Lnk/Sh2b3 Controls the Production and Function of Dendritic Cells and Regulates the Induction of IFN-γ–Producing T Cells

Dendritic cells (DCs), which were originally named after their characteristic morphology, are sparsely but widely distributed cells of hematopoietic origin. They are professional APCs and have crucial functions in the initiation of innate and adaptive immunity in infection and inflammation and in the induction of tolerance under steady-state conditions (1). The number of DCs in the periphery is maintained by the continuous generation of precursors in the bone marrow (BM) as well as by local expansion of resident DCs and their apoptosis. Macrophage DC progenitors (MDPs), originally defined as lineage (Lin)⁻ cells expressing a CX3CR1 promotor–driven GFP transgene and low levels of c-Kit/CD117 (2, 3), as well as common DC precursors (CDPs) defined with a Lin⁻IL-7Rα⁻c-Kit^intFlt3/CD135⁺ M-CSFR/CD115⁺ immunophenotype (4), have been shown to be DC progenitors in BM. Both cell types are components of granulocyte–macrophage progenitors (GMPs). Commitment to the DC lineage occurs at the MDP stage, and MDPs give rise to monocytes and to CDPs that exclusively produce plasmacytoid DCs (pDCs) and pre-DCs, a circulating DC-restricted progenitor that gives rise exclusively to conventional DCs (cDCs) in both lymphoid and nonlymphoid tissue DCs (2–10).

Lnk, recently designated as SH2B adaptor protein 3 (Sh2b3), belongs to an adaptor protein family that includes SH2-B (Sh2b1) and APS (Sh2b2). They share the presence of a homologous N-terminal domain with putative proline-rich protein interaction motifs, followed by the pleckstrin homology and Src homology 2 (SH2) domains and a conserved C-terminal tyrosine phosphorylation site. Lnk/Sh2b3 negatively regulates cytokine and growth factor signals involved in lymphohematopoiesis (11–14). Lnk^−/− mice are characterized by overproduction of B cells and expansion of hematopoietic stem cells (HSCs), as well as overactive megakaryocytopoiesis and erythropoiesis, owing to the absence of negative regulation of stem cell factor, thrombopoietin, and erythropoietin signaling pathways (13–19). Analysis of Lnk^−/− HSCs has shown that Lnk/Sh2b3 controls thrombopoietin-induced self-renewal, quiescence, and proliferation of HSCs (20, 21). Accordingly, aged Lnk^−/− mice manifest some characteristics of myeloproliferative disease (22). In humans, mutations in the LNK/SH2B3 gene have been found in a portion of myeloproliferative disease patients (23–25). Additionally, Lnk/Sh2b3 regulates cytoskeletal rearrangement. Lnk^−/− megakaryocytes cultivated on VCAM-1 (a ligand for α₄β₁ and α₄β₇ integrins) showed altered cell shapes and proplatelet formation compared with wild-type (WT) cells (19). We have reported that Lnk/Sh2b3 promotes stabilization of the developed thrombus, mainly through integrin α_IIbβ₃-mediated actin cytoskeletal reorganization (26).

Recent genome-wide association studies have demonstrated the presence of a nonsynonymous single nucleotide polymorphism in LNK/SH2B3 as a risk factor for several autoimmune diseases, including type 1 diabetes and celiac disease (CD) (27–30). CD is a common intestinal inflammatory disorder resulting from intolerance to gluten (31), and increased production of IL-15 by intestinal epithelial cells has been reported in CD patients. Activation by IL-15 and the killing of intestinal epithelial cells expressing stress- and inflammation-induced nonclassical MHC class I molecules has been suggested as an etiologic event (32). The functions of Lnk/Sh2b3 that enhance the risk for autoimmune inflammation, however, have been largely unrevealed.

DCs show high motility and morphological diversity. They capture Ags in the periphery and migrate to lymph nodes (LNs). They form protruding dendrites and extending lamellipodia in response to various stimuli. Considering those characteristics of DCs, regulation of cytoskeletal rearrangement as well as cytokine responses might play important roles in their development and functions. In this study, we demonstrate that Lnk/Sh2b3 is expressed in DCs and regulates the production and functions of DC lineage cells. In Lnk-deficient mice, the number of DCs in the spleen and peripheral LNs was increased. Lnk-deficient BM-derived DCs (BMDCs) proliferated better than did normal cells in response to GM-CSF. In the mature DC fraction, GM-CSF–mediated induction of enzymes and surface molecules was enhanced. Moreover, IL-15–mediated induction of IL-12Rβ1 and IFN-γ production were augmented in Lnk^−/− DCs. They supported IFN-γ–producing CD4⁺ T and CD8⁺ T cells significantly more efficiently than did normal DCs. Thus, a lack of Lnk/Sh2b3-mediated regulation in DCs might lead to inflammation or protective immunity mediated by IFN-γ–producing T cells.

