The nonreceptor tyrosine kinase SYK drives caspase-8/NLRP3 inflammasome-mediated autoinflammatory osteomyelitis

Chronic recurrent multifocal osteomyelitis (CRMO) in humans can be modeled in Pstpip2cmo mice, which carry a missense mutation in the proline–serine–threonine phosphatase–interacting protein 2 (Pstpip2) gene. As cmo disease in mice, the experimental model analogous to human CRMO, is mediated specifically by IL-1β and not by IL-1α, delineating the molecular pathways contributing to pathogenic IL-1β production is crucial to developing targeted therapies. In particular, our earlier findings support redundant roles of NLR family pyrin domain-containing 3 (NLRP3) and caspase-1 with caspase-8 in instigating cmo. However, the signaling components upstream of caspase-8 and pro-IL-1β cleavage in Pstpip2cmo mice are not well-understood. Therefore, here we investigated the signaling pathways in these mice and discovered a central role of a nonreceptor tyrosine kinase, spleen tyrosine kinase (SYK), in mediating osteomyelitis. Using several mutant mouse strains, immunoblotting, and microcomputed tomography, we demonstrate that absent in melanoma 2 (AIM2), receptor-interacting serine/ threonine protein kinase 3 (RIPK3), and caspase recruitment domain–containing protein 9 (CARD9) are each dispensable for osteomyelitis induction in Pstpip2cmo mice, whereas genetic deletion of Syk completely abrogates the disease phenotype. We further show that SYK centrally mediates signaling upstream of caspase-1 and caspase-8 activation and principally up-regulates NF-κB and IL-1β signaling in Pstpip2cmo mice, thereby inducing cmo. These results provide a rationale for directly targeting SYK and its downstream signaling components in CRMO.

Autoinflammatory bone diseases, including chronic recurrent multifocal osteomyelitis (CRMO), 3 osteoporosis, Paget's disease, arthritis, and periodontal disease, are increasingly per-vasive contributors to severe chronic pain, physical disabilities, and morbidity (1). CRMO is primarily a pediatric chronic inflammatory bone disease, with at least 80% of patients experiencing primary symptoms, including osteomyelitis and debilitating bone pain (2). Treatment of CRMO is currently limited to nonsteroidal anti-inflammatory drugs with escalation to corticosteroids or bisphosphonates for pain relief (3). However, all current therapeutic options have limited specificity to the pathophysiology underlying CRMO.
To study the molecular mechanisms underpinning disease manifestation, CRMO in humans can be modeled in mice that carry the L98P missense mutation in the Pstpip2 gene. Prolineserine-threonine phosphatase-interacting protein 2 (PSTPIP2), a Fes/CIP4 homology domain and Bin-Amphiphysin-Rvs (F-BAR) family protein involved in regulating membrane and cytoskeletal dynamics (4), is encoded by Pstpip2 on chromosome 18 in both humans and mice and is predominantly expressed in the myeloid lineage (5). The L98P mutation in mice is termed chronic multifocal osteomyelitis (cmo), and Pstpip2 cmo mice are phenotypically characterized by autoinflammatory disease involving the bones and skin, resulting in osteomyelitis and bone deformities. The bone lesions in both cmo disease and CRMO are associated with increased IL-1 signaling, osteoclast-mediated resorption, and an elevation of osteoclast precursors (6), but the specific inflammatory pathways critical for disease are not known.
IL-1␤ has been established as the principle driver of dysregulated cellular homeostasis, extracellular matrix composition, proinflammatory cytokine production, and osteolysis in a diverse array of autoinflammatory, hematologic, and bone diseases, including osteoarthritis (7) and multiple myeloma (8). Inhibition of IL-1␤ and IL-1 receptor (IL-1R) signaling has been shown to completely protect against disease in Pstpip2 cmo mice (9), suggesting that inhibition of IL-1␤, IL-1R, or their upstream regulators could provide significant benefit to patients with autoinflammatory bone disease. It is known that caspase-1mediated cleavage of pro-IL-1␤ is activated by the nucleotidebinding oligomerization domain-like receptor family, pyrin domain-containing 3 (NLRP3) inflammasome (10), and previous studies have established a redundant role of caspase-1 or NLRP3 with caspase-8 in mediating this cleavage and disease progression (11,12). However, the signaling cascade involved in caspase-8 activation remains not well-understood.
The nonreceptor tyrosine kinase SYK is a central regulatory molecule in innate immune Toll-like receptor and nucleotide-binding oligomerization domain-like receptor signaling pathways (13,14) and inflammatory cytokine secretion (15). SYK is also known to play a role in activating caspase-8, resulting in IL-1␤ processing (16). Based on the involvement of SYK in the caspase-8 pathway and the importance of caspase-8 in mediating cmo disease, we sought to determine the role of SYK signaling in regulating cmo disease. Here we discovered the mechanistic basis underpinning SYKdependent induction of autoinflammatory osteomyelitis. Specifically, we show that SYK critically up-regulates the pro-IL-1␤ production responsible for cmo disease progression and proinflammatory NF-B signaling, which contributes to pro-IL-1␤ up-regulation.

