Deficiency of the Hemoglobin-Haptoglobin Receptor, CD163, Worsens Insulin Sensitivity in Obese Male Mice

Maintaining proper iron balance is critical for metabolic health. Imbalanced tissue iron levels are linked to metabolic diseases such as nonalcoholic fatty liver disease and type 2 diabetes (1). Iron is essential for many cellular functions, including ATP generation, DNA synthesis and repair, and oxygen transport. Therefore, various cellular and tissue-specific regulatory pathways are critical for tissue and systemic iron homeostasis (2). Dietary iron uptake is regulated by duodenal enterocytes, which engage with transferrin (Tf) for cellular uptake through the Tf receptor (Tfr) to support erythropoiesis (3). Iron is retained systemically through the phagocytosis of senescent erythrocytes by splenic red pulp macrophages leading to recycling of Tf-bound iron into the circulation (2). Tissue-resident macrophages are specialized with resolving inflammatory mediators, such as cytokines and chemokines in tissues, caused by factors such as obesity and infection (4). Tissue-resident macrophages also facilitate storage or transfer of iron, thus contributing to the maintenance of local tissue iron homeostasis (5,6).

Macrophages serve homeostatic needs of cells in response to environmental cues and are often binned into two classifications: proinflammatory (M1-like) and anti-inflammatory (M2-like). This generic classification is governed by in vitro polarization states, and macrophage populations present in vivo span a variety of phenotypes that can simultaneously express M1 and M2 markers (7,8). Macrophage phenotypes also underlie their iron-handling capacity. M2-like tissue-resident macrophages are noted for their intrinsic capacity to provide bioavailable iron within the tissue microenvironment (5,9). Macrophages import Tf-bound iron by receptor-mediated endocytosis via Tfr proteins (TFR1/2) and iron-containing molecules such as hemoglobin-haptoglobin (Hb-Hp) and heme-hemopexin via the scavenger receptors CD163 and CD91, respectively. Importantly, CD163 serves as a cell-surface marker for tissue-resident macrophages (10). Therefore, the anti-inflammatory function of CD163 is defined by hemoglobin-bound iron uptake and catabolism, thereby limiting metabolic tissues from iron overload. Hence, the ability of tissue-resident macrophages to effectively respond to environmental stimuli while also fulfilling their anti-inflammatory functions is crucial for tissue homeostasis.

Metabolic complications in obesity are often associated with inflammatory activation of macrophages in metabolically significant tissues such as adipose tissue (11,12). Macrophages are the most abundant immune cell type in adipose tissue and can expand from ∼10% to ∼40% of all immune cells in the lean versus obese states (13,14). The majority of macrophages recruited to obese adipose tissue are proinflammatory and release cytokines such as IL-1β, TNF-α, and IL-6 (13–15). These cytokines promote adipose tissue insulin resistance and increase lipolysis (16). During inflammation, macrophage iron-handling capacity is also forfeited (9,17). In obesity, the presence of iron-handling CD163⁺ tissue-resident macrophages is diminished in favor of macrophages specialized in managing the lipid-rich environment [e.g., recruitment of lipid-associated macrophages (18)]. This shift in macrophage phenotypes may explain why adipocytes are susceptible to iron overload in obesity (17,19). Adipocyte iron overload is speculated to initiate the temporal response to insulin resistance by inducing systemic dyslipidemia and reducing adiponectin release (19–22). Therefore, interventions aimed to counter the relative loss of CD163⁺ tissue-resident macrophages therapeutically may be advantageous to limit excessive free hemoglobin, iron overload, and inflammatory stress from obesity.

Here, we report loss of CD163 increased adipocyte iron in inguinal white adipose tissue (iWAT) in male mice fed a high-fat diet (HFD). Furthermore, CD163 deficiency hindered the polarization of macrophages toward an M2-like phenotype and reduced hemoglobin scavenging in vitro. Overall, these studies suggest that CD163⁺ tissue-resident macrophages are critical in preventing severe metabolic dysfunction from diet-induced obesity, preserving an anti-inflammatory iron-handling phenotype in M2-like macrophages, and limiting adipocyte iron overload.

