Selective depletion of HSCDAs was achieved by engineering bone marrow HSC donor mice in which adipocyte-specific expression of both a diphtheria toxin subunit A (ROSA^DTA) gene and a mTomato/mGFP reporter gene (ROSA^mTmG) were controlled by the adipocyte-specific adiponectin gene promoter driving Cre recombinase (AdipoQcre) ( Figure 1A ). These transplant donor mice, harboring AdipoQcre, ROSA^mTmG and ROSA^DTA transgenes were designated “AdipoQ DTA”. Littermates that lacked ROSA^DTA but contained the AdipoQcre and ROSA^mTmG transgenes served as control donors and were designated “AdipoQ”. This combination was necessary rather than Cre-negative control mice because Cre recombinase was required for GFP expression and verification of successful HSC transplant. AdipoQ and AdipoQ DTA mice served as bone marrow HSC donors for wild-type recipients allowing us to explicitly monitor the presence and function solely of HSCDAs.

The AdipoQ DTA donor pups were significantly smaller than litter- and age-matched AdipoQ donors ( Figure 1B ), and completely lacked all adipose depots including the subcutaneous inguinal (ventral) fat around the hind limbs and interscapular (dorsal) fat between the shoulder blades, which was present in normal quantity in AdipoQ pups ( Figure 1C ). Additionally, microscopic evaluation of skin sections from AdipoQ DTA mice revealed a complete lack of subcutaneous adipocytes, which were present in AdipoQ skin sections ( Figure 1D ). Thus, targeted ablation of adipocytes in AdipoQ DTA mice is efficient and complete.

HSCs were isolated by flow cytometry ( Supplementary Figure 1 ) from whole bone marrow recovered from the femurs and tibias 28-35 day old female littermate AdipoQ and AdipoQ DTA donors and transplanted into separate cohorts of lethally irradiated 12-week old wild-type female recipient mice and allowed to engraft for approximately 4 weeks. Mice receiving HSCs from AdipoQ DTA donors were predicted to produce only conventional adipocytes while HSCDA production would be blocked or reduced. In mice receiving transplants with bone marrow HSC from AdipoQ donors, production of both conventional adipocytes and HSCDAs were predicted to be normal. Flow cytometry 20 weeks post-transplant ( Supplementary Figure 2 confirmed AdipoQcre-driven GFP expression in the adipocytes of AdipoQ recipients (~16% in gonadal adipocytes and ~11% in subcutaneous adipocytes) while GFP-expressing adipocytes were greatly diminished in mice transplanted with AdipoQ DTA marrow (<5% in both depots) ( Figure 1E ). HSCDA depletion in the Ablated mice was verified by quantitative RT-PCR for the EGFP reporter gene, which was significantly lower in Ablated mice than in the Competent cohort ( Figure 1F ). The GFP^POS cells detected by flow cytometry and quantitative RT-PCR correlated with the presence of GFP-expressing cells containing large lipid droplets observed by fluorescence microscopy or imaging flow cytometry ( Supplementary Figures 3A, B , respectively) in the Competent, but not the Ablated mice. Mice that received HSCs from AdipoQ donors were designated “Competent”, and recipients of HSCs from AdipoQ DTA donors were designated “Ablated” in all figures.

We anticipated blockade of new HSCDA production might reduce adipose tissue mass. Interestingly, we found that the mass of gonadal and subcutaneous adipose depots also did not differ between transplant-naive, Competent or Ablated cohorts ( Figure 2A ). Thus, we explored potential changes in adipocyte size and/or abundance. Morphometric analysis of fixed adipose tissue sections from gonadal and subcutaneous depots (representative images are shown in Supplementary Figure 4 ) demonstrated an increase in the abundance of small adipocytes and a decrease in larger adipocytes in the gonadal depot from the Ablated compared to the Competent cohort ( Figure 2B ), but not in the subcutaneous fat. This was even more evident when the data was averaged for the entire adipocyte populations ( Figure 2C ). Likewise, analysis of free-floating adipocytes from collagenase-digested adipose tissue revealed significantly more small adipocytes ranging from 30 to 40 μm but significantly fewer adipocytes between 60 and 80 μm ( Figure 2D ) in gonadal adipose tissue of ablated mice, but no significant differences in adipocyte size in subcutaneous fat between cohorts. The increase in small adipocytes and concomitant decline in larger adipocytes in Ablated mice, while adipose depot weights remained the same suggested that production of new conventional fat cells was normalizing adipose tissue mass due to loss of HSCDAs.

