Tristetraprolin regulation of interleukin 23 mRNA stability prevents a spontaneous inflammatory disease

We identified several AREs within the 3′UTR regions of human and murine IL12p35, IL12/23p40, and IL23p19 mRNAs, suggesting that these genes could be regulated at the posttranscriptional level. In a first set of experiments, we generated BM-derived DCs (BMDCs) from WT or TTP^−/− mice to assess the impact of this ARE-binding protein on IL12 and IL23 production in response to TLR ligands such as LPS (TLR4 ligand). In comparison with their WT counterparts, TTP^−/− BMDC cells produced fourfold higher levels of IL23. Although IL12p70 levels were also increased in TTP^−/− BMDCs, the observed differences did not reach statistical significance (Fig. 1 A). In marked contrast, production of IL12/23p40 was comparable in both groups. We reached similar conclusions when BMDCs or macrophages were stimulated with TLR7/8 ligand (R848), TLR9 ligand (CpG), or an LPS+R848 combination (unpublished data). Next, we analyzed the impact of TTP on steady-state mRNA levels. In WT BMDCs, IL23p19 mRNA was rapidly and transiently expressed, reaching peak concentrations 2 h after stimulation before returning to near-baseline levels 6 h later. IL23p19 mRNA levels were strongly increased and maintained for a longer timeframe in TTP^−/− cells. We also observed higher IL12p35 mRNA levels in TTP^−/− versus WT BMDCs, but kinetics were similar in both groups. Finally, consistent with protein production, absence of TTP did not affect IL12/23p40 mRNA levels (Fig. 1 B).

To analyze the impact of APC-expressed TTP on IFN-γ and IL17 production, splenocytes from TLR4 KO mice were incubated in the presence of supernatants from control or LPS-activated TTP^+/+ or TTP^−/− BMDCs. Supernatants from LPS-stimulated TTP^−/− BMDCs were found to promote increased IL17A and IFN-γ productions when compared with supernatants from their WT counterparts (Fig. 1 C). Addition of both anti-p40 and anti-p19 neutralizing antibodies blocked IL17 production, whereas only antibodies directed against p40 were able to inhibit IFN-γ production (Fig. 1 D), confirming the molecular identity and bioactivity of, respectively, IL23 and IL12 . Collectively, these results indicate that TTP is implicated in the negative regulation of IL23p19 and, to a lesser extent, of IL12p35 mRNA expression. In the absence of TTP, overproduction of bioactive IL23 and IL12 has a functional impact on the production of IL17 and IFN-γ, respectively.

To assess the role of TTP in the control of IL12p35 and IL23p19 mRNA stability, we analyzed the mRNA half-lives of these subunits. Blockade of transcription by actinomycin D led to a rapid decay of IL12p35 mRNAs in both TTP^+/+ and TTP^−/− BMDCs, whereas IL12/23p40 and IL23p19 mRNA levels were found to be stable in both groups (unpublished data). To reveal an effect of TTP, we treated cells with actinomycin d and SB2202190, a p38-MAPK inhibitor used to abrogate the inhibitory action of p38-MAPK on TTP activation (Carballo et al., 2001). The presence of both inhibitors resulted in the rapid destabilization of IL23p19 mRNA, which was completely attributable to TTP. Addition of the p38-MAPK inhibitor further accelerated IL12p35 mRNA decay in a TTP-dependent manner but did not modify IL12/23p40 mRNA levels (Fig. 1 E).

