed murine anti-MPO IgG, obtained from murine MPO-immunized Mpo −/− mice, into recipient C57Bl/6 mice causes vasculitis. In this model, systemic injection of anti-MPO IgG resulted in urinary abnormalities (hematuria, leukocyturia, and albuminuria), early glomerular neutrophil accumulation, and focal necrotising crescentic glomerulonephritis in all recipient animals (Figure 1). In a subset of animals, vasculitic lesions were also observed in lungs, spleen, and ears. Shortly after this murine model was developed, Little and colleagues [20] developed a rat model of systemic anti-MPO-associated vasculitis ('experimental autoimmune vasculitis' , or EAV) that was characterized by generation of an immune response to exogenously administered human MPO in adjuvant, followed over a period of 4 to 8 weeks by progressive pauci-immune cres entic glomerulonephritis and lung hemorrhage ( Figure 2). Th e initial immune response in this model is to the foreign human MPO protein, but the anti-MPO anti bodies that develop cross-react with rat MPO. Th e histopathological fi ndings in both of these models resemble, to a large extent, those in human ANCA SVV. For example, in agreement with the pauci-immune nature of the glomerular capillary lesions observed in human ANCA-associated glomerulonephritis, only a limited amount of immunoglobulins and complement factors is detected in the glomeruli of mice and rats with anti-MPO-mediated glomerulonephritis.
Both the mouse and rat models of MPO-ANCA SVV convincingly demonstrate the pathogenic potential of MPO-ANCAs but clearly also have their limitations. Essentially, both cannot be regarded as genuine autoimmune models, as they rely on active immunization strategies that depend upon the use of adjuvants for disease induction. Indeed, in the murine model, the MPO-defi cient mouse has never been exposed to any MPO molecule before, so the MPO molecule is eff ectively a xeno-antigen and there is no requirement for breaking tolerance. As a consequence, high-affi nity antibodies recognizing diverse epitopes are induced.
Second, the renal phenotype of both models is mild, thereby limiting the ability to use them to adequately test novel therapies. Without the use of additional infl am matory stimuli, such as lipopolysaccharide (LPS), crescent fraction is of the order of 5% to 10% and excretory renal function is preserved. Th is contrasts sharply with the relentless loss of kidney function observed in human ANCA-associated vasculitis, in which crescents often aff ect 100% of glomeruli. Th erefore, an agent that successfully treats rodent vasculitis may not be eff ective in treating the established human condition. One important step in making the model's vasculitis severity more representative of human disease was published recently in abstract form by Xiao and colleagues [21], who made use of a diff erent strain of mouse (129S6) that is known to be more sensitive to induction of glomerulonephritis than the C57/Bl6 strain used in the original experiments. Using the same approach of passive transfer of anti-murine MPO antibodies, the authors showed that the 129S6 strain develops crescents in 50% to 60% of glomeruli, refl ecting the human situation more closely. Th ese fi ndings have not yet been replicated.
Finally, the passive transfer model developed by Xiao and colleagues [19] is induced by a single injection of anti-MPO IgG. Th erefore, the model is useful for studying the induction of acute vascular injury by anti-MPO antibodies but is less suited for studying the chronic phase of the disease because of the lack of sustained autoantibody production. To address this issue, Schreiber and colleagues [22] employed a bone marrow (BM) transplantation approach to develop a model in which the eff ects of longer-term exposure to anti-MPO responses can be investigated. In these experiments, murine MPOimmunized Mpo −/− mice were irradiated and transplanted with BM from either MPO-defi cient mice or wild-type mice. In the recipient mice, anti-MPO antibody production was preserved but only engraftment of MPO + BM cells resulted in crescentic glomerulonephritis, thereby demon strating that MPO + BM cells are essential for the development of anti-MPO-mediated glomerulonephritis. However, in this model, the disease is again relatively mild, and the contribution of radiation-induced tissue injury and MPO + T cells is unclear.