Studies from several laboratories (8, 24) and our group (13, 15) have generally defined conventional adipocyte progenitors as adipose tissue stromal cells lacking hematopoietic lineage markers (defined as Lin^NEG) while co-expressing mesenchymal markers including CD29, Sca-1 and PDGFRα. We quantified conventional adipocyte progenitors (Lin^NEG/CD29^POS/PDGFRα^POS) cells in the stroma of gonadal and subcutaneous adipose tissue from transplant-naïve, Competent and Ablated mice ( Supplementary Figure 5 ). These cells were significantly more abundant in the gonadal fat of Ablated than in Competent or transplant naive mice ( Figure 3A ). Flow cytometry sorted conventional adipocyte progenitors from Competent and Ablated mice were both capable of spontaneous adipogenesis in vitro ( Figure 3B ).

We recently reported that HSCDAs are produced from hematopoietic stem cells through intermediate progenitor cells simultaneously expressing both myeloid markers (CD45 and CD11b) and mesenchymal progenitor markers (CD29, Sca-1 and PDGFRα) (15). Adipose stromal cells bearing these markers were also isolated from transplant-naïve, Competent and Ablated mice and found to be significantly more abundant in the gonadal fat of the Ablated mice ( Figure 3C ), with no difference noted in subcutaneous fat. When these cells were cultured in fibrin clots to induce their conversion to HSCDAs (13, 15), lipid-engorged adipocytes were only produced from HSCDA progenitors from Competent but not Ablated mice ( Figure 3D ), further confirming the ablation model.

We originally surmised that ablation of HSCDAs would cause a modest reduction in body weight and adiposity. We found that both Competent and Ablated mice lost body weight for the first 2-3 weeks post-transplant compared to transplant-naïve control mice ( Figure 4A ). However, these cohorts regained body weight over the next 5-6 weeks, after which their weights plateaued equally. Likewise, no significant body composition differences were noted between cohorts for adiposity (% Fat, Figure 4B ), fat-free mass ( Figure 4C ) or body water content ( Figure 4D ). There was also no sign of excess lipid deposition in liver or skeletal muscle ( Supplementary Figures 6A, B , respectively). As noted earlier, HSCDA ablation was associated with decreased adipocyte size, but no change in adipose depot weight/volume. These results suggest that blockade of HSCDA production elicits the generation of new conventional adipocytes, perhaps as a mechanism to normalize adiposity and adipose tissue cellularity.

Global gene expression analysis previously indicated that HSCDAs might possess a detrimental phenotype characterized by elevated transcript levels of inflammatory cytokines and an absence of leptin (9, 10). We translated these findings to our in vivo model and analyzed the production of adipokines using a Luminex assay ( Figure 5A ). Circulating leptin levels were significantly higher (>3-fold) in Ablated animals. Interleukin-6 (IL-6) expression was substantially reduced and adiponectin was elevated in the Ablated mice although these differences did not achieve statistical significance when the analysis (by ANOVA) included the transplant-naïve cohort. When the transplant-naïve cohort was excluded from the analysis, and data analyzed by unpaired t test, the differences between Competent and Ablated cohorts approached but were shy of p≤0.05. No significant differences were observed between cohorts for serum insulin, macrophage chemoattractant protein-1 (MCP-1), tissue plasminogen activator inhibitor-1 (tPAI-1) or resistin. Tumor necrosis factor-α (TNF-α) was undetectable in all samples.

Because both cohorts received myeloablative conditioning in preparation for HSC transplant, we compared cytokine production between non-irradiated mice and the Competent cohort. Circulating levels of insulin, IL-6 and MCP-1 did not differ in these cohorts ( Figure 5A ). Leptin, tPAI-1 and resistin were all significantly lower in the non-irradiated mice, while adiponectin was significantly higher in this cohort. TNF-α was again not detected. Thus, some inflammatory adipokines (tPAI-1 and resistin), but not all (IL-6 and MCP-1) were elevated in myeloablated mice.

