Loss of Hematopoietic Cell–Derived Oncostatin M Worsens Diet-Induced Dysmetabolism in Mice

Obesity is a major risk factor for insulin resistance and type 2 diabetes and represents a worldwide health burden (1). The expansion of adipose tissue (AT) during development of obesity elicits an acute sterile inflammatory response triggered by adipocyte death, local hypoxia, and mechanical stress (2) that stimulates recruitment of macrophages (3). This leads to a chronic low-grade inflammation that propagates to other metabolically relevant tissues, including liver (4).

Oncostatin M (OSM) belongs to the interleukin-6 (IL-6)/Gp130 family, which includes IL-6, IL-11, leukemia inhibitory factor (LIF), cardiotrophin-1, and ciliary neurotrophic factor. In humans, OSM binds the gp130/LIFRβ or the OSMRβ/gp130 complexes. In mice, OSM binds almost exclusively the OSMRβ/gp130 receptor (5). OSM intracellular signaling pathways include JAK2/STAT3, MAP kinases, and PI-3 kinase (6). OSM is produced by neutrophils, eosinophils, and macrophages in mice and also by activated T cells in humans (6). It is associated with inflammatory diseases, such as lung inflammation, multiple sclerosis, and rheumatoid arthritis. In the bone marrow (BM), OSM regulates hematopoiesis, myelopoiesis, and hematopoietic stem cell (HSC) traffic (7,8) by instructing stromal cells of the hematopoietic niche to secrete Cxcl12 and retain HSCs within the BM (9,10).

OSM has been studied in the setting of obesity and type 2 diabetes for its role inhibiting adipogenesis (8,11) and for being associated with inflammation. Some reports show that Osm and Osmrß mRNA are upregulated in the AT of mice fed a high-fat diet (HFD) or Lep^Ob/Ob mice (12). Yet, the causal meaning of such regulation is unclear because Osmrß^−/− mice fed a HFD have a more severe metabolic phenotype than wild-type (Wt) controls (12,13). Macrophages accumulating within the stromal vascular fraction of AT during obesity have been described as the major source of OSM in humans and mice, whereas adipocytes were found to express and release little to no OSM (14). Lep^Ob/Ob mice and HFD-fed mice treated with high-dose OSM displayed improved glucose tolerance, reduced hepatic steatosis, and attenuation of AT inflammation (12,15). In adipocyte-specific Osmrß knockout (Osmr^FKO) mice, HFD induced a mild phenotype characterized by increased AT inflammation (16,17). Conversely, inhibiting OSM with a polyclonal antibody reduced blood glucose in HFD-fed animals (18).

In the liver, OSM is mainly produced by Kuppfer cells (19), but its expression is not modified in Lep^Ob/Ob or HFD mice (12). However, hepatic overexpression of OSMRβ or STAT3 blunted steatosis and insulin resistance in Lep^Ob/Ob mice or after an HFD (20).

Reconciling these contrasting findings is challenging. Inconsistency of evidence is likely due to the diverse cellular sources of OSM, the broad expression of its receptors with different subunit combinations, and the resulting tissue-specific activities. In fact, insight gathered from studies on Osmrß^−/− mice might be biased by OSM signaling through the LIF receptor at higher concentrations (21). Furthermore, mouse OSM requires LIFR? to modulate bone formation (22), which, in turn, can affect systemic glucose metabolism through the release of osteokines (23). Here, we tested the hypothesis that hematopoietic cells are a major source of circulating OSM that contributes to the regulation of systemic metabolism.

For all experiments we used male age-matched animals on a C57BL/6J background randomly assigned to the various experimental groups. For the HFD protocol, male mice at 8 weeks of age were randomly assigned to a standard diet (68% of calories from carbohydrates, 23.5% from proteins, 8.5% from fat) (4RF21; Mucedola, Milan, Italy) or an HFD (60% of calories from fat, 21% from carbohydrates, 19% from proteins) (EF acc. D12492 (I) mod.; Ssniff, Soest, Germany) for 12 weeks. Mice were housed with a maximum of 5 animals per cage with environmental enrichments at 23°C with a 12-h light/dark cycle. Additional details are provided in the Supplementary Materials. All studies were performed at the Veneto Institute of Molecular Medicine and were approved by the Veneto Institute of Molecular Medicine Animal Care and Use Committee (authorization no. 175/2002A), and by the Italian Health Ministry (authorization no. 310/2019-PR).

