Adipose tissue (AT) expands by a combination of two fundamental cellular mechanisms: hypertrophic growth of existing adipocytes or through generation of new adipocytes, also known as hyperplastic growth. Multiple lines of evidence suggest a limited capacity for hyperplastic growth of AT in adulthood and that adipocyte number is relatively stable, even with fluctuations in AT mass. If the adipocyte number is stable in adulthood, despite well-documented birth and death of adipocytes, then this would suggest that birth may be coupled to death in a regenerative cycle. To test this hypothesis, we examined the dynamics of birth of new fat cells in relationship to adipocyte death by using high-fidelity stable isotope tracer methods in C57Bl6 mice. We discovered birth of new adipocytes at higher frequency in histological proximity to dead adipocytes. In diet-induced obesity, adipogenesis surged after an adipocyte death peak beyond 8 weeks of high-fat feeding. Through transcriptional analyses of AT and fractionated adipocytes, we found that the dominant cell death signals were inflammasome related. Proinflammatory signals were particularly evident in hypertrophied adipocytes or with deletion of a constitutive oxygen sensor and inhibitor of hypoxia-inducible factor, Egln1. We leveraged the potential role for the inflammasome in adipocyte death to test the adipocyte death-birth hypothesis, finding that caspase 1 loss of function attenuated adipocyte death and birth in murine visceral AT. These data collectively point to a regenerative cycle of adipocyte death and birth as a driver of adipogenesis in adult murine AT.

Postnatal expansion of adipose tissue (AT) mass occurs by increased lipid storage in preexisting adipocytes (hypertrophy) and by expansion of adipocyte number (hyperplasia) due to replication and differentiation of resident adipocyte progenitor cells (adipogenesis) (1). Developmentally impaired adipogenesis, as observed in some lipodystrophies, is associated with insulin resistance (2,3). Less extreme abnormalities in adipogenesis conferred by genetic variation at key developmental loci are associated with increased risk of diabetes (4). Therefore, the mechanisms that regulate adipogenesis in AT may also impact systemic metabolic homeostasis.

Studies in rodents and humans using diverse methodologies, including radiolabeling, stable isotope labeling, 14C birth-dating, and genetic lineage tracing, collectively demonstrate ongoing formation of new adipocytes in adult AT (510). Retrospective modeling of human adipocyte population dynamics using cytometric analyses coupled with 14C birth-dating of adipocytes suggests relative stability of adipocyte number with an ∼10% yearly turnover rate in adulthood even with fluctuations in adiposity (6). Genetic lineage tracing with the “Adipochaser” mouse, designed to capture the prospective formation of new adipocytes, demonstrated a delayed induction of adipogenesis in perigonadal AT after 8 weeks of high-fat feeding but no augmentation of adipogenesis in inguinal AT with up to 12 weeks of high-fat feeding (10).

We developed stable isotope tracer methods, using [15N]thymidine to capture replicating and differentiating adipocyte progenitor cells, the incorporation of which is then quantified at high fidelity with one of two mass spectroscopic methods: with isotope ratio mass spectrometry (IRMS) in pooled adipocyte fractions or with imaging mass spectrometry in individual adipocyte nuclei in AT sections (1,8,9). The imaging method, which we call multi-isotope imaging mass spectrometry (MIMS), uses a high-resolution ion nanoprobe coupled to a mass spectrometer (NanoSIMS) to precisely measure isotope ratios and hence stable isotope tracer incorporation with a lateral resolution of down to ∼30 nm (11,12). With this method, we detected rare birth of new adipocytes in inguinal AT (1% during 8 weeks of labeling, projected 6% per year), a rate similar to the 10% yearly estimate by 14C modeling (8). Adipogenesis was not augmented with diet-induced obesity in inguinal AT, but similar to the Adipochaser mouse, we detected a delayed induction in perigonadal AT between 8 and 14 weeks after initiation of high-fat feeding (8). While it is often suggested that augmentation of adipogenesis directly contributes to the AT expansion of obesity, these studies collectively call into question whether a dietary obesigenic stimulus during adulthood drives adipogenesis through an immediate and direct mechanism.

One hypothesis that could account for the coexistence of adipogenesis, adipocyte death, and stable adipocyte numbers is linkage of adipocyte birth to adipocyte death through a regenerative cycle as occurs after genetic or drug-mediated adipocyte death (13,14). Here we applied stable isotope tracer methods to establish temporal and histological associations between adipocyte death and birth in adult AT of C57Bl6 mice. We then investigated candidate adipocyte death mechanisms involving hypertrophy and activation of inflammasome-mediated cell death pathways, including caspase 1. We leverage this insight to demonstrate attenuation of adipocyte birth by interrupting caspase 1–mediated adipocyte death in perigonadal visceral AT, establishing proof-of-concept for a regenerative adipocyte death-birth cycle as a driver of adipogenesis.

Animal Experiments

Experiments were approved by the Brigham and Women’s Hospital Standing Committee on Animals. C57BL6/J mice were purchased from Charles River Laboratories. Casp1−/− mice and Casp1+/+ on the C57BL6/NJ background were purchased from Jackson Laboratories. Egln1F/F mice were generated and bred as previously described (15). Egln1 was deleted by treating Egln1F/F mice harboring a TAM-regulated Cre (CreER) with tamoxifen (1 mg/day i.p.) for 7 days; mice were sacrificed 1 week later. For diet-induced obesity, mice were fed a high-fat (Research Diets, D12492) or control diet (Research Diets, D12450J). Mice were housed at 22°C ± 2°C, with a 12-h light (0700–1900 h), 12-h dark (1900–0700 h) cycle and free access to food and water.

Stable Isotope Experiments

[15N]thymidine (Cambridge Isotope Laboratories) was administered by twice daily subcutaneous injection (25 mg/kg) or via drinking water (100 mg/L), as previously described (1,8). Stromal vascular and adipocyte fractions were isolated and analyzed by IRMS. The supernatant containing the floating adipocytes was collected after passage through a 300-μm nylon mesh, followed by two serial centrifugation (400g, 10 min) and wash steps with PBS in 50-mL conical tubes. Briefly, samples were dried in tin cups and introduced to an elemental analyzer (Vario PYRO Cube, Elementar) coupled to an IRMS (Isoprime 100, Elementar). Tuning was confirmed with urea standards, and samples obtained from unlabeled mice were used as an additional control for the natural background ratio (15N-to-14N = 0.37%). 15N-labeling was detected as signal above the natural 15N-to-14N ratio.