Next, we measured the abundance of tissue-resident immune cell populations in the Competent and Ablated cohorts by flow cytometry. No differences were observed in CD45^POS (pan-hematopoietic), CD11b^POS (myeloid), B220^POS (B lymphocyte), or CD3^POS (T lymphocyte), populations in either gonadal or subcutaneous fat depots ( Figure 5B ). Neutrophils (Siglec F^POS) were significantly reduced in both gonadal and subcutaneous depots of Ablated mice, while granulocytes (Gr-1^POS) were significantly elevated in the gonadal fat of Ablated animals. Finally, diphtheria toxin elicits cell death through rapid inhibition of protein translation via ADP ribosylation of elongation factor 2 (17). Whether adipocyte death by this mechanism promotes macrophage recruitment to and engulfment of dying adipocytes was assessed by immunohistochemical staining of adipose tissue sections from Competent and Ablated mice with antibodies to the macrophage marker, CD68. Macrophage-enshrouded adipocytes (crown-like structures) were essentially absent in sections of gonadal and subcutaneous adipose tissue from both cohorts ( Supplementary Figure 7 ).

Because myeloablation could also modulate the recruitment of inflammatory cells to adipose tissue, we compared the abundance of pan-hematopoietic and myeloid cells in the gonadal fat of non-irradiated and Competent mice ( Figure 5B ). There was no difference in the levels of these cells between the groups, suggesting that post-irradiation recruitment of inflammatory cells was minimal or had completely resolved by study end.

Indirect calorimetry showed no difference in energy intake or energy expenditure (Total, Resting and Non-Resting), or calculated energy balance (25) between Competent and Ablated cohorts when values were averaged over several days ( Figure 6A ). However, when the data was assessed daily, we noted that energy balance and energy intake were significantly reduced every fourth day, but otherwise stable on the remaining three days ( Supplementary Figure 8 ). This pattern was consistent with data reported by Giles et al. (25) linking similar changes in these parameters to the murine estrus cycle. Furthermore, both energy balance and energy intake were constant in cohorts of transplanted ovariectomized mice, which supports the contention that the changes observed in surgery-naive Competent and Ablated mice were related to intact ovarian function. Although energy expenditure was unchanged, physical activity was significantly diminished in the Ablated mice ( Figure 6B ). Both ambulatory and non-ambulatory physical activity were diminished in the Ablated mice.

There was no significant difference in circulating glucose ( Figure 7A ), non-esterified fatty acids ( Figure 7B ) and triacylglycerol ( Figure 7C ) levels between Competent and Ablated mice. Likewise, oral glucose tolerance tests revealed no difference in glucose clearance between the two cohorts ( Figure 7D ). Insulin-stimulated glucose uptake in fasted mice was lower in the Ablated cohort than in Competent animals ( Figure 7E ). When the data was expressed as Area Under the Curve, the decrease in insulin-stimulated glucose uptake was statistically significant.

Blocking the production of new HSCDAs did not affect body weight, adiposity, adipose depot weights or other measures of body composition. This may be due in part to normal production of HSCDAs in both cohorts prior to the adoptive transfer procedure. Because HSCDA ablation was restricted to those fat cells generated from the transplanted bone marrow HSCs, HSCDAs produced prior to transplant were spared. However, previous analysis of HSCDA abundance with either adoptive transfer or nonmyeloablative models has indicated that their turnover (production versus loss) is fairly constant or increases slowly over time (10, 12, 26). Thus, HSCDAs may be completely ablated as pre-transplant HSCDAs are turned-over or die, and production of new HSCDAs is blocked. It also remains to be determined whether HSCDA turnover rates are comparable to conventional-lineage adipocyte turnover.

The loss of HSCDAs post-transplant in the Ablated cohort may also elicit their replacement with new conventional adipocytes. This idea is supported by the adipocyte sizing analysis which revealed an increase in the number of small adipocytes (30-40 µm) while the percentage of larger adipocytes (60-80 µm) was reduced. If this was simply due to lipolysis and shrinkage of adipocytes that were previously larger, then one would have anticipated a decrease in depot size and adiposity, which was not observed. Moreover, conventional adipocyte progenitors (Lin^NEG/CD29^POS/PDGFR^POS stromal cells) were more abundant in Ablated versus Competent mice, perhaps as a compensatory mechanism to maintain adipose depot cellularity during HSCDA depletion.

HSCDA were also significantly more abundant in the gonadal fat of Ablated mice (9, 10). We have \accumulation is stimulated by high fat feeding and thiazolidinediones (9), and suppression of estrogen signaling in female mice (26). We surmise that expansion of the HSCDA progenitor population in the Ablated animals may reflect an attempt to normalize HSCDA production and adipose tissue cellularity.

