Myeloid-Cell Protein Tyrosine Phosphatase-1B Deficiency in Mice Protects Against High-Fat Diet and Lipopolysaccharide-Induced Inflammation, Hyperinsulinemia, and Endotoxemia Through an IL-10 STAT3-Dependent Mechanism

Protein tyrosine phosphatase-1B (PTP1B) is a nonreceptor tyrosine phosphatase, identified as an attractive drug target for conditions associated with metabolic syndrome as well as potential anticancer therapeutic due to its diverse involvement in regulation of various cell signaling cascades. Evidence to support the notion that PTP1B inhibition may be beneficial in states of overnutrition and insulin resistance was demonstrated in mice with a global- (1,2) as well as tissue-specific PTP1B deletion (3–6).

Insulin resistance in adipose, muscle, and liver is exacerbated in obese states, due to the underlying presence of chronic low-grade inflammation and macrophage infiltration into these tissues. White adipose tissue (WAT) from obese subjects contains greater numbers of infiltrating proinflammatory macrophages in comparison with lean counterparts, and these cells secrete cytokines such as tumor necrosis factor-α (TNF-α), interleukin (IL)-1β, and IL-6, which can impair systemic insulin sensitivity (7–11). This process is enhanced further by the release of chemokines, such as monocyte chemoattractant protein-1, which recruits more activated tissue-macrophages into WAT (12). Saturated fatty acids are able to directly induce the expression of these proinflammatory cytokines via activation of the nuclear factor (NF)-κB pathway, and the receptor primarily responsible for potentiating this effect has been identified as the macrophage toll-like receptor (TLR)-4 (13). Mice with myeloid-TLR-4 deletion are protected against high-fat diet (HFD)-induced inflammation, adipose macrophage infiltration, and insulin resistance (14).

It has been postulated that PTP1B may act as a negative regulator of TLR-4 signaling in macrophages since in vitro PTP1B knockdown in the RAW 264.7 macrophage cell line resulted in elevated production of TNF-α, IL-6, and interferon (IFN)-β following challenge with a variety of TLR ligands (15). PTP1B overexpression in the same cell line caused a concomitant decrease in proinflammatory cytokine production in response to lipopolysaccharide (LPS) and palmitate challenge (15,16). Similarly, splenic macrophages isolated from global PTP1B^−/− mice were found to be highly sensitive to LPS-induced inducible nitric oxide synthase (iNOS) expression and nitric oxide production, in addition to elevated IFN-γ–induced phosphorylation of STAT1 (17). In the same study, in vivo LPS-challenge increased sensitivity to endotoxic-shock in PTP1B^−/− mice, associated with increased systemic production of IL-12 and IFN-γ. However, since these studies were performed in mice with global PTP1B deletion, and PTP1B regulation of cell function/signaling varies in a cell-specific manner (3–6), we set out to definitively establish the in vivo role of macrophage PTP1B by assessing the regulation of inflammation and whole-body metabolism in LysM PTP1B knockout mice.

All animal procedures were performed under a project license approved by the U.K. Home Office under the Animals (Scientific Procedures) Act 1986 (PPL60/3951). PTP1B^fl/fl and mice expressing Cre recombinase (Cre) under the control of the LysM promoter have been described previously (3). To generate myeloid PTP1B^−/− mice, PTP1B^fl/fl mice were crossed with LysM Cre mice (18) and back crossed nine times to pure C57BL/6J mice. DNA extraction and genotyping were performed as described previously (3). Age-matched male mice were studied and compared with PTP1B^fl/fl and LysM Cre control littermates. Mice were group housed and maintained at 22–24°C on 12-h light/dark cycle with free access to food/water. At weaning (21 days), mice were placed on standard 3.4% fat chow-pellet diet (Rat and Mouse Breeder and Grower, Special Diets Services, DBM Food Hygiene Supplies, Broxburn, U.K.) or HFD (Adjusted Calories Diet, 55% fat, Harlan Teklad) for 29 weeks and weight recorded weekly. The approximate fatty-acid profile of Adjusted Calories Diet (percentage total fat) was 28% saturated, 30% trans, 28% monounsaturated (cis), and 14% polyunsaturated (cis). For endotoxemia experiments, 50-week-old chow-fed and high-fat (HF)-fed mice were fasted for 2 h and injected intraperitoneally with LPS (1 and 0.5 mg/kg, respectively; Merck). Mice were observed for signs of sepsis, including reduced mobility, fur ruffling, and conjunctivitis (19). At 24 or 3 h post-LPS injection (stated in figure legends), mice were killed by cervical dislocation and tissues/blood were harvested (3).

