Background
Rheumatoid arthritis (RA) is a prevalent chronic inflammatory autoimmune disease that begins with dysregulation of the immune system and culminates in progressive joint destruction leading to increased morbidity and mortality. While lymphocytes are crucial for the initiation of this disease [
1], professional phagocytes of the innate immune system, including macrophages and dendritic cells (DCs), are also emerging as critical cell populations in the pathogenesis of RA. There are increased numbers of circulating monocytes in the peripheral blood of patients with RA [
2], which in the joint differentiate into pro-inflammatory macrophages, leading to the elevation in macrophage numbers that is associated with articular destruction. Synovial macrophages are highly activated in RA, express elevated levels of toll-like receptors (TLR) 2, 3, 4 and 7 [
3,
4] and contribute to inflammation and cartilage and bone destruction through the production of degradative enzymes and cytokines/chemokines. Approved therapeutic agents decrease inflammation, bone destruction and macrophage numbers in the synovial sublining [
5‐
7]. Further, reduced macrophage number correlates with better outcomes in RA and is a potential biomarker for efficacious therapies [
8,
9]. While synovial macrophages are critical for the pathogenesis of RA, the impact of DCs in the arthritic joint is less defined. DCs contribute to the marked increase in leukocyte infiltration into the synovial tissue in patients with RA [
10], where they may contribute to the initiation of disease by producing cytokines and presenting arthritogenic antigens, which in combination activate autoreactive T cells [
11,
12]. Evidence from human synovial tissue in RA and murine models of RA indicates that DCs drive the formation of ectopic lymphoid organs commonly found in synovial biopsies in RA [
13]. A substantial portion of patients with RA present with a clear type I interferon signature, thereby potentially implicating plasmacytoid DCs in disease pathogenesis, as this population is a major producer of type I interferon [
14,
15]. Irregular expression of Fc-γ-RII and hyperactive responses to stimulation of TLR2 and TLR4 are observed in dendritic cells from patients with RA with increased disease activity [
16,
17]. Lower levels of circulating DCs in patients with RA suggest that plasmacytoid and myeloid DCs may selectively home to the inflamed joint [
18‐
22]. Moreover, this decrease correlates with the presence of a population of DCs enriched for a high T cell stimulatory capacity in the inflamed synovium [
18‐
22]. Although macrophages and DCs are clearly implicated in the pathogenesis of RA, relatively little is known about the mechanisms behind their involvement.
Caspase-8 is a cysteine-aspartic acid protease critically involved in two essential death pathways, apoptosis and necroptosis, responsible for the fate of a cell. Stimulation of a death receptor (Fas or tumor necrosis factor receptor 1 (TNFR1)) by its ligand recruits Fas-associated death domain protein (FADD) [
23]. This protein aggregation then recruits pro-caspase-8, which upon dimerization becomes active. Active caspase-8 initiates the degradative phase of apoptosis through caspase-3/7 activation or blocks necroptosis via receptor-interacting serine-threonine kinase (RIPK) 1-RIPK3 suppression, depending upon the availability of cellular FADD-like interleukin (IL)-1β-converting enzyme (FLICE)-inhibitory protein (cFLIP) [
23,
24]. Should levels of cFLIP be low, caspase-8 homodimers form and apoptosis ensues [
24]. Conversely, high levels of cFLIP enable formation of caspase-8-cFLIP heterodimers, which limit RIPK signaling for necroptosis and prevent apoptosis [
24]. In the absence of caspase-8, apoptosis cannot occur but RIPK signaling proceeds unchecked, leading to necroptosis [
24]. While
RIPK1
-/- mice die perinatally [
25] and
RIPK3
-/- mice show no gross defect in development [
26], global deletion of either RIPK rescues the embryonic lethality associated with global knockout of caspase-8 [
24,
27].
Aside from its prescribed role in necroptosis, RIPK signaling has also been implicated in cell-death-independent activities in innate immune cells. Examination of a kinase-dead mutant of RIPK1 (D138N) in macrophages suggests that RIPK1 kinase activity promotes acute inflammatory responses to lipopolysaccharide (LPS) [
28]. Evaluation of an alternate kinase-dead mutant of RIPK1 (K45A) in macrophages shows that RIPK1 kinase activity plays a critical role in promoting host responses to inflammatory stimuli and cytokine signaling [
29]. It has also been shown that
RIPK3
-/- DCs are highly defective in LPS-induced expression of inflammatory cytokines and contribute to a reduced response to injury-induced tissue repair in a colitis model [
30]. Similar to RIPK, mounting evidence implicates caspase-8 in death-independent activities [
31‐
36] that may also require RIPK. Our own studies supplement death-independent functions for caspase-8. Studies of
Cre
LysM
Casp8
flox/flox mice, where caspase-8 is deleted in lysozyme M-expressing cells, reveal that caspase-8 associates with RIPK1 and RIPK3 to limit its signaling following TLR activation by gut microflora and prevents the continued activation of these populations to keep systemic inflammation in check [
37]. Further,
Cre
CD11c
Casp8
flox/flox mice, where caspase-8 is deleted in CD11c-expressing cells, develop systemic autoimmunity independent of DC lifespan, indicating that caspase-8 signaling in CD11c-expressing populations maintains tolerance [
38]. Although RIPK3 is dispensable for this process in CD11c-expressing cells, uncontrolled TLR activation in an RIPK1-dependent manner is responsible for the enhanced functionality of caspase-8-deficient DCs [
38]. Collectively, these data connect caspase-8 and RIPK not only to death, but also to death-independent inflammatory processes.
