In Vivo Adipogenesis in Rats Measured by Cell Kinetics in Adipocytes and Plastic-Adherent Stroma-Vascular Cells in Response to High-Fat Diet and Thiazolidinedione

Thirty-two 50- to 55-day-old male Long-Evans rats (Harlan Laboratories, Indianapolis, IN) were fed ad libitum rodent LabDiet (5% of calories from fat) and water for 1 week prior to the experiments. The rats were weighed twice a week during the following dietary and ²H₂O-drinking interventions. All procedures were approved by the institutional animal care and use committee at the Pennington Center.

The rats were divided into three groups and fed ad libitum the following diets (Research Diets, New Brunswick, NJ) for 21 days: 1) low-fat diet (LFD) (10% of calories from fat, cat. no. D12450B), n = 10; 2) HFD (60% of calories from fat, cat. no. D12492), n = 11; and 3) HFD-P containing 0.033% pioglitazone (providing 20 mg ⋅ kg⁻¹ ⋅ day⁻¹ given a daily consumption of 15 g of food), n = 11.

On the first day of the dietary intervention, all rats received an intraperitoneal bolus injection of 35 mL/kg body wt 0.9% NaCl in 100% ²H₂O (Cambridge Isotope Laboratories, Andover, MA) after isoflurane anesthesia. Previous studies have shown that loading this dose achieves a 5% body-water enrichment (13). For maintenance of this level of ²H₂O enrichment, rats were given ad libitum access to 8% ²H₂O for the 21-day dietary intervention (13,14). Rats were killed, and epididymal adipose tissue (eAT) was collected. One part was snap-frozen in liquid N₂ for RNA extraction, and another was fixed in 4% paraformaldehyde, embedded in paraffin, and sectioned at 5 μm for immunohistochemistry. The remaining tissue was collagenase digested to isolate adipocytes and stroma-vascular cells as previously described (20).

The proximal ends of the removed femurs were clipped, placed in 2-mL vials supported by plastic inserts, and centrifuged for 5 s at 2,000 rpm as previously described (21,22). The extruded bone marrow cells were frozen in liquid N₂ and stored at −80°C.

The isolated adipocytes from LFD (n = 5), HFD (n = 6), and HFD-P (n = 6) rats were divided into two aliquots. One aliquot was frozen in liquid N₂ and stored at −80°C. The other aliquot was incubated with a cocktail of mouse monoclonal antibodies against markers of rat endothelial cells (CD31, cat. no. 1MCA1334G; AbD Serotec, Oxford, U.K.), nucleated hematopoietic cells (CD45, clone OX-1; Invitrogen, Camarillo, CA), and stem cells (nerve growth factor receptor p75NTP, i.e., CD271 [23], cat. no. MC192, Novus Biologicals, Littleton, CO] for 20 min at room temperature. Cells attached to the antibodies were discarded by using magnetic Dynabeads Pan Mouse IgG (Invitrogen, Camarillo, CA). The immunopurified adipocytes were frozen in liquid N₂ and stored at −80°C. The isolated adipocytes from the remaining animals were directly frozen in liquid N₂ and stored at −80°C until DNA extraction.

For isolation of an enriched population of adipocyte progenitor cells, a protocol for the isolation of mesenchymal stem cells from bone marrow cells using a short-term culture was used (24). The stroma-vascular fraction isolated from eAT was reconstituted in erythrocyte lysis buffer (0.154 mol/L NH₄Cl, 10 mmol/L KHCO₃, and 1 mmol/L EDTA) for 5 min at room temperature, centrifuged, reconstituted in 10% FBS in Dulbecco’s modified Eagle’s Medium/F-12 medium, and cultured for 3 h. The nonadherent cells were washed twice with PBS. The adherent stroma-vascular (aSV) cells were detached (0.25% trypsin/EDTA) and centrifuged, and the pellet was frozen in liquid N₂ and stored at −80°C.