The presence of AREs in the 3′UTRs is known to cause rapid decay of the corresponding mRNAs, resulting in a tightly controlled secretion of many inflammatory mediators (Anderson, 2008). Because IL23p19 mRNA contains five AREs within its 3′UTR (Fig. 1 H), we decided to evaluate the role of this region in the control of p19 mRNA degradation. Interestingly, four of these AREs were conserved in the 3′UTR region of the human IL23p19 mRNA, suggesting that they might play an important role. We cloned the p19 3′UTR immediately downstream of the mIL2 coding sequence and transfected this reporter construct into HEK-293 cells. mIL2 production was greatly reduced in cells transfected with the 3′UTRp19-containing plasmid, consistent with the notion that 3′UTRp19 displays regulatory functions (Fig. 1 F). Overexpression of TTP, but not of a truncated, inactive form of TTP (TTP NTD), led to a further decrease of IL2 production driven by the 3′UTRp19-containing plasmid (Fig. 1 F). Site-directed mutagenesis revealed that all five AREs had to be mutated to completely abrogate the destabilizing effect of TTP (Fig. 1 G). These results are consistent with recent observations from Qian et al. (2011) indicating that multiple AREs mediate IL23 mRNA instability in response to IFN-γ. TTP-mediated mRNA decay requires the recruitment of a poly(A) nuclease complex containing CAF1 (CCR4-associated factor 1; Zheng et al., 2008). We depleted CAF1 protein in RAW cells by knocking down the Cnot7 gene and observed increased LPS-induced IL23p19 mRNA expression in this condition (Fig. 1, I and J). We repeated these experiments in WT BMDCs and confirmed this trend. Importantly, IL23p19 mRNA expression was not enhanced upon CAF1 depletion in TTP^−/− BMDCs. Collectively, our experiments confirm the direct role of TTP on IL23p19 mRNA decay through CAF1-dependent deadenylation and indicate that such a mechanism is critical to control IL23 production in LPS-stimulated BMDCs.

TTP deficiency leads to the spontaneous development of a complex inflammatory syndrome that combines cachexia, dermatitis, arthritis, conjunctivitis, and myeloid hyperplasia (Taylor et al., 1996). Dermatitis is characterized by orthokeratotic hyperkeratosis, acanthosis, and neutrophil infiltrates, features which are compatible with local IL23 production (Chan et al., 2006). Moreover, ubiquitous expression of IL23p19 in transgenic mice results in weight loss, infertility, and inflammation in multiple organs, including skin (Wiekowski et al., 2001). Based on these observations, we analyzed IL23p19 mRNA and protein expression within skin lesions of TTP^−/− mice. We observed increased expression of IL23p19 mRNA in comparison with healthy skin from WT mice (Fig. 2 A). By immunohistochemistry, we were able to detect IL23p19-positive cells in the inflamed epidermis of TTP^−/− mice, which were absent in the skin of TTP^+/+ mice (Fig. 2 B). In contrast, IL12/23p40 mRNA expression in the skin of TTP^−/− mice was comparable to WT mice (Fig. 2 A). Accordingly, we also detected high IL17A and IL22 mRNA levels within skin lesions and draining LNs (dLNs; Fig. 2 C) of mutant mice. Of note, these two cytokines are directly implicated in IL23-induced skin inflammation (Zheng et al., 2007; Rizzo et al., 2011). In contrast, IFN-γ mRNA levels were comparable to that observed in healthy skin and dLNs, whereas increased TNF mRNA expression was confined to inflamed skin (Fig. 2 C). TNF has been considered as the major effector cytokine responsible for the inflammatory syndrome observed in TTP-deficient mice. Although in vitro data largely confirm the negative influence of TTP on TNF expression, our observations concur with previous studies indicating that in vivo, absence of negative regulation by TTP leads to a selective elevated expression of TNF in inflamed tissues (skin, in the present study, or liver, as shown in Kaplan et al., 2011). Notably, expression of TNF in lymphoid tissues (dLN, this study, and spleen and BM in Kaplan et al., 2011) was not affected by lack of TTP expression (Kaplan et al., 2011). Herein, we demonstrate a significant up-regulation of TNF mRNA expression at the inflammation site.

