Loss of Phosphatase and Tensin Homolog in APCs Impedes Th17-Mediated Autoimmune Encephalomyelitis

Multiple sclerosis (MS) is the most common autoimmune disease of the CNS leading to progressive paralysis in adult humans. Genome-wide association studies support the hypothesis that a genetic predisposition affecting mostly immune genes exerts one of the greatest risks to develop MS (1, 2). The emerging inflammatory plaques are one of the hallmarks of this disease and lead to neurologic disturbances, which occur due to a blockage of signal conduction (3).

The most widely used animal model for MS is experimental autoimmune encephalomyelitis (EAE), which recapitulates many features of MS immunopathology (4, 5). Organ-specific autoimmunity, like MS, psoriasis, or rheumatoid arthritis, has long been ascribed to Th1 effector mechanisms. However, the observation that IFN-γ^−/−, IFN-γR^−/−, and IL-12p35^−/− mice are susceptible to EAE, but IL-23p19^−/− mice are completely protected, has spurred the concept that Th17 cells, maintained by IL-23, are the main disease mediators (6–11). In particular, IL-17, one of the signature cytokines released by Th17 cells, exerts many inflammatory effects, including recruitment of leukocytes and in particular neutrophils to the perivascular white matter and disruption of the blood–brain barrier by modulating the endothelial tight junctions (12–14). Additionally, patients with MS disclose elevated levels of IL-17 in their blood mononuclear cells and MS lesions (15–17).

The activation of autoreactive T cells, which target the myelin sheath and nerve axons due to an abrogation of immunologic tolerance, highlights the significance of APCs (18). Bearing in mind that APCs are instrumental in defining the generation of T cell subsets by the cytokines they produce and therefore own the capacity to stimulate the proliferation of pathogenic Th17 cells, it is important to progressively elucidate signaling pathways in APCs (19). Whereas in the early phase of MS, classically activated (M1) leukocytes, producing proinflammatory cytokines such as IL-12, IL-6, and TNF-α, infiltrate into the CNS to induce the disease, in the late phase, predominantly alternatively activated (M2) macrophages are present and resolve the inflammation (20, 21). M2 cells are known to be beneficial in the disease course of EAE and thus signaling pathways, leading to a protective phenotype of APCs that need to be assessed (22).

The PI3K pathway orchestrates a wide range of cellular processes including inflammation and immunity (23), but its contribution to autoimmunity, in particular in myeloid cells, is poorly understood (24, 25). Phosphatase and tensin homolog (PTEN) is the negative regulator of PI3K, dephosphorylating specifically the D3-position of the inositol ring (26). Studies about deletion of PTEN in thymocytes, mature T cells, or B cells have provided knowledge of paramount importance, but the role of PTEN in APCs still requires further evaluation (27–29).

In this study, we demonstrate that PTEN deficiency in APCs leads to a regulatory, immune-suppressive cellular phenotype, determined by reduced MAPK signaling, lower expression of proinflammatory cytokines IL-12/23, IL-6, and TNF-α, and upregulation of Arginase I. In in vitro MLR studies, constitutively active PI3K selectively in APCs reduced the release of Th17 signature cytokines such as IL-17A and IL-22. In the murine EAE model, clinical scores as well as the disease incidence in mice deficient for PTEN specifically in APCs were significantly lower. The generation of myelin oligodendrocyte glycoprotein (MOG)–specific pathogenic Th17 cells in the inguinal draining lymph nodes (dLN) and the spleen was impaired, accompanied by a shift from classic to alternative activation of CD11b⁺CD11c⁺ APCs in the dLN of EAE-induced PTEN-deficient mice. As a consequence, CNS infiltration of myeloid as well as nonmyeloid cells in EAE mice was almost abolished. Splenic CD11b⁺ myeloid cells as well as CNS-infiltrating monocytes/macrophages exhibited significantly reduced markers for classic activation such as inducible NO synthase (iNOS), IL-12a, and IL-23 in myeloid PTEN-deficient mice.

Our findings unequivocally show that tight regulation of PI3K signaling by PTEN in APCs is necessary for the development of Th17-dependent autoimmune pathology, and hence, targeting of PTEN signaling may therefore open new therapeutic avenues for pharmaceutical intervention.

Pten^flox/flox mice were provided by Dr. Tak W. Mak (University Health Network, Toronto, ON, Canada) and crossed with mice expressing the Cre recombinase under the control of the Lysozyme M (LysM) promoter (LysMCre pten^flox/flox or myeloid PTEN^−/−) or the CD11c promoter (CD11cCre pten^flox/flox or dendritic cell [DC] PTEN^−/−). Backcrossing was performed at least eight generations onto the C57BL/6J background. All animal studies were approved and comply with institutional guidelines.