Tail-blood glucose from fasted (5 h) mice was measured using glucometers (Accu-Chek, Burgess Hill, U.K.) (6). Serum TNF-α, IL-6, monocyte chemoattractant protein-1, insulin, leptin, plasminogen activator inhibitor-1, and resistin were measured using adipokine multiplex kit (Millipore, U.K.), and serum IL-10 levels were determined by ELISA (R&D Systems, Minneapolis, MN). Alanine aminotransferase (ALT; BioVision) and serum glucose (Thermo Scientific) levels were assayed per the manufacturer’s instructions. Glucose tolerance tests (GTTs) and insulin tolerance tests (ITTs) were performed as previously described (2,6). For hematocrit-level determination, whole blood was collected in heparinized microhematocrit capillary tubes and centrifuged for 3 min. The percentage hematocrit value was determined by measuring the length of the red blood cell layer and calculating as a percentage of the total height of the column sample.

Single-cell suspensions were made from spleens and lymph nodes by pressing through a 70 μm strainer (BD Falcon). After red blood cell lysis (Sigma), cells were labeled with anti-L3T4 (CD4) microbeads (Milteny Biotec) and sorted using LS-Columns (Milteny Biotec).

Bone marrow–derived macrophages (BMDMs) were obtained by flushing out femurs and tibiae with sterile PBS (Lonza), as described previously (6). Mature macrophages were seeded in six-well tissue-culture plates (1 × 10⁶ cells/well) and serum starved for 16 h prior to stimulation with 100 ng/ml LPS (InvivoGen), 10 ng/ml IL-10 (Peprotech), or 20 ng/ml IL-6 (Peprotech).

Stable knockdown of PTP1B in RAW 264.7 cells (murine-macrophage cell line) was performed using short-hairpin RNA constructs as previously described (20).

Cells and tissues were lysed in radioimmunoprecipitation assay buffer containing fresh sodium orthovanadate and protease inhibitors (21). Proteins were separated by 4–12% SDS-PAGE and transferred to nitrocellulose membranes. Immunoblots were performed using antibodies from cell signaling (New England Biolabs, Hitchin, U.K.; unless stated otherwise) against phosphorylated extracellular signal–related kinase (ERK)1/2, mitogen-activated protein kinase (MAPK) T202/Y204, phosphorylated IκB kinase (IKK) α/β S176/S180, IκBα, p-c-jun S63, phosphorylated P38 T180/Y182, phosphorylated STAT1 Y701, phosphorylated STAT3 Y705, STAT3, BCL-2, iNOS, SHP1, phosphorylated Akt/PKBS473, phosphorylated S6 ribosomal protein S235/236, phosphorylated GSK 3α/β, phosphorylated IR Y1162/1163 (Invitrogen), SHP2 (Santa Cruz), ERK2 (Santa Cruz), T-cell protein tyrosine phosphatase (R&D Systems), phosphorylated JNK/SAPK T183/Y185 (R&D Systems), β-actin (Sigma), and PTP1B (Millipore). Immunoblots were visualized using enhanced chemiluminescence and quantified by densitometry scanning using Bio1D-software (PeqLab, Fareham, U.K.).

Cells/tissues were homogenized in TriFast reagent (Peqlab, Sarisbury Green, U.K.) (20). cDNA synthesis was carried out from 1 µg of RNA using Tetro cDNA-synthesis kit (Bioline). Quantitative real-time PCR was performed using Light-Cycler 480 (Roche), and gene expression of iNOS, TNF-α, IL-6, MCP1, IL-1β, IL-1α, IL-10, and BCL2 was determined relative to the most stable reference gene (YWhaz, NoNo, or β-actin), which was identified using a Web-based reference gene assessment tool (http://www.leonxie.com/referencegene.php?type=reference). Primer sequences are provided in the Supplementary Data online.