We next analyzed ex vivo cytokine production of CD4 T cells from the spleen and dLNs of TTP^+/+ and TTP^−/− mice. We observed that the frequencies of IL17A⁺ and IL22⁺ CD4 T cells were increased in dLNs but not the spleen of TTP^−/− mice when compared with their WT counterpart. (Fig. 2 D). The slight increase in the frequency of IFN-γ–producing CD4 T cells did not reach statistical significance, whereas numbers of TNF⁺ CD4 T cells were comparable in both groups (Fig. 2 D). As TTP could directly influence cytokine production in a T cell–intrinsic fashion, we sorted naive CD4 T cells from WT or TTP^−/− mice and cultured them in neutral (Th0) or polarizing conditions as indicated. After 3 d of culture in Th17- or Th22-promoting media, activated T cells produced similar amounts of IL17A and IL22 in both groups at the protein or mRNA levels (Fig. 2 E and not depicted). Consistent with previous results (Ogilvie et al., 2009), absence of TTP in CD4 T cells led to increased IFN-γ production in Th1 cells (Fig. 2 E). These results indicate that TTP^−/− CD4 T cells are not intrinsically biased toward Th17 or Th22 differentiation.

IL17A-producing γδ T cells are important effectors of inflammation in psoriatic skin (Cai et al., 2011). We found increased frequencies of IL17A⁺ and IL22⁺ γδ T cells in dLNs and skin lesions of TTP^−/− mice (Fig. 2 F). Furthermore, infiltrating γδ T cells also produced higher levels of IFN-γ and TNF, confirming the highly inflammatory nature of these cells (Fig. 2 F). Interestingly, we observed that a significant proportion of CD3⁻CD4⁺ cells produced IL17A and IL22 in the spleen and dLNs of TTP^−/− mice (Fig. 2 G). These cells might correspond to IL23-responsive innate lymphoid populations (Spits and Di Santo, 2011). Altogether, these observations indicate that inflammation that develops in absence of TTP is characterized by local IL17A and IL22 overproduction, a finding consistent with a potential pathogenic role of IL23 in this syndrome.

To elucidate the role of dysregulated IL23 production in the development of TTP deficiency syndrome, we crossed TTP^−/− mice with IL23p19^−/− mice. A cardinal feature of TTP^−/− mice is their low body weight and fat content, as illustrated by their lack of weight gain after 3 wk of age (Fig. 3 A). Remarkably, IL23p19^−/−TTP^−/− mice maintained growth curves identical to TTP^+/+ (Fig. 3 A). Furthermore, none of the IL23p19^−/−TTP^−/− mice displayed signs of dermatitis or arthritis, whereas all TTP^−/− mice developed at least one of these clinical manifestations during the observation period (6 mo). Histological analysis of the skin indicated that hyperkeratosis and cell infiltration observed in TTP^−/− mice were completely absent in the skin of IL23p19^−/−TTP^−/− mice (Fig. 3 B). Similarly, histological analyses revealed a lack of typical joint inflammation in IL23p19^−/−TTP^−/− mice when compared with TTP^−/− mice (Fig. 3 C). We obtained similar results when TTP^−/− mice were treated with neutralizing IL-12/23p40 antibody from 2 wk of age (250 µg, twice a week; not depicted), strongly supporting the notion that the absence of pathology in IL23p19^−/−TTP^−/− mice was secondary to the loss of IL23 expression and not a consequence of genetic background differences. To confirm the important role of IL23 in this model, we generated double KO mice lacking expression of both TTP and IL17A. IL17A^−/−TTP^−/− mice did no display significant weight loss nor signs of arthritis and dermatitis (Fig. 3, A–C), confirming the functional role of the IL23–IL17 axis in this inflammatory model.

The phenotype of IL23^−/−TTP^−/− mice was highly unexpected given the critical and direct role of TTP on TNF mRNA decay. Indeed, TNF neutralization or genetic ablation completely prevents the development of this pathology (Taylor et al., 1996; Carballo and Blackshear, 2001). There is an intimate cross talk between TNF and IL23 in the context of autoimmunity. Treatment of psoriasis or arthritis by TNF neutralization leads to decreased IL23p19 expression at the lesion site (Zaba et al., 2007; Lina et al., 2011), whereas mouse models indicate that IL23-driven epidermal hyperplasia is mediated by TNF (Chan et al., 2006). Given that both cytokines represent direct targets of TTP, this model of spontaneous inflammation represents an ideal setting to further study the interplay between these two cytokines. To this end, we analyzed mice lacking expression of TTP and TNF receptor 1 and 2 expression (TNFR1/2^−/−TTP^−/− mice). Absence of TNF signaling resulted in a significant decrease in IL23p19, but not IL12/23p40, expression in skin lesions and dLNs, revealing a positive and specific feedback loop between TNF and IL23p19 in TTP^−/− mice (Fig. 4 A).