Active EAE was induced in myeloid PTEN^−/− or DC PTEN^−/− mice and their respective wild-type (wt) littermate controls. At day 0, immunization was performed by s.c. injection of 75 μg MOG_aa35–55 (Charite Berlin) in 75 μl H₂O emulsified in 75 μl CFA, which was enriched with 10 mg/ml Mycobacterium tuberculosis (H37Ra; Difco/BD Pharmingen). A total of 200 ng pertussis toxin from Bordetella pertussis (List/Quadratech) was administered i.p. at days 0 and 2 postimmunization. Clinical signs of EAE were assessed with the following disease scores: 0, no disease; 1, tail weakness; 2, paraparesis; 3, paraplegia; 4, paraplegia with forelimb weakness; and 5, moribund or dead animals. At the day of euthanization, splenocytes and lymphocytes (inguinal dLN) were harvested and restimulated in vitro with 30 μg/ml MOG_aa35–55 peptide or anti-CD3 (eBioscience) for 72 h. Supernatants were used for determination of T cell cytokine production.

Inguinal lymph nodes from EAE-immunized myeloid PTEN and respective littermate control mice were harvested at days 0, 6, and 10. CD11b- and CD11c-positive (eBioscience) lymph node cells were FACS sorted (FACSAria; BD Biosciences) and analyzed by quantitative RT-PCR (RT-qPCR) for the expression of markers for M1 or M2 activation.

Mice were anesthetized by s.c. injection with ketamine/xylazine and perfused intracardiacally with PBS. Brain was isolated, cut in small pieces, and digested with Collagenase D/DNAse I mix (0.17 U/ml, 0.01 mg/ml; Roche) by incubating for 30 min at 37°C in a rotating incubator. For further disruption, EDTA was added (final concentration 2 mmol), and the samples were filtered through a 70-μmol cell strainer. Cells were then washed with 1× PBS and centrifuged at 400 × g for 8 min at 4°C. Isolated cells were subjected to FACS analysis to detect surface markers MHC class II (MHC II; BD Biosciences), CD11b (eBioscience), F4/80 (eBioscience), CD45 (BioLegend), CD11c (BD Biosciences), and CD206 (BioLegend) or enriched for CD11b-positive cells (Miltenyi Biotec). MACS-separated CD11b-positive cells were further stained for CD80 (BD Biosciences), CD86 (BioLegend), MHC class I (eBioscience), MHC II (eBioscience), CD11b (eBioscience), CD11c (eBioscience), and CD45 (eBioscience) and analyzed via an LSR2 flow cytometer (BD Biosciences).

Mice were anesthetized by s.c. injection with ketamine/xylazine and perfused intracardiacally with PBS. Brain and spinal cords were isolated and fixed in 4% buffered formalin. Fixed tissues were dissected and embedded in paraffin before sectioning. Sections were stained with Klüver–Barrera (KLB) and H&E using standard procedure. For immunohistochemistry, the following Abs were used: anti-CD3 (Serotec) and anti-F4/80 (Serotec). At least three cross sections from each animal were used for histological evaluation. The inflammatory index represents the total number of mononuclear infiltrates per cross section. For evaluation of demyelinated area, total and demyelinated area of each cross section in the KLB myelin staining was measured. For immunohistological evaluation, all positive cells were counted. Image J (National Institutes of Health, Bethesda, MD) was used for the above-mentioned histological analysis. We used the nonparametric Mann–Whitney U test for statistical analysis of histological evaluations.

On day 8 after EAE induction, inguinal lymph nodes and spleens were harvested to prepare single-cell suspensions. A total of 100,000 cells was seeded in quadruplicates and restimulated with 5 μg/ml MOG_aa35–55 peptide (Charite Berlin), 1 μg/ml lymphocytic choriomeningitis virus (LCMV) peptide 61–80 (AnaSpec), or 5 μg/ml anti-CD3/CD28 Dynabeads (Life Technologies) in 96-well MultiScreenHTS-IP Filter Plates (MSIPS4W10; Millipore) using CTL Test serum-free medium (C.T.L.). IL-17 and IFN-γ ELISPOTs were performed according to the manufacturer’s protocol (Mabtech) using BCIP/NBT Liquid Substrate System (Sigma-Aldrich).