TNF‐α, IL-6, and IL-10 concentrations in BMDM supernatants were quantified by ELISA (R&D Systems) or multiplex ELISA (Millipore), according to the manufacturers’ instructions.

Nitric oxide production by BMDMs was quantified by determining the concentration of nitrite present in supernatants using Griess reaction (22).

Ethanol-fixed, paraffin-embedded adipose tissue sections were stained as previously described (23) using rat antimouse F4/80 (AbD Serotec MCA497RT), CD68 (Abcam ab31630), and iNOS (Cell Signaling).

Liver and spleen tissue was fixed in formaldehyde, embedded in paraffin, sectioned, and stained with hematoxylin and eosin.

Single-cell suspensions were made by pressing spleens through cell strainers, and cell suspensions were centrifuged at 900g for 5 min. Red blood cells were lysed for 1 min with red blood cell lysis buffer (Sigma) and washed by centrifugation at 900g for 5 min in PBS without Ca²⁺ and Mg²⁺, 5% fetal bovine serum, and 5 mmol/L EDTA. We blocked 1 × 10⁶ cells per label for 30 min in the dark with rat anti-mouse CD16/CD32 (BD, 0.5 µg/reaction) and then labeled them with optimized quantities of antibodies (PerCP-Cy5.5 [BD 550954], CD8 V450 [BD 560469], CD3 AF488 [BD 557666], CD45 PE [BD 553089], CD11b AF488 [BD 557672], CD11c V450 [BD 560521], Gr1 APC-Cy7 [BD 557661], MHC-II PE [BD 557000], Ly6C PerCP-Cy5.5 [BD 560602], Ly6G APC [BD 560595], F4/80-AF647 [AbD Serotech]). Data were acquired on BD LSR II and analyzed using DiVa software.

Data are expressed as mean ± SEM. Statistical analyses were performed using correlation analyses, one-way ANOVA with Tukey’s multiple comparison post-tests, two-way ANOVA with Bonferroni multiple comparisons post-tests, and two-tailed Student’s t tests, as appropriate, using GraphPad Prism 5 statistical software.

PTP1B deletion was achieved in BMDMs isolated from LysM PTP1B mice, without affecting PTP1B levels in WAT, liver, muscle (Fig. 1A) or CD4⁺ T-cells (Fig. 1B). A deletion efficiency of ∼75% was determined in LysM PTP1B BMDMs (n = 4) compared with controls (n = 5) (Fig. 1C and D). There were no body weight differences (Fig. 1E), and glucose (GTT) and insulin (ITT) tolerance were unaltered between LysM PTP1B and control mice on chow diet (Fig. 1F and G, respectively). Fed and fasted metabolic parameters were unchanged in the absence of myeloid PTP1B (Table 1). To establish whether macrophage PTP1B plays a role in TLR-4 signaling, LysM PTP1B mice with confirmed PTP1B deletion (Fig. 1C) and littermate controls were injected with low-dose LPS (1 mg/kg), and their response to endotoxin was monitored. Twenty-four hours post-LPS, control mice exhibited obvious signs of endotoxemia, including closed eyes, shivering, reduced mobility, and ruffled fur (Fig. 1H and Supplementary Videos 1 and 2), whereas LysM PTP1B littermates were completely protected (Fig. 1I and Supplementary Videos 3 and 4). LysM PTP1B mice had decreased serum TNF-α and IL-6 levels compared with controls, although this did not reach significance (TNF-α, 0.04 ± 0.005 vs. 0.1 ± 0.03 ng/ml, respectively [P = 0.18; two-tailed t test]; IL-6, 3.7 ± 0.9 vs. 14.3 ± 5.3 ng/ml, respectively [P = 0.06; two-tailed t test]) (Table 2). At 3 h post-LPS, there was also a trend toward increased IL-10 levels in LysM PTP1B mice (0.89 ± 0.12 vs. 0.67 ± 0.03 ng/ml [P = 0.12; two-tailed t test]), although this was not observed at 24 h post-LPS (Table 2). Furthermore, levels of hepatic STAT1 and STAT3 phosphorylation were significantly lower in LysM PTP1B mice, indicative of decreased LPS-induced hepatic inflammation (Fig. 1J and K).