C57BL/6 mice were purchased from Japan SLC (Shizuoka, Japan) or CLEA Japan (Tokyo, Japan). C57BL/6 mice congenic for the Ly5 locus (CD45.1) as well as Rag2^−/−, Lnk^−/−, and Lnk^−/− CD45.1 mice (13, 21) were maintained under specific pathogen-free conditions. Rag1^−/− OT-I and Rag1^−/− OT-II mice were purchased from Taconic and bred under specific pathogen-free conditions. Experiments were performed with age- and sex-matched mice at 6–14 wk of age. All mice were handled in accordance with the Guidelines for Animal Experiments of the Research Institute, National Center for Global Health and Medicine.

BMDCs were generated using standard protocols. Femurs and tibias were harvested and single-cell suspensions were prepared from flushed BM cells. Cells were then cultured in RPMI 1640 supplemented with 10% FBS, 50 μM 2-ME, 1% nonessential amino acids (Life Technologies, Carlsbad, CA), antibiotics, and either 10 ng/ml recombinant murine GM-CSF (PeproTech, London, U.K.) or 100 ng/ml recombinant human Flt3L (PeproTech). Medium was removed and replaced with fresh medium containing GM-CSF on days 2 and 4. For cultures supplemented with Flt3L, medium was not changed. At day 6 or 7, BMDCs were harvested and used for experiments.

For DC isolation, mesenteric LNs (MLNs) or spleens were collected and digested with 100 U/ml collagenase type D (Roche Applied Science, Mannheim, Germany) for 30 min at 37°C. DCs were purified by positive immunomagnetic selection or by sorting with the FACSAria (BD Biosciences, Franklin Lakes, NJ). Immunomagnetic selection was performed using biotinylated anti–CD11c Ab (N418) (BioLegend, San Diego, CA) and anti-biotin–conjugated beads and the MACS system (Miltenyi Biotec, Gladbach, Germany) according to the manufacturers’ recommendations. Purification yielded up to 80% CD11c⁺ cells. DC sorting with the FACSAria was performed to purify the CD3⁻CD19⁻CD11c⁺ population. Purification yielded up to 95% CD11c⁺ cells. For T cell isolation, spleens were collected and disrupted through a 70-μm cell strainer. Naive T cells were purified by using the FACSAria to sort for the CD4⁺CD62L⁺CD44^lo population.

Cells were incubated with anti–CD16/32 mAb (2.4G2, BD Biosciences) to prevent nonspecific binding of Abs via FcR interactions without CD16/32 staining. In general, 1 × 10⁶ cells were incubated on ice for 20 min with FITC-, PE-, PE-Cy7–, allophycocyanin-, allophycocyanin-Cy7–, and biotin-conjugated mAbs for cell surface staining. The following conjugated Abs were purchased from eBioscience (San Diego, CA): CD3ε (145-2C11), CD11b (M1/70), CD16/32 (93) , CD19 (1D3), CD34 (RAM34), CD40 (1C10), CD86 (GL1), CD115 (M-CSFR) (AFS98), CD103 (2E8), CD127 (IL-7Rα) (A7R34), Foxp3 (FJK-165), MHC class II (MHC-II; M5/114.15.2), NK1.1 (PK136), Sca-1 (D7), SIRP-1α (PB4), and TER119 (TER119). The following conjugated Abs were purchased from BioLegend: B220 (RA3-6B2), CCR9 (242503), CD4 (RM4-5), CD8α (53-6.7), CD44 (IM7), CD45.1 (A20), CD45.2 (104), CD62L (MEL14), CD207 (4C7), Gr-1 (RB6-8C5), IFN-γ (XMG1.2), Ly-6C (HK1.4), PDCA-1 (927), and Siglec-H (551). The following conjugated Abs were purchased from BD Biosciences: c-Kit (2B8), CD11c (HL3), CD135 (Flt3), CD212 (IL-12Rβ1) (114), phospho-STAT4 (38/p-Stat4), phospho-STAT5 (47/Stat5 [pY694]), and I-A^b (AF6-120.1). Aldehyde dehydrogenase (ALDH) activity in individual cells was estimated using Aldefluor staining kits (StemCell Technologies, Vancouver, BC, Canada), as described previously (33). Flow cytometry analysis was performed with a FACSCanto II (BD Biosciences).