BM cells were isolated by flushing femurs and tibias with ice-cold MACS Buffer (Miltenyi Biotec GmbH, Gladbach, Germany). AT and livers were processed as described (26). Cells were incubated with antibodies for 15 min at room temperature. Data were acquired with a FACSCanto (BD Biosciences) cytometer or sorted with a FACSCalibur cytometer followed by analysis using FlowJo software (BD Biosciences). The list of antibodies is provided in Supplementary Table 1.

Hematoxylin and eosin staining was performed on 4-μm-thick paraffin-embedded sections by standard methods. Adipocytes count and area were determined using the SMASH plugin. Oil Red O (ORO) staining was performed as previously described (24). Masson’s trichrome staining was performed using a commercially available kit (Bio-Optica, Italy). Peripheral blood triglyceride and cholesterol levels were measured using an Allegro analyzer (Nova Biomedical, Waltham, MA). Plasma free fatty acid levels were measured using a commercially available kit (ab65341; Abcam). AST was measured using the Adaltis Pchem2 Analyzer (Adaltis S.r.l., Italy). Liver lipids were extracted with a hexane-based method (ab211044; Abcam, Cambridge, UK) and quantified using enzymatic kits according to the manufacture’s instruction (ab65336 and ab65359; Abcam). Plasma OSM was quantified with a commercial ELISA assay (R&D Systems, Inc.). Metabolic hormones and adipokines were quantified using a multiplex immunoassay (Mouse Diabetes 8-Plex; Bio-Rad Laboratories).

BM cells were obtained by flushing with sterile ice-cold PBS both femurs and tibia of 3-month-old mice. Macrophages were differentiated and polarized as previously described (10).

Mice received two 5-Gy doses 3 h apart, followed by an intravenous injection of BM cells (4 × 10⁷/each). For BM reconstitution, animals were housed in sterile individually ventilated cages and provided sterile water and food. BM reconstitution was assessed 28 days after transplantation by performing flow cytometry and determining differential white blood cell and red blood cell counts, hemoglobin concentrations, and hematocrit on fresh EDTA-treated blood using the CELL-DYN Emerald analyzer (Abbott). No adverse events were recorded.

RNA was isolated using the Total RNA Purification Micro Kit (Norgen Biotek) and quantified with a NanoDrop 2000 Spectrophotometer (Thermo Fisher Scientific, MA). cDNA was synthesized using the SensiFAST cDNA Synthesis Kit (Bioline, London, UK). qPCR was performed using the SensiFAST SYBR Lo-ROX Kit (Bioline) via QuantStudio 5 Real-Time PCR System (Thermo Fisher Scientific).

Data are reported as the mean ± SE for continuous variables. Normal distribution was checked using the Kolmogorov–Smirnov test. Non-normal continuous variables were log-transformed before analysis. Comparison of data between two or more groups was performed using a two-tailed unpaired Student t test or two-way ANOVA (with post hoc Bonferroni correction), respectively. Statistical significance was set at P < 0.05. GraphPad Prism 9.0 (GraphPad Software, La Jolla, CA) was used for data analysis. The number of biological replicates is reported in the figure legends or shown as individual data points in the figures. Outliers were determined using the robust regression and outlier removal (ROUT) method and removed if the false discovery rate was ≤0.01. Gene expression data were analyzed with gene set enrichment analysis (GSEA) software (Broad Institute).

All data generated or analyzed during this study are included in the published article and its online supplementary files.

