Nuclear Localization of Suppressor of Cytokine Signaling-1 Regulates Local Immunity in the Lung

Suppressor of cytokine signaling 1 (SOCS1) is a negative feedback inhibitor of cytoplasmic Janus kinase and signal transducer and activator of transcription (STAT) signaling. SOCS1 also contains a nuclear localization sequence (NLS), yet, the in vivo importance of nuclear translocation is unknown. We generated transgenic mice containing mutated Socs1ΔNLS that fails to translocate in the cell nucleus (MGLtg mice). Whereas mice fully deficient for SOCS1 die within the first 3 weeks due to excessive interferon signaling and multiorgan inflammation, mice expressing only non-nuclear Socs1ΔNLS (Socs1−/−MGLtg mice) were rescued from early lethality. Canonical interferon gamma signaling was still functional in Socs1−/−MGLtg mice as shown by unaltered tyrosine phosphorylation of STAT1 and whole genome expression analysis. However, a subset of NFκB inducible genes was dysregulated. Socs1−/−MGLtg mice spontaneously developed low-grade inflammation in the lung and had elevated Th2-type cytokines. Upon ovalbumin sensitization and challenge, airway eosinophilia was increased in Socs1−/−MGLtg mice. Decreased transepithelial electrical resistance in trachea epithelial cells from Socs1−/−MGLtg mice suggests disrupted epithelial cell barrier. The results indicate that nuclear SOCS1 is a regulator of local immunity in the lung and unravel a so far unrecognized function for SOCS1 in the cell nucleus.

Suppressor of cytokine signaling 1 (SOCS1) is a negative feedback inhibitor of cytoplasmic Janus kinase and signal transducer and activator of transcription (STAT) signaling. SOCS1 also contains a nuclear localization sequence (NLS), yet, the in vivo importance of nuclear translocation is unknown. We generated transgenic mice containing mutated Socs1ΔNLS that fails to translocate in the cell nucleus (MGL tg mice). Whereas mice fully deficient for SOCS1 die within the first 3 weeks due to excessive interferon signaling and multiorgan inflammation, mice expressing only non-nuclear Socs1ΔNLS (Socs1 −/− MGL tg mice) were rescued from early lethality. Canonical interferon gamma signaling was still functional in Socs1 −/− MGL tg mice as shown by unaltered tyrosine phosphorylation of STAT1 and whole genome expression analysis. However, a subset of NFκB inducible genes was dysregulated. Socs1 −/− MGL tg mice spontaneously developed low-grade inflammation in the lung and had elevated Th2-type cytokines. Upon ovalbumin sensitization and challenge, airway eosinophilia was increased in Socs1 −/− MGL tg mice. Decreased transepithelial electrical resistance in trachea epithelial cells from Socs1 −/− MGL tg mice suggests disrupted epithelial cell barrier. The results indicate that nuclear SOCS1 is a regulator of local immunity in the lung and unravel a so far unrecognized function for SOCS1 in the cell nucleus.
Keywords: rodent, inflammation, signal transduction, transgenic mice, lung, sOcs1 inTrODUcTiOn The suppressor of cytokine signaling (SOCS) family is important for negative feedback inhibition of Janus kinases (JAKs) and signal transducer and activator of transcription (STAT) signaling. All eight members, namely SOCS1-7 and CIS, share key structural elements such as the central cytokine-inducible Src-homology 2 (SH2) domain and the shared C-terminal SOCS-box (1,2). Both SOCS1 and SOCS3 additionally contain a kinase inhibitory region Nuclear Suppressor of Cytokine Signaling-1 Frontiers in Immunology | www.frontiersin.org November 2016 | Volume 7 | Article 514 (KIR) by which they can act as a pseudosubstrate for JAKs (3). SOCS1 was first described in 1997 as a negative feedback inhibitor of cytoplasmic JAK/STAT signaling (4)(5)(6). By the means of the extended SH2 subdomain (eSS) and the KIR, SOCS1 directly binds to JAK2, thereby inhibiting its catalytic activity (3). SOCS1 has further been shown to occupy binding sites for STATs by interacting with interferon receptor domains (7,8). Finally, the SOCS box in SOCS1 mediates interactions with elongins B and C and acts as an E3 ubiquitin ligase that targets JAKs or cytokine receptor complexes for proteasomal degradation (9,10). In addition to JAK/STAT signaling, SOCS1 has been shown to act as a cross talk inhibitor for TLR signaling pathways (11,12). Besides indirect paracrine inhibition of TLR signaling by IFNβ (13), SOCS1 additionally contributes to direct negative regulation by interacting with components of the TLR signaling pathway (12,14). Canonical IFNγ signaling functions by binding to the IFNγ receptor complex, activating Janus kinases (JAK1/2), and subsequent phosphorylation of STAT1 that dimerizes and translocates into the nucleus. Once in the nucleus, activated pY-STAT1 binds to gamma-activated sequence (GAS) elements within the promoters of IFNγ-responsive genes (ISGs) -mostly, "canonical" antiviral genes (15,16). While pY-STAT1 is thought to be pivotal for the IFNγ response, a number of studies have shown that pY-STAT1-independent pathways also exist (17)(18)(19). There is emerging evidence that STAT-independent pathways play important roles in mediating signals for the generation of IFNγresponses such as the mitogen-activated protein kinase (MAPK) or PI3K/AKT pathway (20). It has been previously shown that IFNAR1, TYK2, and STAT1 may translocate into the nucleus (21,22). The presence of unphosphorylated STAT1 in the cell nucleus has been shown to increase expression of only a subset of "non-canonical" IFNγ-induced genes that are pY-STAT1independent (23).
In 2008, a nuclear localization sequence (NLS) has been identified in SOCS1 located between the central SH2 domain and the SOCS box (amino acids 159-173). The NLS resulted in translocation of the protein into the cell nucleus (33,34). Substitution of this sequence with the respective region of SOCS3 showed loss of nuclear localization, whereas fusion of the SOCS1-NLS to the cytoplasmic SOCS family member CIS induced nuclear localization (33). It has been shown that SOCS1 directly interacts with the tumor suppressor p53 leading to activation of p53 via phosphorylation (35). Moreover, SOCS1 induces proteasomal degradation of NFκB (36,37) and, in particular, it interacts with the NFκB subunit p65 in the cell nucleus, thereby limiting induction of a subset of NFκB dependent genes (38).
However, the function of SOCS1 in the cell nucleus in vivo remains elusive. Therefore, we generated a transgenic mouse that only expresses a non-nuclear mutant SOCS1. Mice with transgenic expression of a bacterial artificial chromosome (BAC) containing a mutated Socs1 locus with non-nuclear Socs1ΔNLS, eGFP, and LuciferaseCBG99 (MGL) were generated and backcrossed to Socs1 −/− mice. Socs1 −/− MGL tg mice survived the early lethal phenotype of Socs1 −/− mice, showed unaltered canonical IFNγ-signaling, yet, displayed signs of low-grade airway inflammation and Th2 deviation. Decreased transepithelial electrical resistance (TER) in trachea epithelial cells from Socs1 −/− MGL tg mice suggests disrupted epithelial integrity. Socs1 −/− MGL tg mice present a valuable tool to study the nuclear function of SOCS1 in vivo and allow investigating local immune regulation in the lung by nuclear SOCS1.