CD11c⁺ BMDCs (1 × 10⁵ cells) were replated in 100 μl medium in 96-well plates and stimulated with various concentrations of GM-CSF. Cells were pulse-labeled with [³H]thymidine (0.2 μCi/well) during 16 h of the 64-h culture period and incorporated [³H]thymidine was measured. Purified MLN DCs (3 × 10⁴) were stimulated for 3 d with 1 ng/ml GM-CSF, 1 ng/ml IL-12 (R&D Systems, Minneapolis, MN), 1 ng/ml IL-15 (PeproTech), or combined with both IL-12 and IL-15. Cultured cell supernatants were collected and evaluated for IFN-γ with a mouse ELISA MAX standard kit (BioLegend).

Purified splenic CD11c⁺ DCs were treated with lysis buffer (2% Triton X-100, 150 mM NaCl, 50 mM Tris-HCl, 10 mM sodium fluoride, 1 mM sodium vanadate) and 1× protease inhibitor cocktail (Roche, Basel, Switzerland). Total lysates or the supernatants precipitated with anti–Lnk/Sh2b3 Ab were separated on 7.5% SDS-PAGE and transferred onto polyvinylidene difluoride membranes. Anti–Lnk/Sh2b3 Abs were generated as described before (12). For signaling in BMDCs, cells purified and starved in medium without cytokines for 6–8 h were stimulated with 1 or 10 ng/ml GM-CSF for 5 or 15 min and lysed. Total lysates derived from 1.5 × 10⁶ BMDCs were separated on a 5–20% gradient SDS-PAGE. The following Abs were obtained from Cell Signaling Technology (Danvers, MA): anti-STAT5, anti–phospho-STAT5 (Tyr⁶⁹⁴), anti-ERK1/2, anti–phospho-ERK1/2 (Thr²⁰²/Tyr²⁰⁴), and anti–phospho-JAK2 (Tyr^1007/1008). For mature DCs, purified cells were stimulated with 0.1 ng/ml IL-12 or 1 ng/ml IL-15 for 20 min. Cells were fixed and permeabilized with Phosflow buffer (BD Biosciences) and stained with anti–phospho-STAT4 or anti–phospho-STAT5.

Purified naive T cells (5 × 10⁴) and MLN DCs (2 × 10⁴) were cocultivated for 3 d on plate-bound anti-CD3ε (1 μg/ml) (eBioscience). Designated recombinant cytokines (listed below) were added to the cultures: TGF-β1 (0.5 ng/ml) (R&D Systems), IL-15 (0.1, 0.25, or 1 ng/ml), and GM-CSF (0.1, 0.25 or 1 ng/ml). Three days after the initial stimulation, cells were restimulated for 4 h with PMA (50 ng/ml) plus ionomycin (500 ng/ml). The cells were fixed and permeabilized with the Foxp3 fixation/permeabilization kit (eBioscience) and were used for detection of nuclear Foxp3 and IFN-γ cytokines. Cultured cell supernatants were collected and evaluated for IL-12/23 p40 and IFN-γ with a mouse ELISA MAX standard kit (BioLegend). For DC-free cultures, 5 × 10⁵ naive CD4⁺ T cells were cultured for 3 d with plate-bound anti-CD3ε (2 μg/ml) (eBioscience) and soluble anti-CD28 (2 μg/ml) (eBioscience) in the presence or absence of TGF-β1, IL-12 (PeproTech), IFN-γ (PeproTech), or retinoic acid (RA) (Sigma-Aldrich, St. Louis, MO).

Total RNA was extracted from purified DCs with the RNeasy kit (Qiagen, Hilden, Germany). cDNA was synthesized with Superscript III (Life Technologies) according to the manufacturer’s instructions. TaqMan real-time quantitative RT-PCR was used to detect IL-12R mRNA. TaqMan probes for IL-12Rβ1 (Il12rb1) (Mm00434189_m1), IL-12Rβ2 (Il12rb2) (Mm00434200_m1) and GAPDH (Mm99999915_g1) were obtained from Life Technologies.

Femoral BM cells were washed with PBS and counted. C57BL/6 (CD45.1) nucleated BM cells (5 × 10⁶/recipient mouse) and CD45.2 Lnk^−/− BM cells (2.5-5 × 10⁵) were i.v. transfused into lethally irradiated (9.5 Gy) C57BL/6 (CD45.1) mice. Six weeks later, PBLs were collected from the recipient mice and chimerism was confirmed by the expression of CD45.1 or CD45.2.

