Takahiro Yamakawa, Robert H. Whitson, Shu-Lian Li and Keiichi Itakura

Previous study showed that mice lacking modulator recognition factor-2 (Mrf-2) were lean, with significant decreases in white adipose tissue. One postulated mechanism for the lean phenotype in Mrf-2 knockout mice is a defect in adipogenesis. In order to investigate this further, we examined the effects of Mrf-2 deficiency on adipogenesis in vitro. In mouse fibroblasts (MEFs) derived from Mrf-2−/− embryos, and in 3T3-L1 cells after knockdown of Mrf-2 by small interference RNA (siRNA) there was a potent inhibition of hormone-induced lipid accumulation, and significant decreases in the expression of the adipogenic transcription factors CCAAT/enhancer-binding protein (C/EBP) α and peroxisome proliferator-activated receptor-γ and the mature adipocyte genes they control. Transduction of Mrf-2−/− MEFs with a retroviral vector expressing the longer Mrf-2 splice variant (Mrf-2B) stimulated both gene expression and lipid accumulation. Because 3T3-L1 cells are committed to the adipocyte lineage, we used this simpler model system to examine the effects of Mrf-2 deficiency on adipocyte maturation. Analyses of both mRNA and protein revealed that knockdown of Mrf-2 in 3T3-L1 cells prolonged the expression of C/EBP homologous protein-10, a dominant-negative form of C/EBP. Consistent with these findings, suppression of Mrf-2 also inhibited the DNA-binding activity of C/EBPβ. These data suggest that Mrf-2 facilitates the induction of the two key adipogenic transcription factors C/EBPα and peroxisome proliferator-activated receptor-γ indirectly by permitting hormone-mediated repression of the adipogenic repressor C/EBP homologous protein-10.

WHITE ADIPOSE TISSUE plays an important role in energy balance by accumulating triglycerides when energy substrates are abundant, and liberating glycerol and fatty acids during starvation. Recent data suggest that both lipogenesis and adipocyte differentiation are dynamically regulated to accommodate changes in energy balance. Therefore, understanding the mechanisms that control these processes may be the key to understanding how excess energy input leads to obesity and its complications.

Modulator recognition factors-1 and -2 (Mrf-1, 2) were cloned in our laboratory by screening an expression library for proteins that recognize specific viral DNA sequences (2, 3). Mrf-2 has two splicing variants, Mrf-2A and -2B. Exons III–VI, which encode the ARID (AT-rich interaction domain) DNA-binding motif are found in both proteins, as are exons II and VII. Mrf-2A is truncated at the N-terminal and lacks exons A-D of Mrf-2B. Little is known about the differences in the functions between Mrf-2A and 2B, but it has been reported that Mrf-2B is a more potent inhibitor of cell growth in NIH-3T3 fibroblasts.

The differences in the two splicing variants, Mrf-2A and Mrf-2B are shown. Exons III-VI encode the ARID DNA-binding motif that is found in both Mrf-2A and Mrf-2B, as are exons II and VII. Mrf-2A lacks exons A-D of Mrf-2B; exon I encodes only the first codon of Mrf-2A. This figure also depicts the locations of small interference RNAs directed to Mrf-2, and of PCR primer sets that specifically detect Mrf-2A (blue), Mrf-2B (purple), or both (red). All five of the siMrf-2 target sites are common to both Mrf-2A and Mrf-2B.

To study the functions of Mrf-2 further, we generated mice lacking both Mrf-2A and 2B. Mrf-2 knockout mice are lean, with reduction in both brown and white adipose tissues, and marked lipodystrophy in both inguinal and gonadal white adipose depots. Mrf-2 is widely expressed in adult mouse tissues and therefore, the lean phenotype of Mrf-2−/− mice could arise from a variety of mechanisms. One hypothesis that could explain both the leanness and the lipodystrophy in these mice is that Mrf-2 expression is required for adipogenesis. To test this hypothesis, we examined the effects of Mrf-2 deficiency on adipogenesis using two different in vitro model systems. In the first model, we compared in vitro adipogenesis in primary fibroblast lines (mouse fibroblasts; MEFs) derived from Mrf-2−/− and Mrf-2+/+ embryos. Here we show that the Mrf-2−/− MEF lines have a significant defect in in vitro adipogenesis, compared with MEFs from their wild-type littermates.