Whole-genome expression analysis
Total RNA from 2.5 × 10 5 BMMs was isolated as described above. The quality of total RNA was checked by gel analysis using the total RNA Nano chip assay on an Agilent 2100 Bioanalyzer (Agilent Technologies GmbH, Berlin, Germany , array signals were developed by a 10 min incubation in 2 ml of 1 μg/ml Cy3-streptavidin (Amersham Biosciences, Buckinghamshire, UK) solution and 1% blocking solution. After a final wash in E1BC, the arrays were dried and scanned. Microarray scanning was done using an iScan array scanner. Data extraction was done for all beads individually, and outliers were removed when >2.5 median absolute deviation (MAD). All remaining data points were used for the calculation of the mean average signal for a given probe, and SD for each probe was calculated (ArrayExpress accession E-MTAB-4938). Data was processed using R, including log2 transformation of the data, significance (p ≤ 0.05), and fold change (log2 ≤ −1 or ≥1) filtered. Data were normalized to remove systematic variation and background subtraction. Pathway annotation was performed using the Protein Analysis through Evolutionary Relationships (PANTHER) classification system and transcription factor binding sites (TFBS) among the differentially regulated genes were analyzed using the overrepresentation analysis tool oPOSSUM.

enzyme-linked immunosorbent assay
The 2.5 × 10 5 CD11c + cells were stimulated as indicated in 96-well plates in 200 μl RPMI supplemented with 10% (v/v) FCS and P/S. Supernatants were harvested and analyzed for cytokines by commercially available enzyme-linked immunosorbent assay (ELISA) kits for TNFα and IL-12p40 (BD Biosciences, Heidelberg, Germany). Cytokines were detected by measuring the absorbance at 490 nm with a 650 nm reference in a photometer (Sunrise reader, Tecan, Salzburg, Austria). Cytokine concentrations were calculated according to a standard dilution of the respective recombinant cytokines using Magellan V 5.0 software (Tecan, Salzburg, Austria).

intratracheal il-13 instillation
Mice were anesthetized with isofluorane for 30 s and allowed to hang vertically with their mouths open, supported by a taut string placed under their canine teeth. Their tongues were gently withdrawn with a blunt forceps to keep them from swallowing. Twenty microliters of PBS with or without 5 μg IL-13 (#210-13, Peprotech, Hamburg, Germany) was applied onto the base of their tongues. When the mice had aspirated the applied solution, they were put on their site until they woke up. This intratracheal instillation was performed on days 1, 2, and 3. Mice were analyzed 24 h after the last treatment (permit no. 35-9185.81/G-35/16).

Bronchoalveolar lavage
Lungs were rinsed with 1 ml fresh, ice-cold PBS containing protease inhibitor (Roche, Basel, Switzerland) via a tracheal canula, and obtained cells were counted using a Neubauer chamber. Cytospins were prepared for each sample by centrifugation of 50 μl BAL fluid plus 150 μl of sterile PBS and subsequently stained with Diff-Quik (Medion Diagnostics, Duedingen, Switzerland). Cells were microscopically differentiated and classified as macrophages, neutrophils, eosinophils, or lymphocytes, using standard morphologic criteria (42).

Preparation of single-cell suspensions
For analysis of lung homogenate by qPCR, lungs were perfused through the right ventricle with PBS. Once lungs appeared white, they were removed and sectioned. Dissected lung tissue was then incubated with Liberase™ (100 μg/ml, #5401119001, Roche, Basel, Switzerland) and DNaseI (200 μg/ml, #11284932001, Sigma, MO, USA) at 37°C for 1 h. Digested lung tissue was gently disrupted by passage through a 19-G needle and afterward through a 70-μm pore size nylon cell strainer. Red blood cells were lysed using Red blood cell lysis buffer (eBioscience, CA, USA). CD11c + or CD4 + T cells were isolated using the positive selection CD11c + beads or the negative selection CD4 + T Cell Isolation Kit (Miltenyi Biotec, Bergisch Gladbach, Germany), respectively. Magnetically labeled cells were isolated via the autoMACS Separator (Miltenyi Biotec). For single cell suspension of splenocytes, spleens were dissected and treated as described above without enzymatic digestion.