SYK, but not CARD9, is required for inflammatory disease progression in Pstpip2 cmo mice
In addition to the role of SYK in innate immune signaling pathways (13,14) and inflammatory cytokine secretion (15), recent evidence has indicated the involvement of SYK in a diverse range of biological functions, including cellular adhesion, platelet activation, and osteoclast maturation (24). The SYK adaptor protein caspase recruitment domain-containing protein 9 (CARD9) is expressed primarily in lymphoid tissues and contributes to innate immune signaling in response to fungal, viral, and bacterial infections (25)(26)(27). Given that SYK and CARD9 are involved in caspase-8 activation and subsequent IL-1␤ processing in bone marrow-derived dendritic cells during fungal infection (16), we explored the respective contributions of SYK and CARD9 to disease progression in Pstpip2 cmo mice. First, we monitored disease progression in Pstpip2 cmo Nlrp3 Ϫ/Ϫ Syk fl/fl LysM cre mice and Pstpip2 cmo Nlrp3 Ϫ/Ϫ Card9 Ϫ/Ϫ mice. Although Pstpip2 cmo Nlrp3 Ϫ/Ϫ Card9 Ϫ/Ϫ mice did not show protection from disease, Pstpip2 cmo Nlrp3 Ϫ/Ϫ Syk fl/fl LysM cre mice displayed nearly complete protection (Fig. 2, A and B). Next we investigated whether deletion of SYK in Pstpip2 cmo mice with intact NLRP3 would be sufficient to provide protection from disease. We found that myeloid-specific deletion of SYK alone in Pstpip2 cmo mice (Pstpip2 cmo Syk fl/fl LysM cre ) provided complete protection from disease (Fig. 2C). The structural bone lesions found by micro-CT and the popliteal lymphomegaly observed in Pstpip2 cmo , Pstpip2 cmo Nlrp3 Ϫ/Ϫ , and Pstpip2 cmo Nlrp3 Ϫ/Ϫ Card9 Ϫ/Ϫ mice Figure 1. RIPK3 and AIM2 are dispensable for disease progression in Pstpip2 cmo mice. A, incidence of disease in WT (n ϭ 5), Pstpip2 cmo Nlrp3 Ϫ/Ϫ (n ϭ 9), and Pstpip2 cmo Nlrp3 Ϫ/Ϫ Ripk3 Ϫ/Ϫ (n ϭ 5) mice over the experimental course and representative footpad images, footpad CT scans, and popliteal lymph nodes from these respective mice. B, incidence of disease in WT (n ϭ 8), Pstpip2 cmo Nlrp3 Ϫ/Ϫ (n ϭ 10), and Pstpip2 cmo Nlrp3 Ϫ/Ϫ Aim2 Ϫ/Ϫ (n ϭ 10) mice over the experimental course and representative footpad images, footpad CT scans, and popliteal lymph nodes from these respective mice.
EDITORS' PICK: Targeting SYK prevents disease in Pstpip2 cmo mice were rescued in Pstpip2 cmo Nlrp3 Ϫ/Ϫ Syk fl/fl LysM cre and Pstpip2 cmo Syk fl/fl LysM cre mice (Fig. 2, A-C). Taken together, these data suggest that SYK functions upstream of both caspase-1 and caspase-8 in inducing cmo disease, that SYK is sufficient and necessary for cmo disease induction, and that NLRP3 and CARD9 are dispensable for cmo disease progression.

