Enhanced TLR-MYD88 signaling stimulates autoinflammation in SH3BP2 cherubism mice and defines the etiology of cherubism

Abstract

Cherubism is caused by mutations in SH3BP2. Studies of cherubism mice showed that tumor necrosis factor α (TNF-α)-dependent autoinflammation is a major cause of the disorder but failed to explain why human cherubism lesions are restricted to jaws and regress after puberty. We demonstrate that the inflammation in cherubism mice is MYD88 dependent and is rescued in the absence of TLR2 and TLR4. However, germ-free cherubism mice also develop inflammation. Mutant macrophages are hyperresponsive to PAMPs (pathogen-associated molecular patterns) and DAMPs (damage-associated molecular patterns) that activate Toll-like receptors (TLRs), resulting in TNF-α overproduction. Phosphorylation of SH3BP2 at Y183 is critical for the TNF-α production. Finally, SYK depletion in macrophages prevents the inflammation. These data suggest that the presence of a large amount of TLR ligands, presumably oral bacteria and DAMPs during jawbone remodeling, may cause the jaw-specific development of human cherubism lesions. Reduced levels of DAMPs after stabilization of jaw remodeling may contribute to the age-dependent regression.

Figure 1. Lack of MYD88 rescues Sh3bp2^KI/KI mice from inflammation

(A) Facial appearance of 10-week-old mice. Note the open eyelids, which represent lack of facial inflammation in double mutant mouse (Sh3bp2^KI/KI/Myd88^−/−, yellow arrows) compared to the inflamed control mouse (Sh3bp2^KI/KI/Myd88^+/+, white arrows). (B) H&E staining of tissues from 10-week-old mice (from top: lung, liver, stomach, and lymph node). Arrows indicate the inflammatory lesions in lung and liver. Double-headed arrow in stomach indicates the submucosal inflammatory cell infiltration rich in macrophages. (C) MicroCT image of the calvarial bone from 10-week-old mice. (D) Histomorphometric analysis of liver lesions in 10-week-old mice. n = 8–16 (E) Quantitative measurement of calvarial bone erosion in 10-week-old mice. n = 7–11 (F) Serum TNF-α levels in 10-week-old mice. n = 10–21. Horizontal bar represents average. Error bars represent ± SD. *p < 0.05. N.D.= not detected. N.S.= not significant. See also Figure S1.

Figure 2. MYD88-mediated pathway in hematopoietic cells is responsible for inflammation in Sh3bp2^KI/KI mice

(A) Facial appearance (top), H&E staining of liver tissue (middle), and calvarial bone microCT image (bottom) of bone marrow chimeric mice (13 weeks after transplantation). Busulfan-treated 4 to 5-week-old recipient mice (recip.) were transplanted with bone marrow (BM) cells from 4-week-old donor mice. Note the open eyes (yellow arrows), and reduced liver inflammation and calvarial bone erosion in Sh3bp2^+/+/Myd88^+/+ mouse transplanted with Sh3bp2^KI/KI/Myd88^−/−BM cells compared to Sh3bp2^+/+/Myd88^−/−mouse transplanted with Sh3bp2^KI/KI/Myd88^+/+ BM cells. White and black arrows indicate closed eyelids due to facial skin inflammation and severe liver lesions, respectively. (B) Histomorphometric analysis of liver lesions (n = 3–11), quantitative measurement of calvarial bone erosion (n = 3–7), and serum TNF-α levels (n = 3–11) of recipient mice 13 weeks after transplantation. Horizontal bars represent average. Error bars represent ± SD. *p < 0.05. N.D.= not detected. N.S.= not significant.

Figure 3. Absence of TLR2 and TLR4 rescues inflammation in Sh3bp2^KI/KI mice

(A) Facial appearance (top), H&E staining of liver and stomach tissues (middle), and calvarial bone microCT image (bottom) of IL-1R1-deficient Sh3bp2^KI/KI, TLR2/TLR4-deficient Sh3bp2^KI/KI, and IL-1R1/TLR2/TLR4-deficient Sh3bp2^KI/KI mice at 10 weeks of age. Yellow arrows indicate open eyelids representing rescued facial inflammation. White arrows indicate closed eyelids due to un-rescued facial inflammation. Black arrows indicate inflammatory lesions in liver and stomach. (B) Histomorphometric analysis of liver lesions (n = 7–17) and quantitative measurement of calvarial bone erosion (n = 7–11) in 10-week-old mice. (C) TNF-α levels in serum of 10-week-old mice. n = 6–22. Horizontal bars represent average. Error bars represent ± SD. *p < 0.05. N.D.= not detected. N.S.= not significant. Tlr4^del/del = Tlr4^{lps-del/lps-del}.