T cell Differentiation
CD4 + T cells were isolated from a single cell suspension as described above. 1 × 10 5 cells per well were plated in 96-well round bottom plate in 100 μl RPMI plus β-mercaptoethanol nFκB p65 activity assay NFκB activity was measured in 1 × 10 6 cells of a lung homogenate by the TransAM™ NFκB p65 protein assay (Active Motif, Carlsbad, CA, USA), an ELISA-based method designed to specifically detect and quantify NFκB p65 subunit activation. As a positive control, Raji nuclear extract was used that was  provided with the kit. Wild-type oligonucleotides were used as an internal specificity control. The assay was performed according to the manufacturer's protocol and analyzed using a microplate absorbance reader (Sunrise reader, Tecan, Salzburg, Austria).

histopathological analysis
Organs were fixed via a tracheal canula under constant pressure of 20 cm H2O using 4% (w/v) phosphate buffered paraformaldehyde overnight. Tissues were embedded in paraffin. For analysis of lung inflammation, 2 μm sections were stained with periodic acid-Schiff (PAS) or with hematoxylin and eosin (H&E), respectively.
statistics All experiments were repeated three times unless stated otherwise. Data are shown as mean + SD. Statistical significance of comparison between two groups was determined by two-tailed unpaired Student's t-test (for data sets following Gaussian distribution), Wilcoxon matched pairs test (for data sets not following Gaussian distribution), or two-way ANOVA including Bonferroni post-test (for multiple comparisons). All statistical analyses were done using GraphPad Prism (GraphPad 6.05, San Diego, CA, USA) software. Differences were considered significant at *p < 0.05, **p < 0.01, and ***p < 0.001.

resUlTs sOcs1Δnls is localized in the cytoplasm
It has been shown that SOCS1 is able to translocate into the cell nucleus due to a functional NLS localized between the SH2 domain and the SOCS-box (amino acid 159-173) (33,34). Confirming these results with murine SOCS1 constructs, NIH3T3 cells transiently transfected with murine eGFP-Socs1 showed nuclear localization of the GFP-tagged protein. In contrast, eGFP-SOCS1ΔNLS, in which the NLS has been replaced by the murine SOCS3 sequence (33), was localized more in the cytoplasm ( Figure 1A). Cells transfected with Socs1wt were also stimulated with IFNγ. Upon stimulation enhanced fluorescence in the cytoplasm could be observed, suggesting that SOCS1 is partly translocating out of the nucleus to inhibit signaling in the cytoplasm (Figure 1B). To verify that SOCS1ΔNLS is still functional in the cytoplasm, inhibition of IFNγ signaling was analyzed by Western Blotting. Therefore, the murine macrophage cell line Raw264.7 was transiently transfected with eGFP, eGFP-Socs1, or eGFP-Socs1ΔNLS. Tyrosine phosphorylation of STAT1 was examined upon treatment with IFNγ 1-6 h post-transfection (Figures 1C,D). Already 1 h after IFNγ treatment, eGFP-Socs1 or eGFP-Socs1ΔNLS transfected cells showed lower levels of phosphorylated STAT1 (59 or 57% as compared to eGFP transfected cells 1 h after IFNγ stimulation). Importantly, we could not observe any differences in phosphorylated STAT1 levels between eGFP-Socs1 and eGFP-Socs1ΔNLS-transfected cells. Data suggest that both SOCS1 and SOCS1ΔNLS were effectively inhibiting IFNγ-induced STAT1 tyrosine phosphorylation, which occurs at the level of receptor activation.
generation and characterization of Mice expressing non-nuclear sOcs1Δnls To analyze the function of SOCS1 in the cell nucleus in vivo, transgenic mice were established using a BAC containing a mutated Socs1 locus with non-nuclear Socs1ΔNLS, eGFP, and LuciferaseCBG99, termed MGL (Figure 2A). 2A peptide sequences between the protein coding regions result in three  separate proteins. Thereby, 21 amino acids remain at the C-terminus of SOCS1ΔNLS. The combined expression of GFP and luciferase together with SOCS1 allows using these mice as reporter mice as well. Quantitative real-time PCR was established to confirm mRNA expression of Socs1ΔNLS in bone marrow-derived macrophages (BMMs) from BAC transgenic mice ( Figure 2B). To exclude founder-specific effects due to different integration sites of the BAC, stable expression and regulation of the mutated Socs1 locus was examined in different founders. Therefore, offsprings of the founders #53, #45, and #29 were analyzed with respect to the expression of Socs1 and Socs1ΔNLS in BMMs upon stimulation with IFNγ for 24 h using the qPCR strategy as described in Figure 2B. After stimulation with IFNγ, mRNA expression of both Socs1 and Socs1ΔNLS were induced to a similar amount in all three founders ( Figure 2C). We observed no differences in expression levels of Socs1wt and Socs1ΔNLS. Due to reporter functions of the mutated Socs1 locus, luciferase assay was performed in BMMs after stimulation with IFNγ, showing similar luciferase activity in BMMs for the three founders ( Figure 2D). In addition, GFP positive BMMs were analyzed by flow cytometry (Figure 2E). There was an increase in the percentage of GFP-positive cells after stimulation with IFNγ, and this was similar for founders #53 (26% ± 9.6), #45 (25% ± 8.7), and #29 (21% ± 9.2). Taken together, all three founders showed similar expression levels of the mutated Socs1 locus. Mutant Socs1 expression was also in the same range as

