Endocrinology and Metabolism

Increased TNFα and CCAAT/enhancer-binding protein homologous protein with aging predispose preadipocytes to resist adipogenesis

Tamara Tchkonia, Tamar Pirtskhalava, Thomas Thomou, Mark J. Cartwright, Barton Wise, Iordanes Karagiannides, Alexander Shpilman, Timothy L. Lash, J. David Becherer, James L. Kirkland


Fat depot sizes peak in middle age but decrease by advanced old age. This phenomenon is associated with ectopic fat deposition, decreased adipocyte size, impaired differentiation of preadipocytes into fat cells, decreased adipogenic transcription factor expression, and increased fat tissue inflammatory cytokine generation. To define the mechanisms contributing to impaired adipogenesis with aging, we examined the release of TNFα, which inhibits adipogenesis, and the expression of CCAAT/enhancer-binding protein (C/EBP) homologous protein (CHOP), which blocks activity of adipogenic C/EBP family members, in preadipocytes cultured from young, middle-aged, and old rats. Medium conditioned by fat tissue, as well as preadipocytes, from old rats impeded lipid accumulation by preadipocytes from young animals. More TNFα was released by preadipocytes from old than young rats. Differences in TNFα-converting enzyme, TNFα degradation, or the presence of macrophages in cultures were not responsible. TNFα induced rat preadipocyte CHOP expression. CHOP was higher in undifferentiated preadipocytes from old than younger animals. Overexpression of CHOP in young rat preadipocytes inhibited lipid accumulation. TNFα short interference RNA reduced CHOP and partially restored lipid accumulation in old rat preadipocytes. CHOP normally increases during late differentiation, potentially modulating the process. This late increase in CHOP was not affected substantially by aging: CHOP was similar in differentiating preadipocytes and fat tissue from old and young animals. Hypoglycemia, which normally causes an adaptive increase in CHOP, was less effective in inducing CHOP in preadipocytes from old than younger animals. Thus increased TNFα release by undifferentiated preadipocytes with elevated basal CHOP contributes to impaired adipogenesis with aging.

  • adipocytes
  • CCAAT/enhancer-binding protein-ζ
  • growth arrest and DNA damage 153
  • differentiation
  • stress response
  • hypoglycemia

fat depots enlarge through middle age and then become smaller in advanced old age (34, 38). Declining depot size in old age is associated with ectopic fat deposition in liver, muscle, marrow, and other sites, decreased fat cell size and function, and increased fat tissue TNFα (51). Fat cells turn over, with new fat cells developing from preadipocytes, which are present in fat depots throughout life (3, 4, 37, 59). Aging is associated with reduced preadipocyte capacity for differentiation into fat cells and decreased expression of the key adipogenic transcription factors CCAAT/enhancer-binding protein-α (C/EBPα) and peroxisome proliferator-activated receptor-γ (PPARγ) (24, 30, 35, 38, 64). TNFα can impede adipogenesis through several mechanisms: causing impaired insulin/insulin-like growth factor I responsiveness, increasing CUG triplet repeat-binding protein activity [leading to increased production of C/EBPβ-liver-enriched inhibitory peptide (LIP), an antiadipogenic factor (31)], impairing PPARγ expression through NF-κB-mediated mechanisms (8, 71), and activating the cellular stress response. Preadipocytes can mount an innate immune response, producing inflammatory cytokines and chemokines in response to lipopolysaccharide (12). Thus, in addition to macrophages, the preadipocytes that constitute 15–50% of cells in fat tissue are capable of releasing TNFα.

Together with C/EBPα, other C/EBP family members regulate adipogenesis, including C/EBPβ, C/EBPδ, and C/EBP homologous protein 10 (CHOP), also known as C/EBPζ, DNA damage inducible transcript 3 (Ddit3), and growth arrest and DNA damage 153 (GADD153). Adipogenic C/EBP family members bind as homo- or heterodimers to C/EBP sites in differentiation-dependent promoters (45, 57). An increase in C/EBPβ expression on induction of differentiation promotes PPARγ and C/EBPα expression, driving adipogenesis (11, 13, 70, 72). Although increases in C/EBPα and PPARγ expression during adipogenesis are blunted with aging, the earlier increase in C/EBPβ mRNA is not (30). This implies that mechanisms that contribute to impaired differentiation with aging act at a point between the increase in C/EBPβ transcription and subsequent increases in C/EBPα and PPARγ.