Insights from animal models into ANCA-SVV pathogenesis: mechanisms of anti-MPO-induced acute vascular injury
Th e MPO-ANCA vasculitis rodent models have proven to be very valuable for in vivo studies of eff ector mechanisms involved in the acute vascular infl ammatory phase and for the evaluation of experimental therapies (summarized in Table 1, which is adapted from [23]). In the mouse model, neutrophils are the main eff ector cells, as neutrophil depletion completely prevented vasculitis induction upon injection of anti-MPO IgG [24]. Furthermore, co-administration of LPS and anti-MPO IgG was found to severely aggravate glomerulonephritis development [25] in a Toll-like receptor 4 (TLR4)-dependent manner [26]. Th ese observations support the contention that, following infection, proinfl ammatory stimuli and MPO-ANCAs synergize to cause full-blown vasculitis. To zoom in on the very early events in the interaction of neutrophils with the endothelium, intravital microscopy analysis of the mouse cremasteric microvasculature has been employed [27]. Th is study showed that, in the presence of a local infl ammatory stimulus, anti-MPO IgG reduced neutrophil rolling while promoting adhesion and transendothelial migration of leuko cytes. Th ese MPO-ANCA-mediated neutrophil-endothe lium interactions were found to depend upon β2 inte grins and Fcγ receptors. Th e ability of anti-MPO antibodies to increase leukocyte adhesion to and transmigration through endothelium is also supported by mesenteric intravital microscopy experiments in the EAV rat model [20]. In addition, workers from Monash University in Australia have used renal intravital microscopy to visualize an acute increase in leukocyte adhesion within a more clinically relevant organ, the kidney, following infusion of anti-MPO antibodies [28]. It is conventionally thought that leukocytes do not roll or adhere in glomerular capillaries, but this group provided evidence to support a nonclassical, α4 integrin-mediated mechanism of neutrophil capture in the glomeruli. Intravital microscopy cannot yet observe events in glomeruli in the normal mouse, and these experiments have used a hydronephrotic kidney model that is likely to markedly alter the glomerular responses, thereby making it diffi cult to interpret.

Insights from animal models in ANCA-SVV pathogenesis: an unexpected role for complement
ANCA SVV is a pauci-immune condition. One does not observe deposition of complement components at vasculitic sites, and levels of complement in the blood remain unperturbed unlike, for example, in systemic lupus erythematosus. In addition, the paradigm describing the pathogenesis pathway mentioned above does not include a role for complement. Th erefore, the fi nding that mice depleted from circulating C3 by cobra venom factor as well as mice defi cient in the common complement pathway component C5, its receptor C5aR, or the alternative pathway component factor B were completely protected against anti-MPO-induced glomerulonephritis was unexpected [29,30]. Moreover, in this model, adminis tration of a C5-inhibiting antibody markedly attenuated glomerulonephritis development even when T cells CD4 + eff ector T cells contribute to anti-MPO-mediated crescentic Mouse (anti-GBM) [55] glomeruloneprhitis.

Anti-MPO IgG increases leukocyte adhesion in glomerular capillaries.
Mouse [27] Genetic susceptibility Rat and mouse strains diff er in susceptibility to anti-MPO-mediated Rat [59] crescentic glomerulonephritis.
treatment was started after disease induction [31]. Th e exact mechanism by which anti-MPO antibodies require complement for their action remains to be worked out, but these in vivo experiments have illuminated a novel therapeutic target.

Insights from animal models in ANCA-SVV pathogenesis: testing of novel therapies
Th e rodent models of anti-MPO-mediated glomerulonephritis described above have proven to be useful tools for testing experimental therapies. For example, therapeutic interventions aimed at blocking the proinfl ammatory eff ects of tumor necrosis factor-alpha (TNFα) have been evaluated in both the MPO-ANCA mouse model [25] and the EAV rat model [32]. In both, anti-TNFα treatment was benefi cial and ameliorated disease severity, although this strategy appears to be more eff ective in rats. More recently, interventions have focused on the signalling pathways involved in ANCA-mediated neutrophil activation. Employing in vitro assays and the BM transplantation anti-MPO mouse model, Schreiber and colleagues [33] demonstrated a pivotal role for phoshatidyl inositol 3 kinase-gamma (PI3Kγ) in MPO-ANCAmediated neutrophil activation and glomerulonephritis development. In these studies, transplantation of BM from PI3Kγ-defi cient mice into irradiated MPOimmunized MPO −/− mice prevented glomerulonephritis. Similar eff ects were observed in mice transplanted with wild-type BM upon oral treatment with a PI3Kγ-specifi c inhibitor (AS605240), suggesting that inhibition of PI3Kγ might be a therapeutic option in ANCA-SVV patients.