To identify the main sources of OSM within metabolically relevant tissues, we used FACS to isolate T lymphocytes (CD45⁺CD3⁺CD11b⁻ cells), neutrophils (CD45⁺CD3⁻Ly6G/C^brightCD11b^bright cells), and macrophages (CD45⁺CD3⁻Ly6G/C^midCD11b^brightF4/80⁺ cells) from the AT and the liver (Fig. 1A). Osm expression was highest in neutrophils both in AT and liver. AT macrophages displayed greater Osm expression than did liver macrophages (Fig. 1B). By interrogating available single-cell RNA–sequencing data sets (27), we confirmed that neutrophils showed the highest Osm expression among mature leukocytes and their hematopoietic progenitors (Fig. 1C). Osmr was expressed at similar levels in the AT of Wt and Osm^−/− mice, but its expression in the liver was lower than in AT, especially in Osm^−/− mice (Fig. 1D). On the other hand, Lifr expression was significantly greater in liver than AT, irrespective of the genotype (Fig. 1E).

Next, we found that plasma OSM was markedly reduced after 12 weeks in Wt mice fed the HFD (Fig. 1F). Because this finding contrasts with some data from the literature (12), we assessed the dynamic changes of OSM concentrations during the development of obesity by a longitudinal analysis of glucose tolerance and plasma OSM in mice fed the HFD. Glucose tolerance rapidly deteriorated within 3–7 days after initiation of the HFD, then recovered partially before worsening again between 8 and 12 weeks (Fig. 1G). Plasma OSM peaked on day 14, when glucose tolerance started to improve, and then remained low during establishment of obesity and overt metabolic dysfunction.

At 8 weeks of age, Osm^−/− mice were 6.9% leaner than Wt mice (Fig. 2A) (body weight 24.33 ± 0.47 g vs. 22.63 ± 0.43 g; P < 0.05). Whole-body lean mass was slightly lower in Osm^−/− than in Wt mice, without gross evidence of growth retardation. Indeed, this difference disappeared over 12 weeks of the standard diet or HFD (Fig. 2C and D), resulting in a comparable weight gain (Fig. 2B and Supplementary Fig. 1A) because the increase in lean mass (Fig. 2C) was counterbalanced by less increase in whole-body fat mass (Fig. 2D). HFD induced a mild fasting hyperglycemia in both Wt and Osm^−/− mice (Fig. 2E), but Osm^−/− mice showed a more severe glucose intolerance (Fig. 2F–H). Glucose tolerance was worsened by Osm deletion already after 1 week of an HFD, supporting an early role for OSM in modulating the metabolic response to the HFD (Supplementary Fig. 1B).

There was no difference in insulin sensitivity, as assessed by the ipITT (Fig. 2I–K) and confirmed by insulin concentrations and HOMA-IR values (Fig. 2L and M). Glucagon and the insulin to glucagon ratio showed similar changes in Osm^−/− and Wt mice after they were fed the HFD (Supplementary Fig. 1C and D). Concentrations of glucagon-like peptide-1 (GLP-1) were strikingly lower in Osm^−/− than in Wt mice fed the standard diet, suggesting impaired function of intestinal L cells (Fig. 2N). As a result, the expected decline in GLP-1 levels after the HFD was observed only in Wt mice (28). Levels of glucose-dependent insulinotropic peptide showed no change due to diet or genotype (Supplementary Fig. 1E). The HFD increased leptin, resistin, and PAI-1, and reduced ghrelin in a similar fashion in Wt and Osm^−/− mice (Supplementary Fig. 1F–I). The RER showed a similar circadian variation in Wt and Osm^−/− mice (Supplementary Fig. 1J and K). As early as 4 weeks after initiation of the HFD, both Wt and Osm^−/− mice lost the circadian profile with the expected RER reduction (Supplementary Fig. 1L and M), indicating a shift of substrate utilization from carbohydrates to fatty acids (29). Superimposable results were obtained then the HFD was stopped (Supplementary Fig. 1N and O). Wt and Osm^−/− mice had a similar energy expenditure profile, showing the expected increase at night (zeitgeber time 12–24) (Supplementary Fig. 1P–T), and during the day after eating the HFD, due to an increase in VO₂ caused by differences in substrate utilization (30).

We also measured glucose uptake in vivo. Osm^−/− mice fed the standard diet had lower 2DG uptake in the liver, in AT, and in fast-twitch tibilias anterior muscle. Such differences persisted with the HFD, particularly for the tibialis anterior muscle, which showed a marked reduction of 2DG uptake in Osm^−/− mice (Fig. 2O).