One regulator that can operate at this point is CHOP, a cellular stress-response protein (2, 66). When CHOP forms heterodimers with adipogenic C/EBP family members, it impairs their ability to activate preadipocyte differentiation-dependent promoters. CHOP blocks promotion of adipogenesis by full-length C/EBPβ in the 3T3-L1 murine preadipocyte cell line (56). The tyrosine kinase inhibitor genistein increases CHOP expression, impeding C/EBPβ transactivating activity, expression of C/EBPα and PPARγ, and adipogenesis in these cells (22). Similarly, the calpain protease inhibitor N-acetyl-Leu-Leu-norleucinal, which increases CHOP, blocks adipogenesis in 3T3-L1 cells (61). Very little is known about CHOP in primary preadipocytes, which differ in many respects from cell lines. For example, a round of replication, an event involving changes in CHOP abundance, is required in 3T3-L1 cells before adipogenesis proceeds (61). However, replication is not required before differentiation can proceed in primary preadipocytes (15).

TNFα increases in fat tissue with aging, macrophages and preadipocytes can produce TNFα, TNFα induces cellular stress responses, CHOP is increased by stress responses, and impaired adipogenesis with aging comes at a point in the transcription factor cascade at which CHOP intercedes. Therefore, we tested the hypothesis that TNFα and CHOP increase with aging in cultured undifferentiated preadipocytes, predisposing to impaired adipogenesis.


Preadipocyte cultures.

Specific pathogen-free male Brown Norway rats [3 (young), 15–17 (middle-aged), and 24–31 (old) mo of age] were obtained from the Harlan Sprague Dawley (Indianapolis, IN) colony maintained under contract by the National Institute on Aging (median survival 32 mo, maximum survival 43 mo). Animals were given NIH 31 chow and water ad libitum and were acclimatized in a 12:12-h light-dark constant-environment facility separately from other animals for 7 days before experiments. All studies were approved by the Boston University Institutional Animal Care and Use Committee. After euthanasia with CO2, epididymal or perirenal fat depots were removed under sterile conditions. Fat tissue was minced into fragments, digested in 1 mg of collagenase per milliliter of Hanks' balanced salt solution for 60 min at 37°C, and filtered through a 100-μm nylon mesh (35–37, 67). After centrifugation of the digests, the pellets were resuspended in a basal medium (α-MEM containing 10% FBS and antibiotics) and plated at ∼4 × 104 cells/cm2. After 12 h, a period before replication occurs (14), adherent preadipocytes were washed, trypsinized, and replated at a density of 4 × 104 cells/cm2. We found that replating 1) reduces mesothelial cell and macrophage contamination and 2) results in accurate plating densities [plating density affects capacity of preadipocytes to differentiate (68)]. Using these methods, we demonstrated that preadipocyte recoveries are similar among age groups and that our isolation procedures result in >90%-pure populations of preadipocytes, irrespective of animal age (35). Markers of macrophages [F4/80 (emr1), scya4, cd11β, and cd45 (ptprc)], endothelial cells [vegfr (flk1), vegfr2 (kdr), cd46 (mcp), and von Willebrand], and multipotent mesenchymal progenitors (cd34, connexin 43, sox2, and smad4) indicated that age-related shifts in abundance of these cell types in preadipocyte cultures did not account for our results. Brown fat preadipocytes are not present in epididymal cultures, since uncoupling protein 1 is not evident in these cultures after treatment with isoproterenol but is observed in interscapular preadipocyte positive controls (36, 41). To induce differentiation, confluent preadipocyte cultures were exposed to a serum-free differentiation-inducing medium (DM) containing 5 μg/ml insulin, 10 μg/ml transferrin, and 200 pM triiodothyronine in 1:1 DMEM-Ham's F-12 (20). Media were changed every 2 days.

Coculture and conditioned medium.

For fat tissue coculture studies, 500-mg perirenal fat tissue aliquots from rats of different ages isolated in parallel were placed in Transwell inserts with 0.24-μm pores. The inserts were placed in wells in six-well plates that contained confluent perirenal preadipocytes from 3-mo-old rats. DM was added to the wells containing the inserts for 7 days. For preadipocyte conditioned medium studies, confluent undifferentiated preadipocytes from 3- and 24-mo-old rats cultured in parallel were washed twice and incubated with Krebs-Ringer phosphate (KRP) buffer for 24 h. The medium was removed and centrifuged at 1,000 g for 10 min. The conditioned medium was added to confluent perirenal preadipocytes that had been treated with DM for 48 h for a further 24 h.

Western immunoblot analyses.

