Human T2D-Associated Gene IMP2/IGF2BP2 Promotes the Commitment of Mesenchymal Stem Cells Into Adipogenic Lineage

Expansion of white adipose mass through adipogenesis is the key contributor of obesity. The development of white adipose depots can be divided into two stages: 1) early commitment, which involves differentiation of ADSC (CD24⁺) into preadipocytes (CD24⁻), followed by 2) terminal differentiation of committed preadipocytes into mature adipocytes (1). The nutrient sensing mTOR pathway is essential for both processes to promote adipogenesis (2). On the other side, it is well-known that Wnt/β-catenin signaling restrains adipogenesis by preventing both the commitment of ADSC into preadipocytes and their final differentiation (3). However, the connection of the two pathway is poorly understood.

The insulin-like growth factor 2 (IGF2) mRNA binding proteins (IMPs or IGF2BPs) are a conserved family of single-stranded RNA binding proteins and composed of six canonical RNA binding domains, including two RNA recognition motif (RRM) domains and four K homology (KH) domains. IMPs participate in the splicing, transport, translation, and stabilization of a wide variety of RNAs in a transcript- and tissue-specific manner (4). Recently, IMPs were identified as readers of N⁶-methyladenosine (m⁶A), the most abundant modification of eukaryotic RNAs. They are capable of recognizing thousands of RNAs through the consensus sequence GG(m⁶A)C (5). IMPs are onco-fetal proteins that are essential for stem cell proliferation and differentiation during embryonic development (6) as well as cancer progress (7). In addition, IMPs are direct substrates of mTOR, and the phosphorylation of IMPs by mTOR regulates the RNA life cycle in response to nutrients (8,9).

Genome-wide association studies have identified single nucleotide polymorphisms (SNPs) in the second intron of the human IMP2 gene to be associated with increased risk for type 2 diabetes (T2D) in many populations (10–12). Follow-up human genetic studies further demonstrated that the association of IMP2 intronic SNPs is affected through modification of the expression of IMP2 gene itself instead of neighboring genes in pancreatic islets, supporting the idea that IMP2 is directly involved in T2D pathophysiology (13,14).

For ascertainment of whether IMP2 participates in the regulation of nutrient metabolism, mice lacking Imp2 were generated and characterized (15). Imp2 knockout mice exhibit a normal birth size and body weight similar to that of control littermates until weaning. However, they gain less weight on normal chow diet (NCD) or on a high-fat diet (HFD), due to slower accumulation of both lean and fat mass. The lower fat mass of Imp2 null mice is especially marked on HFD and is accompanied by considerably better glucose tolerance and insulin sensitivity (15). The beneficial but complicated phenotypes of Imp2 null mice prohibit further understanding of IMP2 functions in individual metabolic tissue and its contributions to T2D pathophysiology.

Here, we explored the function of IMP2 during adipogenesis and inquire whether the reduced fat mass of Imp2 null mice is due to the tissue-autonomous effect of white depots. Imp2ff mice that bear a Cre recombinase driven by platelet-derived growth factor receptor α (Pdgfrα) promoter (PIMP2-KO) exhibit a marked loss of Imp2 expression in ADSC and a notable ∼40% decrease in white adipocyte mass. However, they have normal glucose metabolism and insulin sensitivity. The reduced white adiposity is attributable to a marked reduction of preadipocyte number in PIMP2-KO mice while their preadipocytes showed normal proliferation and terminal differentiation. Mechanically, deletion of Imp2 in ADSC increases Wnt receptor Frizzled class receptor 8 (Fzd8) mRNAs (an Imp2 client) and polypeptides. Imp2 destabilized m⁶A-modified Fzd8 mRNA by recruiting CCR4-NOT deadenylase complex in an mTOR-dependent manner. As a summary, our studies reveal that Imp2 deletion in mesenchymal stem cell (MSC) protects mice from diet-induced obesity without affecting glucose and insulin tolerance. Imp2 is critical for the early commitment of ADSC into preadipocytes through regulating Fzd8 mRNA stability but dispensable for preadipocyte proliferation and terminal differentiation. The interaction of Imp2 and Fzd8 mRNA acts as a bridge between mTOR signaling and Wnt pathway to determine the lineage fate of ADSC.