DNA from bone marrow cells was extracted by using a QiaAmp micro-DNA extraction kit (QIAGEN, Germantown, MD). DNA from the adipocytes and the aSV cells was extracted following the protocol for samples with a low number of cells (13). DNA was enzymatically hydrolyzed to free deoxynucleosides, and their deoxyribose moiety was converted to the tetrafluorobenzylpentosetriacetate derivative (13). Enrichment analysis of three replicates was performed on an Agilent 6890/5973 GC/MS equipped with a 30-m DB-17 column (250 μm inner diameter, 25-μm film thickness) using methane NCI and collecting ions in selected ion-monitoring mode at mass-to-charge ratios 435 and 436 (13). Unenriched DNA standards were used to correct for abundance sensitivity, and enrichment of the M+1 ion above natural abundance values were used to calculate the f_AD and f_aSV.

The aSV cells isolated from the eAT of three additional male Long-Evans rats on rodent LabDiet (5% of calories from fat) and at a weight (∼400 g) comparable with that of the experimental rats were cytospun as previously described (25). The cytospun cells were incubated overnight (4°C) with primary antibodies (at 1:100) against mouse adipocyte protein (aP)2 (equivalent to human fatty acid–binding protein 4) (Dr. D. Bernlohr, Department of Biochemistry, Molecular Biology and Biophysics, University of Minnesota, Minneapolis, MN), rat CD34 (cat. no. AF4117; R&D Systems), rat CD31 (cat. no. 550300; BD Pharmingen, San Diego, CA), and rat CD11b/c (cat. no. MCA275EL, clone OX-42; AbD Serotec, Oxford, U.K.)] followed by 20-min incubations (at room temperature) with species-relevant fluorescently labeled secondary antibodies. The nuclei were stained with DAPI. Images were obtained on a confocal microscope (Leica SP5) at 200× magnification. The percentages of adipocyte progenitors (CD34⁺CD31⁻ cells), small immature adipocytes (aP2⁺ cells), endothelial cells (CD34⁺CD31⁺ cells), and monocytes/macrophages (CD11b/c⁺ cells) were determined.

Digital images from hematoxylin-eosin–stained sections were collected using a Zeiss Axioskope microscope (Carl Zeiss, Goettingen, Germany). The adipocyte area was measured in 4–5 fields using ImageJ software (http://rsbweb.nih.gov/ij/).

Sections were incubated overnight (4°C) with primary antibodies against the lipid droplet-associated protein perilipin (at 1:200, cat. no. ab3526; Abcam, Cambridge, MA) or the macrophage marker CD11b/c (at 1:100, clone OX-42; AbD Serotec), followed by an incubation with a fluorescein isothiocyanate–conjugated secondary antibody (at 1:400, cat. no. F2765; Invitrogen, Carlsbad, CA). Transferase-mediated dUTP nick-end labeling (TUNEL) staining was performed the following day per the manufacturer’s protocol (cat. no. 12156792910; Roche Applied Science, Indianapolis, IN). Images were captured using a Zeiss Axioplan 2 upright microscope and Slidebook software, version 2.0 (both from Intelligent Imaging Innovations, Denver, CO). Two to five different areas within one to three sections from 7–10 different animals for each treatment group were analyzed using ImageJ.

The assay was performed to understand whether in vitro proliferation of cultured aSV cells reflects f_aSV. The stroma-vascular cells were seeded at a density of 0.5 × 10⁴ cells/cm² in triplicate in 20% FBS in Dulbecco’s modified Eagle’s medium/F-12 medium. After 1, 4, 6, and 8 days, the nuclear content of the cultures was determined using a CyQuant NF Cell Proliferation Assay kit (Invitrogen) and converted to cell number using a standard curve with a known number of cells. The number of cell doublings at each time point was determined according to the following formula: CD = ln(Nf:Ni)/ln2, where CD is cell doublings, Nf is final number of cells, and Ni is initial (day 1) number of cells (26). The doubling time was calculated by dividing the culture time by cell doublings (26).

