Results

Characterization of Sharpin 
Orthologs of SHARPIN protein are found in various species, including human, mouse and rat. Motif prediction programming, using COILS [11] and MotifScan [12], suggests that SHARPIN contains a coiled-coil (CC) domain, a ubiquitin-like (UBL) domain, and a zinc-finger Ran-binding protein 2 (ZFRBP) domain. These functional motifs constitute similar domain profiles that are present in the SHARPIN protein of all three origins (Fig. 1A), suggesting that SHARPIN exerts highly conserved functions across species. Spontaneous mutations in the mouse Sharpin gene results in a complex inflammatory phenotype characterized by severe dermatitis (Fig. 1B), systemic inflammation and an enlarged spleen (Fig. 1C) caused by extramedullary hematopoiesis [3]. The endogenous expression of Sharpin mRNA in BMDC was determined by quantitative real time-PCR (qRT-PCR) following culture in medium only or after stimulation with LPS. Sharpin mRNA was present in BMDC generated from WT mice (Fig. 1D) and its level was modestly decreased by LPS stimulation. There was a significant reduction of Sharpin mRNA (6–7-fold) in BMDC generated from cpdm mice. Transfection of Flag-tagged Sharpin in fibroblasts (NIH3T3) and macrophages (RAW264.7) indicated cytoplasmic localization of the SHARPIN protein (Fig. 1E).
10.1371/journal.pone.0031809.g001 Figure 1  In vivo and in vitro features of SHARPIN.
(A) COILS and MotifScan programs were used to predict the presence of CC (coiled-coil) domain, UBL (ubiquitin-like) domain and ZFRBP (zinc-finger Ran-Binding protein 2) domain which form similar motif patterns in the SHARPIN protein of human, mouse and rat origins. (B) Eight week-old females of WT and cpdm mice. The mutant mice (above) develop progressive skin inflammation starting at about four weeks. (C) Extramedullary hematopoiesis causes marked enlargement of the spleen of cpdm mice. (D) BMDC were incubated in the absence or presence of 100 ng/ml LPS for 4 hours. The mRNA level of Sharpin was measured by qRT-PCR and presented relative to the mRNA expression in non-stimulated WT BMDC. Bars represent the mean ± s.d. of 3 mice. * P<0.05; ** P<0.001. (E) Fibroblast (NIH3T3) and macrophages (RAW264.7) cells were transfected with the expression plasmid pFLAG-SHARPIN. After 48 hours, cells were fixed and probed with anti-FLAG and FITC-conjugated secondary antibody. Nuclei were stained with DAPI. In both transfected cell lines, FLAG-SHARPIN was found to be cytoplasm-localized.

Phenotyping splenic DC and BMDC from WT and cpdm mice
DC are heterogeneous and can be categorized into multiple subtypes based on surface markers [13]. To determine if the Sharpin mutation affects DC development in lymphoid tissues, mouse spleens were examined for the distribution of conventional DC (cDC; CD11c+CD8α+ and CD11c+CD8α−) [13] and plasmacytoid DC (pDC; CD11c−PDCA-1+) [14]. The percentages for splenic cDC and pDC were both reduced in cpdm mice when compared with WT controls (Fig. 2A). However, when gated on CD11c+ cDC, the percentages of CD8α+ and CD8α− cells were not affected by SHARPIN deficiency (Fig. 2A). Since the spleen of cpdm mice is markedly enlarged and contains three times as many cells (Fig. 1C), the different percentages of splenic cDC and pDC between WT and mutant mice reflect the increased number of total spleen cells rather than a reduction in cDC and pDC numbers. These data indicate that the Sharpin mutation does not affect the distribution of the examined DC subsets in the spleen.
10.1371/journal.pone.0031809.g002 Figure 2  Effect of Sharpin mutation on DC subpopulations and maturation.