Figure 4. Increased response to TLR ligands in Sh3bp2^KI/KI macrophages

(A–H) TNF-α levels in culture supernatants of BMMs stimulated with TLR ligands. BMMs from Sh3bp2^KI/KI (KI/KI), Sh3bp2^+/+ (+/+), and Sh3bp2^−/−(−/−) mice were cultured in the absence of M-CSF for 4 to 6 hours, then stimulated for 24 hours with (A) Pam3CSK4, (B) FSL-1, (C) heat killed Porphyromonas gingivalis (HKPG), (D) heat killed Listeria monocytogenes, (E) Poly(I:C), (F) LPS, (G) ssRNA, and (H) CpG oligo DNA. Dose-dependent effects of TLR stimulation on TNF-α production (upper) and the critical role of MYD88 in TNF-α production in Sh3bp2^KI/KI BMMs (except Poly(I:C)) (lower) are presented. n = 3. Representative results from two independent experiments. (I) Quantitative PCR analysis of TNF-α mRNA expression in BMMs stimulated with Pam3CSK4 or LPS. Average expression level in Sh3bp2^+/+ BMMs at 0 hour was set as 1. β-Actin was used as an endogenous control. Data are presented as means ± SEM from 3 independent experiments. (A–H) Error bars represent ± SD. *p < 0.05 vs. Sh3bp2^+/+ BMMs. Tlr4^del/del = TLR4^{lps-del/lps-del}.

Figure 5. P416R cherubism mutation increases TNF-α production through SYK-mediated NF-κB activation

(A) Effect of IKK and SYK inhibitor on TNF-α production. BMMs were starved for M-CSF for 4–6 hours, followed by the stimulation with Pam3CSK4 or LPS for 24 hours in the presence or absence of different concentrations of BMS-345541 or BAY-613606. *p < 0.05 (KI/KI vs. +/+), n = 3. Representative results from 3 independent experiments. (B) Western blotting analysis of BMMs stimulated with Pam3CSK4 or LPS. Before stimulation BMMs were starved for serum and M-CSF for 4–6 hours. (C) Effect of SYK inhibitor (1 μM) on the activation of NF-κB pathway in the BMMs at 240 minutes after Pam3CSK4 or LPS stimulation. (D) Effect of SH3BP2 tyrosine 174, 183, and 446 residues on TNF-α overproduction. RAW264.7 cells overexpressing wild-type (WT) or P416R mutant SH3BP2 with or without substitution of the tyrosine residues (Y) with phenylalanine (F) were seeded. Supernatants were collected after 24 hours (n = 5). SH2: C-terminal SH2-domain of SH3BP2. 3Y: Y174/183/446Y (no tyrosine substitutions). 3F: Y174/183/446F. (E) Western blotting analysis of BMMs stimulated with Pam3CSK4 or LPS with antibody against Y183-phosphorylated SH3BP2 (pY183, arrowheads) (F) Effect of BAY-613606 (1 μM) on SH3BP2 Y183 phosphorylation in BMMs at 240 minutes after Pam3CSK4 or LPS stimulation. (G) Y174 phosphorylation of VAV1 (pVAV1) after Pam3CSK4 stimulation. (H) Effect of BAY-613606 (1 μM) on VAV1 Y174 phosphorylation following Pam3CSK4 stimulation for 240 minutes. Numbers indicate relative ratio of pVAV1/total VAV1. For stimulation, 10 ng/ml of Pam3CSK4 or LPS was used. Pam3 = Pam3CSK4. BMS = BMS-345541. BAY = BAY-613606. Error bars represent ± SD. *p < 0.05. See also Figure S2, S3, and S4.