Analyses of AT specimens by MIMS were performed as previously described using the prototype NanoSIMS 50 instrument and the NanoSIMS 50 L (Cameca) (8,11,16). Samples were fixed with 4% paraformaldehyde, embedded in LR white or EPON, sectioned (0.5 μm), and mounted on silicon wafers. Samples were analyzed in automated chain analysis mode, with each tile acquired at 256 × 256 pixels, and 50-μm × 50-μm or 60-μm × 60-μm field sizes. Nuclei were visualized and identified, using OpenMIMS (17), a customized plugin to ImageJ, and assigned an identity of adipocyte, stromal vascular cell, or possible adipocyte by an observer unaware of fat depot, experimental group, or labeling status of the cell. A number of features in the high-resolution quantitative mass spectrometry images enable the observer to identify adipocyte nuclei, based on histological features similar to what is achievable with transmission electron microscopy. The lateral resolution down to <30 nm is sufficient to distinguish an adipocyte nucleus in close association to a dominant lipid droplet with a stereotypical signet appearance and distinct from adjacent stromal or vascular cells (8,9). Adipocytes often have small, nondominant lipid droplets that are observable between the nucleus and the plasma membrane (9,18). In addition, unlike standard transmission electron microscopy, MIMS imaging provides complementary mass images that can reveal specific cellular structures. For example, the 31P images reveal nuclei and plasma membranes, due to the dense phosphorus content of chromatin and membrane phospholipids, respectively (8,9,11,12). A minority of nuclei (∼ ≤5%) are typically designated “possible adipocytes” during this first-pass blinded analysis. Typical reasons for the designation of “possible adipocyte” included imaging artifact (e.g., nucleus found at the margin of two adjacent imaging fields) or inability to clearly distinguish a close association with a lipid droplet and shared plasma membrane. In such instances, the cell in question was reimaged at higher resolution (512 × 512 pixels, 10- to 25-μm raster size), and the images were reevaluated by a blinded observer. An additional subset of adipocytes was captured at higher resolution for independent verification, such that ultimately, ∼20% of the adipocytes were imaged at the higher resolution (e.g., Fig. 1A). These higher-resolution images are at a resolution in the 30- to 50-nm range; whereas other images reported here are in the 100- to 250-nm range.

Adipocyte Isolation

Adipocytes were isolated with a modified version of the classic Rodbell method (19), as previously described (1,8). ATs were minced and digested in an enzyme cocktail of collagenase type D (Roche) and Dispase 2 (ZenBio) in PBS. The suspension was incubated in a shaking water bath (37°C, 225 rpm, 40 min). The digest was filtered (300-μm nylon mesh; Spectrum Laboratories) and centrifuged at 400g for 10 min. The supernatant containing the floating adipocytes was collected. For experiments in which adipocytes were fractionated as a function of size, the adipocyte suspension was subjected to serial cell strainers of descending size: 100 μm (for fraction >100 μm), followed by 70 μm (for fraction 70–100 μm), and then 40 μm (for fraction 40–70 μm).

Cell Culture

3T3-L1 preadipocytes (ZenBio) were cultured in DMEM supplemented with 100 units/mL of penicillin-streptomycin containing 10% FBS (Gibco). Differentiation of 3T3-L1 preadipocytes into adipocytes was achieved by the following procedure: cells were cultured in DMEM containing 10% FBS and supplemented with a cocktail of 10 μg/mL insulin (Sigma-Aldrich), 0.5 mmol/L 3-isobutyl-1-methylxanthine (Sigma-Aldrich), and 1 μmol/L dexamethasone (Sigma-Aldrich) for 4 days; after which cells were maintained in DMEM containing 10% FBS and 10 μg/mL insulin for 6–8 additional days. Culture medium was exchanged every other day.

For cell death experiments involving adipocytes derived from primary cells, ATs were collected from mice aged 4 to 6 weeks old. ATs were excised, minced, and digested in an enzyme cocktail consisting of collagenase D (Roche) and dispase II (ZenBio) in PBS, as described previously (1,20). After centrifugation (4°C, 400g, 10 min), the pelleted stromal vascular fraction was placed into culture or subjected to additional negative selection by column-based magnetic-assisted cell sorting (MACS) (Miltenyi Biotec). Monoclonal anti-CD31 microbeads were used to negatively select endothelial cells, and a biotin-conjugated monoclonal anti-lineage cocktail, followed by anti-biotin MicroBeads (Miltenyi Biotec), were used to negatively select hematopoietic cells. The targeted lineage-positive and CD31+ cells were then depleted by retaining them on a MACS Column in the magnetic field of a MACS Separator. Cells were cultured to near confluency in DMEM-F/12 GlutaMAX medium (Gibco, Thermo Fisher Scientific) with 10% Premium FBS (Corning) and penicillin/streptomycin. Cells were differentiated into adipocytes using the same reagent cocktail used for 3T3L1 adipogenesis, described above.

MTT Assay

After serum starvation for 12 h, differentiated adipocytes were exposed to each candidate cell death ligand for 48 h. The following reagents were used: tumor necrosis factor (TNF)-α at 20 ng/mL (PeproTech), interleukin (IL)-1β at 20 ng/mL (R&D Systems), FAS Ligand at 20 ng/mL (R&D Systems), CD40 Ligand at 20 ng/mL (R&D Systems). An MTT assay (Roche) was performed according to the manufacturer’s protocols. Absorbance was measured at 550 nm and 690 nm for the reference.

Oil Red O Staining

Oil Red O (ORO) staining was performed to assess viable adipocytes after exposure to candidate death ligands or as an assessment of adipocyte differentiation. For death analyses, the cells were treated with ligands as described for the MTT assay above. ORO staining was then performed as previously described (1). Briefly, cells were washed three times with PBS, fixed with 4% paraformaldehyde for 10 min, and incubated with ORO (Sigma-Aldrich) for 60 min at room temperature. The cells were then rinsed three times with water. ORO was extracted with isopropanol, and absorbance was quantified at 520 nm.