(A) Spleens from WT and cpdm mice were isolated and subject to collagenase and DNase digestion. The obtained splenic homogenates were centrifuged over a Percoll gradient (35% and 55% density) for 15 minutes. The bands at the 35%-medium and the 35–55% interface were pooled, washed and stained with a combination of various antibodies to stain different DC subsets, conventional CD11c+CD8α+, CD11c+CD8α− and plasmacytoid DC (CD11c−PDCA-1+).The top panels were gated on FSChiSSClo cells to show separate populations of CD11c+PDCA-1− and CD11c−PDCA-1+ cells. Further gating on the CD11c+PDCA-1− subpopulation gave the bottom panel that showed two distinct pools of CD11c+CD8α+ and CD11c+CD8α− cells. Percentages were calculated based on the parental population and were additionally shown as bar graphs (n = 2) (B). (C) WT and cpdm BMDC (5×105) cells were stimulated with medium, 100 ng/ml LPS or 25 µg/ml poly I:C for 24 hours. The cells were labeled with PE-labeled anti-CD40, anti-CD80, and anti-CD86, and subjected to flow cytometry analysis. The populations shown in histograms were gated on CD11c+ cells. Unstained cells served as negative controls. Results are representative of two independent experiments.   BMDC from in vitro cultures functionally resemble non-lymphoid tissue DC and monocyte-derived inflammatory DC [15], [16]. The yields of BMDC from WT and cpdm mice were similar. BMDC were CD11c+ and MHC II+ with low expression of co-stimulatory molecules CD40, CD80, and CD86. The TLR3 ligand poly I:C and the TLR4 ligand LPS each activate overlapping but different signaling pathways and were used to induce DC maturation [17], [18]. Incubation with the TLR agonists for 24 hours resulted in increased expression of CD40, CD80, and CD86 on BMDC; however, there was no difference in the expression levels of these markers between WT and cpdm BMDC (Fig. 2B). Thus, SHARPIN deficiency did not influence the expression of co-stimulatory molecules by BMDC.

Production of proinflammatory mediators by cpdm BMDC is impaired
Incubation with LPS or poly I:C resulted in secretion of IL6, IL12P70, and GMCSF from both WT and cpdm BMDC; however, BMDC from cpdm mice produced significantly less of all three cytokines compared with WT BMDC (Fig. 3A–C, E–F). The amount of nitric oxide generated by mutant BMDC was also significantly reduced compared with WT cells (Fig. 3D), indicating severely disrupted production of proinflammatory mediators from Sharpin-deficient BMDC. In vivo complementation with a Sharpin gene-containing BAC reversed the phenotype of the mutant mice [2], and BMDC generated from these rescued mice secreted significantly more IL12P70 than BMDC from cpdm mice (Fig. 3G), supporting a necessary role of SHARPIN for the production of IL12P70. In addition, the transcript levels of the inflammatory cytokines Il6, Il12p40, Gmcsf, and Ifnb were examined, and these were all significantly reduced in stimulated cpdm BMDC when compared with WT controls (Fig. 4).
10.1371/journal.pone.0031809.g003 Figure 3  Defective production of pro-inflammatory mediators from stimulated cpdm BMDC.
(A–D) Cultured WT and cpdm BMDC (1×105 cells in 0.1 ml complete medium) were washed and stimulated with medium, 100 ng/ml LPS or 25 µg/ml poly I:C for 24 hours. Supernatants were collected for ELISA of IL12P70, IL6, and GMCSF, and for quantification of nitric oxide (NO). (E–F) Gradient numbers (1×104, 2×104, and 4×104 cells in 0.1 mL complete medium) of BMDC were used for100 ng/mL LPS stimulation. After 24 hours, the amounts of IL12P70 and IL6 from the supernatant were measured. (G) The cpdm mice rescued by Sharpin-containing BAC had complete remission of the inflammatory phenotype [2]. BMDC developed from cpdm and rescued cpdm mice were plated (1×106 cells in 0.3 mL) and stimulated with 100 ng/mL LPS. After 24 hours, supernatants were collected for analysis of IL12P70 production. Data are representative of three independent experiments. * P<0.01; ** P<0.005.