Expression of IL17A and IL22 mRNA (Fig. 4 B), and frequencies of IL17A- and IL22-producing CD4 T or γδ T cells (Fig. 4, C and D) within peripheral LN or skin were found to be completely dependent on both IL23p19 and TNF signaling. Similarly, TNF mRNA levels within skin lesions were reduced in the absence of IL23, IL17A, and TNFR expression (Fig. 4 B). No impact was observed on numbers of TNF-producing cells at this site of inflammation (unpublished data), suggesting that other cell types, such as myeloid cells, constitute the source of this cytokine. Regarding IFN-γ production, both mRNA levels and CD4⁺ T cells proportions were comparable between groups (Fig. 4 B and not depicted).

Absence of TTP also leads to myeloid hyperplasia, characterized by increased proportions of F4/80⁺ macrophages and GR1⁺ neutrophils in the BM and the spleen (Taylor et al., 1996), accompanied by a slight reduction in B cell numbers in these lymphoid organs. This spontaneous granulocyte hyperplasia was recently shown to be secondary to the general inflammatory environment and associated with elevated circulating G-CSF levels (Kaplan et al., 2011). As previously shown, myeloid hyperplasia was absent in mice lacking both TTP and TNFR1/2 (Fig. 4, E and F). In keeping with our previous observations, lack of IL23p19 or IL17A expression also completely reverted to the hematological dysfunctions caused by TTP deficiency (Fig. 4 E). We observed that the increase in G-CSF levels in the plasma of TTP^−/− mice was also dependent on IL23/IL17 (Fig. 4 F).

Initial studies of TTP^−/− mice focused on local inflammation of joints, but Adamopoulos et al. (2011) demonstrate that systemic exposure to IL23 by hydrodynamic delivery in adult mice causes severe arthritis accompanied by a general bone loss. We therefore evaluated the overall mineralization of the skeleton by positron emission tomography (PET)–CT scan using [¹⁸F]-NaF. We observed that TTP^−/− develop strong osteopenia, associated with increased uptake of the tracer, suggestive of active bone remodeling. Bone density was globally normalized in absence of IL23. In comparison with TTP^−/− mice, [¹⁸F]-NaF uptake was decreased in TTP^−/−IL23p19^−/− animals, suggesting that IL23 also contributes to bone remodeling in this model (Videos 1–6).

In conclusion, most of the aspects of the TTP deficiency syndrome were found to be IL23 dependent. In addition to this striking observation, we also identified IL17A as a key IL23-induced cytokine implicated in the amplification of the inflammatory processes. Our experiments suggest that TTP directly regulates the production of both TNF and IL23, which act together in a nonredundant fashion to promote IL17A-driven inflammation. Interestingly, we found that IL23p19 and TNF expression in the skin (Fig. 4, A and B) were greatly reduced in absence of IL17A. This observation is consistent with the capacity of IL17A to enhance TNF-induced expression of IL23p19 by human synoviocytes and dermal fibroblasts derived from rheumatoid arthritis patients (Goldberg et al., 2009). In consequence, IL22 mRNA levels (Fig. 4 B) and frequencies of IL22-producing CD4 T or γδ T cells (Fig. 4, C and D) within peripheral LN or skin were decreased in absence of this cytokine. Altogether, these results indicate that IL17A acts as key effector cytokine in the TTP deficiency syndrome. This model reemphasizes TTP and regulators of TTP as potential therapeutic targets in the context of arthritis and psoriasis, as it will affect expression of both inflammatory cytokines.