Cultures were washed with PBS, scraped into RIPA buffer (73), and subjected to SDS-PAGE (44). The gels were stained for visualization of banding, or the protein was transferred to Immobilon P polyvinylidene difluoride membranes for probing. Blotting paper was blocked for 1 h at room temperature in Tris-buffered saline containing 5% milk, 0.5% BSA, and 0.1% Tween 20. Blots were incubated for 1 h at room temperature with primary antibody {catalog nos. 793 (CHOP) and sc-6416 [TNFα-converting enzyme (TACE)], Santa Cruz Biotechnology, Santa Cruz, CA} (49) and for 30 min at room temperature with secondary antibody conjugated to horseradish peroxidase. Binding of the horseradish peroxidase-conjugated secondary antibody was visualized by chemiluminescense. Loading of 40 μg of protein in each lane gave results in the linear range for detection of CHOP. Total protein contents (pg/cell) were 337 ± 32, 347 ± 35, and 361 ± 39 in undifferentiated epididymal preadipocytes from 3-, 17-, and 30-mo-old rats, respectively (n = 16 in each group). Densitometry data are expressed as a function of cell number. Data are expressed as a percentage of mean intensities within each replication of experiments and shown as means ± SE. For example, if n = 6, the result in young preadipocytes is the mean ± SE of replications 1–6, where the value for replication 1 is young1/(young1 + middle-aged1 + old1) expressed as a percentage after correction for cell number.


Confluent, undifferentiated preadipocytes were held at confluence for 48 h and then washed twice with PBS. KRP buffer was added for a further 48 h (3 ml per T-25 flask). TNFα was assayed by ELISA (catalog no. KRC3014, BioSource International).

TNFα degradation.

Preadipocytes that had been confluent for 48 h were washed twice with PBS. 125I-labeled TNFα (0.16 μCi/ml) in KRP buffer containing 4% BSA was added for 24 h (3 ml per T-25 flask). After 24 and 48 h, 50-μl aliquots were run on 12% polyacrylamide gels, with 1 μl (0.002 μCi) of undiluted stock run for controls. After electrophoresis, gels were exposed to X-ray film for 24 h at −80°C. Cells were harvested, and aliquots were assayed for DNA (43).

RNA analysis.

Total RNA was extracted from preadipocytes using the guanidinium thiocyanate-phenol method (10). RNA integrity was verified using 1% formaldehyde-containing denaturing agarose gels. Total RNA (1 μg) was reverse transcribed with Moloney murine leukemia virus reverse transcriptase (Invitrogen, Carlsbad, CA) and random oligonucleotides (Amersham, Piscataway, NJ). mRNA was measured by real-time (see below) or relative quantitative RT-PCR, in which target genes were coamplified with an internal 18S rRNA control sequence (63). The quantitative nature of this approach was confirmed by measurement of CHOP mRNA in serially diluted samples. RNA preparations were checked for DNA contamination by amplification of control aliquots that had not been reverse transcribed. The following primers were used for CHOP: 5′ GCGGCTCAACGAGGAAATCG 3′ (sense) and 5′ AAGCCCCTCTCTCCTTTGGTCTAC 3′ (antisense; accession no. U30186). Real-time PCR were performed using an ABI 7500 real-time PCR system and software (Applied Biosystems, Foster City, CA). Amplification was performed in a final volume of 20 μl containing 1 μl of cDNA from the reverse-transcribed reaction mixture. CHOP and TNFα mRNA expression was analyzed during the exponential phase of amplification using the TaqMan fluorogenic 5′-nuclease PCR assay (Applied Biosystems) for each target gene (23) [Applied Biosystems Assay catalog nos. Rn00492098 (for CHOP) and Rn00562055 (for TNFα)]. 18S rRNA was used as an endogenous control and was detected using a dual-labeled fluorogenic (5′VIC/3′TAMRA) probe (catalog no. 4310893E, Applied Biosystems). Data are expressed as ratio of target mRNA to 18S rRNA.

DNA transfection.

Preadipocytes were cotransfected using the calcium phosphate precipitation method (69) with plasmid mCHOP10.pCDNA1 (provided by D. Ron) and pMSV-β-Gal (Invitrogen) at a molar ratio of 5:1 (transcription factor-reporter construct) to ensure that the pMSV-β-Gal-transfected cells detected by staining for β-galactosidase (β-Gal) would also express CHOP. Approximately 15% of transfected cells were positive for β-Gal. Mock-control cells were transfected with a plasmid, PVG 9607(GAPDH) (42), that expresses GAPDH and had no effect on lipid accumulation compared with untransfected cells, also at a 5:1 molar ratio. After 48 h of culture in DM, cells were fixed in 3% formaldehyde and stained with the chromatographic β-Gal substrate X-Gal for identification of transfected cells (58). Observers unaware of cell origin assessed extent of lipid accumulation by examining transfected cells for the presence of doubly refractile lipid inclusions by low-power phase-contrast microscopy (31, 35, 67). The lipid nature of these inclusions was confirmed by staining with oil red O (62, 67). In previous experiments, we demonstrated that inclusions of this type appear only in differentiating preadipocytes, and not in other cell types (e.g., lung fibroblasts), in the DM we used. The proportion of such cells correlates with the adipogenesis markers glycerol-3-phosphate dehydrogenase activity and the fatty acid-binding protein aP2 (fatty acid-binding protein 4) expression (35, 62, 67). This assay permits delineation of lipid accumulation in the individual cells that are transfected as evident by β-Gal staining. The proportion of transfected cells containing such inclusions was compared with that of parallel mock-control cultures.