A genome-wide association study meta-analysis in more than 100,000 subjects of European and Asian ancestry (29,880 were diagnosed with RA), evaluated nearly 10 million single-nucleotide polymorphisms (SNPs) and identified a SNP associated with risk of RA development within the locus containing the gene encoding for both caspase-8 and cFLIP [
39]. Despite this newly discovered link between caspase-8 and RA, the cellular mechanisms by which caspase-8 mediates this predisposition to RA are unknown. Here, we investigated how caspase-8 signaling impacts development and progression of the acute K/BxN serum-transfer-induced arthritis model of inflammatory arthritis that resembles the effector stage of RA. The K/BxN serum-transfer-induced arthritis model entails initiation, developmental/propagation and resolution phases and is advantageous because it is not T and B cell dependent but rather depends on innate immune cells, including macrophages and DCs, to mediate disease [
40‐
44]. We showed that
Cre
LysM
Casp8
flox/flox mice resolve K/BxN serum-transfer-induced arthritis more rapidly than control mice, which suggests that caspase-8 in this context prolongs the inflammatory response. In stark contrast to
Cre
LysM
Casp8
flox/flox mice,
Cre
CD11c
Casp8
flox/flox mice exhibited a more rapid and severe onset of arthritis, indicating that in this caspase-8-deletion construct, caspase-8 controls the magnitude of the initial inflammatory response. Further, in
Cre
CD11c
Casp8
flox/flox mice, caspase-8 is implicated in the maintenance of synovial tissue-resident macrophages that can limit arthritis development. We observed that global deletion of RIPK3 in both of our caspase-8 deletion constructs (
Cre
LysM
Casp8
flox/flox and
Cre
CD11c
Casp8
flox/flox) caused the response to K/BxN serum-transfer-induced arthritis to revert back to that of control
Casp8
flox/flox mice, potentially independent of controlling necroptosis. These data suggest that deletion of caspase-8 leads to unchecked action of RIPK3, and this delicate balance maintains homeostasis within the joint.
Discussion
Rheumatoid arthritis (RA) affects nearly 1% of the world’s population, making it one of the most prevalent autoimmune diseases. While aberrant monocyte/macrophage and DC function have been detected in the RA synovium, the underlying mechanisms remain largely a mystery. We show here that caspase-8 in lysozyme M-expressing cells promotes a prolonged inflammatory response, as Cre
CD11c
Casp8
flox/flox mice exhibit reduced severity and accelerated resolution of K/BxN serum-transfer-induced arthritis. In contrast, loss of caspase-8 in CD11c-expressing cells controls the magnitude of the initial inflammatory response, as Cre
CD11c
Casp8
flox/flox mice have accelerated induction and exacerbated severity of the effector phase of disease. These data suggest that intact caspase-8 signaling maintains opposing roles in lysozyme M-expressing and CD11c-expressing cells in the pathogenesis of RA. Interestingly, in both cre recombinase constructs, caspase-8 is deleted in both of the synovial macrophage subpopulations and the CD11b+ DCs in the naïve joint. However, in the spleen, a secondary lymphoid organ, caspase-8 is restricted to the neutrophil and monocyte/macrophage populations in Cre
LysM
Casp8
flox/flox mice and the conventional DC populations in Cre
CD11c
Casp8
flox/flox mice. We postulate that caspase-8 deletion in the neutrophils and monocytes of Cre
LysM
Casp8
flox/flox mice that enter the joint under inflammatory conditions may contribute to the differing responses to K/BxN serum-transfer-induced arthritis in Cre
LysM
Casp8
flox/flox and Cre
CD11c
Casp8
flox/flox mice. These data suggest that the specificity of the Cre
LysM and Cre
CD11c deletion constructs are not so clear within tissue-resident populations.
We previously demonstrated that circulating Ly6C
lo monocytes are critical for the induction of K/BxN serum-transfer-induced arthritis. Further, we showed that naïve murine joints contain both MHC II
+ and MHC II
– macrophages, with the majority being MHC II
– tissue-resident macrophages that are capable of limiting the initiation of K/BxN serum-transfer-induced arthritis [
50]. Here we show that
Cre
CD11c
Casp8
flox/flox mice possess an increased population of Ly6C
lo monocytes, which potentially facilitate the observed accelerated initiation of K/BxN serum-transfer-induced arthritis.