Histology

AT specimens were fixed with 4% paraformaldehyde, paraffin-embedded at Harvard Pathology Core, sectioned, and stained using standard methods. Cross-sectional areas of adipocytes in hematoxylin and eosin-stained sections were calculated by tracing the adipocyte periphery in ImageJ. For immunohistochemistry, an antigen retrieval step was used by heating samples in a citrate-based buffer (Dako) at 95°C for 20 min. A rat monoclonal primary antibody to murine F4/80 (Abcam) was used at a concentration of 1:250 for 1 h at room temperature. Endogenous peroxidase activity was quenched with a 20-min incubation in 0.3% H2O2. A mouse adsorbed and biotinylated goat anti-rat IgG secondary antibody was used at a concentration of 1:250 for 1 h, followed by ABC reagent and 3,3′-diaminobenzidine (Vector Laboratories). Adipocytes displaying a halo of F4/80 staining cells were designated as “crown-like structures (CLS)’’ and counted.

Quantitative Real-Time PCR

RNA from adipocytes or cultured cells was isolated using TRIzol Reagent (Ambition Technology) following the manufacturer’s instructions. RNA was reverse-transcribed (High-Capacity cDNA Reverse Transcription Kit; Thermo Fisher Scientific) and quantitative (q)PCR was performed with Power Sybr Green PCR master mix (Applied Biosystems) using a QuantStudio 5 Real-Time PCR System (Applied Biosystems). Gene expression was normalized to GAPDH. The ΔΔCt method was used to calculate the fold change in transcript levels. The qPCR arrays (SA Biosciences) in 96-well format were conducted according to the manufacturer’s protocol using an iCycler (Bio-Rad) instrument.

Statistical Analysis

Statistical analyses were performed with GraphPad Prism 8.0 or 9.0 software. The statistical significance of differences between two groups was determined using Student t tests, differences among more than three groups was evaluated using ANOVA, and post hoc analysis described in figure legends or differences among different cell size groups was calculated using ANOVA and posttest for linear trend. For data that did not pass a normality test (Shapiro-Wilk), we performed nonparametric testing (Mann-Whitney). Values of P < 0.05 were considered significant.

Data and Resource Availability

The data generated during this study are available from the corresponding author upon reasonable request. Any resources generated during this study are available from the corresponding author upon reasonable request.

New Adipocytes Histologically Identified Near Dead Adipocytes

We hypothesized that adipocyte death leads to adipocyte birth as part of a physiological regenerative cycle, in which case we predicted that new adipocytes might arise near dead adipocytes. To test for such an association, we reanalyzed MIMS imaging data of inguinal white AT (iWAT) and gonadal white AT (gWAT) sections from a prior study (8) in which adult C57BL/6 mice were labeled with [15N]thymidine during an 8-week period, starting at 10 weeks of age (n = 3 mice) (Fig. 1). Dead adipocytes are identifiable in AT sections by light microscopy as CLS based on an associated dense halo of macrophages and other leukocytes (21), stereotypical features that also rendered them identifiable by MIMS. Indeed, in an adult visceral AT depot (male perigonadal, gWAT), which is known to have a high frequency of CLS relative to subcutaneous AT, we discovered scattered adipocytes displaying features consistent with CLS (Fig. 1B). MIMS imaging is conducted on resin-embedded tissues, which makes correlative immunostaining challenging, particularly when hydrophobic resins are used (e.g., EPON). However, similar to the preparation of thin sections for electron microscopy, toluidine blue staining provides the possibility of examining adjacent sections with light microscopy prior to MIMS analysis (22). In addition, toluidine blue staining has been used to identify CLS (23). We found that the putative CLS identified with MIMS imaging also demonstrated the typical intense toluidine blue staining (Supplementary Fig. 1). The inflammatory cell “crown” was also composed of cells that were frequently [15N]thymidine labeled, and in an analysis of AT after [2H]glucose labeling, we found that the putative CLS were intensely glucose avid,. as expected of inflammatory cells (Supplementary Fig. 1). Therefore, in a blinded fashion, we extracted the x-y coordinates of CLS from MIMS images. We also extracted the x-y coordinates of 15N-labeled adipocyte nuclei identified in our prior published work (Fig. 1A, and Supplementary Fig. 2), which allowed calculation of the linear proximity of adipocyte nuclei to the nearest CLS. If no CLS was evident in the analytical field, we took the proximity to the edge of the field as proxy. We observed that new adipocytes (15N-labeled) occurred at a higher frequency near CLS (Fig. 1D).

Since adipocyte hypertrophy may drive cellular stress pathways, we also examined the relationship between putative dead adipocytes (CLS) and adipocyte size. We found that adipocytes immediately surrounding CLS were larger, as indicated by their cross-sectional areas, relative to more distant adipocytes in the imaging plane (Fig. 1E). In contrast, we did not detect a size difference between adipocytes adjacent to putative new adipocytes (15N-labeled) relative to more distant adipocytes (Fig. 1F). While these data demonstrate proximity of hypertrophied adipocytes to dead adipocytes and proximity of dead adipocytes to new adipocytes, we cannot infer a causal relationship between these processes. In addition, the identification of putative new adipocytes after an 8-week 15N-labeling protocol does not reveal the temporal dynamics of adipocyte death and birth.

Adipocyte Hypertrophy and Adipocyte Death Temporally Precede Augmentation of Adipogenesis in Diet-Induced Obesity

Prior studies demonstrate an association between murine and human obesity and adipocyte death, particularly in visceral depots (21,2426). We hypothesized that if adipocyte death is a driver of adipocyte birth, then augmentation of adipogenesis would temporally follow an increase in adipocyte death. We first confirmed the association between adipocyte death and diet-induced obesity, finding increased CLS in the AT of C57Bl6 mice at a single time point after 8 weeks of high-fat feeding (Supplementary Fig. 3). We then examined the temporal dynamics of adipocyte hypertrophy and death as measured histologically and as a function of the duration of high-fat feeding (Fig. 2A). In both iWAT and gWAT, we observed increased adipocyte size after 2 weeks of high-fat feeding. The frequency of CLS in both increased slightly at 4 weeks and became much more evident at 8 weeks and beyond. Consistent with prior studies, we observed a higher frequency of CLS in gWAT relative to iWAT (8,21).