10.1371/journal.pone.0031809.g004 Figure 4  Decreased mRNA levels of inflammatory cytokines from cpdm BMDC.
Cultured WT and cpdm BMDC (5×105 cells in 0.2 ml complete medium) were washed and stimulated with 100 ng/ml LPS (A,C,E,G) or 25 µg/ml poly I:C (B,D,F,H). At 0, 1 and 2 hours, total RNA was extracted and subject to qRT-PCR to measure the expression of Il12p40 (A,B), Il6 (C,D), Gmcsf (E,F), and Ifnb (G,H) mRNA. Bars represent mean ± SD. Data are representative of two independent experiments.

Impaired cytokine production by cpdm BMDC is correlated with selective defects in NF-κB signaling
There are a number of possible explanations for the defective cytokine secretion in stimulated cpdm BMDC, including 1) reduced surface expression of the LPS receptor complex, 2) increased production of anti-inflammatory mediators, 3) increased expression of negative regulators of TLR pathways, and 4) impaired TLR-induced signaling activation.
We determined the surface expression of the LPS receptor complex that comprises TLR4, the accessory proteins CD14 and myeloid differentiation factor 2 (MD2/LY96) [19]. Flow cytometric analysis shows that the expression levels of CD14 and TLR4/MD2 between WT and cpdm BMDC were similar (Fig. 5A). We then quantified the secretion of the suppressive cytokines IL10 that can inhibit IL12 secretion in an autocrine manner [20], [21]. The supernatants from LPS-stimulated cpdm BMDC contained significantly lower levels of IL10 than stimulated WT BMDC (Fig. 5B), suggesting that IL10 was not responsible for decreased secretion of IL12P70 by cpdm BMDC. Increased expression of a negative regulator of TLR signaling such as A20 [22] may also suppress cytokine secretion. However, the transcript level of A20 was lower in LPS-activated cpdm BMDC than WT controls (Fig. 5C), thereby ruling out overexpression of A20 as a factor in the reduced cytokine production.
10.1371/journal.pone.0031809.g005 Figure 5  Normal TLR4 and MD2 expression and decreased IL10 secretion and A20 expression by cpdm BMDC.
(A) Unstimulated WT and cpdm BMDC (5×105) were labeled with PE-labeled anti-TLR4/MD2 or anti-CD14 and then subject to flow cytometry analysis. Unstained cells serve as negative controls. (B) Cultured WT and cpdm BMDC (1×105 cells in 0.1 ml complete medium) were washed and stimulated with medium, 100 ng/ml LPS or 25 µg/ml poly I:C for 24 hours. Supernatants were collected for ELISA of IL10. (C) Cultured WT and cpdm BMDC (5×105 cells in 0.2 ml complete medium) were washed and stimulated with 100 ng/ml LPS or 25 µg/ml poly I:C. At 0, 1, and 2 hours, total RNA was extracted and subject to qRT-PCR to measure the production of A20.    The transcription of TLR3/4-induced proinflammatory intermediates is tightly regulated by cellular signaling pathways, in particular NF-κB, TBK1/IRF3, and MAPK [23]–[27]. We next determined if disrupted NF-κB, TBK1/IRF3, and/or MAPK signaling may underlie the impaired cytokine production from stimulated Sharpin-deficient BMDC. Stimulus-induced phosphorylation of the IκB kinase (IKK1/2) is an essential step in NF-κB signaling, allowing phosphorylation and proteasome-mediated degradation of the NF-κB inhibitor IκBα to release the NF-κB transcription factors into the nucleus. The amount of phosphorylated IKK1/2 (p-IKK1/2) and IκBα (p-IκBα) following incubation with LPS or poly I:C was severely decreased in cpdm BMDC as compared with WT controls (Fig. 6). The cpdm BMDC exhibited similar levels of TBK1, ERK1/2, and p38 phosphorylation to those of WT cells (Fig. 6). These results indicate that the absence of functional SHARPIN decreased NF-κB activation but did not affect TBK1/IRF3, ERK1/2, and p38 signaling in BMDC.