Short interference RNA-mediated TNFα knockdown.

TNFα short interference RNA (siRNA) was synthesized (catalog no. M-100312-00, Dharmacon, Lafayette, CO). TNFα siRNA was labeled with a Cy3 labeling kit (Mirus, Madison, WI). The labeled siRNA was transfected into epididymal rat preadipocytes at 75% confluence with the RNAifect transfection reagent (Qiagen, Valencia, CA). As a control, transfection with nonsilencing, fluorescein-labeled siRNA was carried out under the same conditions (catalog no. 1022079, Qiagen). After 6 h, the cells were washed with PBS, and α-MEM containing 10% FBS was added. Undifferentiated cells were harvested after 12 h for determination of CHOP and TNFα mRNA. For studies of adipogenesis, differentiation was initiated 12 h after the siRNA transfection for up to 8 days. Medium was changed every other day.

Statistical analysis.

Values are means ± SE. Each n represents sets of cultures from separate groups of animals. Significance was determined by t-tests or single-factor ANOVA, as appropriate (29, 32). Appropriate post hoc tests were used to determine differences among groups. In transfection and siRNA experiments, numbers of transfectants containing lipid inclusions were compared with control cells by logistic regression analysis, with P values determined from log-likelihood ratios (39). P < 0.05 was considered significant.


Fat tissue and preadipocytes from old rats inhibit lipid accumulation by preadipocytes from young animals.

To test the hypothesis that factors released by fat tissue from old rats impede lipid accumulation by preadipocytes from young animals, fat tissue fragments of equal weight from 3-, 18-, and 31-mo-old rats were cocultured with target preadipocytes from young animals. DM was added. Less lipid was accumulated by the target preadipocytes from young animals by exposure to fat tissue from old animals than exposure to fat tissue from young animals (Fig. 1A). To test whether factors released by preadipocytes from old animals contribute to this inhibition of adipogenesis, conditioned medium was prepared by incubation of preadipocytes from young or old animals in Krebs buffer for 24 h. Treatment of differentiating young rat preadipocytes with the conditioned medium from old rat preadipocytes inhibited lipid accumulation to a greater extent than treatment with conditioned medium from young rat preadipocytes (Fig. 1B). Preadipocytes from old animals also impeded adipogenesis in target preadipocytes from young rats in coculture experiments (not shown). In these experiments, Transwell inserts with confluent preadipocytes from young or old animals were placed into wells containing target cells from young animals and then treated with DM for 48 h (n = 5).

Fig. 1.

Fat tissue and preadipocytes from old rats inhibit adipogenesis. A: representative photomicrograph of target preadipocytes exposed to 3-, 18-, or 31-mo-old rat fat for 7 days. Equal aliquots of perirenal fat tissue isolated from 3-, 18-, and 31-mo-old rats were placed in Transwell inserts, which were placed into 6-well plates containing confluent target preadipocytes from 3-mo-old rats, and differentiation-inducing medium was added. Fat tissue from 31-mo-old rats resulted in less target preadipocyte lipid accumulation than fat from 3-mo-old rats (n = 3, P < 0.05). B: conditioned medium prepared from preadipocytes isolated from old animals impedes lipid accumulation by preadipocytes from young animals. Krebs-Ringer phosphate buffer was incubated with equal numbers of undifferentiated preadipocytes from 3- and 24-mo-old rats for 24 h, removed, and added to differentiated rat preadipocytes. Scale bar, 20 μm. Photomicrographs represent results from 5 experiments.

Preadipocyte TNFα release increases with aging.

TNFα was assayed in medium in which confluent, undifferentiated perirenal and epididymal preadipocytes from 3- and 24-mo-old rats had been cultured for 48 h (Fig. 2A). Cellular TNFα production by preadipocytes from old animals was greater than that by preadipocytes from young animals (n = 22, P < 0.0001 for effect of age, by ANOVA). More TNFα was released by epididymal than by perirenal preadipocytes (n = 12, P < 0.001 for effect of depot, by ANOVA), consistent with the greater capacity for lipid accumulation and differentiation-dependent marker expression by perirenal than epididymal cells (14, 35, 36). By 24 h, conditioned medium TNFα concentrations were 471 ± 75 pg/ml from 24-mo-old rat epididymal preadipocytes and 158 ± 33 pg/ml from 3-mo-old cells (n = 20, P < 0.0005, by t test). The increased TNFα in medium from preadipocytes cultured from old animals is not likely a consequence of increased release of membrane-bound TNFα by TACE, since TACE protein was similar in preadipocytes from old and young animals and was not affected appreciably by depot origin or adipogenesis (Fig. 2B; P = 0.96, by ANOVA). TNFα is labile. To test whether decreased degradation accounts for increased TNFα in medium with aging, labeled TNFα was added to medium in which preadipocytes from young, middle-aged, or old animals were cultured. Degradation was similar across the age groups (Fig. 2C). Thus preadipocyte TNFα release is among the mechanisms potentially contributing to impaired adipogenesis with aging in preadipocyte primary cultures.