Experimental protocols were approved by the Vanderbilt University Institutional Animal Care and Use Committee. Male and female CD163^−/− mice on a C57BL/6N background were purchased from Jackson Laboratory (Cd163^{tm1.1(KOMP)Vlcg}) and backcrossed to 99% purity with the C57BL/6J background. CD163^+/− mice were bred, and CD163^−/− or CD163^+/+ (wild-type [WT] littermates) were randomized to either a low-fat diet (LFD; 10% fat, 3.85 kcal/g; Research Diets) or HFD (60% fat, 5.24 kcal/g; Research Diets) for 16 weeks. All mice were fed ad libitum in a 12-h light/dark cycle at 22°C.

Fat-free mass (FFM) and fat mass (FM) were measured by nuclear magnetic resonance (Bruker Minispec) in conscious mice at 16 weeks after initiating their assigned diet.

The day after body composition assessments, intraperitoneal glucose tolerance tests (ipGTTs) were performed (2 g/kg FFM, 20% dextrose). Mice were fasted for 4 h, and blood glucose was collected using a Contour Next EZ blood glucose monitoring system at 0, 15, 30, 45, 60, 90, and 120 min after glucose injection. The glucose area under the curve was calculated using the trapezoidal rule.

Hyperinsulinemic-euglycemic clamp studies were performed by the Vanderbilt Mouse Metabolic Phenotyping Center. Procedures are detailed in Supplementary Methods and at https://vmmpc.org.

The isolation of epididymal white adipose tissue (eWAT) and iWAT adipocytes from the stromal vascular fraction (SVF) was completed by digesting the tissue in 2 mg/mL collagenase type IV (Worthington), as detailed in the Supplementary Methods and previously (23,24).

Adipose tissue SVF cells were analyzed and sorted on a FACS Aria III cell sorter (BD Biosciences) at the Vanderbilt University Medical Center Flow Cytometry Shared Resource Core. A complete antibody list is provided in Supplementary Table 1. Cells from eWAT and iWAT were blocked for 30 min at 4°C in anti-mouse CD16/CD32 Fc Block, then stained for 30 min at 4°C with following antibodies: BV510 anti-CD45, PerCP/Cyanine5.5 anti-CD11b, APC/Cyanine7 F4/80. DAPI was used for viability. Stained cells were washed twice with PBS plus 1% FBS and isolated by FACS into 15-mL, metal-free conical tubes. Samples were suspended in 200 μL of 70% nitric acid (HNO₃) for inductively coupled plasma mass spectrometry (ICP-MS) analysis.

Male and female mice fed chow diets were euthanized between 8 and 16 weeks of age. Bone marrow–derived macrophages (BMDMs) were extracted from the femurs and tibia, using a 22-gauge needle to flush marrow using cold RPMI medium. After suspension, cells were treated with red blood cell lysis buffer and neutralized with PBS plus 1% FBS. Cells were plated in BMDM medium for 7 days (∼5 × 10⁶ cells/flask). After 7 days of differentiation, M1-like BMDMs were treated with 10 ng/mL lipopolysaccharide (LPS) plus 100 ng/mL IFN-γ for 24 h, and M2-like BMDMs were treated with 10 ng/mL IL-4 plus 10 ng/mL IL-13 for 72 h.

Tissues or cells were lysed, then DNase treated. RNA was purified using a Qiagen RNeasy Mini Kit, then reverse transcribed into cDNA (iScript RT; Bio-Rad). RT-qPCR was performed using FAM-conjugated TaqMan Gene Expression Assay primers (Thermo Fisher) and iQ Supermix (Bio-Rad). Samples were normalized to GAPDH or β-actin and quantified by the comparative cycle threshold method. Complete lists of primers are in Supplementary Table 2.

RNA from unpolarized (M0) and M1- or M2-like polarized BMDMs was purified with the RNeasy Plus Mini Kit. Library preparation included Poly-A enrichment using polyA magnetic beads (New England Biolabs; catalog no. E776), and cDNA library prep using the NEBNext Ultra Kit (New England Biolabs; catalog no. E7760). Paired-end sequencing was completed using the Illumina NovaSeq 6000 sequencing platform targeting an average of 50 mol/L reads per sample.

Differential expression for individual gene read counts were analyzed with DESeq2 1.36 (25), using R, version 4.3.1, and explained previously (26). Differential expression analyses between WT and CD163^−/− BMDMs were corrected for multiple comparisons using Benjamini-Hochberg methods (27).