The HFD resulted in increased median cross-sectional area (CSA) of adipocytes and skewed their distribution in a similar fashion in Osm^−/− and Wt mice (Fig. 3A–C). Crown-like structures (CLSs), which represent infiltrating macrophages around dying adipocytes (31), were increased in both Osm^−/− and Wt mice as a result of the HFD, without significant differences between the two groups (Fig. 3D). The HFD resulted in increased abundance of macrophages and neutrophils (Fig. 3E and F) in Wt and Osm^−/− mice to a similar extent, but Osm^−/− mice had less accumulation of T lymphocytes than the Wt mice (Supplementary Fig. 2A).

Macrophages were the most abundant inflammatory cells in AT and they likely contributed to the majority of OSM gene expression of AT. Consistently, the HFD resulted in reduced Osm expression in macrophages (Fig. 3G), matching reduced Osm expression in AT (Fig. 3H). Deletion of Osm did not impinge upon macrophage phenotype, as assessed by in vitro polarization of BM-derived macrophages (Supplementary Fig. 2B–D). Osmr and Lifr mRNA levels in AT were significantly higher in standard diet Osm^−/− mice compared with Wt mice, whereas the HFD abolished these differences (Supplementary Fig. 2E and F).

In the liver, ORO staining showed expected lipid accumulation induced by the HFD (Fig. 4A) without evidence of fibrosis (Supplementary Fig. 3A). Liver triglyceride content was lower in Osm^−/− mice than in Wt mice under the HFD regimen (Fig. 4B), with no difference in cholesterol content (Supplementary Fig. 3B). The core enrichment of several genes involved in lipid homeostasis was deregulated in HFD Osm^−/− mice compared with Wt mice (Fig. 4C).

Il6 mRNA concentration was higher in Osm^−/− than in Wt mice under both the standard diet and HFD regimens, and Il1b mRNA concentration was increased by HFD in Osm^−/− mice compared with Wt mice (Fig. 4D and E). The HFD resulted in increased abundance of liver macrophages, but not of T lymphocytes, similarly in the two groups (Fig. 4F, and Supplementary Fig. 3C), and neutrophils were increased as a result of the HFD only in Wt mice, not in Osm^−/− (Fig. 4G). Osm expression was not modified by the HFD in inflammatory cells and at tissue level (Supplementary Fig. 3D and E). Osmr was increased as a result of the HFD only in Osm^−/− mice (Supplementary Fig. 3F), whereas Lifr expression was similar among experimental groups (Supplementary Fig. 3G). Serum levels of AST were increased in Osm^−/− mice fed the HFD (Fig. 4H), as a sign of hepatic inflammation and metabolic impairment (32). Furthermore, OSM deletion induced a small increase in serum triglyceride levels independently of the diet, with no difference in the increase of serum cholesterol between Wt and Osm^−/− mice (Supplementary Fig. 3H and I). Plasma free fatty acids showed no difference (Supplementary Fig. 3J).

In summary, deletion of Osm induced no overt change in adiposity due to the HFD, but it worsened liver inflammation and damage. Analysis of single-cell sequencing of the data set (33) showed that Osmr in the liver and AT was expressed almost exclusively by stromal and endothelial cells (Supplementary Fig. 4) suggesting that metabolic effects of OSM might be due to a cross talk between immune cells and the stroma of such organs.

We performed BM transplantation (BMT) experiments to assess whether postnatal deletion of Osm in hematopoietic cells recapitulated the metabolic phenotype of constitutive whole-body Osm^−/− mice (Fig. 5A). After myeloablation, 3-month-old Wt male mice were transplanted with either Wt (BMT^Wt) or Osm^−/− (BMT^OsmKO) BM cells, which yielded a similar degree of reconstitution (Supplementary Fig. 5A and B), as we have previously shown (34). Accordingly, in BMT^OsmKO mice, plasma OSM was reduced to ∼20% the level seen in BMT^Wt mice (3.9 ± 0.4 pg/mL vs. 18.5 ± 4.1 pg/mL; P < 0.05). As reported for Wt mice, plasma OSM declined after HFD in BMT^Wt mice (Fig. 5B), suggesting that blood cells are major contributors to systemic OSM levels. Residual detection of OSM in BMT^OsmKO mice might be due to long-lived tissue-resident cells of monocyte/macrophage lineage that persist after radiative myeloablation (35–37). The HFD had a similar impact on body weight and whole-body composition in BMT^Wt and BMT^OsmKO mice (Fig. 5C and Supplementary Fig. 5C and D). According to these data, we ruled out that hematopoietic-derived OSM is a major regulator of fat accumulation and weight gain induced by the HFD.