Another signalling pathway implicated in ANCAmediated neutrophil activation is the P38 mitogenactivated protein kinase (MAPK) pathway. In vitro, inhibition of P38 MAPK abrogates ANCA-induced neutrophil activation, and there is evidence that the P38 MAPK pathway is activated in glomerular lesions of ANCA-SVV patients [34,35]. Using the anti-MPO IgG/LPS model, van der Veen and colleagues [36] tested the eff ects of an orally administered P38 MAPK inhibitor on glomerulonephritis development. In this study, P38 MAPK inhibition was found to ameliorate disease severity, although the eff ects were rather moderate, reducing glomerular crescent formation by approximately 30%. Th ese data suggest that, besides p38 MAPK activity, other signalling pathways, such as the PI3Kγ (see above) and SYK [37] pathways, are activated in MPO-ANCA-mediated infl am mation and are perhaps more important in the disease process.
An alternative strategy to block the pathogenic eff ects of MPO-ANCAs was described recently by van Timmeren and colleagues [38], who focused on the autoantibodies themselves. In this study, the bacterial enzyme endoglycosidase S (EndoS) was used to specifi cally hydrolyze the conserved asparagine-linked glycans on IgG heavy chains, abolishing Fc receptor-mediated activation of leukocytes and complement [39]. In vitro, EndoS treatment of ANCA IgG markedly attenuated ANCA-mediated neutrophil activation without aff ecting antigen binding capacity whereas injection of EndoS-pretreated anti-MPO IgG in mice prevented glomerulonephritis development. Moreover, systemic EndoS administration early after disease induction rescued mice from disease progres sion. Overall, these data suggest that modulation of IgG glycosylation by EndoS is a promising strategy to interfere with the early ANCA-mediated infl ammatory processes [38].
Finally, as mentioned above, therapeutic approaches aimed at inhibiting complement activation may prove to be fruitful. However, the applicability of results from these and other novel therapies to human disease is hampered by the relatively mild phenotype in both models, both of which have renal disease that is much less severe than the kidney-threatening disease seen in human SVV.

Insights from animal models in ANCA-SVV pathogenesis: investigating anti-Pr3 antibodyassociated vasculitis
Strategies to develop Pr3-ANCA vasculitis models, in contrast to MPO-ANCA vasculitis models, have been unsuccessful so far. Using an approach similar to that of the murine anti-MPO antibody model, Pfi ster and colleagues [40] obtained anti-Pr3 antibodies from Pr3/ elastase double-knockout mice by immunizing with murine recombinant Pr3. Th e anti-Pr3 antibodies were passively transferred to wild-type recipient mice and found to aggravate subcutaneous panniculitis induced by intra dermal injection of TNFα. However, the presence of circulating anti-Pr3 antibodies, in contrast to that of anti-MPO antibodies, did not lead to vasculitic lesions in the lungs or kidneys. Along the same line, van der Geld and colleagues [41] showed that immunization of mice and rats with chimeric human-mouse Pr3 elicited an antibody response to mouse Pr3 and rat granulocytes. Again, however, no signs of vasculitis development were observed in these animals. More recently, Primo and colleagues [42] attempted to use splenocyte transfer from recombinant Pr3-immunized mice to immunodefi cient NOD/SCID (nonobese diabetic/severe combined immunodefi ciency disease) mice that lack a functioning endogenous immune system. Th is splenocyte transfer approach was also adopted by Xiao and colleagues [19] in the anti-MPO model but was abandoned because of the development of numerous immune deposits in the vasculitic lesions, thereby rendering this model nonrepresentative of human vasculitis. Anti-Pr3 antibodies were detectable at high levels in recipients of splenocytes from Pr3-immunized mice, all of which developed necrotising glomerulo nephritis. Th ese experiments support a possible patho genic eff ect of anti-Pr3 antibodies but cannot be considered defi nitive, because of the issue of immune complex deposition following splenocyte transfer. Th erefore, as convincing animal models are still lacking, our knowledge of the pathogenesis of vasculitis induced by anti-Pr3 antibodies remains limited.