10.1371/journal.pone.0031809.g006 Figure 6  Inhibition of NF-κB signaling in cpdm BMDC.
WT and cpdm BMDC (2×106 cells in 0.5 mL complete medium) were stimulated with 100 ng/mL LPS (A) or 25 µg/mL poly I:C (B). At 0, 15, 30, and 60 minutes, whole-cell lysates were obtained and subject to immunoblots with antibodies against proteins involved in NF-κB, TBK1/IRF3, ERK1/2, and p38 signaling pathways. Beta-actin was used as loading control. (C) Cellular levels of p-IKK1/2 and p-IκBα in LPS- or poly I:C-stimulated BMDC were quantitated with ImageJ (NIH) and presented as trend lines. Results are representative of at least two independent experiments.

Th2-biased immunogenicity of stimulated cpdm BMDC
The defective IL12 production (Fig. 3A) and Th2-dominant cytokine profile in cpdm mice [5] suggest that the absence of SHARPIN affects the ability of cpdm BMDC to induce T cell differentiation into effector cells. Co-culture of allogeneic naïve CD4+ T cells with WT BMDC stimulated with LPS or poly I:C elicited robust IFNγ production, whereas the concentration of IFNγ in cpdm BMDC-T cell cultures was significantly lower after LPS stimulation (Fig. 7A), indicating impaired Th1-polarizing abilities of cpdm BMDC. In addition to TLR3/4 agonists, the TLR2 ligand Pam3CYS was used since it has been shown to induce both Th1 and Th2 responses [28]–[30]. Pam3CYS-matured WT BMDC induced robust IFNγ production at a significantly higher level than cpdm BMDC (Fig. 7A). The reduced Th1 differentiation following Pam3CYS stimulation is consistent with the recent report of decreased IL12 production in cpdm macrophages following TLR2 stimulation [31]. In contrast, more Th2-specific IL4 cytokine was produced in the cpdm BMDC co-cultures than the WT control (Fig. 7B), suggesting Th2-skewed immunogenicity of cpdm BMDC. The production of Th17-specific cytokine IL17A following LPS stimulation of dendritic cells was similar between stimulated WT and cpdm BMDC cocultures (data not shown). Despite the distinct Th1- and Th2-stimulating abilities of WT and cpdm BMDC, they were equally effective in IL2 production from BMDC-T cell co-cultures except for poly I:C stimulation where WT BMDC induced more IL2 than cpdm BMDC (Fig. 7C). The Th2-biased stimulating capability of cpdm BMDC when co-cultured with allogeneic naïve CD4+ T cells is consistent with the Th2-dominant cytokine phenotype observed in cpdm mice.
10.1371/journal.pone.0031809.g007 Figure 7  Stimulated cpdm BMDC induced Th2-biased cytokine production from naïve CD4+ T cells.
WT and cpdm BMDC (5×104 cells in 0.1 mL complete medium; MHC haplotype: H-2b) were incubated with medium, 100 ng/mL LPS, 25 µg/mL poly I:C or 5 µg/mL Pam3CYS. After 24 hours, cells were washed with PBS and incubated with freshly isolated allogeneic naïve CD4+ T cells (2.5×105 cells in 0.1 mL complete medium; MHC haplotype: H-2d). After 5 days, supernatants were collected and the secretion of IFNγ, IL4, and IL2 were measured by ELISA. Negative controls are 1) stimulated BMDC without co-culture with allogeneic CD4+ T cells; 2) allogeneic CD4+ T cells without co-culture with stimulated BMDC. Samples from both negative controls had no detectable production of the aforementioned cytokines (not shown). Results are analyzed based on 2–4 mice per group. * P<0.05.

Di