To assess whether the anti-inflammatory phenotype observed in LPS-challenged LysM PTP1B mice is due specifically to the absence of macrophage PTP1B, isolated BMDMs were stimulated with 100 ng/ml LPS for varying durations up to 24 h in vitro.

Elevations in IL-10 mRNA expression in LysM PTP1B BMDMs following LPS stimulation were observed at both 4 and 24 h (Fig. 2A), which was associated with increased STAT3 mRNA expression 4 h post-LPS treatment (Fig. 2B). TNF-α transcript levels were concomitantly downregulated following 4 h of LPS treatment in LysM PTP1B cells (Fig. 2C). Supplementary Table 1 displays additional measures of proinflammatory cytokines and markers of M1 and M2 macrophages.

In keeping with previously published data showing that isolated macrophages from global PTP1B^−/− mice are highly sensitive to LPS-induced iNOS expression (17), we also found increased levels of LPS-stimulated iNOS mRNA transcript in the absence of macrophage PTP1B (Fig. 2D). Supernatants harvested from LysM PTP1B BMDMs challenged with LPS for 6 h contained increased IL-10 (Fig. 2E) and decreased TNF-α levels (Fig. 2F). Supernatant nitrite (Fig. 2G) and iNOS protein levels (Fig. 2H and I) were higher in LPS-stimulated LysM PTP1B macrophages compared with respective control cells.

We found increased levels of phosphorylated STAT1 α and β isoforms in LysM PTP1B macrophages treated with LPS for 24 h (Fig. 3A and C), which is in agreement with previous reports and is a likely explanation for the elevated iNOS and nitrite levels produced by these cells (17). The level of phosphorylated STAT3 was also significantly increased in LysM PTP1B cells following 24 h of treatment with 100 ng/ml LPS (Fig. 3B and C). The ability of endogenous IL-10 to directly induce STAT3 activation is a likely cause for the heightened levels of phosphorylated STAT3 observed in PTP1B-deficient macrophages since the levels of IL-10 detected in respective supernatants were also elevated (24).

To assess this, LysM PTP1B and control macrophages were stimulated with IL-10 for varying durations, and levels of STAT3 phosphorylation were determined (Fig. 3D). IL-6, which also activates STAT3, was used to allow for a direct comparison (Fig. 3E,). An elevation in IL-10–induced phosphorylated STAT3 was observed following 60- and 120-min stimulation (Fig. 3F) in the absence of macrophage PTP1B, which was not the case for IL-6–induced STAT3 phosphorylation (data not shown). There was a negative correlation between PTP1B protein levels present in each cell batch (Fig. 3G) and the quantity of phosphorylated STAT3, following 120-min IL-10 treatment (Fig. 3H), suggesting a dose-dependent effect of STAT3 dephosphorylation by PTP1B. RAW 264.7 cells with PTP1B stable knockdown (Fig. 3I) also showed increased LPS-induced phosphorylated STAT1-Y701 (Fig. 3J), phosphorylated STAT3-Y705, and iNOS (Fig. 3K) and increased IL-10–induced phosphorylated STAT3-Y705 (Fig. 3L).

Previous studies implicated macrophage PTP1B in negative regulation of MAPK and NF-κB signaling cascades, initiated by various TLR ligands (15). In our study, macrophage PTP1B deficiency in vitro did not affect phosphorylation of c-Jun (component of MAPK signaling) (Supplementary Fig. 1) or phosphorylation of p65 (data not shown) and associated degradation of IκBα (components of NF-κB signaling) (Supplementary Fig. 1), following LPS stimulation at various time points. More detailed analysis of these pathways revealed no alterations in the levels of phosphorylated ERK1/2 (Supplementary Fig. 2), p38, JNK1/2, or IKK (Supplementary Fig. 3).