We previously observed a modest induction of adipogenesis in adult gWAT, only after 8 weeks of high-fat feeding, an effect that we did not see in iWAT (8). Since iWAT is less prone to adipocyte death, we hypothesized that cumulative death of adipocytes in iWAT with prolonged high-fat feeding would be associated with augmentation of adipogenesis. To test this, we administered a high-fat or control diet for a total of 20 weeks, with [15N]thymidine administered for the final 4 weeks (Fig. 2B). First, in a pooled analysis inclusive of both iWAT and gWAT, we observed an increase in the frequency of 15N-labeled adipocytes with high-fat feeding. When we focused on the individual depots, we observed a significant increase in 15N-labeled adipocytes with high-fat feeding in iWAT (6 of 135 vs. 0 of 126, P = 0.03) and a directionally consistent trend in gWAT (7 of 98 vs. 1 of 71, P = 0.14). The modest number of adipocyte nuclei analyzed by MIMS reflects a convergence of the low analytical throughput of MIMS with AT-specific histological challenges. The MIMS imaging plane is <1 nm in the z-axis. When this is coupled with adipocyte hypertrophy and the associated high lipid droplet-to-nucleus ratio, the resultant frequency of adipocyte nuclei in an image is low. While we detected an augmentation in adipocyte labeling with diet-induced obesity, this analysis also illustrates the challenge of using MIMS to analyze adipocyte nuclei, particularly in obese mice.

In a second experiment we used a longer [15N]thymidine labeling period and analyzed the bulk adipocyte fractions by IRMS, a higher throughput method (1,8). We administered high-fat feeding for 25 weeks with [15N]thymidine starting at 4 weeks at the onset of the CLS surge (Fig. 2C). We observed augmentation of 15N-labeling in visceral adipocytes with high-fat feeding (Fig. 2C). We also observed increased 15N-labeling in iWAT adipocytes (Fig. 2C), which was in contrast to our prior labeling studies in which high-fat feeding was administered for a shorter period (8). Collectively, these data suggest that prolongation of diet-induced obesity beyond the induction of adipocyte death is associated with recruitment of new adipocytes in C57BL/6 mice.

Inflammatory Death Pathways Are Active in Adipocytes With Diet-Induced Obesity

The possibility that adipocyte death is a stimulus for birth of new adipocytes provides rationale to dissect mechanisms of adipocyte death. To identify candidate death mechanisms, we leveraged the observation that crown-like structures—a histological manifestation of adipocyte death—are more prevalent after high-fat feeding, particularly in gWAT. We performed an interdepot transcriptional comparison, using a qPCR array designed to identify activity of cell death pathways. In whole gWAT samples, we observed significant induction of Bcl2a1a and Casp1(Fig. 3A). The induction of Casp1 is consistent with inflammasome cell death pathways, which have been previously linked to adipocyte death (27). In iWAT, which demonstrates a less dramatic increase in adipocyte death, we did not identify inflammatory pathways, instead observing induction of Atg12, Pvf, App, and Fas. This interdepot discrepancy could reflect the underlying differences in adipocyte death rate and/or a preference for noninflammatory death mechanisms.

We next performed additional qPCR analysis of cell death and inflammatory genes, but restricted to the adipocyte fractions (Fig. 3B). We compared adipocytes isolated from iWAT and gWAT after high-fat feeding compared with adipocytes from chow-fed controls, finding significant induction of Casp1 and other genes associated with the inflammasome and inflammatory cell death, such as Nrlp3, Aim2, Pycard, IL-1b, and Tnf. We also observed induction of Cd40l and its receptor, Cd40. A similar analysis of iWAT adipocytes demonstrated directionally similar effects of high-fat feeding; however, the scale of the expression in iWAT was consistently lower than that observed in gWAT adipocytes. Therefore, inflammatory death pathways represented the most consistently detectable transcriptional signal in AT or isolated adipocytes, particularly in the gWAT from diet-induced obese mice.

Induction of Inflammatory Pathways With Adipocyte Hypertrophy

Given the prevalence of hypertrophied adipocytes adjacent to putative dead adipocytes (Fig. 1D) and prior work demonstrating an association between adipocyte hypertrophy and AT inflammation, we next tested whether increasing adipocyte size was associated with transcription of inflammatory signals (28). We isolated adipocytes from iWAT and gWAT depots and then used sequential filtration to separate them into pools of three different size ranges: >100 μm, 70–100 μm, and 40–70 μm (Supplementary Fig. 4). We performed qPCR on the fractions, with attention to genes related to inflammatory cell death pathways. In adipocytes collected from gWAT, there was a significant relationship between ascending adipocyte size and the expression of several genes related to inflammation and inflammatory cell death, including Casp1, Nrlp3, Tnf, and Il1b (Fig. 4A). None of the effects seen in gWAT reached statistical significance in adipocytes isolated from iWAT. We did not observe an increase in immune cell markers by qPCR with ascending adipocyte size (Supplementary Fig. 4). However, given the challenge of isolating an adipocyte fraction that is 100% pure, the effects of contaminating immune cells cannot be definitively excluded. Relevant immune cells, such as lipid-laden macrophages, are generally smaller relative to adipocytes and therefore less likely to account for an effect that was most evident in the largest adipocyte fraction.