SYK mediates cmo disease by promoting proinflammatory signaling but not inflammasome activation
Disease in cmo mice is mediated by the cytokine IL-1␤ (9). To investigate the role of SYK in regulating IL-1␤ up-regulation in cmo, we first measured pro-IL-1␤ expression and SYK activation in the footpads of WT and Pstpip2 cmo mice. Footpads from Pstpip2 cmo mice had increased pro-IL-1␤ expression and SYK activation with respect to those of WT mice (Fig. 3A). Myeloidspecific deletion of SYK in Pstpip2 cmo mice reduced the expres-sion of pro-IL-1␤ in footpads to a level similar to that of WT mice without affecting the expression of caspase-1 or caspase-8 (Fig. 3A). Consistent with these data, the expression of pro-IL-1␤ induced by lipopolysaccharide (LPS) treatment was increased in bone marrow-derived macrophages (BMDMs) isolated from Pstpip2 cmo mice relative to that of BMDMs from WT mice (Fig. 3B). The increased pro-IL-1␤ expression in Pstpip2 cmo mice correlated with activation of SYK. Myeloidspecific deletion of SYK in Pstpip2 cmo mice abolished the increased induction of pro-IL-1␤ in BMDMs upon LPS stimulation relative to Pstpip2 cmo BMDMs without affecting the expression of caspase-1 and caspase-8 (Fig. 3B). These findings suggest a primary role for SYK in mediating pro-IL-1␤ production and cmo disease progression.
We next sought to identify additional intracellular signaling pathways mediated by SYK signaling that contribute to induction of pro-IL-1␤ expression and excessive inflammation in Pstpip2 cmo mice. Recent evidence has demonstrated that the mitogen-activated protein (MAP) kinases ASK1 and ASK2 centrally regulate NF-B and downstream MAP kinases, including JNK, ERK, and p38, to drive autoinflammatory disease progression in the Ptpn6 spin mouse model of neutrophilic dermatosis (28). We hypothesized that NF-B and MAP kinase signaling promote cmo disease progression and that SYK plays a role in regulating this signaling. Although there was more activation of NF-B and ERK in the footpads of Pstpip2 cmo mice compared with WT mice, JNK and p38 were activated similarly (Fig. 3C). However, deletion of SYK reversed the elevated NF-B but not ERK activation in Pstpip2 cmo mice, suggesting that NF-B plays an important role downstream of SYK to mediate persistent inflammation in cmo disease.
Furthermore, SYK has been shown to regulate inflammasome activation and IL-1␤ maturation downstream of dectin-1 signaling (16). We therefore asked whether SYK regulates both NLRP3 inflammasome and caspase-8 activation upstream of IL-1␤ production. We observed similar caspase-1 and caspase-8 cleavage in BMDMs derived from WT, Pstpip2 cmo , and Pstpip2 cmo Syk fl/fl LysM cre mice in response to the classical NLRP3 inflammasome trigger LPS ϩ ATP, which was further supported by the similar gasdermin D (GSDMD) activation observed among these genotypes (Fig. 3D). In addition, we further noticed that SYK deficiency did not affect the expression of pro-IL-1␤, NLRP3, and ASC, all of which are crucial components of inflammasome signaling (Fig. 3E). These data suggest that SYK does not regulate the caspase-1 and caspase-8 activation mediated by the classical NLRP3 trigger. Overall, our data indicate that SYK regulates NF-B signaling, but not inflammasome activation, for the induction of pro-IL-1␤ to mediate disease progression in Pstpip2 cmo mice.