Since LysM PTP1B mice were protected from an acute inflammatory challenge, we tested in vivo effects of myeloid PTP1B deficiency under states of chronic low-grade inflammation, caused by long-term (29 weeks) HFD feeding (23).

Despite no alterations in body weight (Fig. 4A), GTTs revealed improved ability to clear exogenous glucose in LysM PTP1B HFD-fed mice (Fig. 4B). The area under the curve (AUC) also confirmed that LysM PTP1B mice were more glucose tolerant than control mice (Fig. 4C). LysM PTP1B mice were also more insulin tolerant, with decreased blood glucose levels at 15, 90, and 120 min post-insulin injection (Fig. 4D) and decreased AUC during ITT (Fig. 4E).

Further, acute proinflammatory LPS challenge led to hyperinsulinemia in HFD-fed control mice (75% increase from basal). Strikingly, LysM PTP1B mice were completely protected against LPS-induced hyperinsulinemia (no increase from basal) (Fig. 4F) and post-LPS insulin levels were significantly lower in LysM PTP1B compared with control mice. Post-LPS blood glucose levels were also lower in LysM PTP1B mice (Fig. 4G). Additional measures of metabolic parameters and circulating cytokines are shown in Tables 1 and 2, respectively.

HFD-fed LysM PTP1B and control mice were injected with either saline or insulin (10 mU/g body weight), and components of the insulin signaling pathway were investigated in muscle, liver, and WAT. There were no significant differences in insulin-stimulated insulin receptor phosphorylation between the two genotypes in muscle (Fig. 4H), liver, or WAT (Supplementary Figs. 4 and 5, respectively).

F4/80 and CD68 immunohistochemical staining of WAT from HFD-fed LysM PTP1B mice revealed a trend toward decreased expression of these markers when compared with controls (Fig. 4I). To confirm this quantitatively, we performed quantitative RT-PCR analysis of WAT, which showed lower transcript levels of F4/80 and TNF-α and a trend for decreased expression of CD68 (P = 0.09; two-tailed t test) in LysM PTP1B mice (Fig. 4J), indicative of decreased macrophage infiltration into WAT and associated inflammation. HFD-fed LysM PTP1B mice also had lower levels of hepatic lipid accumulation (Fig. 4K and L), which was associated with decreased serum ALT levels in these mice (Fig. 4M,), indicative of improved liver function. Furthermore, LysM PTP1B mice exhibited increased expression of the antiapoptotic gene BCL2 (Fig. 4N).

Basal circulating levels of the anti-inflammatory cytokine IL-10 were elevated in HFD-fed LysM PTP1B compared with control mice (Fig. 5A). Acute LPS challenge led to an induction of IL-10 in both control and LysM PTP1B mice, which led to a further 120% elevation in IL-10 levels in LysM PTP1B (Fig. 5B). Furthermore, we found a negative correlation between serum IL-10 and ALT levels (Fig. 5C), as well as serum IL-10 and insulin levels (Fig. 5D) in HFD mice following LPS injection, suggesting that elevated circulating IL-10 levels are strongly associated with the anti-inflammatory role of myeloid PTP1B deficiency.

Despite the strong anti-inflammatory phenotype of LysM PTP1B mice, these mice (but not control mice) developed splenomegaly on HFD (Fig. 5E), with a mean spleen weight of 0.48 ± 0.05 g (n = 14) compared with 0.14 ± 0.02 g (n = 13; P ≤ 0.0001; two-tailed t test), which was also significant when corrected for body weight (Fig. 5F). To assess if these mice were anemic, hematocrit testing was performed on whole blood, and no differences were found between LysM PTP1B (36.6 ± 1.1%; n = 9) and control mice (38.9 ± 1.0%; n = 3). Hematoxylin and eosin staining of spleen sections revealed unremarkable histology (Fig. 5G). Spleen size was found to strongly correlate with circulating IL-10 levels under basal and LPS-stimulated states (Fig. 5H and I). Splenic IL-10 mRNA levels correlated with circulating IL-10 levels following LPS-treatment, suggesting that the elevation in systemic IL-10 seen in HFD-fed LysM PTP1B is, at least in part, spleen derived (Fig. 5J).