Given that one potential mechanism of inflammatory activation would be due to changes in oxygen tension with adipocyte hypertrophy, we also considered whether ascending adipocyte size was associated with augmented expression of the canonical hypoxia response factor, Hif1a. We found a significant relationship between ascending adipocyte size and Hif1a expression in adipocytes collected from gWAT, but not from iWAT (Fig. 4B). To test whether hypoxia-sensing pathways may drive inflammatory pathways in adipocytes, we turned our attention to adipocytes isolated from mice in which tamoxifen treatment induces deletion of Egl1, an oxygen sensor and constitutive negative regulator of HIF (15). Egln1FF;CreER mice or controls were treated with tamoxifen, as previously described (15). Within 1 week of completion of the tamoxifen treatment, mice were sacrificed, and the adipocyte fractions were isolated from both iWAT and gWAT. In gWAT adipocytes, we detected augmented expression of inflammatory death pathway genes, including Casp1, IL1b, and Aim2, an inflammasome cofactor (Fig. 4C). Adipocytes from iWAT did not display a similar effect. Collectively, these data suggest that adipocyte hypertrophy is associated with augmentation of inflammatory pathway genes in adipocytes from gWAT. The absence of similar effects in iWAT suggests that additional signals specific to gWAT may work in concert with the hypertrophy and/or hypoxia signals to drive proinflammatory signals pathways in adipocytes. These data also suggest that adipocyte hypertrophy may not be sufficient to drive inflammation in a cell autonomous manner, at least in all depots.

Attenuation of Visceral Adipocyte Death and Visceral Adipocyte Birth With Caspase 1 Genetic Deletion

Adipocyte death has been described as being dependent on inflammatory signaling, such as those mediated by canonical proinflammatory cytokines (TNF) and caspase 1–mediated pyroptosis, a form of programmed cell necrosis (27,29). Therefore, we first tested the degree to which proinflammatory cytokines promote adipocyte death, in vitro, relative to the effect to other death ligands, FASL and CD40L. We generated adipocytes by differentiating primary stromal vascular cells from C57Bl6 AT, using subcutaneous AT as the source of stromal vascular cells due to the efficiency with which this population differentiates in vitro. We first tested the effects of death ligands on viability, using an MTT assay. IL-1 treatment resulted in a reproducible reduction in cell viability (Fig. 5A). We next performed a similar experiment but quantified ORO staining as an orthogonal assessment of adipocyte attrition, finding a similar effect of IL-1 treatment (Fig. 5B). 3T3L1 preadipocyte-derived adipocytes represent an alternative and more efficient source of adipocytes. Therefore, we also tested the effect of death ligands on 3T3L1-derived adipocytes, finding similar effects, with IL-1 having the most consistent effect, an effect that was additive to TNF (Supplementary Fig. 5). While the effect of TNF was directionally consistent with IL-1, it did not reach statistical significance. Collectively, these data confirm the sensitivity of adipocytes to proinflammatory signaling. The effect of IL-1 appeared more robust than TNF in promoting adipocyte attrition, despite the fact that we used a concentration of TNF that exceeds levels secreted by obese AT, at least as measured in cultured explants (30).

We next tested whether caspase 1 loss of function would attenuate IL-1–mediated adipocyte death in vitro. To test this, we isolated stromal vascular cells from Casp1−/− or wild-type control mice, differentiated them into adipocytes, and measured the effect of IL-1 on cell viability with an MTT assay. Similar to our prior experiments, IL-1 produced a reproducible reduction in viability when applied to wild-type cells. In contrast, the effect was attenuated in Casp1−/− cells (Fig. 5C). Therefore, these experiments in cultured cells provided evidence for the sensitivity of adipocytes to inflammatory death pathways involving caspase 1 when cultured in vitro.

Given that inflammatory death pathways involving caspase 1 were most consistently associated with adipocyte death in our collective analyses, we hypothesized that genetic caspase 1 loss of function would attenuate the adipocyte death-birth cycle in vivo. We performed histological assessment of CLS and observed a reduced frequency of CLS in Casp1−/− mice relative to wild-type controls in the visceral depot (gWAT), but not in the subcutaneous depot (iWAT) (Fig. 5D). We then performed [15N]thymidine labeling of adult mice (age 8–10 weeks at start) and found a reduction in adipocyte labeling in the visceral depot (Fig. 5E). When we performed a similar analysis of Casp1−/− and Casp1+/+ mice concomitantly fed a high-fat diet, we detected no significant effect of caspase 1 loss of function (Supplementary Fig. 6). Therefore, the attenuation of adipogenesis through targeting of caspase 1–mediated adipocyte death in gWAT is consistent with an adipocyte death-birth cycle; however, the absence of a similar cycle in iWAT or in gWAT after high-fat feeding suggests additional mechanisms, including the possibility of alternative adipocyte death pathways in vivo.

One limitation of the in vivo approach is the use of a mouse with global caspase 1 loss of function, including the potential confounding effects of caspase 1 function in nonadipocytes. Indeed, caspase 1 loss of function has previously been shown to augment adipogenic differentiation (31). Therefore, we also examined the adipogenic efficiency of cells isolated from Casp1−/− and Casp1+/+ mice, using a negative selection protocol to enrich for progenitors by depletion of lineage-positive cells and CD31+ cells. We did not find a consistent and significant effect of caspase 1 loss of function on adipogenic efficiency, even though there was a trend (P = 0.06) (Supplementary Fig. 7).

An alternative potential mechanism of modulating the adipocyte life cycle would be through modulation of paracrine signaling by a nonadipocyte (e.g., inflammatory cells), as a central function of caspase 1 is the proteolytic activation of IL-1 and IL-18. To assess the plausibility of this, we tested the effect of both cytokines on 3T3L1 adipocyte differentiation, finding that IL-1 treatment attenuated adipogenesis, as previously demonstrated (32). In both respective instances—through cell autonomous effects on adipocyte precursors or modulation of caspase 1-dependent paracrine signals—caspase 1 loss of function would be predicted to augment adipogenesis, thereby biasing toward the null hypothesis. When these data are viewed together with prior data suggesting multiple mechanisms of adipocyte death, it underscores the experimental challenge of efficiently neutralizing adipocyte death in a targeted and specific manner.