A major concern in these studies was the reliance on adoptive transfer to ablate endogenous HSCDA progenitors (HSCs and myeloid intermediates) in the recipient mice so they could be replaced with GFP^POS HSCs. Differentiation of reporter-labeled HSCs to HSCDAs allowed us to confirm the production of HSCDAs and quantify their abundance. We previously detected HSCDAs using several transplant and nonmyeloablative models (14). However, these experiments required adoptive transfer to limit ablation solely to HSCDAs while sparing conventional adipocytes.

Myeloablation by lethal irradiation engenders several problems (27). It effectively kills reproducing cells thus primarily targeting the highly reproductive cells of the hematopoietic/immune system. Survival of irradiated mice can be achieved by HSC/BM transplant which rescues hematopoiesis, including red cell production within six weeks. Although post-transplant survival approaches 100%, low-level body-wide inflammation is often observed for at least several months. We found that expression of three inflammatory cytokines, IL-6, resistin and tPAI-1 were all elevated in irradiated AdipoQ recipient mice compared to non-irradiated control mice. However, this increase in cytokine production was not associated with changes in the abundance of CD45^POS pan-hematopoietic cells, nor CD11b^POS myeloid cells in the adipose tissue of irradiated mice. Nor did we observe increased macrophage recruitment or crown-like structures in the adipose tissue of myeloablated, Competent or Ablated cohorts, a hallmark of elevated adipose tissue inflammation and apoptotic (28, 29) or necrotic (30) adipocyte death. Because HSCDAs produced prior to HSC transplant are spared diphtheria toxin-induced cell death, only HSCDAs produced post-transplant are ablated. The rate of DTA-induced HSCDA death may have been too low to elicit a detectable immune response.

We previously reported that the expression of several inflammatory cytokine genes was higher in HSCDAs than conventional adipocytes (10). However, our current analysis of adipose-tissue-related cytokines showed no significant differences in circulating levels of several adipokines, even when the analysis excluded the transplant-naïve cohort. This may be due to the myeloablation-induced proinflammatory environment in both Competent and Ablated cohorts. Measurement of cytokine levels at the whole-organism level may also mask changes in cytokine levels in individual cell populations and/or tissues. Circulating cytokine levels obviously reflect cytokine production not just by adipocytes, but by other cell populations throughout the body. Likewise, circulating adipokine levels often do not reflect adipokine gene expression levels (31). Thus, we currently conclude that HSCDAs are not the pro-inflammatory agents we previously proposed based on cytokine gene expression analysis alone.

Our previous analysis of cytokine gene expression also revealed that HSCDAs produce no or low levels of the adipokine, leptin (10). Here we show that depletion of HSCDAs leads to a significant increase in circulating leptin. This supports a model in which dying HSCDAs are replaced by leptin-producing conventional adipocytes. However, leptin production in the Ablated cohort far exceeded (~245% increase over Competent cohort levels) the level anticipated by the replacement of HSCDAs (~10-20% of total adipocytes) with conventional adipocytes. Thus, leptin production by conventional adipocytes or other non-adipose sources must have been increased, or its transport across the blood-brain barrier was suppressed as observed in common models of leptin-resistance.

Leptin is recognized for its role in the maintenance of body weight and composition by promoting satiety and increasing energy expenditure. However, these effects are most notable under conditions of excess calorie intake, but are less evident under calorie-restricted, low calorie or eucaloric conditions. In our experiments the mice were fed a eucaloric, low fat/low sucrose diet. Thus, the elevated leptin levels in the Ablated cohort did not correlate with increased or reduced energy intake, increased energy expenditure or overall changes in energy balance. However, physical activity was significantly reduced in the Ablated animals. The diminished physical activity in this cohort was not accompanied by changes in body weight, adiposity or other measures of body composition. Changes in caloric expenditure due to physical activity are frequently masked by decreases in other components of energy expenditure (32).

Among agents that stimulate leptin production, no changes were observed in circulating insulin or IL-6 or circulating glucose or non-esterified fatty acids. All animals were female and none were lost due to infection, removing both estrogen (33) and bacterial lipopolysaccharide (34) as potential causes of increased leptin expression. The reduction in physical activity in the Ablated cohort, in spite of elevated circulating leptin levels, suggests that HSCDA ablation perhaps induces leptin resistance. This conclusion is not supported by the absence of body weight, adiposity, energy intake and energy expenditure differences between Ablated and Competent cohorts. However, it is worth noting that the impact of leptin on body composition and energy metabolism are most notable in models of high calorie feeding and obesity, and less so in low calorie or calorie-restricted models (35). Thus, calorie excess models may be required to reveal the full impact of HSCDAs on energy metabolism.