Freshly dissected spleens from WT C57BL/6J mice or Lnk^−/− mice were embedded in 4% carboxymethyl cellulose and frozen in n-hexane chilled in dry ice. Cryostat sections of frozen tissues were fixed with acetone, air-dried, and stained with FITC-conjugated anti-CD3ε and PE-conjugated anti–CD11c Abs. Confocal microscopic analysis was performed with an Olympus FV-500 confocal microscope.

Mice were anesthetized and hair was removed from each chest. A total volume of 400 μl 2% FITC isomer-I (Sigma-Aldrich) in a vehicle consisting of acetone/dibutylphthalate (1:1) was applied to the stripped skin region. Mice were killed and axillary LNs were collected for analysis 48 h later.

Mice were i.p. injected with 200 μl 10 mg/ml BrdU solution (Sigma-Aldrich). Incorporation of BrdU into cells in BM and spleen was analyzed 2 or 24 h after injection by using a BrdU flow kit (BD Biosciences).

Differences between groups were analyzed for statistical significance with a Student t test and considered significant when p values were <0.05.

Polymorphisms in the LNK/SH2B3 gene locus are reportedly associated with several autoimmune diseases (27–30). Because DCs greatly affect immune responses and Lnk/Sh2b3 is expressed in various hematopoietic cells and progenitors, we investigated the expression of Lnk/Sh2b3 in DCs. We purified mature CD11c⁺ DCs from spleens (>95% purity) of WT or Lnk^−/− mice, and the cell lysates were subjected to immunoprecipitation and immunoblotting using anti–Lnk/Sh2b3 Abs. The 68-kDa band corresponding to the Lnk/Sh2b3 protein was detected in lysates from WT splenic DCs, but not from the Lnk^−/− splenic DCs (Fig. 1A, left). To rule out the possibility of contamination of a small fraction of B cells that highly express Lnk/Sh2b3 in the spleen (13), we prepared BMDCs by culturing BM cells in the presence of GM-CSF, and the lysates were subjected to immunoblotting. We confirmed the expression of Lnk/Sh2b3 protein in BMDCs, and the expression level was not changed during LPS-induced maturation (Fig. 1A, right).

We next assessed DC production in Lnk^−/− mice. The total live cells and absolute number of CD11c⁺ DCs were increased in the spleens and peripheral LNs of Lnk-deficient mice compared with normal mice (Fig. 1B, Supplemental Fig. 1A, 1B). The expression levels of costimulatory molecules such as CD40 and CD86 on CD11c⁺ DCs were similar in Lnk^+/+ and Lnk^−/− mice (Supplemental Fig. 1C). Lnk^−/− mice showed mild splenomegaly because of enhanced lymphohematopoiesis (13, 15). To examine whether increased DCs in the spleen of Lnk^−/− mice accumulated within specific regions, we undertook an immunohistochemical analysis. Splenic CD11c⁺ DCs resided diffusely in the marginal zone border (34), and the pattern was not largely disturbed in Lnk^−/− mice (Supplemental Fig. 1D). We asked whether the enlarged size of the spleen or the increased number of lymphocytes in the absence of Lnk/Sh2b3 might affect the survival or production of DC lineage cells. We generated lymphocyte-deficient Lnk^−/−Rag2^−/− mice and confirmed that the mice also had increased numbers of DCs in the spleen (Supplemental Fig. 1E). Thus, the increased number of DCs was independent of lymphoid lineage cells.

To better understand how DC numbers increased in the absence of Lnk/Sh2b3, we examined DC precursors in BM. We studied the expression of Lnk/Sh2b3 in DC precursors by the use of the Lnk-Venus knockin mice, in which a part of the Lnk gene was replaced by a modified GFP Venus and the Lnk gene expression was monitored by Venus expression on flow cytometry (35). The Lnk expression levels were gradually upregulated as DC precursors matured from Lin⁻Sca-1⁻IL-7Rα⁻c-Kit⁺CD34⁺CD16/32⁻ common myeloid progenitors (CMPs) in BM, and they further increased in cDCs in spleen (Supplemental Fig. 2A). We evaluated the number of Lin⁻Sca-1⁻IL-7Rα⁻c-Kit⁺CD34⁺CD16/32⁺ GMPs as well as CMPs, from which GMPs branch. The percentage of CMPs in Lnk^−/− BM was higher than that in WT BM. There were more CMPs in BM of Lnk^−/− mice compared with that of WT mice. There was also an increased tendency in GMPs, although this did not achieve statistical significance. (Fig. 1C, 1E). GMPs differentiate into Lin⁻Sca-1⁻IL-7Rα⁻c-Kit⁺Flt3⁺ MDPs and Lin⁻Sca-1⁻IL-7Rα⁻c-Kit^intFlt3⁺M-CSFR⁺ CDPs that give rise to cDCs and pDCs (4). We evaluated MDP and CDP fractions and found that the number of MDPs increased in Lnk^−/− mice (Fig. 1D, 1E). Thus, DC lineage precursor cells in BM tended to increase in the Lnk-deficient strain compared with normal DC development.