Fig. 2.

Preadipocyte TNFα release increases with aging. A: confluent epididymal and perirenal preadipocytes from 3- and 24-mo-old rats were incubated for 48 h in Krebs-Ringer phosphate buffer. TNFα, which was assayed by ELISA, increased with aging (P < 0.0001) and was higher in epididymal than perirenal preadipocytes (P < 0.001). B: TNFα-converting enzyme (TACE) remains similar (or tends to decline) with aging, indicating that increased TACE does not account for increased TNFα with aging. TACE was assayed in undifferentiated (Undiff) and differentiated (Diff) epididymal or perirenal preadipocytes from young (3-mo-old) and old (24-mo-old) rats in parallel. Top: representative Western blot. Bottom: densitometric analyses. Values are means ± SE (n = 6, P = 0.96 for effect of age or depot). C: TNFα degradation is similar in the presence of preadipocytes from old and young rats. TNFα degradation was assayed by addition of labeled TNFα to medium in which epididymal (E) and perirenal (P) preadipocytes from 3-, 17-, and 24-mo confluent preadipocytes were cultured. Gel represents results from 6 experiments.

TNFα increases primary rat preadipocyte CHOP.

To test whether TNFα induces primary preadipocyte CHOP, undifferentiated, confluent epididymal preadipocytes from young rats were exposed for 24 h to rat TNFα at 500 pM, which is close to concentrations in conditioned medium from preadipocytes cultured from old animals. TNFα increased cellular CHOP (Fig. 3; n = 3, P < 0.05, by t-test).

Fig. 3.

TNFα increases CCAAT/enhancer-binding protein homologous protein (CHOP) in primary rat preadipocytes. Undifferentiated epididymal preadipocytes from 3-mo-old rats were treated with 500 pM TNFα for 24 h after confluence. Top: representative Western blot analysis of CHOP. Bottom: densitometric analyses. Values are means ± SE; n = 3 sets of animals. *P < 0.05.

Undifferentiated preadipocyte CHOP increases with aging.

Since TNFα increases CHOP and TNFα production increases with aging, we tested the hypothesis that preadipocyte CHOP increases with aging. CHOP levels were measured in undifferentiated epididymal preadipocytes cultured in parallel from young, middle-aged, and old rats 48 h after reaching confluence (Fig. 4). Preadipocytes from the old rats expressed more CHOP than cells from middle-aged or young rats (n = 6, P < 0.05, old vs. young or middle aged, by Duncan's multiple range test), as occurred in preadipocytes from young rats exposed to TNFα.

Fig. 4.

CHOP is higher in undifferentiated preadipocytes from old than younger animals. CHOP levels were measured in confluent preadipocytes cultured in parallel from young (3-mo-old), middle-aged (15-mo-old), and old (27-mo-old) rats that had been maintained at confluence for 48 h. Top: Western immunoblot representative of 6 separate experiments. Bottom: densitometric analyses of CHOP abundance. Values (means ± SE) represent percentage of summed optical densities (OD) across age groups within each experiment. *P < 0.05 vs. 3- or 15-mo-old rats.

CHOP impedes primary preadipocyte lipid accumulation.

To establish whether, as in preadipocyte cell lines, CHOP blocks adipogenesis in euploid, primary cultured preadipocytes, we overexpressed CHOP in perirenal preadipocytes isolated from young animals (Fig. 5). CHOP and β-Gal expression vectors were cotransfected at a 5:1 molar ratio, so that cells transfected with CHOP would be detected by β-Gal staining (30, 62). Control cells were transfected with a GAPDH expression vector, also with the β-Gal expression vector at a 5:1 molar ratio. The transfected cells were treated with DM for 48 h. The proportions of β-Gal-expressing cells that contained the doubly refractile lipid inclusions visible by low-power phase-contrast microscopy [characteristic of differentiating preadipocytes (35)] were determined in CHOP- and control-transfected cultures by observers who were unaware of culture treatments. CHOP transfection prevented lipid accumulation. More preadipocytes transfected with the control vector accumulated lipid than cells transfected with the CHOP construct (odds ratio = 2.5, 95% confidence interval = 1.8–3.5, n = 4 rats, 100 CHOP- and 100 mock-transfected cells/rat, 33 ± 7% of mock-transfected cells acquired doubly refractile lipid inclusions vs. 17 ± 6% of CHOP-transfected cells, P < 0.05, by t-test). Thus lipid accumulation in primary rat preadipocytes is inhibited by overexpression of CHOP, as occurs in cell lines.