Mice in which HSCDAs were ablated also exhibited mild, but significant decline in insulin -stimulated glucose uptake. Unstimulated glucose uptake, and fasting blood glucose and insulin levels were the same in both Competent and Ablated cohorts suggesting that the decrease in insulin-stimulated glucose uptake was due to inhibition of liver and skeletal muscle insulin sensitivity rather than diminished pancreatic insulin production. The ability of peripheral tissues to respond to insulin can be suppressed by circulating free fatty acids (36, 37) and/or inflammatory agents like IL6. TNFα and MCP-1 (38–41), but these parameters did not differ between our cohorts. Moreover, adiponectin, an adipokine that improves insulin sensitivity and attenuates inflammation was elevated in Ablated mice. Hyperleptinemia in the absence of high fat diet and/or obesity improves insulin sensitivity and glucose uptake (42, 43) and only elicits leptin- and insulin-resistance in the context of diet-induced obesity (44). Thus, while hyperleptinemia and hyperinsulinemia are frequently implicated in metabolic problems of obesity, their appearance together in our non-obese/non-high calorie diet model remains puzzling.

Animal Studies were approved by the Institutional Animal Care and Use Committee at the University of Colorado Anschutz Medical Campus. AdipoQ cre [Jax #028020, B6.FVB-Tg(Adipoq-cre)1Evdr/J], ROSA^mT/mG [Jax #007676, B6.129(Cg)-Gt(ROSA)26Sor^{tm4(ACTB-tdTomato,EGFP)Luo}/J] and ROSA^DTA [Jax #009669, B6.129P2-Gt(ROSA)26Sor^tm1(DTA)Lky/J] and wild-type C57BL6/6 (Jax 000664, C57BL/6J) were purchased from The Jackson Laboratory (Bar Harbor, ME). Mice were group housed in polycarbonate cages in the Center for Human Nutrition Satellite Animal Facility (12:12 hr light:dark cycle, thermoneutral at 27°C) with free access to water. They were fed ad libitum a chow a low-fat diet free of sucrose and phytoestrogens (Envigo Teklad global soy protein-free extruded diet #2920X).

Ablation of adipocytes was achieved by crossing AdipoQcre mice with animals in which Cre recombinase expression was directed by the adiponectin promoter (AdipoQcre) with mice in which a diphtheria toxin subunit A transgene was placed downstream of a floxed stop codon in the ROSA26 locus (ROSA^DTA). Cre-mediated excision of the stop codon initiated DTA expression and cell death, which was restricted to mature adipocytes. All donor mice were crossed with flox-stop-flox mTmG reporter transgenic mice (ROSA^mTmG) resulting in either triple transgenic mice (AdipoQcre-ROSA^mTmG/^DTA, designated AdipoQ DTA) or double transgenic mice (AdipoQcre-ROSA^mTmG/⁺, designated AdipoQ) to be used as bone marrow donor animals. The use of these animals as donor mice allow us to reconstitute wild-type animals in which the production (or lack thereof) of adipocytes from bone marrow stem cells could be quantified by flow cytometry or fluorescence microscopy. Control mice carried the AdipoQ cre allele and a ROSA^mTmG reporter gene as this combination was necessary rather than cre-negative control mice, because Cre recombinase was required for GFP expression and adipocyte tracking.

Selective ablation of HSCDAs was achieved by transplanting HSCs from either female AdipoQ or Adipoq DTA mice into myeloablated female wild type C57BL/6. HSCs were purified from BM recovered from the femurs and tibia of euthanized donor mice. Bone marrow cells were collected from the donor animals by dislocation and removal of the hind leg. The lower leg and tissue were dissected from the femurs and the epiphyses cut off. Using a 27 gauge needle the marrow was flushed with 5ml cold HBSS + 2% FBS into a collection tube then filtered through a 100µm cell strainer and washed in HBSS, 2% FBS media.

Fluorescent antibodies to the cell surface markers Lin, Sca-1 and c-Kit were added to the resuspended bone marrow cells at 0.25 micrograms/10⁶ cells. Samples were incubated at 37°C in the dark for 25 minutes. Following incubation samples were centrifuged to remove unbound antibodies, and the cells were resuspended in PBS containing 5% FBS. Bone marrow cells were sorted within 15 minutes using a Moflo XDP cell sorter with Summit 4.3 software. The sheath fluid was Isoflow. The sample and collection tubes were maintained at 5°C. The sample flow rate was set to a pressure differential of less than 0.4 psi. Sort mode was set to Purify 1. Appropriate signal compensation was set using single color control samples. Lin^NEG/Sca-1^POS/cKit^POS cells were isolated as the murine HSC fraction for subsequent transplant into recipient mice.