GM-CSF and Flt3L support development of DCs in the steady-state as well as the inflammatory state. We examined the proliferation of BMDCs with GM-CSF as well as Flt3L. We found that more CD11c⁺ DCs were generated from Lnk-deficient BM cells than from WT BM (Fig. 2A). The difference in the cumulative proliferation curve became evident after day 7, when 60–70% of cultured BM cells became CD11c⁺. This result could reflect the increased proliferative capacity of DC precursor cells or altered growth responses to cytokines. To evaluate the cytokine responsiveness of committed BMDCs, we harvested and purified CD11c⁺ BMDCs after 7 d of cultivation. The cells were reseeded in the presence of various concentrations of GM-CSF, and we investigated their proliferation by measuring thymidine uptake. Lnk^−/− CD11c⁺ BMDCs showed an augmented proliferative response to GM-CSF (Fig. 2B). The phosphorylation of JAK2 (necessary for all of the biological functions after activation of GM-CSFR) (36) was enhanced and prolonged in BMDCs in the absence of Lnk/Sh2b3 (Fig. 2C). Phosphorylation of STAT5 induced by JAK2, as well as phosphorylation of ERK1/2, was also augmented and prolonged in Lnk^−/− CD11c⁺ BMDCs. Thus, in developing DCs, responses induced by GM-CSF were augmented and resulted in enhanced proliferation in the absence of Lnk/Sh2b3. We assessed increased proliferation of DC progenitors in vivo by BrdU incorporation assay. Most DC progenitors in BM rapidly incorporated BrdU, and BrdU⁺ fractions in splenic pre-cDCs and cDCs were increased in Lnk^−/− mice compared with WT mice within 1 d after bolus injection of BrdU (Supplemental Fig. 2B).

We inquired whether any of the functions of DCs were altered in the absence of Lnk/Sh2b3, and we first examined Ag capture and migration of DCs by painting FITC on skin and measured DCs carrying FITC in draining LNs (37). Although the ratio of FITC-bearing CD11c⁺ DCs seemed to be slightly reduced in Lnk^−/− mice, reflecting the increased total DC number (Supplemental Fig. 3A), the absolute numbers of migrating FITC⁺MHC-II⁺ DCs in draining LNs were comparable in Lnk^+/+ and Lnk^−/− mice (Supplemental Fig. 3B). Dermal MHC-II^high DC subpopulations determined by the expression of CD207, CD11b, and CD103 (38) were largely normal in the skin of Lnk^−/− mice (Supplemental Fig. 3C). These results suggested that Ag capture and migration ability was not significantly compromised in Lnk^−/− DCs.

Next, we examined whether responses to GM-CSF were also affected in mature DCs by Lnk deficiency. GM-CSF reportedly induces ALDH activity essential for RA production in mature DCs. ALDH activity can be monitored by use of Aldefluor dye (33). The number of Aldefluor-derived fluorescence-positive (Aldefluor⁺) DCs, that is, freshly isolated DCs carrying ALDH activity, was increased in MLNs and slightly elevated in spleens from animals lacking Lnk/Sh2b3 (Fig. 3A, 3B). We examined the induction of ALDH activity in splenic DCs following cultivation with GM-CSF. Positive Aldefluor staining of splenic DCs was more strongly induced by GM-CSF in cells from Lnk^−/− animals than those from WT DCs (Fig. 3C).