Cre
CD11c
Casp8
flox/flox mice are also predisposed to a reduced proportion of MHC II
– macrophages in the naïve joint, suggesting that the lack of a sufficient population of these cells at the onset of disease may contribute to the accelerated initiation of arthritis. Further, we show that caspase-8 potentially controls the endocytic capacity of macrophages, as caspase-8-deficient synovial macrophages in
Cre
CD11c
Casp8
flox/flox mice express reduced CD206. Although the M1/M2 macrophage classification system may not be entirely relevant beyond in vitro settings, elevated CD206 expression has been associated with M2 alternatively activated macrophages that participate in wound healing and remission and/or prevention of disease [
54]. It is possible that the observed reduction in CD206 may render caspase-8-deficient macrophages less capable of endocytosing cellular debris arising from the damage induced by the arthritic inflammatory assault, and are therefore unable to control the ensuing inflammation. Further, based on in vitro studies, M2 macrophages exhibit poor antigen-presentation capabilities unlike classically activated macrophages [
54]. This reduction of CD206 may be indicative of a smaller proportion of M2-like macrophages in the joints of
Cre
CD11c
Casp8
flox/flox mice. Therefore, caspase-8-deficient synovial macrophages may show increased antigen-presenting capabilities. Thus, further analysis is required to determine how reduced CD206 affects synovial macrophage function. Taken together, these data suggest that within the joint, the caspase-8/RIPK3 signaling axis controls macrophage function, potentially not through death-related mechanisms.
The function of caspase-8 extends to inhibition of signaling through RIPK, a family of enzymes that turn on programmed necrotic cell death, or necroptosis. However, RIPK inhibition by caspase-8 not only leads to suppression of necroptosis but also potentially death-independent, cell-specific processes including inflammation. We have previously shown that caspase-8 controls the response to TLR activation in monocytes/macrophages, while caspase-8 limits DC activation and prevents a break in tolerance, and both of these functions are RIPK-dependent. In vitro studies implicate RIPK1 in the hyper-activation of bone marrow-derived macrophages and bone marrow-derived DCs from
Cre
LysM
Casp8
flox/flox and
Cre
CD11c
Casp8
flox/flox mice following TLR activation [
37,
38]. These data suggest that uncontrolled RIPK1 activity might contribute to the systemic lupus erythematosus (SLE)-like symptoms of caspase-8-deficient mice. Further, symptoms of systemic inflammation in myeloid cell-specific caspase-8-deficient mice are ameliorated by deletion of RIPK3 [
37]. However, we find that RIPK3 is not involved in the aggressive SLE-like disease in
Cre
CD11c
Casp8
flox/flox mice [
38,
55]. While a recent report shows that
RIPK3
–/– mice display more rapid resolution of K/BxN serum-transfer-induced arthritis compared to control mice [
56], we did not observe this pattern in our study, which may be the result of K/BxN serum differences, colony environment and/or diet. We did find that
RIPK3
–/–
Casp8
flox/flox mice have slower induction of arthritis than
RIPK3
–/– mice, potentially owing to interactions from the remaining 129 background present in both
Casp8
flox/flox and
RIPK3
–/– strains despite being backcrossed to B6 for over 12 and 4 generations, respectively [
26,
37,
38,
46]. However, we observed that global deletion of RIPK3 in
Cre
LysM
Casp8
flox/flox and
Cre
CD11c
Casp8
flox/flox mice causes the response to K/BxN serum-transfer-induced arthritis to revert to that of
Casp8
flox/flox control mice, indicating that the aberrant responses observed require RIPK3 action. Further, global deletion of RIPK3 in
Cre
CD11c
Casp8
flox/flox mice reversed the response to levels below that of
Casp8
flox/flox mice, potentially indicating a caspase-8-independent effect of RIPK3 in
Cre
CD11c
Casp8
flox/flox mice; future studies are required to understand this phenomenon. Caspase-8 initiates the degradative phase of the apoptotic cascade. However, in our mixed bone marrow chimera mice,
Cre
CD11c
Casp8
flox/flox-derived synovial macrophage subsets were detected at decreased proportions compared to WT-derived subsets, indicating that caspase-8-deficient synovial macrophage subsets do not accumulate and/or persist in the joint due to a lack of apoptosis. One interpretation of these data is that synovial macrophage subsets are succumbing to RIPK3-mediated necroptosis, since caspase-8 is not present to inhibit RIPK3. However, the proportion of
RIPK3
–/–
Cre
CD11c
Casp8
flox/flox-derived macrophage subsets were not restored to WT proportions, indicating that caspase-8-deficient macrophages are not undergoing necroptosis in the naïve joint. These data point to a death-independent function for the caspase-8/RIPK3 signaling axis within the joint; however, further investigation will be required to determine potential mechanisms.