ANCA SVV: outstanding issues
Clinical and experimental studies in the fi eld of ANCA SVV have contributed much to our current knowledge of disease pathogenesis, but many questions remain. First, despite several attempts and approaches (vide supra), no convincing animal model for Pr3-ANCA SVV has yet been established. Th is raises the fundamental question of whether the pathogenesis of MPO-ANCA vasculitis and that of Pr3-ANCA vasculitis are the same. It has been recognized that Pr3-and MPO-ANCA SVV patients diff er to some extent in their clinical presentation and histopathological characteristics of the vasculitic lesions [43]. Compared with patients with MPO-ANCAs, patients with Pr3-ANCAs more frequently present with extrarenal manifes tations and respiratory tract granu lomas and tend to have a higher rate of disease relapses. Moreover, systemic injection of high-affi nity anti-Pr3 antibodies, unlike that of anti-MPO antibodies, in mice does not cause vascu litis. Th e underlying mechanisms accounting for these diff erences are not easily explained but may be due to diff erences in the ability of MPO-ANCAs and Pr3-ANCAs to interact with their target antigens, to activate their target cells, or to evoke cellular immune responses [43]. Th e discrepancy in pathogenic potential between anti-Pr3 and anti-MPO antibodies in animal models may also be caused by diff erences in physicochemical properties of the antigens themselves. For example, the isoelectric points of murine and human Pr3 (approximately 7) are far less than that of MPO (greater than 10) [40], and this theoretically could result in diff erential inter actions of the antigens with negatively charged cell structures. Th us, clinical and experimental fi ndings suggest that the pathogenesis of MPO-ANCA SVV and that of Pr3-ANCA SVV may not be the same, although adequate animal models need to be established to confi rm this.
Second, why ANCA SVV primarily targets small-to medium-sized blood vessels and aff ects susceptible organs such as the kidneys and lungs is unclear. However, it is likely that the target of these diseases, the endothelium, actively participates in the induction and progression of vasculitis. Th rough expression of adhesion molecules and generation of cytokines and chemokines, activated endothelial cells are important players in driving the infl ammatory response. It is also well appreciated that endothelial cells from diff erent vascular beds are quite heterogeneous in their response to infl ammatory stimuli, and this is most likely due to organ-and function-specifi c adaptations [44]. With this in mind, it will be interesting to compare the phenotype of endothelial cells in vascular beds that are aff ected in ANCA SVV with those that are resistant. One approach could be to analyze gene and protein expression profi les of aff ected vascular beds in human tissues and experi mental models of ANCA SVV [45]. In the end, such analyses may reveal new vascular bed-specifi c targets for treatment. Th ird, the reason that ANCAs develop in the fi rst place remains unknown, although a number of theories on ANCA immunogenesis have been proposed [46], In a concept analogous to the idiotype network championed by Shoenfeld [47], Pendergraft and colleagues [48] described the presence of antibodies to a peptide encoded by (a portion of) the antisense DNA to the neutrophil granule constituent, Pr3. In this 'theory of autoantigen complementarity' , the Pr3-ANCAs are part of an idiotypic network, and the authors postulated that some antigens expressed on infectious agents (for example, Staphylococcus aureus) may act as the comple men tary antigen. Several units around the world are in the process of replicating this fi nding of antibodies to complementary Pr3 in patients with crescentic glomerulo nephritis.
A second theory invokes molecular mimicry between exogenous proteins and the ANCA antigens. Th is theory assumes that an initial immune response is evoked against pathogen-derived peptides that are highly homolo gous to peptide sequences within the ANCA antigens, resulting in a cross-reactive immune response against the ANCA selfantigens. A recent study by Kain and colleagues [49] suggests that such a mechanism may operate in ANCA SVV. In that study, the authors ob served that circulating autoantibodies against lysosomal-associated membrane protein 2 (LAMP-2), a heavily glycosylated type 1 membrane protein involved in cellular adhesion and homeostasis, are highly prevalent in patients with active focal necrotising crescentic glomerulonephritis, most of which were also seropositive for either MPO-or Pr3-ANCAs. Subsequent experiments revealed potential pathogenic eff ects of anti-LAMP-2 antibodies. In vitro, anti-LAMP-2 antibodies caused neutrophil and endothelial cell activation, and injection of polyclonal rabbit anti-LAMP-2 antibodies induced a mild form of pauci-immune crescentic glomerulo nephritis in rats. Of particular interest is the observation that a major epitope recognized by anti-LAMP-2 antibodies has strong homology with FimH-1, an adhesin of common Gramnegative bacteria. Upon immunization with FimH-1, rats developed antibodies directed against FimH-1 that crossreacted with LAMP-2 and caused crescentic nephritis. Overall, these intriguing data suggest that infections with Gram-negative bacteria may provoke an autoimmune response to LAMP-2 that in turn induces vasculitis. However, since bacterial Gram-negative infections are common and ANCA SVV is rare, other factors must also be involved in disease induction [50]. Clearly, more studies are needed to extend these observations, and confi rmation in other patient cohorts is eagerly awaited.