Fig. 5.

CHOP overexpression blocks adipogenesis in preadipocytes from young rats. Perirenal preadipocytes from 3-mo-old rats were cotransfected with CHOP and β-galactosidase (β-Gal) expression vectors at a 5:1 molar ratio to ensure that cells staining for β-Gal were transfected with CHOP. CHOP expression reduced the proportion of cells that contained lipid droplets large enough to be doubly refractile by low-power phase-contrast microscopy (n = 4, P < 0.05). Scale bar, 20 μm.

Reducing TNFα decreases old rat preadipocyte CHOP and increases lipid accumulation.

Lowering TNFα mRNA by treating preadipocytes from old animals with TNFα siRNA decreased CHOP to <25% of levels in control cultures (Fig. 6A; n = 5, P < 0.005 for decreased TNFα and P < 0.0005 for decreased CHOP, by t-test). Furthermore, introducing fresh culture medium decreased CHOP, as occurs in 3T3-L1 cells (6, 25), consistent with a contribution of gain or loss of autocrine factors (e.g., TNFα), metabolites, or nutrients to increased CHOP in preadipocytes from old rats. Epididymal primary preadipocytes from young animals were kept in medium for 48 h after confluence without medium change (control cultures), or medium was changed in parallel cultures every 2 h three times before protein isolation. Changing medium reduced CHOP abundance to 56.7 ± 9.5% of that in control cultures (n = 4, P < 0.05, by t-test), with a similar decrease in old rat preadipocyte CHOP after the medium changes (59.1 ± 10.3%). Additionally, TNFα siRNA treatment enhanced lipid accumulation in differentiating perirenal preadipocytes from old rats (Fig. 6B). The proportion of cells containing doubly refractile lipid inclusions was 2.15 times larger in TNFα siRNA-treated cells than in cells treated with a control nucleotide (95% confidence interval = 1.26–3.67, n = 4 animals; logistic regression analysis). Thus TNFα increases primary preadipocyte CHOP and reduction of preadipocyte TNFα reduces CHOP in undifferentiated preadipocytes and augments subsequent lipid accumulation.

Fig. 6.

Inhibition of preadipocyte TNFα production blunts the increase in CHOP with aging and partially restores adipogenesis. A: reducing TNFα mRNA lowers CHOP in preadipocytes from old rats. Epididymal undifferentiated preadipocytes from 31-mo-old rats were incubated with TNFα short interference RNA (siRNA) or a control scrambled nucleotide (Cont) for 18 h. TNFα siRNA decreased TNFα (n = 7, *P < 0.00005, by t-test) and CHOP mRNA (**P < 0.0005), indicating that increased preadipocyte CHOP with aging is due, in part, to increased preadipocyte TNFα. B: lowering TNFα production enhances lipid accumulation by preadipocytes from old rats. Preadipocytes from 31-mo-old rats were treated with TNFα siRNA (a) or scrambled nucleotide (b) as described in A. More (2.15-fold) cells transfected with TNFα siRNA accumulated doubly refractile lipid inclusions visible at low power than cells transfected with scrambled nucleotide (95% confidence interval = 1.8–3.5, n = 4). Scale bar, 20 μm.

Increase in CHOP late during differentiation does not change substantially with aging.

In 3T3-L1 cells, CHOP increases after the differentiation-dependent increase in C/EBPα (6). This late differentiation-dependent increase changed little with aging (Fig. 7A). CHOP levels 48 h after initiation of differentiation in preadipocytes cultured in parallel from 3- and 27-mo-old rats were similar (Fig. 7B; n = 7, P = 0.37, by t-test). The differentiation-dependent increase in CHOP was reflected in fat tissue, with CHOP mRNA abundance being similar in freshly isolated whole fat tissue from the inguinal depots of young and old rats (Fig. 7C; n = 4, P = 0.9, by t-test). Thus CHOP is increased in undifferentiated preadipocytes with aging, predisposing to impaired adipogenesis, but is not increased in differentiated fat cells.

Fig. 7.