All animal procedures were approved by the Institutional Animal Care and Use Committee of Massachusetts General Hospital and were performed in accord with the National Research Council guidelines for laboratory animal care (16). Mice were maintained on the C57BL/6J background in a specific-pathogen-free facility with a 12-h light/12-h dark cycle and fed irradiated chow (Prolab 5P75 Isopro 3000, with 5% crude fat; PMI Nutrition International) or HFD (D12492i, 60% of kcal fat; Research Diets). After weaning at 4 weeks of age, sex-matched littermates were housed in the same cage with a maximum number of four mice per cage. Cages were changed every 2 weeks, and nesting material was provided. Animals were sacrificed if they had wounds from fighting, developed severe dermatitis or tumors, or showed other signs of morbidity, such as unresponsiveness to touch or not eating. Imp2 floxed mice have previously been described (17–19). PDGFRa-Cre mice were purchased from The Jackson Laboratory (stock no. 013148). All mice studies were conducted for both sexes. Data are available on request where only data for males are presented.

For analysis of body weight, body composition, and food intake, mice were kept at a normal ambient temperature of 23°C and had free access to water and food. Measurements of energy expenditure were performed with a customized indirect calorimetric system. Mice were individually housed and placed in a Promethion metabolic analyzer (Sable Systems International, Las Vegas, NV) for 3 days before recording started. Body composition was evaluated in live, conscious animals in triplicate with quantitative nuclear magnetic resonance spectroscopy (EchoMRI analyzer; Echo Medical Systems, Houston, TX).

For glucose tolerance tests, mice were fasted 6 h. D-glucose (20%) (1 g/kg body wt i.p.; Sigma-Aldrich, St. Louis, MO) was administered. At 0, 15, 30, 60, and 120 min after administration, blood was collected through tail vein bleeding. Glucose levels were measured with a One Touch Ultra AlphaTRAK 2 glucometer (Zoetis, Parsippany, NJ). For insulin tolerance tests, high-fat-fed mice were fasted for 6 h. Human insulin (Eli Lily) at 0.75 units/kg i.p. was injected. Blood was drawn from the tail vein after injection, and glucose levels were measured as described above.

Blood was collected into EDTA-coated tubes (Sarstedt, Newton, NC). Serum was separated by centrifugation at 4°C, frozen in liquid nitrogen, and assayed at the Vanderbilt University Mouse Metabolic Phenotyping Center.

Subcutaneous and gonadal white adipose tissue (WAT) from age- and sex-matched PIMP2-KO and control mice was dissected. For histological analysis, tissue slices were fixed in buffered formalin and paraffin embedded. Fat cell size was determined with the Coulter counter after osmium tetroxide fixation essentially as described by Hirsch and Gallian for both sexes (20).

Subcutaneous and gonadal WAT was dissected from PIMP2-KO and control mice, minced, and digested with use of 0.5% collagenase (Sigma-Aldrich) at 37°C for 1 h. Cell suspensions were filtered through a cell strainer with 100 μm nylon mesh (BD Biosciences, San Jose, CA) and washed twice with PBS by centrifugation for 3 min at 1,500 rpm at 4°C. Cell pellets were suspended in red blood cell lysis buffer (Sigma-Aldrich) and incubated for 5 min. Cells were washed and centrifuged twice in PBS and incubated in DMEM (Gibco, Carlsbad, CA) with 10% FBS (Gibco) and 1% penicillin/streptomycin (Life Technologies, Carlsbad, CA) at 37°C in 5% CO₂. The media were replaced after 24 h and every 3 days thereafter. Cultures were incubated for 4–7 days until 70–80% confluence was attained. ADSC were passaged and replated at initial concentration of 5 × 10⁵ cells per 100-mm dish. All experiments were performed with passage 2 ADSC. Cell proliferations were analyzed with WST-1 proliferation assay kit according to the manufacturer’s instructions (Roche, Mannheim, Germany). In each experiment, 4 × 10³ cells/well were plated in 96-well plates, and the proliferation was measured in quadruplicate at 0, 24, 48, 72, and 96 h.