Recipient mice were irradiated with two split doses separated by 4 hours of 6 Gy per dose using an X-ray source. Immediately following the second irradiation dose, mice were injected via the retroorbital venous plexus with 1 x 10⁶ isolated bone marrow cells from AdipoQ or AdipoQ DTA donor mice. Wild-type mice transplanted with HSC from AdipoQ donors were designated “Competent”, and mice transplanted with HSC from Adipoq DTA donors were designated “Ablated”. Engraftment was evaluated approximately 4 weeks post-transplant by flow cytometry of PBMC for membrane-localized Tomato expression from the ROSA^mTmG reporter present in both Competent and Ablated cohorts, and routinely approached 95% of circulating cells.

Adipocyte and stromal fractions were prepared by collagenase digestion and flotation/differential centrifugation as previously described (9). PBMCs were isolated from defibrinated blood by centrifugation over Ficoll-Paque medium. Approximately 3 ml of blood were mixed with an equal volume of PBS and gently layered over 3 ml of Ficoll-Paque. The tubes were centrifuged at 500 x g for 30 minutes and the supernatant carefully removed. The PBMCs layered on top of the Ficoll-Paque were recovered, washed once with PBS and resuspended in PBS or culture medium as necessary.

For metabolic phenotyping, mice were moved and singly housed in a Comprehensive Laboratory Animal Monitoring System (CLAMS) Indirect Calorimetry System for a period of 7 days. The first 3 days were a ‘run-in’ period during which the animals acclimated to the new housing conditions as well as different water lixits and food hoppers. No measurements were analyzed during the run-in phase. The run-in period allowed for all necessary checks and calibration to be performed on the instruments. Following the 3 day ‘run-in’ period, metabolic measurements were collected for 4 consecutive days.

Mice were weighed for total body weight, then placed into a measurement cylinder. The mouse was guided to the end of the cylinder using a plunger with enough space for the mouse to turn around, but no more. The assembly was inserted into the qMRI instrument and body composition measurements were recorded over a period of approximately 2 minutes.

Approximately 0.75 ml blood samples were recovered from anesthetized mice immediately prior to euthanasia. Blood glucose levels were determined using a OneTouch Ultra Glucometer. Serum was prepared from each blood sample for subsequent analyses. Circulating insulin levels were measured by ELISA according to the manufacturer’s instructions. Blood serum non-esterified fatty acids and triglycerides were determined using standard clinical reagents. Plasma adiponectin, interleukin-6 (IL-6), tumor necrosis factor-α (TNF-α), interleukin-1β (IL-1β), tissue plasminogen activator inhibitor-1 (tPAI-1), leptin and macrophage chemoattractant protein-1(MCP-1) levels were measured by Luminex multiplex bead assay.

Five-micrometer sections of paraformaldehyde-fixed, paraffin-embedded adipose tissue were deparaffinized and rehydrated in a graded ethanol-water series. Sections were subjected to antigen retrieval in citrate buffer in a pressure cooker for 20 min. Sections were blocked with PBS containing 10% goat serum for 30 min at room temperature. Sections were then blocked with anti-mouse IgG (H+L) antibody to prevent non-specific secondary antibody binding to endogenous antibodies in the tissue. Sections were incubated overnight in PBS-10% goat serum at 4° C with a mouse anti-CD68 antibody. The sections were then washed and incubated with the indicated Alexa Fluor-conjugated goat anti-mouse secondary antibodies (1:250 dilution) for 1h at room temperature. Coverslips were affixed with ProLong Gold fluorescent mounting medium containing DAPI.

Microscopy was performed on a Nikon TE2000-U inverted epifluorescent microscope with a motorized, remote focus/Z-axis stage controller. Brightfield and fluorescent images were captured to a desktop computer with either a DS-Fi2 color camera or a DS-QiMc black & white camera both managed by a DS-U3 PC-based camera control unit. Image data was acquired with NIS-Elements software.

Statistical analysis was performed using the Prism 9.2.0 program. Comparisons were performed using unpaired t tests with Welch’s correction. The data presented in graphs represent averages of values obtained from at least two experiments with 4 animals per group. Data were considered statistically significant with a P value <0.05.