IL-1β is a key inflammatory cytokine in the pathogenesis of RA, as highlighted by its role in driving cartilage destruction and the efficacy of its blockade in both mice and humans [
57]. IL-1β is predominantly produced by innate immune cells through activation of the NLRP3 inflammasome [
58]. Humanized mice expressing disease-associated mutations in NLRP3 develop normally but acquire progressive and debilitating arthritis with age [
59]. Numerous studies have implicated caspase-8 in controlling activation of the NLRP3 inflammasome. Loss of caspase-8 in DCs and macrophages facilitates LPS-induced and Pam
3Cys-induced NLRP3 activation through RIPK3 [
55,
56,
60]. Further, in the absence of both inhibitors of apoptosis (IAPs) and caspase-8, RIPK3-mediated NLRP3 inflammasome activation can occur [
56]. However, in contrast, under certain circumstances, caspase-8 can directly cleave pro-IL-1β into its mature form [
61‐
63], suggesting that caspase-8 is necessary for IL-1β production. Indeed, a recent study suggests that whole blood cells from patients with RA show increased expression of NLRP3 and secretion of NLRP3-mediated IL-1β via TLR3 and TLR4, but not TLR2, activation that is driven by caspase-1 and caspase-8 [
64]. While we do not see differences in serum IL-1β between our strains, we have not yet examined IL-1β within the joint. Taken together, depending on the stimulus, the delicate balance of caspase-8/RIPK3 signaling axis may be controlling NLRP3 inflammasome activation within the arthritic joint and warrants further investigation.
A recent study examined an SNP within
Caspase-
8 in the context of a large Chinese-based cohort (615 patients with RA and 839 controls) and found no association between this particular SNP and susceptibility to RA development [
65]. However, a prior genome-wide association study identified an SNP associated with risk of RA development within the locus containing the gene encoding for both caspase-8 and the catalytically inactive homolog of caspase-8, cFLIP [
39]. It has been shown that mice with targeted deletion of cFLIP in CD11c-expressing populations develop spontaneous erosive inflammatory arthritis that resembles RA and is accompanied by the production of autoantibodies to joint antigens [
66]. However, a report by a different research group shows that mice lacking cFLIP in CD11c-expressing cells were found to develop neutrophilia (caused by excessive production of granulocyte colony-stimulating factor receptor, (G-CSF)) and splenomegaly, but do not spontaneously develop arthritis, potentially owing to either differences in the efficiency of cFLIP deletion or variability in colony environment [
67‐
70]. In contrast to both models of CD11c-specific deletion of cFLIP, caspase-8 deletion in CD11c-expressing populations does not result in the spontaneous development of arthritis or neutrophilia. This suggests that although cFLIP is a catalytically inactive form of caspase-8, these molecules possess differing functions within CD11c-expressing populations, and further examination is warranted to determine if the RA-risk SNP affects cFLIP or caspase-8.
Caspase-8 is a downstream signaling mediator of the death receptor Fas, which has been implicated in inducible murine models of RA-like disease [
71,
72]. The onset of K/BxN serum-transfer-induced arthritis in lysozyme-M-specific
Fas-knockout (
Cre
LysM
Fas
flox/flox) mice is comparable to that of control
Fas
flox/flox mice. However, arthritis resolution is accelerated in the chronic phase in
Cre
LysM
Fas
flox/flox mice, as evidenced by the reduction of inflammation and neutrophil infiltration [
73]. Consistent with this rapid resolution of disease, higher levels of IL-10 and reduced CXCL5 (a neutrophil chemotactic chemokine) and TLR2 ligand, endoplasmin (also known as GRP94) [
73], are expressed in the joints of
Cre
LysM
Fas
flox/flox mice. Here we show that
Cre
LysM
Casp8
flox/flox, similar to
Cre
LysM
Fas
flox/flox mice, exhibit accelerated resolution of K/BxN serum-transfer-induced arthritis. This finding suggests that in lysozyme-expressing cells, Fas and caspase-8 function may be involved in the same pathway to control IL-10 expression to enable sufficient response to an inflammatory insult.
We believe that future studies will elucidate new cell autonomous mechanisms by which caspase-8 regulates arthritis pathogenesis. Here, we provide a potential mechanism between the link between caspase-8 and RA susceptibility and the cellular mechanisms by which this predisposition takes effect that includes suppression of inflammation induced by RIPK3. Therefore, these studies substantiate critical and opposing cell-specific roles for the caspase-8/RIPK3 signaling axis in arthritis pathogenesis and highlight the need for further mechanistic insight.