Adipose tissues were dissected from subcutaneous and gonadal fat and stromal vascular fraction was isolated. Cells labeling and sorting were performed as previously described (21). Briefly, the isolated SVF was resuspended in ice-cold DMEM with 2% FBS for labeling. Antibodies labeling was performed on ice for 20 min. Cells were washed and resuspended in sorting buffer containing 0.5g/mL propidium iodide (Sigma-Aldrich) for sorting. Cells were sorted on a BD SORP FACSAria and analyzed on a BD LSR II (BD Biosciences), each running with BD FACSDiva 6 software. For sorting, cells initially were selected by size and then by exclusion of dead cells based on propidium iodide uptake. Next, cells were separated by cell surface markers.

FACS-isolated ADSC cells (CD24⁺) from pooled SVF of gonadal and subcutaneous fat were grown to confluence and induced to differentiate with use of adipogenic cocktail consisting of 10 μg/mL insulin (Sigma-Aldrich), 1 μmol/L dexamethasone (Sigma-Aldrich), and 500 μmol/L isobutylmethylxanthine (Sigma-Aldrich). After 48 h adipogenic induction, cultures were maintained in DMEM containing 10% FBS and 10 μg/mL insulin for another 7 days. Differentiated adipocytes were fixed in 4% paraformaldehyde for 45 min at room temperature. After washing with 1× PBS twice, the cells were stained with Oil Red O working solution (0.5% iso-propanol; Sigma-Aldrich) at room temperature for 30 min. The Oil Red O solution was removed, and the cells were washed with 1× PBS before imaging.

Mouse embryonic fibroblast cells C3H10 T1/2 were purchased from ATCC (CCL-226) and cultured according to the supplier’s protocol. Briefly, cells were grown in high-glucose (4.5 g/mL) DMEM supplemented with 10% FCS and 0.5% penicillin-streptomycin (v/v) at 37°C and 5% CO₂. Cells were cultured until they reached 80–90% confluence. Cells between passages 4 and 10 were used for the experiments. Flag-tagged full-length IMP2 WT, IMP2-AA, and IMP2-DD were obtained as previously described (22). IMP2-AA abolishes phosphorylation, while IMP2-DD functions as a phosphomimetic. The Fzd8 mutant plasmids were generated with QuikChange II Site-Directed Mutagenesis Kit (Agilent Technologies). Plasmids were transfected into C3H/10T1/2 cells with Lipofectamine (Thermo Fisher Scientific) according to the manufacturer’s protocol. The CRISPR/Cas9 method was used to delete Imp2 (gRNA: ACAAGAACAATTCCTGAGCT) and Cnot1 (gRNA: GCTTGAGAACTTTACTCAGC) in C3H/10T1/2 cells as previously described (22). The lentivirus was prepared in human embryonic kidney (HEK) 293T cells, and the viral supernatant was then used to infect C3H10T1/2 cells. The cells were selected with 1 μg/mL puromycin for gene knockout and assayed with Western blots. For NSC654259 (PC-60868; ProbeChem) treatment, the compound was dissolved in DMSO to make 10 mmol/L stock solution and the final concentration of 1 μmol/L was used for experiments.

For differentiation assay, C3H10 T1/2 cells were plated (2 × 10⁵ cells/cm²) in collagen I–coated 24-well plates. One day postconfluence, differentiation medium containing 10 μg/mL insulin (Sigma-Aldrich), 1 μmol/L dexamethasone (Sigma-Aldrich), 500 μmol/L isobutylmethylxanthine (Sigma-Aldrich), and 1 μmol/L rosiglitazone (Sigma-Aldrich) was added. After 48 h adipogenic induction, cultures were maintained in DMEM containing 10% FBS and 10 μg/mL insulin for another 7 days. Every 2 days, the maintenance media was replaced. Cells were lysed after 96 h for gene expression assay and after 1 week for Oil Red O staining.