RNA from the eAT of eight animals per group was extracted with TRIzol (Invitrogen) and an RNeasy kit (QIAGEN, Valencia, CA). The yield and purity of RNA were determined using the Nanodrop ND1000 Spectrophotometer (Agilent Technologies, Palo Alto, CA). One microgram of RNA was treated with DNase I (Life Technologies, Carlsbad, CA) and reverse transcribed to cDNA (High Capacity cDNA Reverse Transcription kit; Applied Biosystems, Foster City, CA). The gene expression of inhibitors (Wnt 10b [27] and Pref-1 [28,29)]) and stimulators (peroxisome proliferator–activated receptor [PPAR]γ, its upstream transcriptional inducer zfp423 [30)] and its target aP2) of adipocyte differentiation was determined by RT-PCR (Table 1). The measurements were performed in triplicate on an ABI PRISM 7900 (Applied Biosystems, Branchburg, NJ) sequence detection system. Glyceraldehyde-3-phosphate dehydrogenase (GAPDH) expression was used for data normalization. The ΔC_T values (C_T value of each sample minus C_T value of GAPDH), the ΔΔC_T values (ΔC_T for each gene minus the mean ΔC_T of the same gene in LFD rats [control]), and the fold changes relative to the control (2^−ΔΔCT) were calculated.

One- or two-way ANOVA followed by post hoc Tukey tests was performed using SAS statistical software (version 9.1.3; SAS, Cary, NC) to determine differences among the groups with α set at 0.05. Group differences in body weight were tested by repeated-measures ANOVA with baseline weight as a covariate. Data are presented as means ± SE unless otherwise mentioned.

Rats on the HFD-P experienced immediate higher weight gain compared with rats on the LFD (P = 0.007 by day 4 [Fig. 1]), which remained higher thereafter. The HFD rats weighed more than LFD rats on day 10 (P = 0.03) and later. There was no statistical difference between the weights of HFD and HFD-P rats at any time during the dietary treatment.

The f_aSV in HFD-P rats was significantly higher than in the LFD and HFD groups (Fig. 2A). The difference in the f_aSV between HFD and LFD rats was of borderline statistical significance (P = 0.058). The f_AD was lower (∼50%) than the f_aSV in all groups (Fig. 2A vs. B–D). Paired comparison of f_AD (nonpurified) and f_AD (immunopurified) from 11 animals from all diet groups was 42.9 ± 2.2 vs. 35.5 ± 1.1%, respectively; P = 0.004 (Fig. 2B), representing a 17% reduction in the fractional replacement value. The f_AD (nonpurified) in the HFD-P rats was significantly greater than that in the LFD group (50 ± 3 vs. 37 ± 3%; P = 0.01), whereas the f_AD (nonpurified) in the HFD rats (46 ± 3%) was not statistically different from that in either the HFD-P or the LFD group (Fig. 2C). In contrast, the f_AD (immunopurified) in HFD-P rats was greater than that in the HFD group (39 ± 1 vs. 33 ± 1%; P = 0.04), and the f_AD in the LFD group (46 ± 3%) was not different from either HFD-fed group (Fig. 2D).

The predominant (∼70%) subfraction in the stroma-vascular cells was adipocyte progenitors, CD34⁺CD31⁻ cells (Fig. 3A). Importantly, ∼10% of the remaining CD34⁻ aSV cells were aP2⁺ cells, demonstrating a separate population of small adipocytes (Fig. 3B). Collectively, ∼80% of the aSV cells are involved in adipogenesis. There were small contaminations with endothelial cells (0.5%, CD34⁺CD31⁺ cells) and macrophages (3.3%, CD11b/c⁺ cells). Approximately 16.2% of the cells were not characterized.

The number of cell doublings of stroma-vascular cultures was not different among the groups and time points. For all time points combined, the number of cell doublings in LFD, HFD, and HFD-P rats were 1.0 ± 0.1, 1.1 ± 0.1, and 0.8 ± 0.1, respectively. The doubling time was also not different among the groups but increased significantly from day 3 to day 5 or day 7 (P = 0.02 and P = 0.0002, respectively), indicating increased contact inhibition. For all groups, the doubling time at days 3, 5, and 7 of culture was 4.3 ± 0.7, 7.1 ± 0.7, and 8.7 ± 0.8 days, respectively.