The Ab against Arginase I was a kind gift from Dr. Morris. Abs for Western blotting against p38, p-p38, JNK, p-JNK, AKT, p-AKT, p-GSK3β, and PTEN were obtained from Cell Signaling Technology, and anti-Actin Ab was purchased from Sigma-Aldrich. Anti-iNOS/NOS Type II Ab was purchased from BD Transduction Laboratories. Abs used for immunohistochemistry and FACS analysis were: CD11b (Serotec); CD11c, CD45, and CD25 (BD Biosciences); CD4 (Beckman Coulter); and Foxp3 (eBioscience). LPS and CpG DNA were obtained from Invivogen. OVA was purchased from Sigma-Aldrich. Abs for ELISA against murine IL-6, IL-12/23, IFN-γ, IL-17A, IL-22, and IL-4 were obtained from eBioscience.

Briefly, bone marrow was flushed from femurs and tibias and cultivated in complete RPMI 1640 medium supplemented with 20 ng/ml recombinant mouse GM-CSF (R&D Systems). On days 3 and 6 after isolation, medium was replaced with fresh medium supplemented with 20 ng/ml GM-CSF.

Total RNA was isolated from cultivated bone marrow–derived DCs (BMDCs) using TRIzol reagent (Invitrogen) according to the manufacturer’s protocol and used for first-strand cDNA synthesis (Fermentas). mRNA expression levels were quantified by means of real-time PCR using Fast SYBR Green Master Mix (Applied Biosystems) with StepOne Real-Time PCR System (Applied Biosystems). Sequences of primers used were: PTEN forward (fwd), 5′-ACA CCG CCA AAT TTA ACT GC-3′ and PTEN reverse (rev), 5′-TAC ACC AGT CCG TCC CTT TC-3′; Arginase I fwd, 5′-GTG AAG AAC CCA CGG TCT GT-3′ and Arginase I rev, 5′-CTG GTT GTC AGG GGA GTG TT-3′; Stabilin I fwd, 5′-CCC TCC TTC TGC TCT GTG TC-3′ and Stabilin I rev, 5′-CAA ACT TGG TGT GGA TGT CG-3′; TNF fwd, 5′-AGC CCC CAG TCT GTA TCC TT-3′ and TNF rev, 5′-CTC CCT TTG CAG AAC TCA GG-3′; iNOS fwd, 5′-AAT CTT GGA GCG AGT TGT GG-3′ and iNOS rev, 5′-CAG GAA GTA GGT GAG GGC TTG-3′, HPRT fwd, 5′-CGC AGT CCC AGC GTC GTG-3′ and HPRT rev, 5′-CCA TCT CCT TCA TGA CAT CTC GAG-3′; IL-23p19 fwd, 5′-ATG CTG GAT TGC AGA GCA GTA-3′ and IL-23p19 rev, 5′-ACG GGG CAC ATT ATT TTT AGT CT-3′; IL-12p35 fwd, 5′-CCC TTG CCC TCC TAA ACC AC-3′ and IL-12p35 rev, 5′-AAG GAA CCC TTA GAG TGC TTA CT-3′; IFN-γ fwd, 5′-TGA GCT CAT TGA ATG CTT GG-3′ and IFN-γ rev, 5′-ACA GCA AGG CGA AAA AGG AT-3′; retinoic acid-related orphan receptor γt fwd, 5′-CCG CTG AGA GGG CTT CAC-3′ and retinoic acid-related orphan receptor γt rev, 5′-TGC AGG AGT AGG CCA CAT TAC A-3′; Tbet fwd: 5′-TCA ACC AGC ACC AGA CAG AG-3′ and Tbet rev: 5′-ATC CTG TAA TGG CTT GTG GG-3′; and IL-17 fwd: 5′-TGA GCT TCC CAG ATC ACA GA-3′ and IL-17 rev: 5′-TCC AGA AGG CCC TCA GAC TA-3′.

BMDCs derived from wt or PTEN^−/− mice were stimulated with LPS or CpG. After stimulation, total cell lysates were prepared (Laemmli buffer), and immunoblots were performed by electrophoresis on 10% SDS-polyacrylamide gels. Proteins were blotted onto a polyvinylidene difluoride membrane and, after blocking with 5% dry milk/0.05% Tween 20, incubated with primary Abs in the same solution. Bound Abs were detected by anti-IgG conjugated with peroxidase and subsequent chemiluminescent detection.

LPS-activated (100 ng/ml), OVA-loaded (50 μg/ml) BMDCs derived from wt or PTEN^−/− mice were cocultured with MACS-isolated responder CD4⁺ T cells from OT-II mice at a DC/T cell ratio of 1:5 (TCR-transgenic T cells specific for Ova_323–339; The Jackson Laboratory). After 72 h, culture supernatants were collected and cytokine concentrations determined via ELISA. Cells were then incubated for further 18 h in the presence of 1 μCi/well [³H]thymidine to quantify proliferation.