Total RNA (1 mg) was used for cDNA preparation with random-hexamer primers using SuperScript III reverse transcriptase (Invitrogen). Steady-state mRNA expression was measured by quantitative real-time PCR using SYBR Green Master Mix (Bio-Rad Laboratories) with a CFX96 real-time PCR system (Bio-Rad Laboratories). All quantitative PCR primer pairs were purchased from OriGene.

WAT from 6-week-old male mice were harvested after sacrifice and extracted directly with a TissueLyser (QIAGEN) in ice-cold lysis buffer (140 mmol/L KCl, 1.5 mmol/L MgCl₂, 20 mmol/L Tris-HCl at pH 7.4, 0.5% Nonidet P-40, 0.5 mmol/L dithiothreitol, 1 units/μL RNase inhibitor, one complete EDTA-free protease inhibitor cocktail tablet) and extracted for 10 min. The lysates were centrifuged for 10 min at 14,000 rpm, and the supernatant was transferred to a fresh 1.5-mL tube. Total protein was measured by a Bradford assay, and 5 mg cytoplasmic lysate protein was subjected to immunoprecipitation. Lysates were incubated with 500 μL protein A Dynabeads magnetic beads precoated with IMP2 antibody and incubated for 6 h at 4°C with rotation. Dynabeads were extensively washed with lysis buffer five times and digested with DNase I and protease K. RNA was extracted with phenol-chloroform and precipitated with ethanol. Real-time PCR was performed to examine RNAs associated with IMP2 from subcutaneous and gonadal WAT.

Subcutaneous and gonadal depots were homogenized in ice-cold lysis buffer (10 mmol/L Tris [pH 7.5], 250 mmol/L KCl, 10 mmol/L MgCl₂, 0.5% Triton X-100, 2 mmol/L dithiothreitol, 100 g/mL cycloheximide, 100 units/mL RNase inhibitor [Thermo Fisher Scientific], and protease inhibitor cocktails [Thermo Fisher Scientific]) with a QIAGEN tissue homogenizer. Homogenates were continuously vortexed for 5 min at 4°C and incubated for 15 min on ice. Samples were then centrifuged at 10,000g for 10 min at 4°C. The resulting supernatant was layered on a 15%–50% linear sucrose gradient (20 mmol/L Tris [pH 7.5], 250 mmol/L KCl, 10 mmol/L MgCl₂) and centrifuged in an SW41 rotor at 40,000 rpm for 160 min at 4°C. After centrifugation, the gradient was displaced through a spectrophotometer, and the optical density at 254 nm was continuously recorded. Two sucrose fractions, representing the subpolysomal and polysomal portions of the gradient, were collected directly into an equal volume of TRIzol reagent (Invitrogen). RNA was extracted, and samples were analyzed for quality with the Agilent 2100 bioanalyzer (Agilent Biotechnologies). An equal quantity of RNA in each fraction was converted to cDNA.

Cells were treated with 1 mmol/L actinomycin D (Sigma-Aldrich) for 12 h to block transcription. After the indicated times, total RNAs were extracted, followed by DNase digestion for eliminating DNA contamination and cDNA syntheses. The concentration of mRNAs was quantified by real-time RT-PCR with the SYBR Green.

Proteins were extracted from WAT from 6- to 8-week-old HFD-fed male mice with a tissue homogenizer (QIAGEN). Subcutaneous and gonadal WAT were homogenized in ice-cold buffer (20 mmol/L Tris [pH 7.5], 2.7 mmol/L KCl, 1 mmol/L MgCl₂, 1% Triton X-100, 10% [wt/vol] glycerol, 1 mmol/L EDTA, 1 mmol/L dithiothreitol) supplemented with protease (Thermo Fisher Scientific) and phosphatase inhibitor cocktails (Millipore). Samples were then centrifuged at 13,000 rpm for 10 min at 4°C, and the supernatants were collected. The protein content of the supernatant was determined with a bicinchoninic acid assay (Thermo Fisher Scientific). Aliquots of each extract containing 50 μg protein were loaded for Western blotting. Proteins were resolved on a 4%–12% gradient SDS Bis-Tris gel (Invitrogen, Carlsbad, CA). Antibodies for immunoblotting are listed in Supplementary Table 1.