Functional impairment of nFκB inhibition in Socs1 −/− MGL tg Mice
To analyze the function of SOCS1 in the cell nucleus, Socs1MGL tg mice were mated with Socs1 +/− mice to generate Socs1 −/− MGL tg mice, expressing Socs1ΔNLS in an otherwise Socs1-deficient background. In order to show that Socs1 −/− MGL tg mice indeed lack SOCS1 in the cell nucleus, we tried staining of endogenous SOCS1 by immunohistochemistry. However, we could not find a sufficiently specific antibody that was not staining sections from Socs1 −/− mice (including newly generated antibodies). Therefore, we decided to do a functional approach to verify non-nuclear expression of Socs1MGL. It has been reported that nuclear SOCS1 limits NFκB signaling by degradation of the NFκB subunit p65 (38). To examine whether NFκB signaling is altered in Socs1 −/− MGL tg mice, CD11c + cells were isolated from lungs and stimulated ex vivo with TLR agonists. Stimulation of CD11c + cells from Socs1 −/− MGL tg mice with CpG-DNA, LPS, and pI:C for 24 h led to an increased protein expression of IL-12p40 as compared to CD11c + cells from Socs1 +/− MGL tg mice ( Figure 2F). The same could be shown in CD11c + cells isolated from spleens ( Figure S1 in Supplementary Material). Data suggest sustained NFκB activation in Socs1 −/− MGL tg mice that was confirmed by transcription factor binding assay for p65 ( Figure S1 in Supplementary Material). We did not detect differences regarding TNFα protein levels ( Figure S1 in Supplementary Material), which is in full accordance with previous findings (38) showing that only a subset of NFκB dependent genes is altered in Socs1 −/− MGL tg mice. In contrast, IL-12p40 induction that needs prolonged binding of NFκB to its promoter (44) was sensitive to SOCS1-induced NFκB inhibition. The results entirely phenocopy in vitro data using non-nuclear SOCS1ΔNLS, suggesting that Socs1 −/− MGL tg mice functionally lack SOCS1 in the cell nucleus.
Socs1 −/− MGL tg Mice survive the early lethal Phenotype as compared to Socs1 −/− Mice Socs1 −/− mice die within 2-3 weeks due to multiorgan inflammation (24)(25)(26). In contrast, Socs1 −/− MGL tg mice survived and showed no early lethality up to 60 days (Figure 3A), suggesting that lack of Socs1wt is rescued by Socs1ΔNLS. Socs1 +/− mice also showed normal survival indicating that one allele of Socs1 is sufficient for rescue of the severe knockout phenotype. In a small cohort (n = 4), survival of Socs1 −/− MGL tg mice was recorded for an extended period ( Figure S2A in Supplementary Material). Up to 38 weeks, Socs1 −/− MGL tg mice appeared healthy without overt abnormalities. Although lethality was rescued in Socs1 −/− MGL tg mice, the mice showed reduced body weight both for female and male 8-week-old mice ( Figure 3B) suggesting that lack of nuclear SOCS1 results in partial functional impairment. In the long-term survival cohort, Socs1 −/− MGL tg mice showed slightly reduced body weight as well ( Figure S2 in Supplementary Material).

similar mrna expression levels, Protein levels, and Protein half-life of sOcs1 and sOcs1Δnls
To verify that BMMs of Socs1 −/− MGL tg mice have similar expression levels of total Socs1 (both Socs1 and Socs1ΔNLS) as compared to BMMs of Socs1 +/− and Socs1 +/− MGL tg mice, cells were stimulated with IFNγ for 24 h and qPCR was performed to detect mRNA of total Socs1 using the qPCR strategy as described in Figure 2B. Similar expression of total Socs1 mRNA could be verified in BMMs of Socs1 −/− MGL tg , Socs1 +/− , and Socs1 +/− MGL tg mice ( Figure 4A). For further analysis, we generated a new SOCS1 antibody. This antibody did not detect a band with the expected molecular weight for SOCS1 in lysates of BMMs of Socs1 −/− mice ( Figure 4B), thus proving specificity in Western Blot analysis. We confirmed expression of SOCS1 and SOCS1ΔNLS protein in lysates of BMMs of Socs1 +/− and Socs1 −/− MGL tg mice stimulated with IFNγ for 6 h. The higher molecular weight of SOCS1ΔNLS likely resulted from the additional 21 amino acids after cleavage of the 2A sequence. In Figure 4C, quantification was done normalized to β-actin expression. Comparable protein levels for SOCS1 and SOCS1ΔNLS were detected ( Figure 4B). The NLS of SOCS1 (RRMLGAPLRQRRVR) resembles a bipartite NLS composed of two basic stretches. Since the basic amino acid lysine is important for marking proteins for the ubiquitin proteasome pathway (45,46), we addressed the question whether exchanging the NLS with the SOCS3 counterpart might alter protein half-life. We, therefore, performed a cycloheximide (CHX) chase experiment (Figures 4D,E). Six hours post stimulation with IFNγ, CHX was added to block nascent protein synthesis. Already after 4 h of CHX treatment, there was only 4% of the SOCS1 protein and 12% of the SOCS1ΔNLS protein remaining. In summary, we did not observe alteration of mRNA expression levels, protein levels, or protein half-life upon mutating SOCS1 into SOCS1ΔNLS.  (Figure 3A). As it is known that Socs1 −/− can be rescued by the administration of anti-IFNγ antibodies in the neonatal period or by using Socs1 −/− IFNγ −/− mice (47,48), we hypothesized that canonical IFNγ signaling is not altered in Socs1 −/− MGL tg mice. To test this hypothesis, tyrosine phosphorylation of STAT1 was examined in BMMs upon treatment with IFNγ for 1-6 h ( Figure 5A). IFNγ signaling was prolonged in Socs1 −/− mice as shown by the sustained levels of phosphorylated STAT1. There was a decline in phosphorylated STAT1 levels for both Socs1 +/− MGL tg mice (to 20%) and Socs1 −/− MGL tg mice (to 22%) already 4 h after IFNγ stimulation as compared to 58% for Socs1 −/− mice (Figures 5A,B). No significant differences were observed between Socs1 +/− MGL tg and Socs1 −/− MGL tg mice. Analyzing mRNA expression levels of classical IFNγ target genes, both iNOS and Irf9 were induced upon stimulation with IFNγ to a similar extent in Socs1 +/− MGL tg and Socs1 −/− MGL tg mice ( Figure 5C). Of note, the expression levels were also similar to Socs1 +/− mice, arguing against gene dosage effects. This allows interpretation of the data from Socs1 −/− MGL tg mice with regards to lack of nuclear SOCS1 and not altered concentration of cytoplasmic SOCS1. There was a minor, but non-significant, increase in the expression level of Icam-1 after 6 h of stimulation in BMMs of Socs1 −/− MGL tg mice. These findings indicate that SOCS1ΔNLS was still able to regulate cytoplasmic signaling pathways and that canonical IFNγ signaling was not altered in Socs1 −/− MGL tg mice.