The adipocyte size (surface area) was comparable (P = 0.2) among the groups: LFD, 3,046 ± 281 μm²; HFD, 3,684 ± 268 μm²; and HFD-P, 3,253 ± 281 μm². For further characterization of the adipocytes of the LFD and HFD groups, cell death was assessed in eAT from all groups. HFD-P adipose tissue exhibited the largest amount of cell death, as evidenced by the higher TUNEL staining in the HFD-P group (282.5 ± 34.5) compared with both the LFD (183.8 ± 20.0) and HFD (146.7 ± 13.1) groups (P = 0.0002 [Fig. 4A and B]). There was no statistical difference in overall cell death between the LFD and HFD groups (Fig. 4B). For further characterization of cell death, TUNEL staining was performed in conjunction with staining for a macrophage marker (CD11b/c) and an adipocyte cell viability indicator (perilipin). As indicated by CD11b/c staining, HFD adipose tissue exhibited a significant increase in macrophage infiltration compared with that in both LFD and HFD-P groups (P < 0.0001), with LFD adipose tissue containing 33.1 ± 3.3% macrophages, HFD 51.1 ± 2.4%, and HFD-P 39.0 ± 2.8% (Fig. 4C). There was no difference in percentage of macrophages between the LFD and HFD-P groups (Fig. 4C). In addition, there was no difference in the percentage of macrophages contributing to overall counts of cell death in the adipose tissue (data not shown). However, of the macrophages present in the adipose tissue, both HFD (86.3 ± 2.2%) and HFD-P (80.7 ± 2.3%) groups showed an increase in the percentage of apoptotic macrophages, determined by dual CD11b/c and TUNEL-positive stained macrophages, compared with the LFD group (68.2 ± 4.2%; P = 0.0003 [Fig. 4D]). Regarding the adipocyte apoptosis, the perilipin intensity in the HFD-P group (22,184 ± 1,812 arbitrary units [AU]) was significantly higher compared with that in the HFD group (14,937 ± 1,578 AU; P = 0.033) but not the LFD group (16,636 ± 2,961 AU) (Fig. 5A and B). There was no difference in perilipin between the LFD and HFD groups (Fig. 5B) Thus, HFD-P treatment may have an anti-inflammatory effect by decreasing macrophage infiltration in response to HFD feeding, as well as increasing macrophage apoptosis in adipose tissue and sparing or improving adipocyte health by preventing adipocyte apoptosis.

The level of PPARγ2 mRNA in HFD-P rats was significantly lower than that in LFD and HFD groups (Table 1), indicating a ligand-receptor negative feedback relationship. The level of aP2 mRNA in rats in both HFD dietary groups was significantly higher than in the LFD rats. The expression of the remaining genes analyzed was similar among the groups.

The main goal of the current study was to refine the method previously described for the measurement of in vivo adipogenesis (13,14). Several potential improvements to this technique were tested in the context of high-fat feeding in the presence or absence of pioglitazone.

First, the stroma-vascular fraction was enriched in adipocyte progenitor cells by short-term culturing of isolated stroma-vascular cells to allow efficient and preferential attachment of cells from nonhematopoietic lineages to plastic (aSV) (24). Phenotyping of the aSV cells in rats fed rodent LabDiet (5% of calories from fat) showed that 70% of aSV cells are adipogenic progenitor (CD34⁺CD31⁻) cells (31). We did not measure the percentage of CD34⁺CD31⁻ in the stroma-vascular fraction from the experimental rats. However, the noticeable reduction of the proportion of macrophages, from 33.1% CD11b/c⁺ cells in adipose tissue of LFD-fed rats to 3.8% in the aSV cells, and the detection of a very small percentage of endothelial cells in aSV cells attest that an enrichment of aSV cells with adipocyte progenitors was achieved. Remarkably, an additional 10% of the aSV cells are aP2⁺ cells. These cells are CD34⁻ cells, indicating that they represent an independent subfraction from the adipocyte progenitor pool. We previously found aP2⁺CD68⁻ cells in human stroma-vascular fractions (25). We have also demonstrated that these cells may contain multiple small (<10 μm in diameter) lipid droplets, confirming that they are early differentiated (immature) adipocytes (25). Thus, we believe that the f_aSV reflects a component of early differentiation of adipocytes in addition to proliferation of adipocyte progenitor cells and possibly recruitment of preadipocytes to the adipocyte lineage, whereas the f_AD primarily indicates the terminal phase of differentiation.