Embryonic fibroblast cultures contain heterogeneous cells populations and may include mesenchymal stem cells, preadipocytes, and other cell types. Therefore, the observed effects of Mrf-2 deficiency in this model could result from a reduction in the commitment of precursors to the adipocyte lineage, or a reduction in adipocyte maturation, or both. Because murine 3T3-L1 preadipocytes are committed to the adipocyte lineage, we elected to use this simpler system to determine whether Mrf-2 is required for adipocyte maturation.

The process of in vitro adipogenesis has been examined in considerable detail in 3T3-L1 cells. When treated with hormone mixtures that contain 3-isobutyl-1-methylxanthine (IBMX), dexamethasone (Dex), and insulin (Ins), these cells divide one or two times in a process called mitotic clonal expansion, and CCAAT/enhancer-binding proteins β and δ (C/EBPβ and -δ) are induced. C/EBPβ and -δ cooperate to induce C/EBPα and peroxisome proliferator-activated receptor-γ (PPARγ). C/EBPα and PPARγ maintain the expression of one another, and activate transcription of many genes that are characteristic of mature adipocytes, such as adipose fatty acid-binding protein (aP2), phosphoenolpyruvate carboxykinase (PEPCK), lipoprotein lipase, and perilipin.

Another member of the C/EBP family, C/EBP homologous protein-10 (CHOP-10; also called growth arrest and DNA damage 153–GADD 153), has inhibitory effects on C/EBPα, -β, and -δ. In CHOP-10, proline replaces the basic region alanine and lysine residues that are essential for DNA-binding activity in other C/EBP family members. In the absence of CHOP-10, the other C/EBPs form homodimers, and activate transcription by binding to DNA; in the presence of CHOP-10, C/EBPs form heterodimers with CHOP-10 that cannot bind to DNA, and consequently the expression of C/EBP-dependent genes is suppressed. Expression of CHOP-10 falls during clonal expansion, when C/EBPβ activity is increasing. The failure to decrease CHOP-10 under some experimental conditions causes inhibition of adipogenesis, primarily due to a decrease in C/EBPβ activity. Here we show that knockdown of Mrf-2 inhibits adipocyte maturation in 3T3-L1 preadipocytes, and provide preliminary evidence that this is due to persistent expression of CHOP-10 in the early phases of this process.

Mrf-2−/− MEFs Have a Significant Deficit in in Vitro Adipogenesis

The propagation of mouse embryo fibroblasts is subject to considerable variation. This may be due to variations in timing of the pregnancies, the fairly crude dissections of mouse embryos or variations in culturing and freezing of the lines. As a result, even wild-type MEF lines show considerable variation in their potential for in vitro adipogenesis. Therefore, in examining the effects of the Mrf-2 knockout on adipogenesis, we sought to minimize these variables by comparing knockout and wild-type MEFs from the same litters. Figure 2A⇓ shows a typical experiment. It can be seen that the number of adipocytes in either of the two Mrf-2−/− MEF cultures was significantly lower than in the Mrf-2+/+ MEF culture from d 6–12 of hormone treatment. These results are typical of similar experiments in which one or more wild-type MEF lines were compared with one or more knockout MEF lines from four separate litters. In each case, the number of fat cells produced was significantly lower for the knockout line than for the paired wild-type line.