Differential expression of a subset of non-canonical iFnγ Target genes by Socs1 −/− MGL tg Mice
To support the hypothesis that canonical IFNγ signaling is not altered in Socs1 −/− MGL tg mice, whole-genome expression analysis was performed. Therefore, BMMs were stimulated with IFNγ for 24 h and RNA was extracted and subjected to whole-genome expression analysis. 1097 genes were differentially regulated between untreated and IFNγ stimulated cells, but only 86 genes  Figure 6A). Most differentially regulated genes were induced rather than repressed in Socs1 −/− MGL tg mice (Figures 6B,C). The top 10 differentially regulated genes ( Figure 6D) included significantly higher expressed genes in BMMs of Socs1 −/− MGL tg mice such as Indoleamine 2,3-Dioxygenase 1 (Indo, 11.47-fold) and SelectinL (Sell, 6.69-fold) as well as significantly lower expressed genes in BMMs of Socs1 −/− MGL tg mice such as Src-Like-Adaptor (Sla, 0.29-fold) and Growth Differentiation Factor 3 (Gdf3, 0.26fold). Those genes were confirmed by qPCR to be differentially regulated in Socs1 −/− MGL tg mice ( Figure 6E). Importantly, no canonical IFNγ target genes were differentially regulated in Socs1 −/− MGL tg mice. Of note, we found 38 genes involved in NFκB signaling to be upregulated in BMMs of Socs1 −/− MGL tg mice and 16 downregulated ones. Pathway annotation was performed using the PANTHER classification system among the 86 genes differentially regulated between IFNγ treated BMMs of Socs1 +/− MGL tg and Socs1 −/− MGL tg mice ( Table 1). Instead of IFNγ signaling pathway, we found TLR and TNF signaling pathways to be dysregulated. For TLR signaling pathway, three genes assigned to the signaling pathway were significantly higher expressed and five were significantly lower expressed in BMMs of Socs1 −/− MGL tg mice. For TNF signaling pathway, five genes assigned to the signaling pathway were significantly higher expressed, whereas only one was significantly lower expressed in BMMs of Socs1 −/− MGL tg mice, arguing for a general induction of the pathway. Moreover, TFBS among the differentially regulated genes were analyzed using the overrepresentation analysis tool oPOSSUM ( Table 2). TFBS for CTCF, IRF2, and NFκB were overrepresented among the differentially regulated genes in Socs1 −/− MGL tg mice with 21, 8, and 56 genes, respectively. STAT1 as classical transcription factor for IFNγ signaling was not overrepresented among the differentially regulated genes, strengthening the hypothesis of functional regulation of canonical IFNγ signaling by Socs1ΔNLS in Socs1 −/− MGL tg mice. In summary, whole genome expression analysis revealed a small subset of non-canonical IFNγ-regulated genes that were differentially regulated in Socs1 −/− MGL tg mice with an overrepresentation of TFBS for NFκB.