However, a limitation to measuring the f_AD as an index of terminal differentiation is that other cell types with fast cell turnover, including hematopoietic, endothelial, and stem cells, may adhere to the floating adipocytes. In rodents, the probability for contamination, particularly with macrophages, is high owing to the considerable accumulation in the eAT (up to 50%) by high-fat feeding in our rats and in mice from other studies (32). Accordingly, a second potential improvement that we tested was to perform negative immune selection of the freshly isolated adipocytes and eliminate possible contamination with other cellular types. Importantly, the immunopurification step led to a significant reduction of the ²H-content in DNA (to 83% of the values observed in nonpurified adipocytes). This reduction was larger in the HFD-fed rats that, interestingly, had the largest macrophage infiltration as evidenced by immunohistochemistry (CD11b/c). Together, our results suggest that contamination of adipocytes with other cell types may have contributed to an overestimation of the f_AD when measured without purification. Moreover, the immunopurification step led to differences between the group profiles of the f_AD (purified) versus f_AD (nonpurified) adipocytes, suggesting that the immunopurification of adipocytes may be beneficial. Finally, in line with previous findings (15), we show that the value of f_aSV is higher than that of f_AD. This potential for higher incorporation of ²H into the DNA of aSV cells compared with adipocytes may allow for its more accurate determination.

The second aim of the study was to understand the relative importance of determining f_aSV and f_AD in reflecting induced adipogenesis. To test this, we included groups of rats fed an HFD in the presence or absence of pioglitazone. Pioglitazone is a member of the thiazolidinedione class of drugs and a ligand for the key adipogenic transcription factor PPARγ. Studies suggest that it may contribute to the adipocyte hyperplastic mechanism of adipose tissue expansion in the context of energy excess (19,33–35). Interestingly, we found higher f_aSV in the HFD-P group versus the HFD alone and LFD groups but comparable f_AD between the HFD-P and LFD groups. These results are in line with previously reported data showing that 16-day treatment of obese Zucker rats with pioglitazone results in the appearance of new adipocytes and increased fat mass, but this is not coupled with a significant increase in newly formed adipocytes as assessed by the ²H₂O incorporation in vivo (36). One possible mechanism explaining the increased f_aSV by pioglitazone is suggested by the increased aP2 gene expression, increased perilipin protein expression, and decreased levels of PPARγ mRNA in the HFD-P vs. LFD rats, indicating potential formation of new, small, and metabolically active adipocytes. That these newly formed small adipocytes remain in the stroma-vascular fractions after centrifugation is also likely, as evidenced by the presence of very small lipid-containing aP2⁺CD68⁻ cells emitting strong fluorescent signal in the stroma-vascular fraction in our previous studies (25,37), as well as in the f_aSV in this study. One limitation of the current study is that the weight of the eAT depot was not determined, precluding an analysis of the relationship between f_aSV or f_AD and the number of adipocytes. However, other studies have shown that HFD-induced weight gain is due to increased total body fat (38) and increased eAT mass (19). Another limitation of our study is that the dynamics of the proportion of small to large adipocytes was not determined.