Figure 6. Microorganism-independent inflammation in Sh3bp2^KI/KI mice and increased response to DAMPs in Sh3bp2^KI/KI BMMs

(A) Facial appearance, H&E staining of tissues (from top: lung, liver, stomach, and lymph node), and microCT image of calvarial bone of 10-week-old germ-free Sh3bp2^+/+ and Sh3bp2^KI/KI mice. White arrows indicate closed eyelids due to skin inflammation. Black arrows indicate inflammatory lesions. Note the development of inflammation in Sh3bp2^KI/KI mice even under germ-free conditions. (B) Histomorphometric quantitation of liver inflammation in 10-week-old Sh3bp2^KI/KI mice under SPF (n = 15) and germ-free (n = 19) conditions. (C) Quantitative measurement of calvarial bone erosion in 10-week-old Sh3bp2^KI/KI mice under SPF (n = 11) and germ-free (n = 12) conditions. (D) Serum TNF-α levels in 10-week-old Sh3bp2^KI/KI mice under SPF (n = 21) and germ-free (n = 18) conditions. (E) TNF-α levels in culture supernatants of BMMs stimulated with serum amyloid A (SAA) (1 μg/ml, top) or hyaluronic acid (100 μg/ml, bottom). (F) Quantitative PCR analysis of SAA mRNA expression in BMMs stimulated with TNF-α (10 ng/ml). Average expression level in Sh3bp2^+/+ macrophages before TNF-α stimulation is set as 1. Results from 3 independent experiments. (G) Quantitative PCR analysis of SAA mRNA expression in liver tissues from 3-week-old Sh3bp2^+/+ (n = 6) and Sh3bp2^KI/KI (n = 6) mice. (H) Histomorphometric analysis of liver inflammation in 8-week-old Sh3bp2^KI/KI mice injected with necrostatin-1 (Nec-1) (n = 6) or DMSO (n = 8). (I) TNF-α levels in serum of 8-week-old Sh3bp2^KI/KI mice injected with Nec-1 (n = 5) or DMSO (n = 7). SPF = specific-pathogen-free. Horizontal bars represent average. Error bars represent ± SD (B, C, E, G, H) or ± SEM (F). *p < 0.05. See also Figure S5.

Figure 7. SYK deletion in macrophages rescues Sh3bp2^KI/KI mice from inflammation and calvarial bone erosion

(A) Facial appearance, H&E staining of lung, liver, and stomach tissues, and microCT image of calvarial bone of 8-week-old Sh3bp2^KI/KI/LysM-Cre (−)/Syk^fl/fl and Sh3bp2^KI/KI/LysM-Cre (+)/Syk^ΔM/ΔM mice. Yellow arrows indicate open eyelids due to the rescue of facial skin inflammation. White arrows indicate closed eyelids due to inflamed skin. Black arrows indicate inflammatory lesions. (B) Histomorphometric analysis of liver lesions (n = 6–7). (C) Quantitative measurement of calvarial bone erosion (n = 6–7). (D) Serum TNF-α levels (n = 6–7). Horizontal bar represents average. (E) TNF-α levels in culture media of BMMs stimulated with Pam3CSK4 (left) or LPS (right) for 24 hours (n = 3). (F) Western blotting analysis of NF-κB pathway in Pam3CSK4- or LPS-stimulated BMMs. (G) Measurement of IL-1β levels in culture supernatants from BMMs stimulated with Pam3CSK4 or LPS (n = 3). (H) qPCR analysis of TNF-α mRNA expression in BMMs stimulated with TNF-α (10 ng/ml). Average expression level in Sh3bp2^+/+ BMMs at 0 hour was set as 1. Representative data from 3 independent experiments (n = 3). (I) Schematic proposal for the pathogenesis of cherubism. Mutant SH3BP2 increases myeloid cell responses not only to M-CSF (Ueki et al., 2007) for initiation of inflammation but also to TLR ligands (this study) for development of inflammation. Mutant SH3BP2 augments NF-κB activation in cooperation with SYK, resulting in the greater TNF-α production. SYK-mediated SH3BP2 phosphorylation at Y183 (pY183) plays a critical role in the process. Mutant macrophages also express more TNF-α and IL-1β in response to TNFR and TLR2/4 stimulation, respectively (dotted red lines). IL-1-mediated TNF-α inflammatory loop (blue lines) stabilizes the inflammation (Yoshitaka et al., 2014). High amounts of PAMPs and DAMPs in jaws contribute to the jaw-specific cherubism phenotype. Stabilization of jaw remodeling as well as age-dependent maturation of MYD88-independent immune system may explain the resolution of lesions after puberty in human patients. (A–D) 8-week-old mice were used for the histological and microCT comparisons. (E–G) Ten ng/ml of Pam3CSK4 or LPS was used for stimulation. Error bars represent ± SD, *p < 0.05. N.S.= not significant.