Interestingly, recent evidence indicates that selective ablation of TTP in macrophage/granulocyte lineages is not sufficient to drive autoimmune inflammation (Kratochvill et al., 2011; Qiu et al., 2012). Previous studies showed that transfer of TTP^−/− BM in RAG^−/− mice could recapitulate the full syndrome, indicating the role of TTP in hematopoietic cells (Carballo et al., 1997). TTP is expressed in T lymphocytes and could mediate mRNA decay of several T cell–derived cytokines, such as IL2, IFN-γ, or IL17 (Ogilvie et al., 2005, 2009; Lee et al., 2012). We confirmed that upon Th1 differentiation, TTP^−/− naive CD4 T cells tend to produce higher levels of IFN-γ. However, neither IL17A nor IL22 production was increased upon polarizing conditions. These results indicate that exacerbated Th17 and Th22 responses in the absence of TTP are not an intrinsic property of CD4 T cells. Other cell populations might therefore be implicated in the development of this autoimmune pathology. Infiltrating DCs are critical to maintain Th17 responses in the context of EAE (Huang et al., 2012). Inflammatory dermal DCs expressing TNF and IL23 are found in psoriatic skin samples (Hänsel et al., 2011). Preliminary experiments indicate that in inflamed TTP^−/− skin, both F4/80- and CD11c-positive cells stain for IL23. Hence, the role of TTP-mediated TNF and IL23 regulation should be addressed in specific DC subpopulations.

The IL23–IL17 axis is also implicated in the regulation of granulopoiesis through induction of G-CSF (Stark et al., 2005). We provide evidence that myeloid hyperplasia is also a feature of the TTP deficiency syndrome that can be the consequence of dysregulated IL23 production.

Both IL12p35 and IL12/23p40 are highly regulated at the transcriptional and epigenetic levels (Goriely et al., 2008). TTP-mediated mRNA destabilization is not implicated in IL12/23p40 regulation and plays a minor role for IL12p35. In sharp contrast, IL23p19 is a primary response gene that strongly relies on posttranscriptional regulation. Given the central role of IL23 in inflammatory bowel diseases, multiple sclerosis, psoriasis, or arthritis, our observations have important implications for the development of antiinflammatory therapies targeting this cytokine.

TTP (Zfp36)-deficient and TNFR1/2^−/− TTP^−/− mice on mixed 129/Sv-C57BL/6 background were obtained from P.J. Blackshear. IL23p19- and IL17A-deficient mice on C57BL/6 background were obtained from Genentech and Y. Iwakura (Institute for Medical Science, Tokyo, Japan), respectively. 129/Sv-C57BL/6 TTP^−/− mice were maintained as a 129/Sv-C57BL/6 TTP^+/− colony, as homozygous mice are not fertile. Experiments on TTP^−/− or IL23p19-deficient or IL17A-deficient backgrounds were performed using littermates as controls. Mice were bred and maintained according to institutional guidelines.

Murine BMDCs were generated as previously described (Goriely et al., 2006). The HEK-293 human kidney cell line was obtained from LGC Promochem. Ultra-Pure LPS from E. coli (0111:B4) was obtained from InvivoGen. Actinomycin D and SB202190 were obtained from Sigma-Aldrich and EMD Millipore, respectively.

A 656-bp fragment of the IL-23(p19) gene (nt 1290/1945) was amplified by PCR from mouse genomic DNA and subsequently cloned as an EcoRI insert into the pCDNA3.1 vector containing the coding sequence of the murine IL2 to generate the UTRp19 reporter plasmid. The plasmids UTRp19m1, UTRp19m12, UTRp19m34, UTRp19m5, UTRp19m345, and UTRp19m12345 are derivatives of UTRp19 in which the ARE site was altered by the QuikChange Lightning Site-Directed Mutagenesis kit (Agilent Technologies). For each ARE element, the two T nucleotides flanking the ATTTA sequence were changed in nucleotide G. pCDNA3.1-mIL2 expression vector contains the murine IL2 coding sequence, pCDNA3.1-TTP expression vector contains the TTP coding sequence, and pCDNA3.1-TTP NTD expression vector contains the first 102 amino acids of the TTP coding sequence.