Nuclear proteins were isolated using the Nuclear Extract kit (78833; Thermo Fisher Scientific) following the manufacturer’s instructions. Briefly, the cells were collected into buffer CER I and were vortexed vigorously for 15 s, followed by a 10-min incubation on ice. The ice-cold buffer CER II was added to the mixture, and the tubes were vortexed for 5 s. The samples were incubated on ice for 1 min, vortexed for another 5 s, and centrifuged for 5 min at the speed of 16,000g at 4°C. The supernatant containing cytoplasmic extract was immediately transferred into a new prechilled tube and further lysed in 1× Laemmli buffer. The pellet (nuclear fraction) was incubated in ice-cold NER buffer for a total of 40 min and vortexed for 15 s every 10 min. The lysate was centrifuged at maximum speed for 10 min at 4°C for collection of the supernatant (nuclear extract).

Cycloheximide was used at a final concentration of 250 mmol/L to inhibit mRNA translation; protein half-life was determined with immunoblots of extracts prepared before and at various times after cycloheximide addition.

Comparisons between the mean ± SD of two groups were calculated with two-way ANOVA or Student unpaired two-tailed t test. P values of ≤0.05 were considered statistically significant.

All data and resources are available on request.

To define IMP2’s function in adipogenesis, we first examined Imp2 expression in WAT. The level of Imp2 polypeptide is mostly enriched in ADSC followed by preadipocytes from SVF; mature adipocytes have the lowest level of Imp2 protein (Fig. 1A). We next crossed Pdgfrα-Cre mice with Imp2^lox+/+ mice to delete Imp2 in ADSC. PDGFRα is a membrane-bound tyrosine kinase receptor expressed in perivascular stromal cells within a variety of tissues (23) and have commonly been used as a cell surface marker for MSC identification (24). As expected, the Imp2 expression in ADSC was greatly reduced in PDGFRα-Cre^+/−/Imp2^lox+/+ (PIMP2-KO) mice but was similar in mature adipocytes compare with controls (Fig. 1B). The fat mass of PIMP2-KO mice fed on HFD is consistently lower after weaning (Fig. 1C), whereas lean body mass was not different from control animals for both sexes (Fig. 1D). The total body weight of male and female PIMP2-KO mice fed HFD is consistently lower than that of controls, with statistical significance reached at ∼30 weeks (Supplementary Fig. 1A). The PIMP2-KO mice displayed normal body size, food intake, energy expenditure, and physical activity (Supplementary Fig. 1B–E). Consistent with reduced white fat mass, serum leptin level is lower in PIMP2-KO mice (Table 1). However, fasting blood glucose and serum insulin levels were similar between PIMP2-KO and control mice (Table 1). Glucose tolerance and insulin sensitivity were also not different on either HFD or NCD (Fig. 1E and F and Supplementary Fig. 1E and F and Supplementary Fig. 2). Based on the above data, we conclude that PDGFRα-Cre^+/−/Imp2^lox+/+ mice are resistant to diet-induced obesity, although to an extent that does not alter glucose metabolism and insulin sensitivity.

Next, we tried to determine the cellular basis for the decreased WAT mass in PIMP2-KO (Fig. 2A) and analyzed hematoxylin-eosin (H-E)-stained histological sections of white fat pads. The results of H-E showed that PIMP2-KO mice have similar mature adipocyte size similar to that of controls (Fig. 2B), suggesting that the decreased white adipocyte mass in PIMP2-KO is due to decreased mature adipocyte cell number. This was confirmed by studies of the osmium tetroxide–fixed adipocytes. While the radius of mature adipocytes is not different (Fig. 2C) between PIMP2-KO and control, the number of mature adipocytes from PIMP2-KO was reduced ∼40% for both gonadal and subcutaneous depots of HFD-fed mice (Fig. 2D). Similar phenotypes have also been observed for male PIMP2-KO mice fed on NCD (Supplementary Fig. 2D–F). Therefore, the WAT depots of PIMP2-KO mice contain reduced number of mature adipocytes of a size similar to that of control white adipocytes.