Purified CD4⁺ T cells were stimulated with plate-bound anti-CD3ε (1 μg/ml) and anti-CD28 (3 μg/ml) on 48-well plates (0.5 × 10⁶ cells/well total CD4⁺ T cells) in 1 ml T cell medium (RPMI GlutaMAX-I supplemented with 10% FCS, antibiotics, and 2-ME; all from Invitrogen). T cell polarization was performed as follows: for Th17 conditions, 20 ng/ml IL-6 (R&D Systems) and 1 ng/ml TGF-β1 (R&D Systems); and for Th1 conditions, recombinant human IL-2 (20 U/ml), IL-12 (5 ng/ml), and anti–IL-4 (3 μg/ml).

For intracellular cytokine staining, detection cells were restimulated with PMA (50 ng/ml) and ionomycin (500 ng/ml) (Sigma-Aldrich) for 4 h in the presence of GolgiStop (BD Biosciences). Cells were incubated with anti-CD4 before fixation with 2% paraformaldehyde and permeabilization with Perm/Wash solution (BD Pharmingen). Cells were stained with FITC-anti–IFN-γ or PE-anti–IL-17. For analysis of Foxp3⁺ T cells in the spleen, cells were stained with the Foxp3/Transcription Factor Staining Buffer Set according to the manufacturer’s instructions (eBioscience).

Statistical significance of data were calculated by use of an unpaired two-tailed Student t test. Two-way ANOVA analyses were used to analyze two groups over time. Statistical analysis was performed using GraphPad Prism software (GraphPad Software, La Jolla, CA). Results are presented as the mean ± SD. The p values <0.05 were considered statistically significant (p values were expressed as follows: *p < 0.05, **p < 0.01).

Animal experimentation is in accordance with institutional guidelines of the Medical University of Vienna; an ethical approval was obtained by the Federal Ministry for Science and Research, Vienna, Austria (BMWF-66.009/0241-II/3b/2011; BMWF-66.009/0055-II/3b/2014 and BMWF-66.009/0216-WF/II/3b/2014).

We and others have shown that the PI3K pathway is involved in the regulation of immune responses by a reduction of proinflammatory cytokines, such as IL-12 and IL-6. However, its role for autoimmune diseases is poorly understood (30–33). The present study was performed to study the impact of a lack of PTEN in macrophages and DCs on the disease course of MOG_35–55-induced EAE, which is a well-established model of MS (34). Myeloid PTEN^−/− and wt mice were immunized with MOG peptide and monitored at regular intervals for their clinical conditions related to EAE. Scoring of clinical signs demonstrated a typical disease course, including weight loss and paralysis 14 d after immunization in the wt group, whereas PTEN^−/− animals displayed a significantly lower disease incidence, delayed onset, and negligible symptoms of the disease (Fig. 1A).

For histopathological assessment, we isolated brain and spinal cord tissues taken from the mice at the day of sacrifice. Histological examination disclosed that the inflammatory indices in myeloid PTEN^−/− mice were significantly less severe than in wt mice with a similar trend for reduced macrophage infiltration and CNS-infiltrating CD3⁺ T cells. Immunohistochemical and molecular assessment in the wt situation revealed that infiltrating cells are clearly iNOS positive, indicating the classic activation of these macrophages in early (Fig. 2E, 2F) as well as late phases of EAE (data not shown). Furthermore, KLB staining of the spinal cord and software-supported quantification thereof displayed reduced demyelination, showing the profound inhibition of the disease (Fig. 1B–F). These results indicate that the absence of PTEN in myeloid cells, comprising macrophages and DCs, directly affects the decision toward the CNS autoimmune disease.