AT expansion is commonly described as being dependent on a combination of two distinct cellular mechanisms: the hypertrophy of existing adipocytes and the hyperplastic recruitment of new adipocytes from resident adipocyte progenitor cells (adipogenesis). In a prior study, we used stable isotope tracer methods to quantify adipogenesis in mice and, surprisingly, failed to detect augmentation of adipogenesis in the early diet-induced obesity (8). It was only after more prolonged high-fat feeding that we detected recruitment of new adipocytes, but only in the visceral depot and at a time point coincident with augmentation of adipocyte death (8). In the current study, we further dissected these birth/death dynamics in C57BL/6 mice. Our data suggest a regenerative cycle of adipocyte death and birth providing one explanation for why the early expansion of AT with diet-induced obesity is uncoupled from augmentation of adipogenesis. We observed the strongest evidence for a regenerative cycle in visceral AT, including 1) a spatial and temporal association between adipocyte death and birth and 2) demonstration that both arms of the cycle—adipocyte death and birth—are attenuated in visceral AT by a genetic loss of function of Casp1, a mediator of inflammatory death in adipocytes.

The premise that adipocyte death drives adipocyte birth is further supported by orthogonal and previously published evidence (1). Multiple studies demonstrate adipogenesis in AT of adult mice and humans (6,810). When viewed together with evidence pointing to stability of adipocyte numbers in adulthood (6,8), this suggests matched attrition of adipocytes (2). The forced, widespread induction of white adipocyte death by transgenic activation of caspase 8, although nonphysiologic, is followed by regenerative growth of new adipocytes (3,14). A prior genetic lineage tracing study demonstrated evidence of adipogenesis, including proliferating platelet-derived growth factor receptor–expressing progenitors in close proximity to putative dead adipocytes (13).

In our study, we tested for the coupling of adipocyte death and birth by targeting inflammatory death pathways with caspase 1 loss of function. Caspase 1 has previously been implicated in adipocyte death by pyroptosis (27). Indeed, we found that exposure to IL-1 led to adipocyte attrition in vitro that was largely attenuated by concomitant caspase 1 loss of function. However, the inhibition of adipocyte death with global Casp1 loss of function, in vivo, was only clearly evident in the visceral depot, an effect that was lost with diet-induced obesity. Previous work demonstrates murine adipocyte death by mechanisms including both pyroptosis and apoptosis (24,27,33,34). Inflammatory death can also operate independent of caspase 1 (35). Moreover, caspase 1 may modulate AT cellularity and global metabolic function via diverse cell autonomous and paracrine mechanisms (31,33). Such redundancies underscore the challenge of neutralizing adipocyte death with a single specific genetic intervention.

A key theme that emerged from this study was the contrast between the inguinal and perigonadal depots, reflected in almost every analysis. The sequential augmentation of adipocyte death and birth with prolonged high-fat feeding or the hypertrophy-related augmentation of inflammatory pathways was evident to a markedly lesser degree in inguinal fat. Egln1 loss of function resulted in augmented transcription of proinflammatory genes in visceral adipocytes, but not in subcutaneous adipocytes. This suggests modifying factors that regulate adipocyte-driven inflammation and death beyond adipocyte hypertrophy. Indeed, while there are multiple lines of evidence suggesting a link between large adipocytes and inflammation or insulin resistance, some reports suggest that small adipocytes are also markers of inflammation, perhaps due to a defect in functional maturation (28,36). Our data demonstrating depot-specific effects of adipocyte hypertrophy further underscore the complexity of the relationship between adipocyte size and inflammation and also add to the growing recognition of fundamental differences in the biology and potential for pathologic changes as a function of the fat depot (37,38).

This study leaves a number of important questions unanswered. First, from a methodological standpoint, adipocytes are identified based on their buoyancy (IRMS) or stereotypical histological features with high resolution MIMS imaging. Stable isotope-based mass spectrometry approaches provide highly quantitative measurements, unencumbered by some of the typical challenges associated with studying AT, such as high propensity for autofluorescence with light microscopy. However, in the future, the method could be strengthened further by coupling the molecular specificity conferred by genetic reporters with the quantitative power of mass spectrometry, as has been achieved in the liver and heart (39,40). Recent advances in tracking genetically encoded labels with MIMS in vitro may provide a viable avenue to achieve this goal in AT (41,42).

Second, we focused on the perigonadal and inguinal fat depots in C57Bl6 mice. Prior studies have demonstrated strain-dependent differences in the dynamics of adipocyte death, adipogenesis, and adipocyte hypertrophy, and therefore, the degree to which our findings are generalizable to other strains of mice or to human ATs is unknown (8,21,43).

Third, while some of our experiments included both male and female mice, this study was not designed to detect sex-specific differences.

Fourth, we have not directly studied the adipocyte progenitor cell, including the anatomic site in which the proximal proliferation and differentiation events transpire. Multiple lines of evidence point to a perivascular niche, raising questions about the modifying effects of vessel proximity and the degree to which progenitors migrate from their niche as they become activated.

Fifth, we used global caspase 1 loss of function to interrupt the death-birth cycle as experimental proof of concept; however, it will be important to verify these results with orthogonal and/or more cell type-specific approaches.

Sixth, the presence of a regenerative death-birth cycle does not exclude coexistent direct stimulatory effects of caloric excess on adipogenesis. Our methods of measuring adipogenesis use [15N]thymidine to capture replicating and differentiating adipocyte progenitors. Although stable isotope tracers are advantageous because they are innocuous and can be tracked with high-fidelity mass spectrometry methods, cell cycle labels only detect new adipocytes that arise from progenitor cell differentiation that is preceded by self-renewal. Indeed, our previous modeling of adipocyte turnover suggests a subpopulation of cells that differentiate without preceding division (8). A good candidate for such a subpopulation is the committed preadipocyte, identified by single-cell transcriptomics studies as residing in a more advanced state of differentiation relative to adipocyte progenitors (44,45).

Lastly, our study does not address the metabolic implications of adipocyte death or adipocyte regeneration. Given the recognition that functional adipocytes are crucial for metabolic homeostasis, it will be important in the future to investigate whether a regenerative defect and attrition of adipocytes in WAT contributes to insulin resistance, particularly in advanced age where loss of subcutaneous fat and visceral redistribution is associated with deleterious metabolic consequences (38,46).

This article contains supplementary material online at https://doi.org/10.2337/figshare.17296316.

Funding. This work was funded by National Institute of Diabetes and Digestive and Kidney Diseases grants K08DK090147, R01DK120659, and R03DK106477 to M.L.S.