Blood glucose levels were slightly higher in BMT^OsmKO mice fed the standard diet, though the difference did not reach statistical significance despite adequate sample size (Fig. 5D) (145.2 ± 4.99 mg/dL vs. 116.6 ± 6.04 mg/dL; n = 8; P = 0.08). Glucose intolerance was induced by the HFD in both BMT^OsmKO and BMT^Wt mice (Fig. 5E–G), but the analysis of area under the curve (AUC) from ipGTTs showed a significant effect of both diet and genotype, suggesting that a worse baseline glucose intolerance of BMT^OsmKO mice was further exacerbated by the HFD (Fig. 5G). BMT^OsmKO mice at the end of the HFD regimen also had slightly worse insulin sensitivity than the BMT^Wt animals (Fig. 5H–J), despite similar insulin concentrations and HOMA-IR (Fig. 5K and L). Changes of RER, glucose uptake, and energy expenditure due to the HFD were similar in the two groups of BMT animals (Supplementary Fig. 6A–K).

The HFD resulted in increased adipocyte median CSAs (Fig. 6A and B), right-skewed their distribution (Fig. 6C) and increased the number of CLSs per adipocyte similarly in BMT^Wt and BMT^OsmKO mice (Fig. 6D). Osm expression in AT was reduced by the HFD in BMT^Wt mice, with a small residual expression in BMT^OsmKO mice (Fig. 6E), matching plasma OSM concentration (Fig. 5B). Leptin, resistin, and PAI-1 levels were increased as a result of the HFD, ghrelin showed no variations, GLP-1 concentrations were low in all groups, and glucose-dependent insulinotropic peptide levels were slightly increased in BMT^OsmKO (Supplementary Fig. 7A–F). ORO staining and triglyceride quantification in the liver showed similar effects of the HFD in both BMT^Wt and BMT^OsmKO mice (Fig. 6F and G), without signs of fibrosis or changes in total cholesterol level (Supplementary Fig. 7G and H). The core enrichment for genes involved in lipid homeostasis displayed no striking differences between BMT^Wt and BMT^OsmKO mice (Fig. 6H). Of note, hematopoietic-restricted OSM deletion recapitulated the increase in AST observed in whole-body Osm^−/− mice after the HFD regimen (Fig. 6I). Elevations in plasma cholesterol and triglyceride levels were equally observed in BMT^Wt and BMT^OsmKO mice fed the HFD (Supplementary Fig. 7I and J) without changes in FFAs (Supplementary Fig. 7K). Expression of Osm, Osmr, and Lifr showed no difference due to either diet or the BMT in the AT or the liver (Supplementary Fig. 7L–P).

Altogether, these data show that hematopoietic cells were major contributors to Osm expression and OSM concentrations in plasma, liver, and AT. Moreover, postnatal, hematopoietic-restricted Osm deletion recapitulated glucose intolerance and liver damage due to whole-body Osm deletion.

We finally transplanted Wt and Osm^−/− BM cells into lethally irradiated Osm^−/− recipients (KO^BMTWt and KO^BMTKO, respectively) (Fig. 7A). Successful engraftment was confirmed by peripheral blood cell count and flow cytometry analysis of the BM (Supplementary Fig. 8A–E). The HFD induced a modest weight gain in both KO^BMTWt and KO^BMTKO mice (Fig. 7B) and a significant increase in whole-body fat mass (Supplementary Fig. 9A and B) but no evidence of excess myelopoiesis (Supplementary Fig. 9C and D). Plasma OSM concentrations were expectedly higher in KO^BMTWt mice than in KO^BMTKO mice but remained fivefold lower than in Wt or BMT^Wt mice and unaffected by the HFD (Fig. 7C). Therefore, rescue of OSM levels was far from complete. The HFD induced glucose intolerance and insulin resistance similarly in KO^BMTWt and KO^BMTKO mice, with no difference in AUC based on donor cell type (Fig. 7D–I). These data indicate that, despite full hematopoietic reconstitution, transplantation of Wt cells into Osm^−/− mice did not replenish full OSM secretion and did not rescue metabolic dysfunction of Osm^−/− mice.