In order to characterize cell subsets that are crucial for the onset (day 10 postimmunization) of disease in the EAE model, we performed a series of experiments in myeloid PTEN-deficient mice and littermate control animals. At that stage, only in some of the wt animals were clinical signs of disease evident (three out of seven). As expected, none of the pten gene-deficient animals were affected. However, FACS analyses revealed that the majority of wt animals (six out of seven) showed varying magnitudes of myeloid cell influx into the CNS (characterized as subset 2: CD11b^highCD45^high cells/monocytes and macrophages), which was absent in the myeloid PTEN-deficient mice (Fig. 2A, 2B). The leukocyte CNS infiltration was also reflected by the clinical score, being more pronounced in overtly diseased animals. At the same time, influx of nonmyeloid cells, most probably T cells, was detected (characterized as subset 3: CD11b^negCD45^high cells) only in wt mice (Fig. 2A, 2C). To phenotype the infiltrating cell subsets that were initially observed by flow cytometry, we performed transcriptional analysis of relevant lymphoid and myeloid signature genes in ex vivo MACS-isolated cells and whole CNS tissue. We found a significant increase in rorγt and il17a transcripts in the CNS of EAE wt mice as compared with myeloid PTEN-deficient mice, which is indicative of accumulation of Th17 cells in the CNS. In parallel, t-bet and ifn-γ were increased in the wt, but clearly absent in gene-deficient animals (Fig. 2D). Next, we analyzed genes mostly expressed in the myeloid cell compartment in whole tissue and in CD11b⁺ MACS-sorted CNS cells. We detected significant upregulation for inos, il23, and il12a mRNA as well as iNOS protein by immunoblotting and immunohistochemistry in wt as compared with myeloid PTEN-deficient mice (Fig. 2E–G). We included LPS/IFN-γ–induced macrophages as positive control (+c) for iNOS expression in the immunoblot (Fig. 2F). Interestingly, we observed concomitant Arginase I expression in the CNS, indicating that monocytes and macrophages are probably skewed to a M1-like, iNOS⁺/ArgI⁺ phenotype. Immunohistological data indicates that Arginase I colocalizes with iNOS on macrophages in the spinal cord of EAE mice (data not shown).

These findings clearly show that myeloid PTEN-deficient mice are protected from clinical symptoms of EAE due to diminished numbers of infiltrating pathogenic myeloid as well as lymphoid cells and reduced or absent expression of disease-related genes.

Because the pathogenesis of EAE depends on the generation of Th17-polarized T cells (35), we studied the influence of PTEN deficiency on the cytokine production in the inguinal dLN at the preclinical stage of EAE (day 8 postimmunization). We measured significantly decreased numbers of IL-17A–producing T cells after restimulation with MOG peptide as measured by ELISPOT assay, whereas there was no difference in IL-17A production after stimulation with an unrelated control peptide (LCMV) (Fig. 3A, 3B). However, we did not observe a clear difference (p = 0.1) in the peptide- induced IFN-γ spot formation in T cells derived from wt and myeloid PTEN^−/− EAE mice (Fig. 3A, 3B). Alternatively, the differential polarization of Th cells to regulatory T cells (T_reg) could have been promoted in myeloid PTEN-deficient mice. However, when we took a closer look at CD4⁺ Foxp3⁺ T_reg, we could not detect differences at this early time point (Fig. 3C).

In line with the ELISPOT data, ELISA analysis of splenocytes restimulated ex vivo with MOG peptide 21 d after immunization for EAE showed significantly reduced levels of both IL-17A and IL-22 in myeloid PTEN^−/− mice compared with wt mice. IFN-γ production was numerically lower but we did not detect significant differences (Fig. 3D). Furthermore, we did not detect differences in the release of the Th2 signature cytokine in IL-4 in that assay (Fig. 3D). Similar results were obtained when we analyzed cytokines after restimulation of splenic and dLN T cells with anti-CD3 Ab, which revealed significant reduction of IL-17A and IL-22, whereas IFN-γ and IL-4 expression were unaltered (Fig. 4A).

Ex vivo restimulation with plate-bound anti-CD3 Ab of dLN cells or spleen cells 8 d after EAE induction showed a slight reduction of IL-17A–producing cells in the dLN cells and a significant reduction in splenocytes from myeloid PTEN^−/− mice compared with littermate control mice in the ELISPOT assay (Fig. 4B). These data clearly show that PTEN in APCs is indispensable for a fully functional Th17 response, whereas only minor effects on the polarization of Th1 cells were detected. These findings offer a potential explanation for the observed diminished CNS leukocyte influx (Fig. 2) and reduced severity of EAE (Fig. 1) in myeloid PTEN-deficient mice.

In order to analyze the cellular phenotype of myeloid cells in vivo, we FACS sorted CD11b⁺CD11c⁺ cells from the dLN of EAE mice on day 6 postimmunization with the MOG_35–55 peptide. These experiments have been performed not only in the PTEN^fl/fl LysM cre mice (myeloid PTEN^−/−) but also in the PTEN^fl/fl CD11c cre animals (DC PTEN^−/−). As a gating strategy for cell sorting, we decided to include CD11b intermediate to high as well as CD11c intermediate to high cells (Supplemental Fig. 1A–C).