Tissue iron was visualized at ×10 (Leica DMi8) using the Perls Prussian Blue perfusion and staining method described previously (17,28). After cardiac perfusion, ∼15 mL of Prussian blue stain (4% paraformaldehyde, 1% potassium ferrocyanide, 1% HCl) was perfused and fixed for 1 h. Tissues were incubated in Perls staining solution for 12 h at 4°C before paraffin embedding.

Tissues, adipocytes, and SVF populations were homogenized and digested by 70% HNO₃ in metal-free conical tubes (VWR). Total iron was quantified by ICP-MS (Agilent); the procedure is detailed in Supplementary Methods.

M0, M1-, and M2-like BMDMs were plated in 96-well Seahorse assay plates (50,000–75,000/well) in Seahorse assay medium, and mitochondrial respiration was analyzed on a Seahorse XFe96 Extracellular Flux Analyzer (Agilent). For mitochondria stress tests, BMDMs were sequentially treated with 10 mmol/L glucose, 1 mmol/L oligomycin, 1 mmol/L carbonyl cyanide 4-(trifluoromethoxy)phenylhydrazone, and 0.5 mmol/L antimycin A plus rotenone. The oxygen consumption rate (OCR) was normalized to cellular protein content.

Student t tests were performed for between-group measurements. For studies that included main effects for treatment or time (e.g., ipGTT, clamps), a two-way ANOVA was completed with Tukey multiple comparisons correction. RNA sequencing (RNAseq) and figure generation were completed using R, version 4.3.1, and the remainder of analyses were performed with GraphPad Prism 10.1.2. Data are expressed as mean ± SEM, and significance was set to α = 0.05.

RNAseq data and the source code used for this publication are accessible through the GitHub data repository (https://github.com/MichaelSchleh/CD163-and-immunometabolic-health).

CD163 is expressed exclusively on human monocytes and macrophages (Fig. 1A) and on tissue-resident macrophages in mice (23). To examine the role of CD163 on metabolic homeostasis, a 16-week LFD and HFD feeding paradigm was implemented (Fig. 1B). Deletion of CD163 was confirmed in both eWAT and iWAT adipose tissue macrophages (ATMs; Fig. 1C). Body mass was tightly linked to diet (HFD vs. LFD), and body weight was similar between CD163^−/− mice and WT littermates (Fig. 1D). After 16 weeks on their diet, male CD163^−/− mice fed the LFD had slightly lower FFM (P = 0.04) and greater FM (P = 0.02), whereas CD163^−/− mice fed the HFD had lower FFM compared with WT littermates (P = 0.04; Fig. 1E). No difference in daily energy intake was observed between CD163^−/− mice and WT littermates in either diet group (Fig. 1F). In female mice, body weight did not differ between genotypes (Supplementary Fig. 1A).

Although there was no difference in glucose tolerance by genotype in LFD or HFD male mice at 4, 8, or 12 weeks (Supplementary Fig. 2), glucose tolerance was worsened in CD163^−/− male mice compared with WT controls after 16 weeks of HFD feeding (Fig. 1G and H). Female CD163^−/− mice and WT controls had similar glucose tolerance at all time points (Supplementary Fig. 1B–E). At 16 weeks, blood glucose levels were similar between the genotypes in fed animals; however, fasting plasma glucose levels were higher in lean and obese CD163^−/− male mice (Fig. 1I). Fasting insulin levels were significantly higher in CD163^−/− male mice fed the LFD and trended greater in the CD163^−/− mice fed the HFD (Fig. 1J). The HOMA for insulin resistance is commonly used to assess insulin sensitivity in clinical settings (29) and was greater in the CD163^−/− mice fed the LFD and HFD (Fig. 1K). Plasma nonesterified fatty acid (NEFA) and triacylglycerol did not differ between genotypes in the fasted or fed state, apart from fed-state triacylglycerol levels, which were lower in the CD163^−/− mice (Fig. 1L and M). Due to the hemoglobin-scavenging properties of CD163, the CD163^−/− mice in both the LFD and HFD groups had higher plasma iron levels (Fig. 1N), and total hemoglobin in circulation was only higher in CD163^−/− mice fed the HFD (Fig. 1O). Hematocrit did not differ between groups (Fig. 1P). These data collectively show that loss of CD163 attenuates glucose tolerance and elevates circulating iron and hemoglobin in obesity.