Flow cytometry analysis of splenocytes established that the percentage of cells expressing the granulocyte/monocyte marker Gr1 and the monocyte marker Ly6C were significantly increased in LysM PTP1B cells compared with controls, and a trend for increased expression of Ly6G was noted (P = 0.09) (Fig. 5K). There were no alterations in the percentage of cells expressing CD3, CD4, CD8, B220, CD11c, CD11b, F4/80, and MHCII; however, when normalized against total cell number, a significant increase in absolute cell numbers expressing all markers was established for LysM PTP1B mice compared with controls (Supplementary Table 1).

Spleen lysates from HFD-fed LysM PTP1B mice revealed an increase in the phosphorylation of STAT3α and STAT3β isoforms compared with control lysates (Fig. 5L and M). Using antibodies against total STAT3, a significant increase in the alternatively spliced STAT3β isoform was noted in LysM PTP1B mice (Fig. 5L and M). Interestingly, the splenic level of the antiapoptotic protein BCL2, which is regulated at the gene level by STAT3, was also increased in these mice, which could contribute toward the splenomegaly phenotype observed (Fig. 5L and M).

Previous reports have implicated macrophage PTP1B in the negative regulation of TLR-4-mediated inflammatory signaling; however, these studies have been restricted to in vitro cell line analysis or characterization of responses in PTP1B global knockout mice (15–17,25). To definitively address the role of macrophage PTP1B in inflammatory signaling, we have generated mice with myeloid-specific PTP1B deletion and interrogated their whole-body physiology and signaling in isolated macrophages. These mice exhibited an LPS-tolerant phenotype in vivo, which was mirrored by the attenuation of proinflammatory cytokine expression in LPS-treated BMDMs lacking PTP1B in vitro. A likely explanation for the downregulation of TNF-α observed in vitro was the concomitant increase in transcription and secretion of the anti-inflammatory cytokine, IL-10. This was further reinforced by heightened levels of phosphorylated STAT3, which is known to mediate IL-10–driven repression of inflammatory targets (26). A similar increase in phosphorylated STAT3 was observed when macrophages lacking PTP1B were challenged with IL-10 in vitro. The IL-10R engage Janus kinase (JAK)1 and Tyk2 following receptor activation, leading to the phosphorylation and activation of STAT3 (27,28), and since Tyk2 is a known substrate of PTP1B (29), this is a plausible reason for the hyperphosphorylation of STAT3 observed in the absence of PTP1B. Furthermore, activated STAT3 has been shown to control the expression of the IL-10 promoter, leading to a positive feedback mechanism (30). In summary, the altered cytokine profile displayed by these cells and their increased sensitivity to IL-10 may be responsible, at least in part, for the state of LPS tolerance observed in LysM PTP1B mice 24 h after receiving low-dose endotoxin in vivo.

Our in vitro analysis also revealed increased STAT1 phosphorylation, leading to increased iNOS and nitrite production in LPS-treated LysM PTP1B BMDMs, which is in agreement with previous findings of LPS-treated spleen-macrophages from global knockout mice (17). Although heightened levels of nitric oxide have been partly implicated as a cause for increased endotoxin sensitivity in PTP1B^−/− mice, no such phenotype was observed in LysM PTP1B mice. These findings do, however, add to a large body of evidence that links PTP1B to the regulation of JAK STAT activation (3,29,31–34).

The anti-inflammatory phenotype observed in endotoxin-challenged LysM PTP1B mice was similarly replicated in our long-term HFD-feeding study, which is known to chronically increase plasma LPS concentration and has thus been termed metabolic endotoxemia (23). Most remarkably, in vivo LysM PTP1B mice exhibited increased basal and LPS-induced levels of circulating IL-10 compared with control animals, which we found to negatively correlate with insulin and ALT levels. The insulin-sensitizing effect mediated by IL-10 has been widely documented before. Low circulating levels of IL-10 have been associated with obese and insulin-resistant states, as have polymorphisms and haplotypes of the IL-10 promoter, and IL-10 has been shown to mediate insulin-sensitizing effects in adipose, liver, and skeletal muscle by suppressing the deleterious effects of TNF-α and IL-6 on insulin signaling (35–40). Furthermore, a shift toward an alternatively activated, anti-inflammatory M2 macrophage phenotype in adipose tissue has been associated with increased expression of IL-10 (38).