In both mouse strains lacking PTEN in APCs, we found a significant reduction of the classic activation marker iNOS, but simultaneously, a significant upregulation of the marker for alternative activation Arginase I (Fig. 5A). This indicates that PTEN deficiency is responsible for a complete phenotypic switch in CD11b⁺CD11c⁺ cells. Similar results were found when we analyzed these cells on day 3 post–MOG immunization (data not shown).

Next, we aimed to analyze gene expression in monocytes and macrophages isolated from spleens of myeloid PTEN-deficient and wt mice on day 6 postimmunization. For that purpose, we MACS-isolated CD11b⁺ cells and subsequently analyzed these cells by FACS. We could clearly show an enrichment of CD11b⁺ cells in the EAE mice. Moreover, we observed a drop in total cell numbers that were obtained after isolation of cells from the spleens of myeloid PTEN-deficient mice (Fig. 5B). Gene expression analysis of selected genes revealed enhanced activation of iNOS, IL-12a, and IL-23 in wt as compared with PTEN-deficient cells, indicating that PTEN is required for classic activation (Fig. 5C, top panel). Conversely, markers for alternative activation, such as Arginase I and Stabilin I were not changed, when we compared myeloid PTEN-deficient isolated cells with wt littermate control cells. Both markers were downregulated in MOG-immunized mice from both groups. In contrast, IL-10 expression was significantly upregulated in naive PTEN-deficient mice but was unchanged in MOG-immunized mice (Fig. 5C, bottom panel). Taken together, we believe that PTEN skews myeloid cells, APCs, and macrophages to a M1-like phenotype in vivo, which supports the findings that we obtained for myeloid cells in the CNS at onset of EAE (Fig. 2).

To further assess the underlying immunological mechanisms, we stimulated bone marrow cells from wt and myeloid PTEN^−/− animals with GM-CSF to generate BMDCs, which resemble monocyte-derived or inflammatory DCs, also similar to those cells that where obtained from the dLN (Fig. 5) (36). The efficacy of PTEN deletion in BMDCs was confirmed by immunoblotting (Supplemental Fig. 2A). To characterize the functional properties of PI3K/PTEN in APCs, we stimulated wt or PTEN^−/− BMDCs with either LPS, a ligand for TLR4, or CpG DNA, a ligand for TLR9 (Fig. 6A). The assessment of several prominent markers of classical and alternative activation of APCs showed that PTEN deficiency in BMDCs had a profound effect on upregulation of Arginase 1 as well as Stabilin 1. In contrast, the expression of iNOS and TNF was significantly lower (Fig. 6A). Because these results suggested an alternative APC activation, we were interested in the molecular mechanisms leading to this regulatory phenotype. We focused our investigation on downstream effectors of the PI3K pathway. Although wt BMDCs did not exhibit detectable p-AKT basally, stimulation via TLR4 or TLR9 resulted in phosphorylation of AKT (Supplemental Fig. 2A). In contrast, PTEN deficiency led to constitutive activation of AKT in unstimulated, naive BMDCs, which was marginally enhanced upon stimulation with LPS or CpG (Supplemental Fig. 2A). We next examined intracellular mechanisms, by which the lack of PTEN promotes the regulatory phenotype. Therefore, we investigated the levels of p-p38 as well as phosphorylated GSK-3β, because both of which are considered to be key signaling check points regulating proinflammatory cytokine production. We found a reduction of p-p38 MAPK and, accordingly, phosphorylation of GSK-3β was increased in DCs activated with either LPS or CpG for 20 or 45 min, respectively (Fig. 6B). There was, however, no apparent difference in the levels of phosphorylated JNK (Supplemental Fig. 2A) or IκB degradation (not shown) after stimulation with the above-mentioned TLR ligands, providing evidence for a selective regulation of the p38 as well as the GSK-3β pathway, without skewing the TLR signaling to a general inhibition. Moreover, we detected significantly reduced IL-23p19 and IL-12p35 mRNA levels in myeloid PTEN^−/− cells, when we measured gene expression of these cytokine subunits in BMDCs or bone marrow macrophages after stimulation with LPS (Fig. 6C). In addition, we analyzed supernatants of GM-CSF–differentiated BMDCs and found significantly reduced IL-6 and IL-12/23p40 release in myeloid PTEN-deficient cells (S. Blüml, E. Sahin, V. Saferding, E. Goncalves-Alves, E. Hainzl, B. Niederreiter, A. Hladik, T. Lohmeyer, J.S. Brunner, M. Bonelli, M.I. Koenders, W.B. van den Berg, G. Superti-Furga, J.S. Smolen, G. Schabbauer, and K. Redlich, submitted for publication).