To assess whether the impaired glucose tolerance in the CD163^−/− mice was attributed to peripheral insulin resistance, hyperinsulinemic-euglycemic clamps were conducted (Supplementary Fig. 3). Data from male mice are shown in Fig. 2, and for female mice are shown in Supplementary Fig. 4. Body mass did not differ between CD163^−/− male mice and WT littermates 1 week after catheterization (Fig. 2A). As designed, insulin levels increased from the basal to clamp states (endogenous glucose production plus exogenous insulin infusion), with no significant difference by genotype (Fig. 2B). Arterial glucose levels were clamped at 150 mg/dL by regulating the variable glucose infusion rate (GIR; Fig. 2C); in HFD-fed mice, the GIR required to maintain euglycemia was nearly twofold lower in CD163^−/− mice compared with WT littermates (Fig. 2D). Consistent with lower GIR necessary to maintain euglycemia, the glucose rate of disappearance (Rd), calculated by the sum of GIR and glucose rate of appearance (Ra), was attenuated in CD163^−/− mice only in the HFD group (Fig. 2E). Glucose Ra did not differ across groups in the basal or clamp states (Fig. 2F). Similarly, glucose Ra suppression (index of hepatic insulin sensitivity) was not different, although it trended lower in CD163^−/− mice (Fig. 2G). CD163^−/− mice had higher levels of NEFA throughout the clamp in the LFD and HFD groups (Fig. 2H and I), suggesting resistance to the suppressive effects of insulin. The NEFA percent suppression from basal to clamp conditions did not differ by genotype in either dietary condition (Fig. 2J).

Differences in glucose disposal are largely attributed to impaired glucose uptake from peripheral tissues. The tissue metabolic index (Rg) estimates the rate of glucose uptake in selected tissues, and the CD163^−/− mice fed the LFD had lower eWAT and brown adipose tissue (BAT) Rg values (Fig. 2K). The CD163^−/− mice fed the HFD had a significantly decreased Rg in the gastrocnemius and vastus lateralis, and the value trended lower in the soleus (Fig. 2K). Female CD163^−/− mice had higher insulin concentrations during the hyperinsulinemic clamp only in the HFD group (Supplementary Fig. 4B) but displayed no differences in GIR or glucose Rd, NEFA response to insulin, or Rg with either diet (Supplementary Fig. 4D–J). These findings indicate that obese CD163^−/− male mice exhibit insulin resistance, elevations in circulating NEFA levels in the presence of high insulin concentration, and decreased skeletal muscle Rg.

Total iron content (⁵⁶Fe) was measured by ICP-MS in eWAT, iWAT, BAT, liver, gastrocnemius, heart, pancreas, thymus, and spleen. CD163^−/− mice in the HFD group had significantly greater amounts of iWAT and BAT tissue iron (Fig. 3A). Furthermore, adipocytes were isolated from the eWAT and iWAT (Fig. 3B). Iron content of eWAT adipocytes was not affected by CD163 deficiency or diet, whereas adipocytes from iWAT of CD163^−/− mice fed the HFD had greater iron concentrations (Fig. 3C). To assess the effects of CD163 on SVF immune-cell iron handling, FACS was implemented to sort nonhematopoietic cells (CD45⁻), nonmyeloid lymphocytes (CD45⁺ CD11B⁻), and macrophages (CD45⁺ CD11B⁺ F4/80⁺; Fig. 3D). Iron content in CD45⁻ cells was greater in iWAT than eWAT, and significantly more iron was detected in iWAT CD45⁺ CD11B⁻ in CD163^−/− mice (Fig. 3E). In contrast, iron in CD45⁺ CD11B⁺ F4/80⁺ ATMs from iWAT of CD163^−/− mice was lower than in controls (Fig. 3E). Additionally, ATMs were isolated from eWAT and iWAT by magnetic sorting (F4/80⁺ fraction), and expression of iron handling genes was analyzed by qPCR. Tfr1 and Cisd1 gene expression was lower in eWAT and iWAT ATMs of CD163^−/− mice, whereas ferroportin (Slc40a1) and Ncoa4 were decreased only in iWAT ATMs (Fig. 3F). Prussian blue staining did not reveal major visual differences in iron deposition or structure from the eWAT or iWAT, whereas BAT from CD163^−/− mice appeared more unilocular compared with that of WT littermates (Fig. 3G). Altogether, these findings illustrate iron handling is altered in iWAT of CD163^−/− mice, as evidenced by decreased ATM iron content and a consequential increase in adipocyte iron.