In spleen, cDCs are divided into two populations by their expression of CD11b and CD8α, namely CD11b⁺CD8α⁻ and CD11b⁻CD8α⁺ DCs. The ratio of CD11b⁺CD8α⁻ to CD11b⁻CD8α⁺ DCs was not skewed by Lnk deficiency. However, the size of the CD103⁺ fraction in CD8α⁺ DCs as well as the absolute number were significantly increased in Lnk^−/− spleens (Fig. 3D, 3E). CD103⁺ DCs, which usually reside in nonlymphoid tissues, are known to develop and expand in response to GM-CSF and have the ability to activate CD8⁺ T cells into IFN-γ–producing cells (39, 40). Overnight cultivation of splenocytes demonstrated that expression of CD103 on peripheral CD8α⁺ DCs was maintained by the presence of GM-CSF (Fig. 3F). These results suggested that GM-CSF–induced responses in DCs were exaggerated in Lnk^−/− mice.

DCs play important roles in the priming and differentiation of effector T cells in immune responses. We next examined whether dysfunction of Lnk/Sh2b3 in DCs might affect their ability to support differentiation of Th cells from naive CD4⁺ T cells. We isolated CD11c^hi DCs from MLNs of Lnk^−/− or WT mice and cultured them with naive splenic CD4⁺ T cells purified from WT mice allotypically distinguished from DCs by the expression of CD45.1 or CD45.2. After 3 d of cultivation, expression of IFN-γ and Foxp3 in T cells was analyzed by intracellular staining. In the presence of GM-CSF, cocultivating with Lnk^−/− DCs generated more IFN-γ–producing T cells compared with cocultivation with WT DCs (Fig. 4A–D). Involvement of IL-15 in the pathogenesis of CD has been suggested (32). IL-15 induces several responses critical for T cell priming in DCs (41, 42). In the presence of IL-15, Lnk^−/− DCs supported more IFN-γ–producing cells compared with normal DCs (Fig. 4A–D). The induction of IFN-γ–producing CD4⁺ T cells was increased in a dose-dependent manner for both GM-CSF and IL-15 (Fig. 4B, 4C). Those two cytokines had an additive effect for the induction of IFN-γ⁺CD4⁺ T cells under these coculture conditions (data not shown). IL-12, produced from DCs and a strong inducer of Th1 responses, was not increased in the supernatants (Fig. 4E).

We tried to clarify the molecular mechanisms by which Lnk^−/− DCs enhanced their capacity to support IFN-γ⁺CD4⁺ T cells. We screened cytokines that were involved in Th1 cell induction when DCs were cultured alone. Lnk^−/− DCs tended to produce more IFN-γ compared with normal DCs (Fig. 5A). Consistent with the amounts of IL-12 in coculture supernatants described above (Fig. 4E), IL-12 production from Lnk^−/− DCs and WT DCs were comparable (data not shown). IFN-γ production was significantly increased by the stimulation of IL-15 from Lnk^−/− DCs but not from normal DCs (Fig. 5A). IL-12 is known to induce IFN-γ production in DCs. We found that IL-12–induced IFN-γ production was exaggerated in the absence of Lnk/Sh2b3 (Fig. 5A). Addition of IL-15 together with IL-12 strongly induced IFN-γ production by normal DCs, and it completely masked the difference between Lnk^−/− and normal DCs. IL-15 is required for IL-12Rβ1 expression by DCs (41). Thus, the results indicated that the increased responsiveness of Lnk^−/− DCs to IL-15 might be responsible for their increased IFN-γ production, particularly if mediated by autocrine IL-12. DCs freshly isolated from MLNs of Lnk^−/− mice expressed more IL-12Rβ1 (Fig. 5B). After starvation in culture, Lnk^−/− DCs tended to show increased phosphorylation of STAT5, although this did not achieve statistical significance (Fig. 5C, 5D). They expressed more IL-12Rβ1 in response to IL-15 (Fig. 5B). Accordingly, IL-12–mediated STAT4 phosphorylation was observed in a higher proportion of Lnk^−/− DCs than in normal DCs (Fig. 5C, 5D).