Based on these observations, we conclude that BMDCs with constitutively active PI3K, due to the lack of PTEN, acquire an alternative, regulatory activation state and are significantly impaired in the induction of cytokines like IL-23 and IL-6, both important cytokines indispensable for the generation of the Th17 type of immune responses (37–39).

Next, we tested the T cell stimulatory capacity of APCs in vitro in a model for DC–T cell interaction. For this purpose, we used TCR-transgenic CD4⁺ T cells (OT-II), which specifically recognize chicken OVA 323–339 presented on MHC II molecules. There was no difference in the capacity of OVA-loaded BMDCs from wt and PTEN^−/− to induce proliferation in OT-II cells, indicating that the capacity of activating primary T cells was not altered in myeloid PTEN^−/− animals when compared with wt animals (Fig. 7A). Importantly, however, T cells encountering PTEN^−/− APCs produced reduced amounts of IL-22, IFN-γ, and, to a lesser extent, IL-17A (Fig. 7B–D), suggesting that PTEN in BMDCs regulates their capacity to polarize T cells. These results provide evidence that PTEN mediates the DC–T cell crosstalk and thus regulates Th17 differentiation.

Intracellular staining of CD4⁺ T cells isolated from wt and myeloid PTEN^−/− mice cultured under Th1- or Th17-polarizing conditions in vitro revealed that percentages of IFN-γ– and IL-17A–producing cells were comparable between the two groups of animals (Supplemental Fig. 3A). This indicates that the T cells derived from myeloid PTEN-deficient mice exhibit the intrinsic capabilities to fully polarize into effector T cells under in vitro conditions.

Altogether, PTEN has a direct role in DCs to promote the development of Th17 responses in vivo and in vitro.

To dissect the selective roles of PTEN in different immune cell subtypes, we generated mice lacking PTEN only in CD11c⁺ innate immune cells (DC PTEN^−/−). These mice have a more narrow cellular deletion spectrum as compared with the LysM cre deleter strain. Analysis of BMDCs derived from DC PTEN^−/− mice resulted in a comparable cellular phenotype (gene deletion efficiency, signaling pattern, M2 marker expression, and cytokine expression and release) as compared with myeloid PTEN^−/− (Fig. 8A–D). After induction of EAE in these animals, clinical scoring revealed significantly reduced clinical symptoms of EAE in DC PTEN^−/− compared with wt animals, similar but somewhat less pronounced as compared with the phenotype we observed in the myeloid PTEN^−/−. This suggests that PTEN deficiency due to gene deletion in APCs, including CD11c⁺ DCs, essentially protected from Th17-mediated disease, corroborating the beneficial effects we observed in myeloid PTEN-deficient mice (Fig. 8E).

Intrinsic signals of innate immune cells triggering pathogenic Th17-mediated inflammation are poorly understood. In this study, we identified the PI3K/PTEN signaling axis as a critical modulator of DC functions important for Th17 subset polarization, which is indispensable for organ-specific inflammatory autoimmune disorders such as MS.

To discern functional differences in vivo, we generated conditional knockout mouse models with PTEN deficiency in the myeloid cell compartment (LysM-Cre) and in CD11c⁺ cells (CD11c-Cre) (40, 41). The evolutionary conserved PI3K pathway orchestrates a broad range of biological processes with protective or deleterious roles in inflammation and immunity depending on the context. The majority of studies concentrates on the inhibition or knockdown of selective PI3K isoforms, showing its link to the activation or proliferation of T and B cells, Th cell differentiation, and chemokine-dependent migration of neutrophils and macrophages (42–48).

PI3K and PTEN are clearly implicated in the differentiation and proliferation of conventional CD8⁺ and CD103⁺ DC in response to Flt3 ligand (49). However, the functional properties of the PI3K/PTEN signaling pathway in APCs, in particular inflammatory or monocyte-derived DCs, have not been defined yet. Phenotyping of APCs with a constitutively active PI3K revealed a regulatory, alternatively activated phenotype, which could also be identified in vivo in lymph node–derived CD11b⁺CD11c⁺ APCs, indicated by the upregulation of Arginase I and Stabilin I and reduced expression of iNOS and TNF-α, which points at a limited potency to develop TNF- and iNOS-producing DCs (50). These CD11c⁺ myeloid cells are relevant for pathogen killing, but may contribute to tissue damage as well.

Analysis of downstream effector signaling pathways disclosed that activation of the mitogen- and stress-activated protein kinase p38, shown to be critical for autoimmune responses (51), was significantly reduced in PTEN^−/− DC, stimulated with pathogen-associated molecular patterns such as LPS or CpG DNA. These data support the notion for a critical involvement of the PI3K/PTEN pathway modulating signaling pathways indispensable for autoimmunity.