Animal models for ANCA SVV: what is next?
Despite their limitations, the current animal models of ANCA vasculitis are likely to be useful for further elucidation of mechanisms and factors involved in disease pathogenesis and for identifying targets for treatment. In this respect, the murine model is likely to be useful in identifying therapeutic targets at the point of acute vascular injury, whereas the EAV rat model may be more useful in identifying therapies that can be administered over a more prolonged period of time to disrupt the MPO-specifi c immune response. Th e most pressing current need is the development of a reliable model of anti-Pr3-associated vasculitis. In addition, we believe that the following issues will be important to address over the coming 5 to 10 years: Th e crucial role of neutrophils in ANCA-SVV pathogenesis is well established but other eff ector cells are likely to contribute as well. Besides neutrophils, ANCAs can activate monocytes in vitro to produce oxygen radicals [51], proinfl ammatory cytokines, and chemokines [52]. In addition, macrophages are important cellular components of the infl ammatory infi ltrate in vasculitic lesions and contribute to glomerular crescent formation [53]. Th us, it is of interest to explore the role of monocytes/macrophages in disease progression in the ANCA vasculitis models. To this end, strategies to deplete monocytes/macrophages or modify their functions could be applied in these models.
A rather unexplored area in the SVV models is the role of T cells in disease pathogenesis. In the original mouse model developed by Xiao and colleagues [19], adoptive transfer of splenocytes from mMPO-immunized MPOdefi cient mice into mice that lack mature B and T cells (RAG1 −/− mice) caused severe glomerulonephritis. In this model, adoptive transfer of pure B cells also induces disease manifestations whereas transfer of pure CD4 + T cells does not [54]. Th ese results indicate that, in this model, MPO-specifi c CD4 + T cells are not required for disease induction but do not rule out a role for these cells in the maintenance and propagation of the immune response. A study by Ruth and colleagues [55] indeed suggests that MPO-ANCA-and MPO-specifi c CD4 + T cells may work together in a unique way. Th ese authors demonstrated that immuni zation of C57Bl6 mice with human MPO in adjuvant induces a humoral (MPO-ANCA) as well as a cellular (MPO-specifi c CD4 + T-cell reactivity) immune response. In these mice, an additional challenge with a subnephritogenic dose of heterologous anti-GBM antibodies caused glomerular MPO deposition and triggered the development of severe crescentic glomerulo nephritis. Interestingly, similar experiments performed in B celldefi cient mice still resulted in crescentic glomerulonephritis despite the absence of MPO-ANCAs in these mice. On the basis of these results, the authors postulated that the eff ector phase of MPO-ANCA-associated glomerulonephritis is a two-step process requiring (a) MPO ANCA-mediated glomerular neutrophil recruitment and release of MPO and (b) CD4 + T-cell eff ector responses to induce crescentic glomerulonephritis [55]. Th e importance of CD4 + eff ector T cells in anti-MPO glomerulonephritis may be confi rmed through studies involving transfer of MPO-specifi c T cells from immunized Mpo −/− mice into wild-type recipients with or without anti-MPO IgG. Using a similar experimental setup, Gan and colleagues [56] recently investigated the role of T helper 17 (Th 17) cells in autoimmune anti-MPO glomerulonephritis. Th 17 cells are a recently identifi ed Th subset characterized by the production of eff ector cytokines such as interleukin (IL)-17A, IL-17F, IL-21, and IL-22. IL-17A is of particular interest because it has a wide range of proinfl ammatory properties promoting neutrophil and monocyte recruit ment and stimulation of release of proinfl ammatory cytokines such as TNF and IL-1 by macrophages. Interestingly, increased serum levels of IL-17 and IL-23 in conjunction with increased percentages of circulating Th 17 cells have been detected in human ANCA SVV [57]. In their studies, Gan and colleagues [56] showed that immunization of C57Bl6 mice with murine MPO resulted in MPO-specifi c dermal delayed-type hyper sensi tivity and systemic IL-17A produc tion. Upon injection of low-dose anti-GBM anti bodies, these mice developed glomerulonephritis. In contrast, IL-17A-defi cient mice were almost completely protected from disease induction, and this was due in part to reduced glomerular neutrophil recruitment. Th ese results identify IL-17A as an impor tant eff ector cytokine in the patho genesis of MPO-ANCA glomerulonephritis and suggest that targeting IL-17A may be a therapeutic option. It should be noted, however, that the models employed by Ruth and colleagues and Gan and colleagues are diff erent from the model originally described by Xiao and colleagues [19] because a subnephritogenic dose of hetero lo gous anti-GBM antibodies is used to trigger disease manifestations. Th us, eff ectively, these are models involv ing immune complex deposition in addition to anti-MPO autoantibodies.