Profile of CHOP expression during differentiation is not substantially affected by aging. A: CHOP levels were assayed in perirenal preadipocytes from 3- and 24-mo-old rats at confluence (C) or 8–72 h after addition of differentiation-inducing medium. Although CHOP expression was more readily detectable before and early during differentiation in preadipocytes from old than young rats, timing of the subsequent differentiation-dependent increase was similar. Results are representative of 5 experiments. B: by 48 h after initiation of differentiation, CHOP abundance is similar in preadipocytes from old (27M) and young (3M) animals (P = 0.37 for effect of aging, n = 7). CHOP abundance was directly compared in samples from young and old animals run on the same gels. Top: representative Western immunoblot. Bottom: densitometric analyses of CHOP abundance. Values are means ± SE. C: CHOP is similar in inguinal fat tissue from 3-mo-old (3M) and 24-mo-old (24M) rats (P = 0.9 for effect of aging, n = 4). Top: representative gel. Bottom: densitometric analyses of CHOP mRNA as a function of the 18S rRNA (18S) PCR internal control. Values are means ± SE.

Primary preadipocyte CHOP increases in response to hypoglycemia.

Primary preadipocyte CHOP was sensitive to low glucose (Fig. 8), as are 3T3-L1 cells (6). In 3-mo-old undifferentiated epididymal preadipocytes, CHOP was more abundant after 24 h of exposure to 0.5 than 6 mM glucose (23 ± 2 vs. 6 ± 3 arbitrary units, n = 3, P < 0.05, by t-test). In preadipocytes from 24-mo-old animals, low glucose was not as effective at inducing CHOP (20 ± 2 and 12 ± 3 arbitrary units after exposure to 0.5 and 6 mM glucose, respectively, n = 3, P = 0.14, by t-test). The blunted response to hypoglycemia with aging may be related to higher baseline CHOP expression (in the 6 mM glucose-treated cells) in the preadipocytes from the 24- than 3-mo-old rats (12 ± 3 vs. 6 ± 3 arbitrary units, n = 3, P < 0.05, by t-test).

Fig. 8.

Representative Western immunoblot showing that induction of CHOP by hypoglycemia in preadipocytes is blunted with aging. Confluent undifferentiated epididymal preadipocytes from 3- and 24-mo-old rats were exposed to basal medium containing 0.5–6 mM glucose for 24 h. Hypoglycemia induced CHOP in preadipocytes from young animals (P < 0.05, n = 3) but was less effective in cells from old rats (P = 0.14). Basal CHOP was higher in preadipocytes from old than young animals in 6 mM (physiological) glucose (P < 0.05).


Aging is associated with increased fat tissue TNFα, impaired preadipocyte differentiation into fat cells, decreased adipogenic transcription factor expression in fat tissue and cultured preadipocytes, and ectopic lipid deposition with systemic lipotoxicity and increased predilection for metabolic disease (7, 21, 24, 33, 38, 50). TNFα inhibits adipogenesis through multiple mechanisms, including generation of antiadipogenic regulators, such as CHOP (as shown here) and C/EBPβ-LIP (31). Other inflammatory mediators, including IL-6 and nitric oxide (through inducible nitric oxide synthase), as well as chemokines that attract macrophages and other inflammatory cells, are produced by preadipocytes (12). These inflammatory mediators and chemokines could also impair adipogenesis and increase with aging.

Most fat tissue TNFα is likely produced by macrophages. Fat tissue macrophage abundance increases in subcutaneous, but probably not visceral, fat depots with aging (28). Our data suggest that preadipocytes, a cell type with expression profiles closer macrophages than to fat cells (9), also produce TNFα. TNFα release per preadipocyte increases over threefold with aging. Preadipocyte numbers per fat depot also increase with aging (37), magnifying their potential contribution to fat tissue TNFα production. Under culture conditions in which macrophages were absent, preadipocytes from old animals produced sufficient TNFα to inhibit their own adipogenesis, as evident from the TNFα siRNA experiments. Thus preadipocytes could make a more substantial contribution to generation by fat tissue, which acts in an autologous fashion, than has previously been appreciated, especially in old age. Furthermore, preadipocyte TNFα release is depot dependent, consistent with the contention that age-related changes in fat tissue function occur at different rates in different fat depots (7). In combination with age-related changes in diet, activity, hormonal milieu, fat tissue macrophages, and other processes extrinsic to preadipocytes or fat cells, inherent processes, some of which act through increased inflammatory mediator generation, appear to contribute to fat tissue dysfunction with aging.

TNFα at concentrations close to those in conditioned medium from old rat preadipocyte cultures increased CHOP expression. This is different from effects of TNFα on CHOP in brown fat preadipocytes, in which CHOP expression is very low and not increased detectably by TNFα (53). The extent of the CHOP increase in white fat preadipocytes in response to these concentrations of TNFα was similar to the increase in basal CHOP expression in old compared with young rat preadipocytes. CHOP blocks adipogenesis in primary rat preadipocytes, as found in the murine 3T3-L1 cell line (2). CHOP expression is increased in undifferentiated preadipocytes from old animals, in which the capacity to differentiate is impaired. Lipid accumulation was partially restored by reducing CHOP through introduction of TNFα siRNA into preadipocytes from old animals before induction of adipogenesis. Decreasing preadipocyte TNFα likely blunts other potentially antiadipogenic processes as well. These findings support the hypothesis that increased CHOP expression, partly in response to increased autologous TNFα, contributes to the predisposition of undifferentiated preadipocytes from old animals to resist adipogenesis.