To confirm that the altered characteristics of Lnk^−/− DCs were cell intrinsic and not a consequence of Lnk deficiency in the microenvironment of the lymphoid organ, we generated BM chimeric mice in which nearly half of the myeloid lineage cells were derived from Lnk^−/− CD45.2⁺ precursors and the remainder were derived from normal CD45.1⁺ precursors. We found that Lnk^−/−-derived CD45.2⁺CD8α⁺CD103⁺ cDCs are increased even in the spleen of BM chimeric mice whereas the percentage of CD4⁺ cDCs, CD8α⁺ cDCs, and pDCs is not largely changed in spleen and MLNs between CD45.1⁺ and CD45.2⁺ (Supplemental Fig. 2C). This result indicated that the intrinsic effect of Lnk deficiency induced CD8α⁺CD103⁺ cDCs development. Lnk^−/− CD45.2⁺ DCs and normal CD45.1⁺ DCs were purified from chimeric mice and IFN-γ production was examined (Fig. 5E). In Lnk^−/− DCs, more Il12rb1 transcripts accumulated in the presence of low concentrations of IL-15 than in normal DCs that had been simultaneously isolated from the same chimeric mice (Fig. 5F). Thus, responses mediated by IL-15 in DCs were enhanced in the absence of Lnk/Sh2b3, resulting in an increased proportion of Il12rb1-expressing DCs. The enhanced responses of Lnk^−/− DCs were due to cell-intrinsic changes and did not depend on environmental changes.

In the presence of TGF-β, naive CD4⁺ T cells differentiate into Foxp3⁺ peripheral regulatory T cells (Tregs). RA facilitates TGF-β–induced Treg production; however, it also has coadjuvant properties that induce Th1 immunity in the presence of IL-15 (42). Because Lnk^−/− DCs showed better responsiveness to GM-CSF in the induction of ALDH activity required for RA production and showed better responsiveness to IL-15 in Th1 T cell induction, we asked whether induction of Foxp3⁺ cells might be altered by cocultivation with Lnk^−/− DCs. As reported, addition of TGF-β to culture medium resulted in induction of Foxp3⁺ T cells (Fig. 6A). Although IFN-γ⁺ cells barely differentiated in cocultivation with normal DCs, some IFN-γ⁺ cells differentiated during coculture with Lnk^−/− DCs (Fig. 6A). Addition of IL-15 increased IFN-γ⁺ cell production in a dose-dependent manner, and induction of IFN-γ⁺Foxp3⁺ cells occurred only in the presence of Lnk^−/− DCs (Fig. 6A, 6B). GM-CSF alone showed only marginal effects. However, the combination of GM-CSF and IL-15 supported apparent differentiation of IFN-γ⁺ cells and IFN-γ⁺Foxp3⁺ cells in coculture with Lnk^−/− DCs but not with normal DCs (Fig. 6A).

As shown above, Lnk^−/− DCs produced more IFN-γ but not IL-12 compared with normal DCs in response to IL-15. To confirm the effects of IFN-γ overproduced by Lnk^−/− DCs in Treg differentiation, we directly added IFN-γ or IL-12 to CD4⁺ T cells in the presence of TGF-β, or TGF-β in combination with RA. As reported (42), IL-12 supported the appearance of IFN-γ⁺Foxp3⁻ cells in the presence of TGF-β and RA (Fig. 6C), although the reduction of Foxp3⁺ cells was not evident under our culture conditions. In the presence of TGF-β, IL-12 induced IFN-γ production both in Foxp3⁻ and Foxp3⁺ cells. Essentially the same results were observed following the addition of IFN-γ (Fig. 6C). Thus, the ability of DCs to support Th1 cells or Tregs was altered by Lnk deficiency in response to inflammatory cytokines GM-CSF and IL-15. Consequently, the deficiency led to enhanced expression of IL-12Rβ1 as well as augmented production of both RA and IFN-γ.

DC homeostasis is dependent on the rate at which DC progenitors enter the blood from BM, cell division, and apoptosis regulated by cytokines (43, 44). Lnk/Sh2b3 expressed in DCs constrained GM-CSF signaling and controlled the number of DCs that differentiated from progenitor cells. The number of CD11c^intCD45RA^loCD43(Ly48)^intSIRP-α^intCD4⁻CD8⁻MHC-II⁻ splenic pre-cDCs (45) was not increased significantly in Lnk^−/− mice (data not shown). Pre-cDCs reportedly expand primarily in the BM and differentiate into cDCs in lymphoid tissues via cell division (8). Interestingly, Lnk deficiency did not affect proliferation in the periphery. The transition from pre-cDCs to cDCs might be prompt, or alternatively the cell division of pre-cDCs might not be strongly affected by DC precursor trophic cytokines.