The source of elevated IL-10 in HFD-fed LysM PTP1B mice suggests it to be splenic in origin, due to the close correlations noted between circulating IL-10, spleen size, and splenic IL-10 mRNA expression. Further analysis of the spleens from these mice revealed an increased expression of the antiapoptotic protein BCL2, which is regulated by STAT3 and may partly explain the splenomegaly phenotype (41) and an elevation in the alternatively spliced STAT3β isoform. Isoform-specific knockout models have demonstrated that STAT3β plays a crucial role in inhibiting the acute inflammatory responses, whereby mice lacking STAT3β showed increased susceptibility and impaired recovery following in vivo challenge with LPS (42).

Increased phosphorylation of splenic STAT3 found in HFD-fed LysM PTP1B mice is in agreement with our in vitro BMDM studies, as well as a recent study confirming STAT3 as a substrate of myeloid PTP1B (34), which demonstrated that global PTP1B^−/− mice are protected against experimental colitis because of a STAT3/JAK2-mediated expansion of myeloid-derived suppressor cells (MDSCs) (34). Our data also provide evidence that this mechanism is, at least in part, applicable to the anti-inflammatory phenotype observed in HFD-fed LysM PTP1B mice. States of chronic inflammation, which could be extended to encompass metabolic endotoxemia, are known to expand immunosuppressive MDSC populations (CD11b+ Gr+ cells) via elevations in factors such as proinflammatory cytokines, prostanoids, and growth factors that lead to persistent STAT3 activation (43). In the absence of myeloid PTP1B, we observe increased phosphorylated STAT3 in the spleen, elevated numbers of splenic cells expressing myeloid markers, and marked splenomegaly, which would be indicative of MDSC expansion (43). In addition to the ability of MDSCs to inhibit T-cell activation (44,45) and NK-cell tumor cytotoxicity (46), these cells are known to interact with macrophages by secreting elevated levels of IL-10, which promote macrophage polarization toward the M2 macrophage phenotype (47). It is therefore plausible to postulate that in the absence of myeloid PTP1B, chronic HFD feeding leads to a STAT3-dependent expansion of IL-10–secreting splenic MDSCs and possibly M2 macrophages, which alleviate hepatic and adipose inflammation and enhance insulin sensitivity. Figure 6 displays a schematic depicting the postulated mechanism responsible for the beneficial effects observed for HFD-fed LysM PTP1B mice.

In summary, this study provides evidence that LysM PTP1B mice exhibit improved glucose tolerance and suppressed inflammatory responses in HFD-fed and endotoxemic mouse models. This is contrary to previous reports that have assigned an anti-inflammatory function to macrophage PTP1B and will therefore help allay concerns relating to the application of PTP1B inhibitors in the clinical setting and open up new avenues for the use of PTP1B inhibitors as anti-inflammatory agents.

Acknowledgments. The authors thank Benjamin Neel (University of Toronto), Barbara Kahn (Harvard Medical School), and Kendra Bence (University of Pennsylvania) for providing PTP1B floxed mice and for their continuous mentoring support. The authors also thank Simon Arthur and Vicky McGuire (University of Dundee) for helpful discussions.

Funding. This work was supported by a British Heart Foundation project grant to M.D. (PG/11/8/28703) and a European Foundation for the Study of Diabetes/Lilly Diabetes Programme grant to N.M. and M.D.

Duality of Interest. No potential conflicts of interest relevant to this article were reported.

Author Contributions. L.G. conceived, designed, and performed the experiments and wrote the manuscript. K.D.S. designed and performed the experiments. A.C., E.K.L., C.O., A.A., J.W., and C.M.-G. performed the experiments. J.V.F. and H.M.W. suggested experiments and reviewed the manuscript. N.M. and M.D. conceived and designed the experiments and wrote the manuscript. M.D. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Prior Presentation. Parts of this study were presented in abstract form at the Diabetes U.K. Professional Conference, Manchester, U.K., 13–15 March 2013.