Duality of Interest. C.G. is currently employed at Zeiss. B.O. is currently employed at Regeneron. The experimental contributions to this manuscript by both coauthors preceded their current employment. M.L.S. has performed consulting for Regeneron and Amgen. No other potential conflicts of interest relevant to this article were reported.

Author Contributions. A.M., Y.Z., G.V.N.K., C.G., S.K., and M.L.S. contributed to the investigation. A.M., B.O., and M.L.S. contributed to conceptualization. A.M. and M.L.S. wrote the original draft. C.G. contributed to study methodology. B.O. contributed resources. M.L.S. supervised the study. All authors reviewed and edited the manuscript. M.L.S. 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.

1.
Zhang
Y
,
Federation
AJ
,
Kim
S
, et al
.
Targeting nuclear receptor NR4A1-dependent adipocyte progenitor quiescence promotes metabolic adaptation to obesity
.
J Clin Invest
2018
;
128
:
4898
4911
2.
Garg
A
.
Acquired and inherited lipodystrophies
.
N Engl J Med
2004
;
350
:
1220
1234
3.
Moitra
J
,
Mason
MM
,
Olive
M
, et al
.
Life without white fat: a transgenic mouse
.
Genes Dev
1998
;
12
:
3168
3181
4.
Majithia
AR
,
Flannick
J
,
Shahinian
P
, et al.;
GoT2D Consortium
;
NHGRI JHS/FHS Allelic Spectrum Project
;
SIGMA T2D Consortium
;
T2D-GENES Consortium
.
Rare variants in PPARG with decreased activity in adipocyte differentiation are associated with increased risk of type 2 diabetes
.
Proc Natl Acad Sci U S A
2014
;
111
:
13127
13132
5.
Klyde
BJ
,
Hirsch
J
.
Isotopic labeling of DNA in rat adipose tissue: evidence for proliferating cells associated with mature adipocytes
.
J Lipid Res
1979
;
20
:
691
704
6.
Spalding
KL
,
Arner
E
,
Westermark
PO
, et al
.
Dynamics of fat cell turnover in humans
.
Nature
2008
;
453
:
783
787
7.
Neese
RA
,
Misell
LM
,
Turner
S
, et al
.
Measurement in vivo of proliferation rates of slow turnover cells by 2H2O labeling of the deoxyribose moiety of DNA
.
Proc Natl Acad Sci U S A
2002
;
99
:
15345
15350
8.
Kim
SM
,
Lun
M
,
Wang
M
, et al
.
Loss of white adipose hyperplastic potential is associated with enhanced susceptibility to insulin resistance
.
Cell Metab
2014
;
20
:
1049
1058
9.
Guillermier
C
,
Fazeli
PK
,
Kim
S
, et al
.
Imaging mass spectrometry demonstrates age-related decline in human adipose plasticity
.
JCI Insight
2017
;
2
:
e90349
10.
Wang
QA
,
Tao
C
,
Gupta
RK
,
Scherer
PE
.
Tracking adipogenesis during white adipose tissue development, expansion and regeneration
.
Nat Med
2013
;
19
:
1338
1344
11.
Steinhauser
ML
,
Bailey
AP
,
Senyo
SE
, et al
.
Multi-isotope imaging mass spectrometry quantifies stem cell division and metabolism
.
Nature
2012
;
481
:
516
519
12.
Gyngard
F
,
Steinhauser
ML
.
Biological explorations with nanoscale secondary ion mass spectrometry
.
J Anal At Spectrom
2019
;
34
:
1534
1545
13.
Lee
YH
,
Petkova
AP
,
Granneman
JG
.
Identification of an adipogenic niche for adipose tissue remodeling and restoration
.
Cell Metab
2013
;
18
:
355
367
14.
Pajvani
UB
,
Trujillo
ME
,
Combs
TP
, et al
.
Fat apoptosis through targeted activation of caspase 8: a new mouse model of inducible and reversible lipoatrophy
.
Nat Med
2005
;
11
:
797
803
15.
Olenchock
BA
,
Moslehi
J
,
Baik
AH
, et al
.
EGLN1 inhibition and rerouting of α-ketoglutarate suffice for remote ischemic protection
.
Cell
2016
;
164
:
884
895
16.
Gyngard
F
,
Trakimas
L
,
Steinhauser
ML
.
High-fidelity quantification of cell cycle activity with multi-isotope imaging mass spectrometry
.
Methods Mol Biol
2021
;
2158
:
257
268
17.
Guillermier
C
,
Poczatek
JC
,
Taylor
WR
,
Steinhauser
ML
.
Quantitative imaging of deuterated metabolic tracers in biological tissues with nanoscale secondary ion mass spectrometry
.
Int J Mass Spectrom
2017
;
422
:
42
50
18.
Steinhauser
ML
,
Guillermier
C
,
Wang
M
,
Lechene
CP
.
Approaches to increasing analytical throughput of human samples with multi-isotope imaging mass spectrometry
.
Surf Interface Anal
2014
;
46
(
Suppl. 1
):
165
168
19.
Rodbell
M
.
Metabolism of isolated fat cells. I. Effects of hormones on glucose metabolism and lipolysis
.
J Biol Chem
1964
;
239
:
375
380
20.
Chu
AY
,
Deng
X
,
Fisher
VA
, et al
.
Multiethnic genome-wide meta-analysis of ectopic fat depots identifies loci associated with adipocyte development and differentiation
.
Nat Genet
2017
;
49
:
125
130
21.
Murano
I
,
Barbatelli
G
,
Parisani
V
, et al
.
Dead adipocytes, detected as crown-like structures, are prevalent in visceral fat depots of genetically obese mice
.
J Lipid Res
2008
;
49
:
1562
1568
22.
Guillermier
C
,
Doherty
SP
,
Whitney
AG
, et al
.
Imaging mass spectrometry reveals heterogeneity of proliferation and metabolism in atherosclerosis
.
JCI Insight
2019
;
4
:
e128528
.
23.
Murano
I
,
Rutkowski
JM