Our previous studies indicated that Lnk/Sh2b3 restricted activation of ERK1/2 in thrombopoietin-dependent signaling pathways of megakaryocytes (19) and in Kit ligand–activated MC9 mast cell lines (14). Tong et al. showed that Lnk/Sh2b3 blocked erythropoietin-mediated erythropoiesis by negatively regulating JAK2, STAT5, Akt, and ERK1/2 activation (17). In DCs, we demonstrated enhanced activation of GM-CSF–mediated signaling pathways, JAK2, STAT5, and ERK1/2 MAPK in the absence of Lnk/Sh2b3. We also found that ALDH activity induced by GM-CSF in DCs was enhanced by Lnk deficiency. ALDH is the enzyme that catalyzes retinaldehyde to RA and plays an important role in the pathway of RA synthesis from vitamin A (retinol). Splenic DCs treated with GM-CSF facilitate TGF-β–dependent differentiation of naive T cells to Foxp3⁺ T cells (Fig. 6A, WT DCs) (33). However, DC responses to IL-15 in addition to those against GM-CSF were also augmented by the absence of Lnk/Sh2b3. The responses to those cytokines made Lnk^−/− DCs strongly support the appearance of IFN-γ–producing T cells even in the presence of TGF-β (Fig. 6A, Lnk^−/− DCs).

Conversion of Tregs into IFN-γ–producing Th1 cells is mediated by DCs that have been stimulated by IL-15 and RA (42). IL-15 and RA activated the JNK pathway, which results in the production of IL-12 by DCs. Under our culture conditions, we did not observe enhanced production of IL-12 from DCs, but we did detect increased production of IFN-γ as a consequence of elevated expression of IL-12Rβ1 and autocrine production of IL-12. In the presence of TGF-β, IFN-γ as well as IL-12 induced IFN-γ production by Foxp3⁺CD4⁺ T cells that had been derived from naive CD4⁺ cells (Fig. 6C). The addition of RA reduced IFN-γ production by most cells, but the appearance of IFN-γ–producing cells was similarly supported by addition of IFN-γ as well as IL-12. Lnk^−/− DCs’ enhanced conversion of CD4⁺ T cells into IFN-γ–producing cells even in the presence of TGF-β could result from a combination of enhanced responsiveness of Lnk^−/− DCs to GM-CSF and IL-15, which results in the elevated production of RA and IFN-γ. Thus, defective or decreased functionality of Lnk/Sh2b3 in DCs affected T cell differentiation toward Th1 responses. The T cell priming skewed toward Th1 cells from naive CD4⁺ T cells and developing Tregs might facilitate autoimmune diseases such as CD and type I diabetes. Alternatively, the skewed priming toward Th1 cells might be beneficial for certain infections. The accumulation of LNK/SH2B3 single nucleotide polymorphisms in European populations might be a result of selective pressure arising from infections. Further experiments will be required to explore the relationship between Lnk/Sh2b3 function and autoimmune diseases or protection from infections.

CD103⁺ DC numbers were increased in Lnk^−/− spleens compared with WT. Those DCs were able to activate CD8⁺ T cells by cross-presentation (39). The induction of IFN-γ–producing CD8⁺ T cells was more prominent in Lnk^−/− CD8α⁺ splenic DCs (Supplemental Fig. 4). We also examined cross-presentation of GM-CSF–induced and OVA-loaded BMDCs. Proliferation of CD8⁺ T cells from OT-I transgenic mice was induced equally well by cocultivating with BMDCs from WT or Lnk-deficient BMDCs (data not shown). The expression levels of cell surface MHC class I were comparable between Lnk^−/− BMDCs and normal BMDCs. We conclude that cross-presentation on DCs was not affected by Lnk deficiency. However, induction of IFN-γ–producing CD8⁺ T cells was enhanced when they were cocultured with CD8α⁺ DCs from Lnk^−/− mice (Supplemental Fig. 4B). Several studies proved that GM-CSF–producing T cells in so-called Th1 and Th17 cells are pathogenetic and responsible for autoimmune-mediated tissue inflammation (46, 47). Enhanced responses of DCs to GM-CSF might contribute to an exacerbated inflammatory loop among DCs, naive T cells, and effector T cells.

In conclusion, our results revealed a formerly unrecognized role of Lnk/Sh2b3. The data clearly showed that Lnk/Sh2b3 regulated the production and function of DCs. The polymorphism of human LNK/SH2B3 is reportedly a risk factor for several autoimmune diseases (27–30). This intracellular adaptor protein controls the production of DC lineage cells by constraining GM-CSF signals. It also regulates GM-CSF and IL-15 signals in mature DCs and affects their ability to prime naive CD4⁺ T cells.

We thank M. Ikutani and R. Hamamichi for helpful discussions and advice and A. Yoshikawa, J. Koyama, T. Norose, and M. Anraku for technical assistance.

The authors have no financial conflicts of interest.