To differentiate the impact of CD163 on the transcriptomic profile between classically activated (M1-like) and alternatively activated (M2-like) macrophages, BMDMs were isolated and polarized for RNAseq (Fig. 4A). Principal component analysis was used for dimensional reduction of M1-like and M2-like BMDMs, and the segregation by genotypes (CD163^−/− vs. WT) was found only in M2-like BMDMs (Fig. 4B). Differential expression analysis in M1-like BMDMs detected 380 upregulated genes (false discovery rate [FDR] <0.05; fold change >2.0) and 170 significantly downregulated genes in CD163^−/− compared with WT BMDMs, whereas M2-like BMDMs presented 351 significantly upregulated genes and 222 downregulated genes in CD163^−/− BMDMs (Fig. 4C). No genes significantly differed between genotypes in M0 BMDMs (Supplementary Fig. 5).

Gene ontology (GO) comparing differentially expressed genes between groups identified several redundant pathways significantly altered in M1- and M2-like BMDMs between genotypes. M1-like CD163^−/− BMDMs diverged in processes related to rRNA processing and translation, whereas M2-like CD163^−/− BMDMs upregulated several immune and inflammatory processes (Fig. 4D). Several M2-like gene signatures (Arg1, Egr2, Acadm, and Cd200r) were downregulated in CD163^−/− BMDMs upon M2 polarization (Fig. 4E). Additionally, 42 genes from the innate immune response biological process (GO:0045087) were differentially expressed between genotypes in M2-like BMDMs, all of which were upregulated in CD163^−/− mice compared with WT littermates (Fig. 4F). Several hemoglobin and iron-handling genes from the iron transport biological process (GO:0006826) were also differentially expressed in CD163^−/− BMDMs upon M2 polarization (Fig. 4F).

OCR and extracellular acidification rate (ECAR) were measured in M0, M1-, and M2-like BMDMs by extracellular flux analysis. Loss of CD163 decreased OCRs with significantly decreased basal, ATP-linked, and maximal respiration in M0 BMDMs (Fig. 5A). M1-like BMDMs from CD163^−/− BMDMs had lower OCRs, respiratory measures, and proton leak (Fig. 5B). M2-like CD163^−/− BMDMs had lower OCR values, including basal, ATP-linked, and maximal respiration (Fig. 5C). Maximal ECAR during the mitochondrial stress test was greater in M0 and M1-like CD163^−/− BMDMs (Fig. 5D).

Hemoglobin uptake was quantified in M0, M1-, and M2-like BMDMs treated with 25 mg/dL fluorescence-conjugated hemoglobin for 24 h (Fig. 6A and Supplementary Fig. 6). Flow cytometry was used to measure median fluorescence intensity (MFI) as a proxy for hemoglobin uptake in hemoglobin-treated BMDMs for 2, 6, and 24 h (Fig. 6B). Over the 24-h treatment period, M2-like BMDMs presented >twofold greater hemoglobin uptake compared with M0 and M1-like BMDMs (Fig. 6C). Furthermore, in M2-like BMDMs, CD163 deficiency attenuated hemoglobin uptake after 24 h (Fig. 6C).

Total hemoglobin-positive area was also measured by fluorescence imaging (Fig. 6D). M0 BMDMs had a greater hemoglobin-positive area at 6 h and returned to near basal levels after 24 h, whereas M1-like BMDMs had increased hemoglobin-positive area after 24 h, and this was greater in WT littermates compared with CD163^−/− mice (Fig. 6E). M2-like BMDMs significantly increased hemoglobin-positive area at 6 h and throughout the 24 h treatment only in WT BMDMs, whereas CD163^−/− BMDMs presented area fractions similar to that of M0 and M1-like BMDMs (Fig. 6E). These data show that loss of CD163 hinders M2-like macrophage potential to buffer extracellular free hemoglobin.