,
Wang
QA
,
Cho
YR
,
Scherer
PE
,
Cinti
S
.
Time course of histomorphological changes in adipose tissue upon acute lipoatrophy
.
Nutr Metab Cardiovasc Dis
2013
;
23
:
723
731
24.
Alkhouri
N
,
Gornicka
A
,
Berk
MP
, et al
.
Adipocyte apoptosis, a link between obesity, insulin resistance, and hepatic steatosis
.
J Biol Chem
2010
;
285
:
3428
3438
25.
Cinti
S
,
Mitchell
G
,
Barbatelli
G
, et al
.
Adipocyte death defines macrophage localization and function in adipose tissue of obese mice and humans
.
J Lipid Res
2005
;
46
:
2347
2355
26.
Strissel
KJ
,
Stancheva
Z
,
Miyoshi
H
, et al
.
Adipocyte death, adipose tissue remodeling, and obesity complications
.
Diabetes
2007
;
56
:
2910
2918
27.
Giordano
A
,
Murano
I
,
Mondini
E
, et al
.
Obese adipocytes show ultrastructural features of stressed cells and die of pyroptosis
.
J Lipid Res
2013
;
54
:
2423
2436
28.
Ye
RZ
,
Richard
G
,
Gévry
N
,
Tchernof
A
,
Carpentier
AC
.
Fat cell size: measurement methods, pathophysiological origins, and relationships with metabolic dysregulations
.
Endocr Rev
8 June 2021 [Epub ahead of print]. DOI: 10.1210/endrev/bnab018
29.
Prins
JB
,
Niesler
CU
,
Winterford
CM
, et al
.
Tumor necrosis factor-alpha induces apoptosis of human adipose cells
.
Diabetes
1997
;
46
:
1939
1944
30.
Hotamisligil
GS
,
Arner
P
,
Caro
JF
,
Atkinson
RL
,
Spiegelman
BM
.
Increased adipose tissue expression of tumor necrosis factor-alpha in human obesity and insulin resistance
.
J Clin Invest
1995
;
95
:
2409
2415
31.
Stienstra
R
,
Joosten
LA
,
Koenen
T
, et al
.
The inflammasome-mediated caspase-1 activation controls adipocyte differentiation and insulin sensitivity
.
Cell Metab
2010
;
12
:
593
605
32.
Simons
PJ
,
van den Pangaart
PS
,
van Roomen
CP
,
Aerts
JM
,
Boon
L
.
Cytokine-mediated modulation of leptin and adiponectin secretion during in vitro adipogenesis: evidence that tumor necrosis factor-alpha- and interleukin-1beta-treated human preadipocytes are potent leptin producers
.
Cytokine
2005
;
32
:
94
103
33.
Wang
H
,
Capell
W
,
Yoon
JH
,
Faubel
S
,
Eckel
RH
.
Obesity development in caspase-1-deficient mice
.
Int J Obes (Lond)
2014
;
38
:
152
155
34.
Feng
D
,
Tang
Y
,
Kwon
H
, et al
.
High-fat diet-induced adipocyte cell death occurs through a cyclophilin D intrinsic signaling pathway independent of adipose tissue inflammation
.
Diabetes
2011
;
60
:
2134
2143
35.
Schneider
KS
,
Groß
CJ
,
Dreier
RF
, et al
.
The inflammasome drives GSDMD-independent secondary pyroptosis and IL-1 release in the absence of caspase-1 protease activity
.
Cell Rep
2017
;
21
:
3846
3859
36.
McLaughlin
T
,
Deng
A
,
Yee
G
, et al
.
Inflammation in subcutaneous adipose tissue: relationship to adipose cell size
.
Diabetologia
2010
;
53
:
369
377
37.
Cohen
P
,
Levy
JD
,
Zhang
Y
, et al
.
Ablation of PRDM16 and beige adipose causes metabolic dysfunction and a subcutaneous to visceral fat switch
.
Cell
2014
;
156
:
304
316
38.
Fox
CS
,
Massaro
JM
,
Hoffmann
U
, et al
.
Abdominal visceral and subcutaneous adipose tissue compartments: association with metabolic risk factors in the Framingham Heart Study
.
Circulation
2007
;
116
:
39
48
39.
Senyo
SE
,
Steinhauser
ML
,
Pizzimenti
CL
, et al
.
Mammalian heart renewal by pre-existing cardiomyocytes
.
Nature
2013
;
493
:
433
436
40.
Zhang
Y
,
Guillermier
C
,
De Raedt
T
, et al
.
Imaging mass spectrometry reveals tumor metabolic heterogeneity
.
iScience
2020
;
23
:
101355
41.
Narendra
DP
,
Guillermier
C
,
Gyngard
F
,
Huang
X
,
Ward
ME
,
Steinhauser
ML
.
Coupling APEX labeling to imaging mass spectrometry of single organelles reveals heterogeneity in lysosomal protein turnover
.
J Cell Biol
2020
;
219
:
e201901097
42.
Narendra
DP
,
Steinhauser
ML
.
Metabolic analysis at the nanoscale with multi-isotope imaging mass spectrometry (MIMS)
.
Curr Protoc Cell Biol
2020
;
88
:
e111
43.
Surwit
RS
,
Feinglos
MN
,
Rodin
J
, et al
.
Differential effects of fat and sucrose on the development of obesity and diabetes in C57BL/6J and A/J mice
.
Metabolism
1995
;
44
:
645
651
44.
Merrick
D
,
Sakers
A
,
Irgebay
Z
, et al
.
Identification of a mesenchymal progenitor cell hierarchy in adipose tissue
.
Science
2019
;
364
:
eaav2501
45.
Hepler
C
,
Shan
B
,
Zhang
Q
, et al
.
Identification of functionally distinct fibro-inflammatory and adipogenic stromal subpopulations in visceral adipose tissue of adult mice
.
eLife
2018
;
7
:
e39636
.
46.
Borkan
GA
,
Hults
DE
,
Gerzof
SG
,
Robbins
AH
,
Silbert
CK
.
Age changes in body composition revealed by computed tomography
.
J Gerontol
1983
;
38
:
673
677
Readers may use this article as long as the work is properly cited, the use is educational and not for profit, and the work is not altered. More information is available at https://www.diabetesjournals.org/journals/pages/license.