Socs1 −/− MGL tg Mice spontaneously Develop low-grade inflammation in the lung
Although inhibition of IFNγ signaling by SOCS1ΔNLS was still functional in Socs1 −/− MGL tg mice, gene expression analysis indicated that differences due to lack of SOCS1 in the cell nucleus were present. We closely analyzed Socs1 −/− MGL tg mice for disease symptoms. Histopathological analysis revealed low-grade inflammation in lung and liver in a significant number of Socs1 −/− MGL tg mice (Table S2 in Supplementary Material). However, no differences were observed in serum AST and ALT levels ( Figure  S3 in Supplementary Material). We, therefore, focused on lung histopathology. Infiltrates in lung tissue were observed in 45% of the lung sections of Socs1 −/− MGL tg mice ( Figure 7A) in three different founders. PAS staining was performed to identify mucus producing cells, revealing a higher number of PAS positive cells in the lungs of Socs1 −/− MGL tg mice ( Figure 7B). In addition,      Socs1 −/− MGL tg mice showed 19.4-fold increased serum IgE levels ( Figure 7C). Since SOCS1 has been shown to be important for T helper cell differentiation (31,49,50), we analyzed whether Socs1 −/− MGL tg mice have a T helper cell bias. Therefore, CD4 + T cells were isolated from lung homogenates and expression of transcription factors for T helper cell subsets was examined by qPCR ( Figure 7D). Socs1 −/− MGL tg mice showed a 4.2-fold increase of Gata3 + cells, suggesting a higher number of Th2 cells.
Using an in vitro differentiation assay, naive CD4 + T cells from Socs1 −/− MGL tg mice tend to express more Gata3 as compared to CD4 + T cells from Socs1 +/− MGL tg mice, even under neutral conditions (T0, RPMI only) ( Figure 7E). Increased mRNA expression of IL-4, IL-5, and IL-13 in complete lung homogenates of Socs1 −/− MGL tg mice compared to both Socs +/− MGL tg and Socs +/− mice confirmed this Th2 bias ( Figure 7F). Notably, we found one population of Socs1 −/− MGL tg mice with a strong expression of Gata3 and Th2 type cytokines in lung homogenates, and the second population showing a weaker Th2 bias, consistent with the fact that we could observe infiltrates in the lung only in 45% of the mice. levels upon challenge were induced, yet, showed no difference with respect to expression of the non-nuclear SOCS1. Similar effects were observed in Socs1 −/− MGL tg mice upon intratracheal IL-13 instillation. IL-13-treated mice of all genotypes developed neutrophilia in the lung. Socs1 −/− MGL tg mice additionally showed enhanced influx of eosinophils (however non-significant) and lymphocytes ( Figure S4A in Supplementary Material). In addition, Socs1 −/− MGL tg mice showed increased mRNA expression of IL-4, IL-5, and IL-13, which was even more pronounced upon IL-13 treatment. Socs1 induction in all three genotypes upon IL-13 instillation was equal ( Figure S4B in Supplementary Material).

Disrupted epithelial integrity in
Since Socs1 −/− MGL tg mice showed enhanced expression of IL-25, IL-33, and Tslp in lung homogenates (Figure 9A), we closer analyzed the airway epithelium. Therefore, tracheas were isolated and trachea epithelial cells were differentiated in an air-liquid interface (ALI) using transwells. Increased IL-33 expression could be verified in isolated trachea epithelial cells from Socs1 −/− MGL tg mice. In addition, Ccl26 expression was examined since it is known for the recruitment of eosinophils (52). Indeed, trachea epithelial cells from Socs1 −/− MGL tg mice expressed significantly more Ccl26 as compared to cells from Socs1 +/− MGL tg mice. Interestingly, we found decreased TER in trachea epithelial cells from Socs1 −/− MGL tg mice as compared to cells from Socs1 +/− and Socs1 +/− MGL tg mice. This suggests disrupted epithelial integrity and might explain low-grade inflammation observed in the lungs of Socs1 −/− MGL tg mice.