Iron is a critical element for numerous biological processes and is exquisitely controlled by tissue- and cellular-dependent processes (2). However, excess iron is harmful to cells and is associated with β-cell dysfunction, nonalcoholic fatty liver disease, and type 2 diabetes risk (1). Obesity recruits proinflammatory macrophages to adipose tissue (13–15), and macrophage polarization is linked with iron handling (9,30). Our group identified a subset of tissue-resident macrophages, MFe^hi cells, with high intrinsic capacity for iron cycling (17,31). These cells present greater CD163 and tissue-resident macrophage gene expression, suggesting the anti-inflammatory influence of this receptor is partly explained by free hemoglobin scavenging. However, the capacity for iron handling in MFe^hi cells becomes diminished upon obesity (17). Here, we show that CD163 is critical to support an anti-inflammatory phenotype in M2-like BMDMs, suggesting tissue-resident macrophages in vivo may become compromised upon CD163 deficiency. These findings support CD163 as a critical receptor for limiting severe insulin resistance upon obesity, partially through its anti-inflammatory contribution to free hemoglobin scavenging.

Hemolysis is a physiological process by which hemoglobin is released into the circulation from senescent erythrocytes. Free hemoglobin has a strong affinity for haptoglobin, primarily produced by hepatocytes, protecting tissues from heme-induced oxidative damage (32,33). However, the antioxidant potential for haptoglobin requires CD163 on tissue-resident macrophages to scavenge Hb-Hp complexes. Additionally, heme catabolism by heme-oxygenase I results in the production of free iron, CO, and biliverdin; and CO production has anti-inflammatory effects resulting from IL-10 release (34). Others have shown that excess heme iron initiates the release of proinflammatory cytokines from macrophages (35–37). Therefore, heme catabolism (and lower levels of circulating heme) appears vital to sustain anti-inflammatory roles of CD163-expressing macrophages. Here, we confirmed M2-like macrophages have a profound role for hemoglobin uptake compared with M0 and M1-like BMDMs, and loss of CD163 nearly abolishes the hemoglobin-regulatory components signature of M2-like macrophages. These data support a critical role for CD163 in buffering iron-containing hemoglobin and limiting iron spillover into key metabolic tissues.

Macrophage metabolic states are linked to their polarization (38). Specifically, M1-like macrophages primarily rely on glycolysis, whereas M2-like macrophages are oxidative (39). Previous reports have shown both mitochondrial iron overload and depletion diminish oxidative metabolism (19). We observed loss of CD163-attenuated mitochondrial oxidation across the BMDM polarization spectra, and M2-like macrophages had the greatest absolute OCR impairment. It is unknown whether OCR impairments in complexes I–III are due to perturbations in Fe-S cluster regulation. Cisd1 (MitoNEET), an inhibitor of mitochondrial iron uptake, was decreased in CD163^−/− eWAT and iWAT ATMs. Recent reports have shown macrophage-specific MitoNEET overexpression lowers mitochondrial matrix iron content and promotes an anti-inflammatory ATM profile, protecting from insulin resistance (19). We speculate the lower heme- and Tf-derived iron uptake in CD163^−/− BMDMs lessens cytosolic iron accumulation but increases mitochondrial iron accumulation, as shown by decreased MitoNEET expression. This effect may explain why M2-like BMDMs upregulate inflammatory gene signatures, because lipid uptake is used as a proinflammatory mediator in M1-like macrophages (e.g., eicosanoids), as opposed to oxidation in M2-like macrophages (40). Whether macrophage mitochondrial metabolism dictates the fate of polarization, or vice versa, and the influence of iron in this process, will be intriguing to uncover in future work.