OSM has been claimed to be responsible for both proinflammatory and anti-inflammatory actions (38,39). Therefore, it is not surprising that its role in obesity and type 2 diabetes is disputed. Here, we show that OSM is mainly produced by macrophages and neutrophils in the blood and tissues and undergoes modulation during a HFD regimen, with an early surge followed by a chronic reduction. This dynamic modulation can be interpreted as an adaptive response affecting the metabolic status because whole-body Osm deletion worsened glucose tolerance from the first weeks of the HFD regimen. Notably, deletion of Osm in adult hematopoietic cells was sufficient to induce glucose intolerance, mimicking constitutive whole-body Osm^−/− deletion. Data obtained from hematopoietic-restricted Osm^−/− mice confirmed that blood cells are the major source of OSM in plasma, liver, and AT, suggesting these mice accurately show the net metabolic effects of inhibiting OSM postnatally.

Despite successful engraftment, the transplantation of Wt BM cells into Osm^−/− mice (KO^BMTWt) yielded low plasma OSM and, consequently, failed to improve metabolic derangement induced by the HFD. Tissue-resident macrophages can persist after irradiation and BMT and are scarcely renewed by circulating cells (40). They are known to release OSM to recruit OSM-producing neutrophils in a paracrine loop needed for tissue regeneration and homeostasis (41). Therefore, we hypothesize that in KO^BMTWt mice, Osm^−/− tissue-resident macrophages failed to elicit the paracrine activity required for instructing Wt cells to release OSM, thus explaining the persistently low plasma OSM concentrations. Conversely, by transplanting Osm^−/− BM into Wt mice, even if tissue-resident Wt macrophages still produced OSM, recruited immune cells would be devoid of OSM, thus recapitulating the metabolic phenotype of Osm^−/− mice. Recruitment of neutrophils in the liver is driven by AT macrophages (42), which also emerge as a major source of OSM that could be instrumental in establishing a meta-inflammatory cross talk between AT and the liver. Indeed, the number of neutrophils was reduced in the liver of Osm^−/− mice after the HFD regimen, where we observed the most striking effect of Osm deletion.

Reduction of Osm expression in AT after the HFD regimen was paralleled by reduction in OSM plasma levels. This finding is consistent with AT-infiltrating hematopoietic cells being the major source of AT Osm. This finding apparently contrasts with prior reports showing increased OSM in AT of different models of obesity but, at the same time, favorable metabolic effects of high-dose OSM administration (12). Although this counterintuitive result may reflect a form of OSM resistance that can be overcome with high OSM doses, we have shown that treatment with OSM activates myelopoiesis (34), suggesting caution in pursuing OSM as a treatment for metabolic diseases. In humans, levels of OSM in the subcutaneous AT have been reported to correlate inversely with glucose disposal, indirectly suggesting that elevated OSM in AT promotes metabolic dysfunction (14). This association, however, does not necessarily imply that OSM impairs metabolism because it can be also explained as a compensatory response to the worsening metabolic status. Furthermore, the metabolic significance of subcutaneous AT is different from that of visceral AT (43), and additional studies in humans are warranted to dissect this point.

In type 1 diabetes, we discovered a pathway whereby hyperglycemia increased Osm expression that, in turn, sustained myelopoiesis and impaired HSC mobilization (34). Hence, OSM inhibition may provide short-term benefit for stem cell-mediated tissue homeostasis. In type 2 diabetes and obesity, OSM is likely subjected to different layers of regulation uncoupled from hyperglycemia, including microbial triggers (44). Our new data in HFD mice suggest that OSM inhibition may have detrimental metabolic effects that can persist long term.