DiscUssiOn
Suppressor of cytokine signaling 1 is a classical negative feedback regulator of cytoplasmic JAK/STAT signaling (4)(5)(6). However, it has been described that SOCS1 is also localized in the cell nucleus (33,34), yet, the function of SOCS1 in the cell nucleus in vivo remains elusive. To study the role of nuclear SOCS1, we generated transgenic mice using a BAC containing a mutated Socs1 (Socs1ΔNLS) that fails to translocate in the cell nucleus, which is expressed together with eGFP and LuciferaseCBG99 (MGL). Using BACs to create transgenic mice is a commonly used approach (53)(54)(55)(56), which allows manipulating genes embedded within their genetic regulatory environment. We aimed for similar gene regulation of Socs1wt and Socs1ΔNLS and therefore carefully controlled that the locus of BAC-vector integration produced similar transcript amounts of Socs1wt and Socs1ΔNLS (Figure 4). There were no detectable differences regarding expression and regulation of Socs1wt or Socs1ΔNLS mRNA between three different founders (Figure 2). Next, we investigated whether there is a gene dosage effect. We cannot fully exclude that increased localization of SOCS1ΔNLS to the cytoplasm contributes to the observed effects although we did not find any indication that this construct is more effective in inhibiting JAK/STAT signaling. In fact, data from Socs1 +/− and Socs1 +/− MGL tg mice were very similar, thus arguing against any side effect due to increased cytoplasmic localization of SOCS1ΔNLS. In contrast to Socs1 −/− mice, Socs1 +/− mice lacked pathological levels of IFNγ (24) and were found to be phenotypically normal (26). We confirm these data by showing that Socs1 +/− mice had normal survival (Figure 3) indicating that one allele of Socs1 is sufficient for rescue of the severe knockout phenotype. Analyzing expression of IFNγ-dependent genes ( Figure 5) and the lung phenotype (Figures 7 and 9), we found no difference between Socs1 +/− and Socs1 +/− MGL tg mice, arguing against a gene dosage effect and for a localization-specific effect resulting in eosinophilic lung inflammation in Socs1 −/− MGL tg mice. As reported previously, we found that SOCS1 is expressed at low levels and is relatively short-lived (57), but can be induced by IFNγ (49). The NLS of SOCS1 (RRMLGAPLRQRRVR, amino acid 159-173) resembles a bipartite NLS composed of two basic stretches. Lysine as a basic amino acid is important for the ubiquitin proteasome pathway, linking ubiquitin chains onto proteins to mark them for degradation via the proteasome (45,46). Therefore, we addressed the question whether exchanging the NLS with the SOCS3 counterpart might alter protein half-life (Figures 4D,E). However, protein half-life was not altered upon exchanging the NLS corresponding part of SOCS1 with SOCS3 (SOCS1ΔNLS). In general, the results confirm previously described expression patterns of SOCS1, indicating that the transgene has integrated in a region accessible for transcriptional regulation and that using BAC transgenic mice is a valid approach to study the function of SOCS1 in the cell nucleus.
In order to show that Socs1 −/− MGL tg mice indeed lack SOCS1 in the cell nucleus, we tried staining of SOCS1 on lung sections by immunohistochemistry. However, we could not find a specific antibody that was not staining sections from Socs1 −/− mice. Therefore, we decided to apply a functional approach to verify non-nuclear expression of Socs1MGL. SOCS1 has been shown to induce proteasomal degradation of NFκB (36,37) by interaction with p65 in the cell nucleus, thereby limiting induction of a subset of NFκB-dependent genes (38). Lack of nuclear SOCS1 leads to sustained activation of NFκB that could be confirmed using a transcription factor assay specifically for p65 ( Figure  S1 in Supplementary Material). Socs1 −/− MGL tg mice indeed showed sustained IL-12p40 protein levels in CD11c + cells of the lung and spleen (Figure 2F). We did not detect differences regarding TNFα protein levels ( Figure S1 in Supplementary Material). Unlike TNFα that shows fast NFκB recruitment to a constitutively and immediately accessible promoter, IL-12p40 is a gene that needs prolonged binding of NFκB to its promoter (44). Only a small subset of NFκB-dependent genes that is dependent on prolonged transcriptional activation is affected by sustained activation of p65 such as IL-12p40. Taken together, findings in Socs1 −/− MGL tg mice are fully consistent with previously described in vitro data using non-nuclear SOCS1ΔNLS (38), suggesting that Socs1 −/− MGL tg mice lack SOCS1 in the cell nucleus. In addition, we found a substantial number of differentially expressed genes annotated to TLR and TNF signaling, and NFκB-binding sites were overrepresented among those genes (Tables 1 and 2).
Socs1 −/− MGL tg mice, expressing only non-nuclear mutant Socs1 (Socs1ΔNLS), survive the early lethal phenotype of Socs1 −/− mice (Figure 3; Figure S2 in Supplementary Material) that otherwise die within 3 weeks due to excessive immune signaling and multiorgan inflammation (24)(25)(26). The data show that SOCS1ΔNLS was sufficient to rescue lack of wild-type SOCS1. The disease in Socs1 knockout mice mainly depends on hyperresponsiveness to IFNγ as it can be prevented in the neonatal period by the administration of anti-IFNγ antibodies or by using Socs1 −/− IFNγ −/− mice (47,48). Therefore, we hypothesized that canonical IFNγ signaling might still be efficiently regulated by Socs1ΔNLS. IFNγ binds to the IFNγ receptor complex, activates JAK1/2, and subsequently leads to tyrosine phosphorylation of STAT1 (pY-STAT1). pY-STAT1 dimers in turn translocate into the nucleus and activate transcription of "canonical" IFNγ-responsive genes (15,16,58). Socs1 −/− MGL tg mice showed functional regulation of canonical IFNγ signaling, as shown by unaltered pY-STAT1 levels and whole-genome expression analysis in BMMs (Figures 5 and 6). In addition to canonical signaling, a number of studies have shown that pY-STAT1-independent pathways also exist (17)(18)(19). Besides their localization on the plasmamembrane, IFNAR1 and TYK2 have been shown to occur in the nucleus as well (21,22). In addition, it has been shown that STATs translocate into the nucleus in a pY-independent manner, where they activate expression of only a subset of "non-canonical" IFNγ-induced genes (23). Indeed, a subset of non-canonical IFNγ target genes were differentially regulated comparing BMMs of Socs1 −/− MGL tg and Socs1 +/− MGL tg mice. Pathway annotation did not reveal IFNγ signaling to be differentially regulated, confirming that Socs1ΔNLS was still able to regulate cytoplasmic signaling pathways and that canonical IFNγ signaling was not altered in Socs1 −/− MGL tg mice. We observed minor, but non-significant differences in Icam-1 expression upon stimulation with IFNγ comparing BMMs of Socs1 +/− MGL tg and Socs1 −/− MGL tg mice (Figure 5). This is in line with literature showing that upregulation of ICAM-1 by IFNγ is inhibited by SOCS1 (59) with its inhibitory capacity depending on the functional NLS of SOCS1 (34).