Compared with results in LFD rats, the HFD in the absence of pioglitazone led to a more gradual increase in the additional weight gain that was accompanied by a borderline increase in f_aSV along with no change in f_AD in HFD-fed rats. This suggests stimulation of clonal expansion and early phases of differentiation as possible mechanisms. We believe that the higher gene expression of aP2 in the face of stable PPARγ in the HFD versus LFD group reflects this possibility through a mechanism of a direct stimulatory effect of fatty acids on aP2 transcription, independent of the induction of PPARγ transcriptional activity (39), along with a positive feedback loop in which aP2 facilitates the trafficking of fatty acids and their metabolites serving as ligands of PPARγ (40). Interestingly, a longitudinal monitoring of the recruitment of new inguinal adipocytes in growing genetically obese Zucker rats reveals a periodic wave of growth of inguinal adipose depot approximately every 55 days, starting with hyperplasia followed by hypertrophy (41). Strissel et al. (42) also demonstrated such waves of hyperplasia-hypertrophy-apoptosis in mice fed an HFD (60% of calories from fat) that were more prominent in the eAT compared with the inguinal depot and had a span of 20 weeks. The adipocyte hypertrophy was evident after 8 weeks of HFD feeding, explaining why it was not observed after the 3 weeks of HFD feeding in our study. Moreover, we generated evidence that apoptosis of hypertrophic adipocytes or restriction through extracellular fibrosis (data not shown) had not changed and thus had not inhibited the adipocyte enlargement. Taken together, these data support the potential hyperplastic mechanism of early phases of fat expansion induced by high-fat feeding. The supplementation of high-fat feeding with pioglitazone enhances both early phases of adipogenesis and terminal differentiation, as the f_AD in the HFD-P rats was significantly higher than in HFD rats. The observed anti-inflammatory effect of the pioglitazone through increased apoptosis of macrophages may be responsible, as it has been demonstrated that the macrophage recruitment and chronic inflammation in adipose tissue has been associated with impairment of adipocyte differentiation (43–45). Thus, our findings of similar weight gain but greater f_aSV and f_AD in the HFD-P vs. HFD group illustrate that measurement of f_aSV and f_AD facilitates the detection of small differences in the processes associated with in vivo adipogenesis. Interestingly, studies in C57BL/6 mice (C. Loe, M. Hellerstein, unpublished data) using the original technique showed increased f_AD and f_aSV with high-fat feeding, which were not further enhanced by adding pioglitazone. Differences in the methods and species may contribute to the disparities in the results.

In conclusion, f_aSV alone could be a reliable index of in vivo overall adipogenesis, while f_AD reflects primarily the adipocyte maturation rates. The measurement of f_aSV may be technically more feasible and sensitive for providing information on the effect of drugs or diets on overall in vivo adipogenesis. Further studies are needed to validate the feasibility and usefulness of the method in humans.

This work was supported by Nutrition Obesity Research Center (NORC) Grant 1P30 DK072476 (to E.R.), entitled “Nutritional Programming: Environmental and Molecular Interactions” and sponsored by the National Institute of Diabetes and Digestive and Kidney Diseases, and by the National Institutes of Health (DK060412 to E.R.). This work used the facilities of the Cell Biology and Bioimaging Core, which are supported in part by a Centers of Biomedical Research Excellence grant (NIH P20-RR021945) and NORC center grants from the National Institutes of Health. M.K.H. has ownership interest in KineMed, Inc. No other potential conflicts of interest relevant to this article were reported.

Y.D.T. planned the study design, performed the experiment, wrote the first draft of the manuscript, and is the guarantor of the study. M.F. provided guidance for the DNA extraction, performed the gas chromatography–mass spectrometry analysis, and provided editorial assistance. P.M.R. performed RT-PCR and provided editorial assistance. J.D.C. performed immunocytochemistry and provided editorial assistance. T.M.H. performed immunohistochemistry and immunostaining, analyzed those data, described those techniques, and provided editorial assistance. J.Y. contributed to discussion and provided editorial assistance. M.K.H. maintained quality assurance of the gas chromatography–mass spectrometry methods, reviewed data, contributed to discussion, and provided editorial assistance. E.R. reviewed the proposal, provided funding, contributed to discussions, provided guidance for the development of the manuscript, and provided editorial assistance.

Parts of this work were presented in abstract form at the Obesity Society 28th Annual Scientific Meeting, San Diego, California, 8–12 October 2010.

The authors thank Shantele Thomas (now at Sanford/Burnham Institute, Orlando, Florida) for technical help.