A role for OSM in hepatic lipid metabolism has been proposed, but the net effect is difficult to predict because OSM downregulates both CPT-1a and FAS, thus modulating both fatty acid oxidation and synthesis (45). Possibly due to their lesser fat mass, Osm^−/− mice fed the HFD had reduced steatosis (12,13,19,20), while showing increased inflammatory gene expression and signs of liver damage. These mixed results might derive from a multiplicity of pathways activated by OSM, as well as from long-term adaptations to constitutive whole-body Osm deletion. In fact, hepatic lipid accumulation induced by the HFD was not modified in BMT^OsmKO mice. OSM is required for development of the fetal liver, which is also the major site of antenatal hematopoiesis (46). Reduced steatosis in Osm^−/− liver might also result from altered intestinal lipid absorption, given the role of OSM in modulating the gut barrier function (39).

The literature is inconsistent on the effects of the HFD on OSMRβ, which we found was expressed at low levels in AT and liver. Data from Osmrß^−/− mice support our findings (12), but the metabolic effects of OSM in vivo may be indirect, because OSMRβ in the liver and in AT is expressed mainly by endothelial and stromal cells (16). Indeed, the metabolic phenotype of OSMR^FKO mice is rather mild (16,17). It is also possible that adipocyte OSMRβ acts as a decoy receptor to buffer OSM concentration at the tissue level (47).

As a limitation, we acknowledge we are still missing the exact mechanism whereby OSM regulates the metabolic response to the HFD. Although insulin sensitivity was unaffected, glucose tolerance was consistently worsened by constitutive whole-body and adult, hematopoietic-restricted Osm deletion. Along with the reduction in muscle and liver glucose uptake, these data suggest that OSM may promote glucose disposal. OSM stimulates GLP-1 expression by ileal L cells, an effect supposed to potentiate insulin secretion (15). In agreement with this, Osm^−/− mice had remarkably lower levels of GLP-1, which could contribute to glucose intolerance after the HFD. Indeed, although mice lacking Glp1r have normal oral glucose tolerance, mice with selective deletion of β-cell Glp1r have impaired intraperitoneal glucose tolerance (48). Therefore, the interplay between OSM and the incretin effect is worth exploring in future studies.

The plethora of pathways influenced by OSM, its dynamic regulation over time, and the tissue specificity of OSM’s effects are evident from our results. Dissecting these pathways will require targeted and conditional knock-out models to ablate OSM or its receptors from several relevant cell populations, including intestinal cells and β-cells.

In conclusion, we found that OSM undergoes a dynamic regulation during development of obesity-associated metabolic dysfunction, and we propose that the coordinated release of OSM by blood cells is part of a physiological response to an HFD, contributing to metabolic homeostasis.

Acknowledgments. The authors thank Kenneth A. Dyar (Institute for Diabetes and Cancer, Helmholtz Diabetes Center, Helmholtz Zentrum Munich) for copyediting and proofreading the manuscript.

Funding. This study was supported by a grant from GlaxoSmithKline, a grant from the University of Padova to M.A. (Finanziato dall’Unione Europea – NextGenerationEU, STARS@UNIPD MalTraDiates), and grants to G.P.F. (PNRR M4C2-Investimento 1.4-CN00000041 Finanziato dall’Unione Europea – NextGenerationEU; PRIN 2017-93XZ5A [Metabolic therapy of immuno-inflammation]).

Duality of Interest. M.A., S.C., and G.P.F. are the inventors of a patent (EP3197480), held by the University of Padova, on the use of pharmacologic OSM inhibition to allow stem cell mobilization. No other potential conflicts of interest relevant to this article were reported.

The external sponsor had no role in the study design, conduction, data analysis, interpretation, and decision to publish.

Author Contributions. M.A. designed and performed experiments, researched and analyzed data, provided funding and wrote the manuscript. S.C., A.R., L.M., F.I.A., C.B., and G.Z. designed and performed experiments, researched and analyzed data, and revised the manuscript. A.R. designed experiments, contributed to discussion, and revised the manuscript. G.P.F. researched and analyzed data, provided supervision and funding, and wrote the manuscript. M.A. and G.P.F. are the guarantors of this work and, as such, had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.

Prior Presentation. Part of this work was presented at the 58th Annual Meeting of the European Association for the Study of Diabetes, Stockholm, Sweden, 19–23 September 2022.