We closely analyzed Socs1 −/− MGL tg mice for disease symptoms and found reduced body weight and spontaneous development of low-grade inflammation in the lung (Figure 7). Expression of SOCS1 in the lung has been reported for alveolar macrophages (60), bronchial epithelial cells (61), and eosinophils (62). Mice fully deficient for SOCS1 show extensive hematopoietic infiltration in the lung (26), arguing that SOCS1 is involved in immune regulation in the lung. Increased serum IgE levels in Socs1 −/− MGL tg mice suggest an allergic airway disease. To analyze whether the Th2 bias observed in Socs1 −/− MGL tg mice is of physiological relevance, mice were challenged by either inhaled OVA or IL-13. Both, upon OVA sensitization and challenge as well as IL-13 instillation, Socs1 −/− MGL tg mice showed increased airway eosinophilia (Figure 8; Figure S4 in Supplementary Material). This is in line with previous data showing that serum IgE levels and infiltrating eosinophils were considerably increased in the lungs of OVA-treated Socs1 −/− IFNγ −/− mice (63). So far, it is unclear how the lack of nuclear SOCS1 leads to airway eosinophilia. One hypothesis is that SOCS1 is crucial to maintain epithelial cell barrier function ( Figure S5 in Supplementary Material). Sustained NFκB signaling might lead to an activation of the epithelium. We observed increased expression of the epithelial cell-derived cytokine IL-33 in primary murine trachea epithelial cells of Socs1 −/− MGL tg mice (Figure 9). Since it has been shown previously (64) that IL-33 has an impact on epithelial integrity, higher IL-33 levels in Socs1 −/− MGL tg mice might result in epithelial barrier disruption. Enhanced barrier permeability in turn might facilitate other immune cells such as DCs in initiate host defense mechanisms resulting in inflammation (65). Triggering of pattern recognition receptors on epithelial cells has been reported to release of IL-33 leading to an activation of DCs (66)(67)(68). Recently, it has been shown that IL-33 is constitutively expressed in the cell nucleus in epithelial cells (69) where direct interaction between SOCS1 and IL-33 might be possible. Furthermore, IL-13 has been shown to downregulate junctional components including E-cadherin in bronchiolar epithelial cells leading to disruptive effects on airway epithelial barrier function (70) and explaining why we see stronger eosinophilia upon IL-13 treatment. Indeed, higher SOCS1 expression has been shown to inhibit IL13 induced CCL26 expression in epithelial cells in vitro whereas reduced SOCS1 expression was correlated with enhanced airway eosinophilia (71). Epithelial cells of Socs1 −/− MGL tg mice produce more CCL26 which in turn attracts eosinophils resulting in airway eosinophilia ( Figure S5 in Supplementary Material). The second hypothesis is that hematopoietic cells are the key players involved in nuclear SOCS1 induced airway eosinophilia. Lee et al. showed that serum IgE levels and infiltrating eosinophils were considerably increased in the lungs of OVA-treated Socs1 −/− IFNγ −/− mice (63). They suggest that regulation of SOCS1 mainly affects hematopoietic cells, not epithelial cells. McCormick et al. showed that reduced expression of SOCS1 has been shown to prolong IL-4induced IRS-2 tyrosine phosphorylation and enhanced M2 differentiation (72). IRS-2 also plays a major role in allergic lung inflammation and remodeling (73). In addition, SOCS1 is important in helper T cell differentiation (5,6,49): it is rapidly induced in response to many cytokines, including IFNγ and IL-4 and it is an important negative feedback inhibitor of both signaling pathways. When Socs1 −/− mice are crossed with either an IFNγ −/− or STAT6 −/− mice, survival is prolonged (47,50), indicating that SOCS1 regulates both IFNγ-driven Th1 and IL-4-driven Th2 responses. Supporting this finding, CD4 + T cells from Socs1 −/− mice spontaneously differentiate into Th1 and Th2 cells, thereby producing IFNγ and IL-4, respectively (31,50). It has previously been shown in vitro that SOCS1 is a negative regulator of Th2-dependent pathways, achieved by inhibition of pSTAT6 (74). In line with this, Socs1 −/− MGL tg mice showed enhanced percentage of Gata3 + CD4 + cells and increased expression of IL-4, IL-5, and IL-13, suggesting that nuclear SOCS1 plays a role in T cell differentiation. Even under neutral conditions, CD4 + T cells of Socs1 −/− MGL tg mice tend to differentiate into Gata3 + cells, arguing for a T cell intrinsic effect of nuclear SOCS1. Increased Th2 cytokines in Socs1 −/− MGL tg mice could in turn act on the epithelial cells. Further clarification will require generating bone marrow chimeras to differentiate between contributions of nuclear SOCS1 in cells sensitive for radiation (hematopoietic cells) or radiation-resistant cells (such as epithelial cells).
There have been several publications linking SOCS1 expression with allergic diseases such as asthma (61,71,74,75). Gielen et al. (61) observed that nuclear SOCS1 suppressed rhinovirus induction of interferons, which is discussed to be associated with increased susceptibility to virus exacerbation in severe asthma. Since induction of interferons occurs via pattern recognition receptors that are linked to NFkB signaling, inhibition of NFkB signaling by nuclear SOCS1 might be a possible mechanism for the control of immunity in the lung. Socs1 gene expression was significantly lower expressed in the airways of severe asthmatics compared with mild/moderate asthmatics, and was inversely associated with airway eosinophilia (71,74), suggesting that the absence of SOCS1 leads to Th2 bias. Using Socs1 −/− MGL tg mice, we have shown for the first time, that not only the presence of SOCS1 but also the localization is crucial for effective regulation of Th2 responses. A study assessing functional variants of Socs1 within a population of adult Japanese asthma patients found a significant association between the Socs1 promoter polymorphism (−1478CA < del) and adult asthma. It is suggested that promoter polymorphism leads to increased SOCS1 and inhibition of interferons, leading to higher susceptibility to virus-induced asthma exacerbations (75). Another study showed that expression of nuclear SOCS1 is increased in atopic asthmatic patients (61), which was associated with suppression of rhinovirus-induced interferons. The findings allow the conclusion that SOCS1 in the cell nucleus plays an important role in the regulation of local immunity in the lung that needs to be further investigated.
Taken together, Socs1 −/− MGL tg mice showed functional regulation of canonical IFNγ signaling, but differential regulation of a subset of IFNγ-dependent genes, possibly due to alteration in NFκB signaling pathway. Socs1 −/− MGL tg mice spontaneously developed low-grade airway inflammation and had increased serum IgE levels and Th2 cytokines in the lung. Upon OVA sensitization and OVA aerosol challenge as well as IL-13 instillation, Socs1 −/− MGL tg mice reacted with augmented influx of eosinophils, arguing for an immune regulatory function of nuclear SOCS1 in the lung. We present a valuable tool to study the nuclear function of SOCS1 in vivo that allows investigating local immune regulation in the lung by nuclear SOCS1.