We next sought to determine the role of Imp2 in the adipogenic development and first isolated preadipocytes (Lin⁺/CD29⁺/CD34⁺/Sca1⁺/CD24⁻ cells) from SVF of pooled samples of gonadal and subcutaneous adipose tissues. Surprisingly, this cell population was reduced by ∼50% in PIMP2-KO mice (Fig. 3A), suggesting that Imp2 is critical for the early commitment of MSC into the adipogenic lineage. Consistent with flow cytometry data, the mRNA expression of Pdgfβ, Dlk1, and Vcam1 greatly decreased in ADSC (CD24⁺ cells) isolated from P-IMP2 KO mice (Fig. 3B–D). Furthermore, Cebpα and Pparγ mRNA from differentiated PIMP2-KO ADSC also decreased (Fig. 3E and F), indicating that the adipogenic potential of PIMP2-KO is impaired. Next, we examined the role of Imp2 in terminal differentiation. When seeded at the same density, preadipocytes from WAT depots of PIMP2-KO mice showed normal proliferation (Supplementary Fig. 3A) and normal terminal differentiation (Supplementary Fig. 3B). In summary, Imp2 deletion in ADSC reduces the number of preadipocytes but does not impair their ability to proliferate or differentiate into mature adipocytes.

We previously performed proteomics and RNA-sequencing analysis of Imp2 null mouse embryonic fibroblast cells (22) and observed significantly upregulated Wnt/β-catenin signaling, which is known to inhibit the commitment of MSC to adipocytes (3). Therefore, we examined the mRNA and protein expressions of key factors (25) of canonical Wnt pathway in PIMP2-KO ADSC (Fig. 4A and B). Consistent with mouse embryonic fibroblast data, the expression of Frizzled (Fzd) polypeptides (Fzd8/Fzd2) and the receptor of Wnt signaling pathway (Wnt10a) was increased in PIMP2-KO ADSC, especially for Fzd8 (Fig. 4C). For identification of Imp2 mRNA clients during adipogenesis, immunoprecipitates (IPs) were prepared from extracts of ADSC isolated from 4-week-old mice using anti-Imp2 antibody. Compared with IPs obtained with nonimmune IgG, the Imp2 IP was substantially enriched in mRNAs encoding Fzd8, whereas the enrichment of other mRNAs examined was not significantly different (Fig. 4D). In line with increased Fzd8 protein abundance, the elevated β-catenin polypeptides from nucleus were observed in PIMP2-KO ADSC (Fig. 4E). Moreover, the transcripts of β-catenin directly regulated genes, such as Ccnd1, C-Myc, Mmp7, and Itf2, are consistently increased in PIMP2-KO ADSC (25) (Fig. 4F).

To understand the mechanisms by which Imp2 regulates Fzd8 mRNA abundance, we inactivated Imp2 by CRISPR in the MSC-like cell line C3H10T1/2 and examined the binding of Imp2 with Fzd8 mRNA (Fig. 5A). Compared with CRISPR GFP control C3H10T1/2 cells, Imp2 CRISPR cells exhibit ∼10-fold increase of the abundance of Fzd8 mRNA and protein (Fig. 5B and C), accompanied by a significantly prolonged mRNA half-life (Fig. 5D), but unaltered protein half-life (Fig. 5E). The polysomal abundance of Fzd8 mRNA was not altered by Imp2 deletion (Fig. 5F), suggesting a normal translation of Fzd8 mRNA. Thus, the increased Fzd8 polypeptide abundance in the IMP2 null C3H10T1/2 cells is due to increased Fzd8 mRNA stability and Imp2 is required to promote Fzd8 mRNA degradation.