Murine cytokines (IL23, IL12/23p40, IL12p70, IL2, IL17, IL22, and IFN-γ) levels were determined in cell-free supernatants using Duoset ELISA (R&D Systems) with detection limits of 30 pg/ml. G-CSF plasma levels were quantified using Quantikine ELISA kit (R&D Systems) with detection limits of 15 pg/ml.

Total RNA from cells and tissues was extracted using a MagNA Pure LC RNA High Performance Isolation kit and MagNA Pure LC RNA Isolation kit III, respectively (Roche). RT and real-time PCR reactions were then performed using Taqman RNA amplification kit (one-step procedure) on a LightCycler 480 Real-Time PCR system (Roche). mRNA levels were expressed as absolute number of copies normalized against β-actin mRNA. This was achieved by generating standard curves from serial dilutions of PCR-generated standards. The sequences of primers and probes were: β-actin: 5′-TCCTGAGCGCAAGTACTCTGT-3′, 5′-CTGATCCACATCTGCTGGAAG-3′, and probe 5′-ATCGGTGGCTCCATCCTGGC-3′; IFN-γ: 5′-GGATGCATTCATGAGTATTGC-3′, 5′-GCTTCCTGAGGCTGGATTC-3′, and probe 5′-TTTGAGGTCAACAACCCACAGGTCCA-3′; IL12/23p40: 5′-GTTCAACATCAAGAGCAGTAGCA-3′, 5′-CTGCAGACAGAGACGCCATT-3′, and probe 5′-CCCTGACTCTCGGGCAGTGACA-3′; IL12p35: 5′-CTTAGCCAGTCCCGAAACCT-3′, 5′-TTGGTCCCGTGTGATGTCT-3′, and probe 5′-TCTGGCCGTCTTCACCATGTCA-3′; IL17a: 5′-GCTCCAGAAGGCCCTCAG-3′, 5′-CTTTCCCTCCGCATTGACA-3′, and probe 5′-ACCTCAACCGTTCCACGTCACCCTG-3′; IL22: 5′-ACAGGTTCCAGCCCTACATG-3′, 5′-GTCGTCACCGCTGATGTG-3′, and probe 5′-TGGTACCTTTCCTGACCAAACTCAGCA-3′; IL23p19: 5′-CCCGTATCCAGTGTGAAGATG-3′, 5′-CCCTTTGAAGATGTCAGAGTCA-3′, and probe 5′-CCACAAGGACTCAAGGACAACAGCC-3′; TNF: 5′-CAGACCCTCACACTCAGATCA-3′, 5′-CACTTGGTGGTTTGCTACGA-3′, and probe 5′-TCGAGTGACAAGCCTGTAGCCCA-3′. The primers and probe mix for the quantification of Cnot7 mRNA (Mm00516123_m1) were purchased from Life Technologies.

HEK-293 cells were transfected using FuGENE-6 (Roche) as previously described (Goriely et al., 2006). Promoter activities were analyzed with the quantification of mouse IL2 production in the cell-free supernatant using Duoset ELISA (R&D Systems).

For siRNA transfection, Cell Line Nucleofector kits T (T-001 program) and V (Y-001 program) were used for RAW cells and BMDCs, respectively, using a Nucleofector 2b apparatus (Lonza). Mouse Cnot7 (encoding CAF1) ON-TARGET plus siRNA (L-059073-00-0005) and nontargeting siRNA (D-001810-10-05) SMART pools were purchased from Thermo Fisher Scientific.

Single cell suspensions of spleen were prepared from TLR4^−/− mice. 2 × 10⁶ cells/ml were cultured with 100 U/ml IL2 (Proleukin) in the presence or absence of supernatant of BMDCs culture for 72 h, after which supernatant were collected and analyzed for cytokines production using ELISA. Anti-IL23p19 (Polyclonal goat IgG, AF1619) and anti-IL12/23p40 (C17.8) antibodies were purchased from R&D Systems and BD, respectively.

Naive CD4⁺ T cells were purified by pre-enrichment of CD4⁺ T cells from spleens by magnetic-activated cell sorting (Dynal CD4⁺ T cell negative isolation kit; Invitrogen) and sorting by flow cytometry for CD4⁺CD62L⁺CD25⁻ T cells (FACSAria; BD) after staining with anti-CD4 (RM4-5), anti-CD62L (MEL-14), and anti-CD25 (PC61; all from BD).