Besides studies into the role of eff ector cells, further elucidation of the pathogenic mechanism of the ANCA autoantibodies themselves is of interest. In the mouse model, the induced polyclonal anti-MPO antibodies are pathogenic, but it is unclear whether disease induction is dependent upon specifi c antibody isotypes or antigen epitopes. Th ese issues could be addressed using monoclonal antibodies generated from murine MPO-immunized Mpo −/− mice combined with heavy-chain switch variants of these monoclonal antibodies [23].
Th e ANCA vasculitis animal models are also likely to contribute to the elucidation of genetic risk factors for disease development. Until now, genetic studies in human ANCA SVV have focused on candidate genes and have been hampered by small sample sizes [58]. Although large multicenter genome-wide studies in ANCA SVV have been initiated and are currently under way, the animal models may also off er opportunities. In the rat EAV model, Wistar Kyoto (WKY) rats have been shown to be highly susceptible to vasculitis development upon immunization with human MPO in complete Freund's adjuvant. In contrast, three other rat strains tested -Lewis, Wistar Furth, and Brown Norway -were found to be resistant to vasculitis development, although similar levels of anti-human MPO antibodies were detected [59]. Interestingly, since WKY and Lewis rats share the same major histocompatibility complex (MHC) 2 haplotype, these observations indicate that susceptibility to vasculitis development in this model is dependent on non-MHC-linked genes [59]. Similarly, preliminary studies in the anti-MPO IgG transfer mouse model have shown that 129S6 mice are much more susceptible to anti-MPO IgG-mediated glomerulonephritis induction than the originally used C57Bl6 mice [21]. Since both strains are of the H2b MHC haplotype, this again indicates that non-MHC genes are involved. Collectively, these observations in rats and mice now pave the way for more detailed genetic studies that will aid in the identifi cation of genetic risk factors for human ANCA SVV.
Finally, the rodent models of ANCA SVV may continue to be used for the discovery and testing of new targets for treatment. Possible target candidates include mediators of signalling pathways other than PI3K and P38 MAPK that have been shown to be involved in ANCA-mediated activation of neutrophils in vitro. Also, new targets may be identifi ed by analyzing vascular bed-specifi c gene and protein expression patterns or via genome-wide gene expression analysis of aff ected tissues. Since the alter native pathway of complement appears to be pivotal in anti-MPO-mediated glomerulonephritis in mice, a potential therapy could involve inhibition of components critical for this pathway, including factor B and properdin. Such therapies have been evaluated recently in other infl ammatory models and could be attractive targets for ANCA vasculitis as well [60,61].

Conclusions
Boosted by the development of various animal models for MPO-ANCA SVV, our knowledge of the unique pathogenic mechanisms involved in ANCA-mediated vasculitis has increased tremendously and this will open new avenues for therapeutic strategies. At the same time, many questions concerning the pathogenesis and immuno genesis of ANCA SVV remain. Th e current MPO-ANCA models will continue to be helpful in providing answers to these questions, although further 'fi ne tuning' of the animal models is necessary. Th e development of a convincing in vivo model for Pr3-ANCA SVV is eagerly awaited.