Discussion
Cmo has been shown to be mediated by pathological IL-1␤ production downstream of NLRP3/caspase-1 and caspase-8 (9,11,12). Disease progression occurs despite single deficiency of either caspase-1 or caspase-8 (12), which suggests that the caspases function as part of distinct complexes that are independently activated. Although caspase-1 and caspase-8 have both been shown to colocalize with AIM2/ASC speck to medi-
EDITORS' PICK: Targeting SYK prevents disease in Pstpip2 cmo mice ate pro-IL-1␤ cleavage (22), AIM2 deficiency did not provide protection in Pstpip2 cmo mice, further supporting that, in cmo disease, caspase-1 and caspase-8 operate and are activated independently in distinct complexes. In this study, we demonstrated that deficiency of SYK in Pstpip2 cmo mice prevented the induction of osteomyelitis. SYK signaling upstream of caspase-1 and caspase-8 to promote pro-IL-1␤ production centrally mediates cmo disease induction. Thus, it is interesting that deficiency of the SYK adaptor protein CARD9 did not provide protection in Pstpip2 cmo mice. In addition to promoting pro-IL-1␤ synthesis, SYK, but not CARD9, has been shown to regulate NLRP3 inflammasome activation during fungal infection (29). This suggests that the CARD9 pathway selectively transduces SYK signaling to promote pro-IL-1␤ synthesis but not inflammasome activation. Additionally, several reports have highlighted the role of SYK in regulation of the NLRP3and caspase-8-mediated inflammasomes (16,29,30). However, our data with the canonical NLRP3 trigger LPS ϩ ATP did not reveal a dependence of caspase-1 and caspase-8 processing on SYK, suggesting an exclusively diverse but specific role for SYK in mediating cmo disease. In this regard, SYK primarily acts as a pivotal regulator of pro-IL-1␤ synthesis but not as a regulator of EDITORS' PICK: Targeting SYK prevents disease in Pstpip2 cmo mice inflammasome activation; however, these two processes both converge toward the production of active IL-1␤. Recent evidence has also established central roles for the NLRP3 inflammasome and IL-1␤ signaling in several additional related disorders of nonbacterial osteomyelitis, including Majeed syndrome; synovitis, acne, pustulosis, hyperostosis, and osteitis syndrome; and deficiency of IL-1R antagonist (3,9,12). Our findings provide an important context for evaluating the role of SYK in mediating these related autoinflammatory bone disorders and for the therapeutic potential of SYK inhibitors in this disease spectrum.
The central regulatory role of SYK is not confined to IL-1␤mediated autoinflammatory disease. We have reported previously that SYK licenses MyD88 to induce IL-1␣-mediated inflammatory disease in Ptpn6 spin mice (31). Similarly, we observed increased activation of SYK in the absence of PSTPIP2, suggesting that PSTPIP2 functions to suppress SYK signaling. However, the regulatory mechanisms behind SYK activation by PSTPIP2 require further investigation. Recent evidence has established that PSTPIP2 interacts with SHIP1, which is encoded by Ptpn6 (32), suggesting that SHIP1 may be able to modulate SYK activation through its phosphatase activity.
SYK signaling is known to be activated downstream of various cell surface receptors, including CD74, integrins, C-type lectin receptors (dectin-1 and dectin-2), and Fc receptors (27). Identification of the specific triggers of SYK activation in these Pstpip2 cmo mice would further clarify the signaling mechanism and provide a deeper understanding of the progression of cmo disease. SYK signaling has also been strongly associated with the recruitment of neutrophils to areas of inflammation (33). The marked reductions in inflammation and lymphomegaly seen in SYK-deficient Pstpip2 cmo mice indicate that SYK signaling potentially mediates neutrophil recruitment in Pstpip2 cmo mice. Although T cell dysregulation has been associated with inflammatory bone diseases, previous studies have characterized the osteomyelitis in cmo disease by increased neutrophil numbers without T cell abnormalities (9,34). As neutrophils have been implicated as major contributors to IL-1␤ production in cmo (11), our findings suggest that SYK-mediated recruitment and activation of neutrophils may also play a role in promoting the boney inflammation characterizing Pstpip2 cmo mice. Previous studies have shown that inhibition of signaling pathways highly associated with caspase-8 activation and inflammatory bone disease, such as tumor necrosis factor signaling, fails to protect against cmo disease (9,12). This also indicates that current guidelines for the therapeutic use of tumor necrosis factor inhibitors in the subset of patients with CRMO and concurrent autoimmune diseases may not be effective in treating CRMO. Therapeutic options for the largely pediatric and adolescent CRMO population are limited by nonspecificity and inadequate control of pain and disease progression, which can result in physical disabilities or permanent deformities. As genetic deletion of Syk in the myeloid compartment of Pstpip2 cmo mice resulted in complete prevention of disease induction and progression, SYK and its downstream signaling components represent promising novel therapeutic targets in CRMO.

Micro-CT
A Siemens Inveon CT scanner (Siemens Healthcare) was used to capture micro-CT images. Mouse footpads were imaged with a 672 ϫ 1344 mm matrix and a field of view of 30.04 ϫ 60.08 mm with one bed position. Projections were obtained at 80 peak kilovoltage and 500 A (1050-ms exposure, 1000-ms settle time) over half rotation (440 projections), giving an isotropic resolution of 44.7 m. Inveon Research Workplace software was used to process the data.

Statistical analysis
Each experiment was repeated at least twice before inclusion in the manuscript. The log-rank (Mantel-Cox) test was used to compare statistical significance between survival curves in the two groups.