All cultures were incubated in IMDM with 10% (vol/vol) FCS, 2 mM l-glutamine, 1 mM sodium pyruvate, 0.1 mM nonessential amino acids, 40 µM β-mercaptoethanol, 100 U/ml penicillin, and 100 U/ml streptomycin (all from Lonza). Culture plates were coated for 2 h with 5 µg/ml anti-CD3 in PBS (145-2C11; BD). Cells were plated at a density of 10⁶ cells per ml and were stimulated with 1 µg/ml anti-CD28 (37.51, BD Biosciences). T cells were cultured for 3 d in Th0 conditions (10 µg/ml anti–IFN-γ [R4-6A2; BD], and 10 µg/ml anti-IL4 [11B11; BD]), Th1 conditions (10 ng/ml recombinant IL12 [R&D Systems] plus 10 µg/ml anti-IL4), Th17 conditions (2.5 ng/ml of recombinant human TGF-β [R&D Systems] and 10 ng/ml of recombinant IL6 [R&D Systems] plus anti–IFN-γ and anti-IL4), or Th22 (20 ng/ml of recombinant IL6 and 10 ng/ml of recombinant IL23 [R&D Systems] plus anti–IFN-γ and anti-IL4).

Skin was incubated in 2.5 mg/ml collagenase I (Sigma-Aldrich). Tissues were dissociated with the gentle MACS Dissociator (Miltenyi Biotec) to obtain single-cell suspensions.

Spleen cells, dLNs, or skin cells were stimulated for 4 h with 25 ng/ml PMA (Sigma-Aldrich) and 500 ng/ml ionomycin (Sigma-Aldrich) with the addition of 5 µg/ml Brefeldin A (Sigma-Aldrich) for the last 2 h. Cells were first stained for surface antigens (CD4 PB [RM4-5], CD3 Percp [SP34-2], and γ/δ TCR FITC [GL3], all from BD) and then treated with Cytofix/Cytoperm (BD) according to the manufacturer’s directions. Intracellular cytokine staining was performed using anti-IL17A AF647 (TC11-18H10; BD), anti-IL22 PE (clone 1H8PW5R; eBioscience), TNF APC (MP6-XT22; BD), or anti–IFN-γ PE (XMG1.2; BD).

Spleen cells were stained using anti-CD11b PE (M1/70), anti-GR1 APC (RB6-8C5), anti-CD3 Percp (SP34-2), and anti-B220 FITC (RA3-6B2). All antibodies were purchased from BD. Antibody directed against F4/80 PB (CI:A3-1) was purchased from AbD Serotec.

Mouse tissues were fixed in 4% buffered formalin. When required, tissues were decalcified in 30% formic acid. Fixed tissues were embedded in paraffin, sectioned, and stained with H&E. Staining was performed using goat anti–mouse IL23p19 antibody (R&D Systems) and biotin-conjugated donkey anti–goat IgG (Jackson ImmunoResearch Laboratories) as secondary antibodies.

Mann-Whitney test or Kruskal-Wallis test (Nonparametric ANOVA) with Dunn’s Multiple Comparisons tests were used for statistical calculation.

Videos 1–6 show that TTP-associated osteopenia and increased bone remodeling are dependent on IL23.

The Institute for Medical Immunology is sponsored by the government of the Walloon Region and GlaxoSmithKline Biologicals. This study was supported by the Fonds National de la Recherche Scientifique (FRS-FNRS, Belgium), the WELBIO, and an Interuniversity Attraction Pole of the Belgian Federal Science Policy. C. Molle is a postdoctoral fellow of the FRS-FNRS. L. Ysebrant and M. Andrianne are supported by a grant from the Télévie. S. Goriely is a research associate of the FRS-FNRS. P.J. Blackshear is supported by the Intramural Research Program of the NIEHS, NIH.

The authors have no conflicting financial interests.