T cells feature a large degree of plasticity (52). APCs provide a specific cytokine milieu that regulates T cell plasticity and subsequent differentiation into distinct T cell subsets (53). Thereby, APCs orchestrate the adaptive immune response. Several studies have shown that IL-6, TGF-β, and, most importantly, IL-23 are required T cell–polarizing cytokines, which promote the expression of IL-23R and the expansion of Th17 cells (39, 54–60). In the current study, we provide in vitro as well in vivo evidence that APCs are strongly dependent on PTEN to generate pathogenic CD4⁺ T cells. Recently, we showed that PTEN deficiency reduces IL-6 and IL-12/23 expression in macrophages and imprints CEBP/β and STAT3-dependent signaling (31). Additionally, we demonstrated previously in lungs of myeloid PTEN-deficient mice, infected with Streptococcus pneumoniae, that TNF-α expression was reduced and IL-10 simultaneously increased, protecting the animal from exacerbated tissue damage (32). In this study, we provide genetic evidence that GM-CSF–differentiated DC, lacking PTEN, disclosed a diminished Th17-polarizing cytokine profile as measured by significantly reduced IL-6 and IL-12/23p40 release. This results in profound effects on T cell polarization without affecting concurrent T cell proliferation. Therefore, we propose that PI3K activation in APCs, which is enhanced and prolonged in case of PTEN deficiency, diverts the cytokine production from a Th17 (Th1)–inducing profile to a rather immune-suppressive cytokine/enzyme environment. PTEN deficiency in APCs clearly skews CD11b⁺CD11c⁺ myeloid cells to an alternative activation state with low iNOS and high Arginase I expression. However, these changes in the cellular phenotype of APCs and the cytokine milieu in the lymphoid organs did not have an impact on the number of T_reg in myeloid PTEN-deficient EAE mice. Although PTEN deficiency in APCs prevents Th17 polarization, we did not find a significant increase in Th2 cytokine production in in vitro as well as in vivo assays.

PTEN deficiency and concomitant activation of PI3K and AKT in T cells lead to dramatic phenotypic changes, with increased proliferation, IL-2 production, and Th1 polarization measured by IFN-γ production (61, 62). In T cells derived from mice with PTEN deficiency in myeloid cells, we did not find an intrinsic disability to polarize into either Th1 or Th17 cells. This observation excludes the possibility that our mice have dysfunctional T cells a priori and underlines our hypothesis that APCs lacking PTEN during the induction of the disease are not able to fully generate pathogenic T cells required to cause EAE pathology.

In the clinically relevant murine EAE model, PTEN in APCs regulated the fate of pathogenic Th17 cells. PTEN gene deficiency in myeloid cells (LysM) or CD11c⁺ cells (CD11c cre) resulted in a decrease of signature cytokines for pathogenic T cells, such as IL-17A or IL-22, suppressed disease severity and incidence. In EAE, CCR2⁺Ly6C^hi monocytes/macrophages infiltrate the CNS and contribute to tissue damage and disease progression (63, 64). In situ, in the spinal cord of MOG-immunized mice, we found a marked reduction of F4/80⁺iNOS⁺ M1 macrophages in PTEN^−/− mice compared with wt animals. However, after analysis of the leukocyte influx data in the early stages of the disease, we assume that the small number of myeloid cells infiltrating the CNS in the myeloid PTEN-deficient mice still resemble M1, iNOS⁺ monocytes/macrophages, albeit at a much lower level as compared with wt littermate control mice. Clearly, we could not detect alternative activation of infiltrating cells in myeloid PTEN-deficient mice at onset as well as recovery phase of EAE. This phenomenon can also be observed in secondary lymphoid organs such as the spleen on day 6 of EAE.

Because we could recently show that PTEN deficiency in macrophages (LysM cre) leads to alternative activation (31), we suggest that in the recovery phase of EAE, the change in macrophage or microglia phenotype due to the PTEN deficiency might also contribute to the reduced disease severity.

These findings indicate that the dysfunctional Th17 responses in myeloid PTEN-deficient mice preclude macrophage recruitment and activation and thereby promote protection from the development of the disease.

In conclusion, the regulatory phenotype observed in PTEN-deficient DCs in vitro and the beneficial effects on EAE disease progression in mice lacking PTEN provides genetic evidence that sustained PI3K activation in APCs abrogates pathogenic Th17 cells polarization in autoimmunity. This underscores a critical role for PTEN in disease pathogenesis and warrants further possibilities for therapeutic intervention.

We thank Susanne Humpeler for excellent technical assistance.

The authors have no financial conflicts of interest.