A, Time-course of adipogenesis. Two Mrf-2−/− MEF lines (red symbols) and one Mrf-2+/+ MEF line (blue symbols) from the same litter were plated at the same densities and treated with adipogenic hormones for 12 d. At the indicated time points, fat cells were counted as described in Materials and Methods. Values are means, ± SE for quadruplicate wells. * and **, Significant differences between both Mrf-2−/− MEF lines and the Mrf-2+/+ MEF line, in pairwise comparisons using a two-tailed Student’s t-test, P < 0.03 and P < 0.001, respectively. The right-hand panels show representative low-power fields of Oil Red O-stained cultures at d 12. The results are representative of independent experiments using paired Mrf-2−/− and Mrf-2+/+ MEF lines from four different litters. B, Gene expression in adipogenesis. The three MEF lines shown in Part A, plus one Mrf-2−/− and two Mrf-2+/+ MEF lines from a different litter were plated in six-well plates and treated with adipogenic hormones. At the indicated time points, RNA was extracted from each MEF line. cDNA was prepared from all of the samples using the same reaction mix, and expression of the indicated genes, normalized to 18S RNA was measured using real-time PCR. *, Significant differences between the Mrf-2−/− and Mrf-2+/+ MEF lines, using an unpaired t test, P < 0.05.

As expected, the expression of multiple markers of mature adipocytes was also reduced in Mrf-2−/− MEFs compared with MEFs from wild-type littermates. Figure 2B⇑ shows real-time RT-PCR data for RNA samples derived from the same cells depicted in Fig. 2A⇑, plus RNA samples that were obtained in a duplicate experiment using MEF lines from another litter. The second litter included one Mrf-2−/− embryo and two Mrf-2+/+ embryo, and therefore, we were able to compare three RNA samples of each genotype at each time point. As can be seen, the expression of the adipocyte markers fatty acid synthase (FAS), perilipin, PEPCK, and aP2 was significantly reduced in the Mrf-2−/− MEFs after 12 d of hormone treatment. Interestingly, the adipogenic transcription factors C/EBPα and PPARγ were also reduced in the Mrf-2−/− MEFs at d 12.

Overexpression of Mrf-2B Stimulates Lipogenesis and Adipocyte Gene Expression in Mrf-2−/− MEFs

Mrf-2−/− MEFs lack expression of both Mrf-2A and Mrf-2B. Therefore, it was not clear whether the deficit in adipogenesis in these cells is due to the lack of Mrf-2A, Mrf-2B or both. To address this, we prepared retroviral vectors that express these proteins and tested their effects on adipogenesis in Mrf-2−/− MEFs. Adipogenesis was not stimulated in Mrf-2−/− MEFs that were transduced with the Mrf-2A expressing retrovirus (data not shown) but was stimulated modestly in Mrf-2−/− MEFs that were transduced with the Mrf-2B-expressing retrovirus, compared with both untreated MEFs and MEFs that were transduced with the LacZ control retrovirus. The Mrf-2B retrovirus also stimulated the expression of C/EBPα, PPARγ, and mature adipocyte markers, including aP2, PEPCK, FAS, and perilipin (Fig. 3B⇓). Interestingly, the expression of the Mrf-2B transgene was transient, reaching a peak at 48 h after transduction (d 0) and declining rapidly by d 2 of hormone treatment. Although Mrf-2B transgene expression remained above background levels at d 12 (Fig. 3B⇓, upper right panel; note that this is a log scale), the peak of expression occurred in the early phases of adipogenesis. By contrast, most of the stimulation of gene expression occurs at d 12. This would suggest that either the modest overexpression of Mrf-2B that remains at d 12 (2-fold over background) is sufficient to stimulate the expression of the late genes, or that these are downstream effects of the high level of Mrf-2B expression at early time points. In this regard, it is interesting to note that transduction with the Mrf-2B retrovirus inhibits CHOP-10 expression before the addition of adipogenic hormones at d 0, compared with either LacZ retrovirus or no treatment. An increase in CHOP-10 expression coincides with the decrease in Mrf-2B expression in Mrf-2B retrovirus-transduced cells. CHOP-10 expression decreases after d 2 of hormone treatment, but it is not lower in the Mrf-2B-transduced cells than in the untreated cells. Taken together, these results suggest that overexpression of Mrf-2B in the early stages of hormone treatment is sufficient to stimulate both adipogenesis and adipocyte-specific gene expression several days later.