CHOP inhibits differentiation by forming heterodimers with adipogenic C/EBP family members, particularly C/EBPβ (1, 56). The truncated, alternatively translated C/EBPβ-LIP isoform also forms heterodimers with adipogenic C/EBP family members, blocking differentiation (31, 65). Activity of CUG-binding protein, which promotes C/EBPβ-LIP translation, increases in undifferentiated preadipocytes with aging, contributing to increased C/EBPβ-LIP during adipogenesis in preadipocytes from old animals (30, 31). Similar to CHOP, CUGBP is stress responsive and increases after treatment with TNFα. Thus CHOP and C/EBPβ-LIP are increased with aging, inhibiting adipogenic C/EBP family members. Furthermore, aging is associated with decreased C/EBPδ induction during early adipogenesis (30). C/EBPβ/δ heterodimers are especially effective in inducing PPARγ (70). Hence, redundant pathways likely predispose preadipocytes to resist adipogenesis with aging. This redundancy may account for the partial, rather than full, restoration of function by individual interventions, such as blocking TNFα production.

After an initial decline, CHOP expression increases as primary preadipocyte differentiation proceeds, as occurs in preadipocyte cell lines (6, 61). This is not substantially affected by aging, nor is subsequent expression of CHOP in fat tissue. The late differentiation-dependent increase in CHOP expression may be involved in regulating progression of adipogenesis but is not required for adipogenesis to proceed in 3T3-L1 cells (6, 56). Perhaps effects of absence of CHOP on fat tissue function only become apparent under certain conditions, for example, after activation of stress responses, during tissue remodeling, or with hypoglycemia. One could argue that using interventions to reduce CHOP in preadipocytes from old rats would further test the prediction that lowering CHOP would restore adipogenesis. However, the lack of an effect of decreasing the late differentiation-dependent increase in CHOP on adipogenesis, the impact of C/EBPβ-LIP and other redundant antiadipogenic pathways, and the biphasic nature of CHOP expression during normal adipogenesis would make such an experiment difficult to interpret.

We found that primary rat preadipocyte CHOP expression is stress responsive (TNFα, hypoglycemia, and infrequent medium changes), as in 3T3- L1 cells (6, 17, 66). In preadipocytes cultured from old animals, basal expression of CHOP is increased. Increased basal CHOP has also been reported in primary hepatocytes cultured from old compared with young rats (26, 27, 48), suggesting that this may be a general phenomenon. Elevated CHOP predisposes to apoptosis in response to further stress in primary hepatocytes from old animals (48) and in other cell types (5, 16, 19, 40, 74). Consistent with this increased predilection to apoptosis, we found increased susceptibility to fatty acid-induced apoptosis in preadipocytes from old rats (21).

Although basal CHOP expression is increased in preadipocytes with aging, further adaptive increases in response to glucose deprivation are blunted, despite lack of change in capacity of preadipocytes to utilize exogenous glucose with aging (21). An analogous increase in basal expression of various stress-response proteins, with blunting of further adaptive stress-induced increases, has been observed with aging in other cell types (18). Basal NF-κB DNA-binding activity increases with aging in spleen and liver cells (54, 60), whereas acute activation of its expression by extracellular signals in T cells is blunted (52). Similarly, basal expression of certain heat shock proteins increases with aging (46, 47), but heat shock protein induction in response to acute stress is attenuated (55). Thus there may be a general increase in basal activity of stress-responsive pathways with aging but a diminished ability to respond vigorously to acute insults. Increased basal CHOP expression, with reduced preadipocyte capacity for adipogenesis, could impede fat tissue expansion in response to nutrients, regeneration of fat tissue, and adaptive responses to starvation or hypoglycemia. Our findings support the contention that basal stress-response pathway activation, combined with reduced capacity of these pathways to respond to additional stress, may contribute to reduced plasticity and resilience of tissues with aging.


This work was supported by National Institutes of Health Grants AG-13925 and DK-56891 and a Glenn/American Federation for Aging Research Scholarship to B. Wise.


We are grateful for the helpful advice of D. Ron and J. Armstrong and the technical assistance of Andrew Cartwright and Gabriel Chan.


  • * T. Tchkonia and T. Pirtskhalava contributed equally to this work.

  • The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.


View Abstract