It is established that macrophage inflammatory activation is related to iron-handling capacity (9). Recent reports have shown depot-specific differences in inflammatory cell profile, where iWAT presents ∼70% higher proportion of M2-like ATMs compared with eWAT (41). We found loss of CD163 increased iron content in BAT and iWAT of mice fed HFD. Increased iWAT adipocyte iron was accompanied by lower iWAT ATM iron, suggesting CD163 deficiency impedes the iron-buffering capacity of iWAT ATMs, resulting in iron spillover into adipocytes and other immune cells. Additionally, iWAT iron overload suggests that metabolic dysfunction resulting from an HFD may be temporally initiated in this “subcutaneous-like” adipose depot and follows with systemic consequences to glucose tolerance. In humans, the majority of lipolysis-derived FAs originate from subcutaneous adipose tissue compared with visceral sources (42). Loss of CD163 also attenuated Tfr1 expression in M2-like BMDMs and eWAT/iWAT ATMs, which is consistent with previous reports showing M2-like macrophages present higher levels of Tfr1 and Cd163 in an effort to buffer extracellular iron and/or support tissue repair processes (9). Therefore, the proinflammatory landscape of CD163-deficient ATMs may also impair iron homeostatic functions by limiting Tf-bound iron uptake, in addition to the observed differences in Hb-Hp receptor scavenging.

The cross-talk linking macrophage and adipocyte iron regulation with systemic insulin sensitivity is still unconfirmed. Here, iWAT adipocytes were burdened with greater iron, likely consequential to diminished ATM iron clearance. Furthermore, NEFA content was elevated during the clamp period of the hyperinsulinemic-euglycemic clamp, which is a consequence of greater postprandial lipolysis and lower FA clearance. In humans, the subcutaneous adipose tissue corresponds to ∼70% of total lipolysis compared with visceral sources (42), and greater lipolysis rates upon hyperinsulinemia are linked with poor insulin-mediated glucose disposal (43). Because skeletal muscle glucose uptake was impaired during the hyperinsulinemic clamp, and skeletal muscle is the primary source for insulin-mediated glucose disposal (44), we propose adipocyte iron overload accelerates postprandial lipolysis in accordance with skeletal muscle FA uptake. Furthermore, we show a proinflammatory phenotype in M2-like BMDMs may also translate to greater proinflammatory cytokine release by ATMs—a hallmark of obesity-induced adipose tissue inflammation (12). These cytokines may also sustain a cycle of proinflammatory immune cell infiltration and local insulin resistance in adipose tissue (45), resulting in excess lipolysis and FA intermediate production known to impair insulin action.

We report that CD163 deficiency results in immunometabolic defects including reduced mitochondrial metabolism, loss of anti-inflammatory phenotype in M2-like BMDMs, and impaired hemoglobin scavenging. To this end, systemic insulin resistance was present, in addition to iWAT adipocyte iron overload. Future studies will delineate whether loss of CD163 induces insulin resistance as a consequence of iron overload in metabolic tissues or by loss of anti-inflammatory function; both roles are key signatures for CD163 in tissue-resident macrophage.

Acknowledgments. The authors acknowledge Marnie Gruen for assistance with animal husbandry, and Gabriel Ferguson for assistance with experiments. The authors thank Stuart Landstreet and Dr. Ciara Shaver for assistance with fluorescence hemoglobin preparation and guidance. The authors acknowledge the following: Vanderbilt University (VU) and Vanderbilt University Medical Center (VUMC) core facilities: VUMC Hormone Assay & Analytical Services Core (NIH DK020593), VU Metabolic Mouse Phenotyping Center (NIH DK059637; www.vmmpc.org), Translational Pathology Shared Resource (NCI/NIH Cancer Center Support Grant 5P30 CA68485-19), VUMC Flow Cytometry Shared Resource Core (DK058404), and Vanderbilt Mass Spectrometry Research Center.

Funding. A.H.H. is supported by the National Institutes of Diabetes and Digestive and Kidney Diseases (grant R01DK121520) and the Department of Veterans Affairs Research Career Scientist grant IK6 BX005649. M.W.S. is supported by the National Institutes of Health Molecular Endocrinology Training Program (grant T32DK007563).

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

Author Contributions. M.W.S. designed the research studies, performed experiments, interpreted results, and drafted the manuscript. A.R. was responsible for mouse phenotyping, animal husbandry, performed experiments, and interpreting results. M.A. conceived and designed the research studies and edited the manuscript. A.H.H. conceived and designed the research studies, interpreted results, and reviewed and edited the manuscript. All authors approved the final version of the manuscript. A.H.H. 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. A.H.H. is now Vice-Provost and Senior Associate Dean for Faculty Affairs and Career Development at University of Texas Southwestern in Dallas, TX.