The m⁶A modification is known to regulate the stability of the modified RNAs through the recruitment of specific “reader” proteins (26,27). IMPs have been shown to selectively recognize and bind to m⁶A consensus sequence RR(m⁶A)CH (R = A, G, or U; R = A or G; and H = A, C, or U) which is near stop codons and enriched in the 3′-untranslated regions (28,29). Therefore, we hypothesized that Imp2 controls Fzd8 mRNA stability through m⁶A recognition. Using Fzd8-specific m⁶A assays, we find that Fzd8 mRNAs preferably bind with m⁶A antibodies (Fig. 5G), indicating m⁶A modification of Fzd8 mRNAs. Notably, the Fzd8 mRNA stop codon containing region has mostly predicted m⁶A modifications that overlap with conserved IMP2 binding sites (Fig. 5H and I). To test whether this region is critical for Imp2-regulated Fzd8 mRNA stability, we inserted the 160 nt wild-type or mutant sequence flanking stop codon of Fzd8 mRNA into Fzd8 coding region (Supplementary Fig. 4) and transfected them into C3H10T1/2 cells. IMP2 RNA immunoprecipitation–quantitative PCR showed a strong binding of IMP2 with wild-type Fzd8 mRNA but not with the mutant form (Fig. 5J). Moreover, the half-life of Fzd8 m6A mutant increased approximately sixfold via the Fzd8 wild-type mRNA (Fig. 5K). Taken together, our data demonstrate that Imp2 selectively binds m⁶A-methylated Fzd8 mRNA and controls its decay.

Intracellular RNA degradation occurs by two major pathways starting from the 5′ end or the 3′ end of the RNA, respectively. IMP1 has been reported to interact with CNOT1 directly (30), which is the scaffold protein of the CCR4-NOT complex, an important deadenylase responsible for poly(A) tail shortening and inducing 3′-5′ decay of numerous RNAs in the cytoplasm. We therefore hypothesized that IMP2 associates with components of the RNA decay machinery to control Fzd8 mRNA degradation. To test this hypothesis, we performed Imp2 IP in C3H10T1/2 cells and confirmed the interaction of IMP2 with CNOT1 (Fig. 6A). If the interaction between IMP2 and CNOT1 is crucial for the Fzd8 mRNA stability, the depletion of CNOT1 will increase the expression level of Fzd8 transcript. As expected, a marked increase in the abundance of Fzd8 mRNA was observed after Cnot1 expression was abolished (Fig. 6B and C). Moreover, the depletion of Imp2 and Cnot1 simultaneously had no additive effect on Fzd8 mRNA upregulation (Fig. 6B and C), indicating that the two proteins are mechanistically linked in determining the abundance of Fzd8 mRNA. Therefore, by associating with CNOT1, IMP2 recruits the CCR4-NOT deadenylase complex onto Fzd8 mRNA and promotes its degradation.

We previously published that mTOR directly phosphorylates IMP2 at Ser162Ser164. To evaluate the functional significance of Ser162Ser164 for the ability of IMP2 to control Fzd8 mRNA stability, we generated IMP2 variants with Ala or Asp at both Ser162 and Ser164, which we refer to as IMP2-AA (nonphosphorylatable) and IMP2-DD (phosphomimetic), and stably expressed them in IMP2-deficient C3H10T1/2 cells. Polyclonal populations were selected. The half-life of Fzd8 mRNA from C3H10T1/2 cells expressing IMP2-DD was substantially shorter than wild type IMP2 (Fig. 6D); whereas the half-life of Fzd8 mRNA from IMP2-AA stable cells showed significantly prolonged half-life compare to wild type IMP2 (Fig. 6D). As IMP2 controls RNA life cycle through binding affinity (4), we examined IMP2 variants’ ability to bind Fzd8 mRNA. Unexpectedly, all IMP2 P-site mutants bind Fzd8 mRNA equally as well as IMP2 wild type (Fig. 6E), indicating that the IMP2 phosphorylation-dependent Fzd8 mRNA degradation is not achieved through affecting the protein-mRNA interaction. Next, we examined the association of IMP2 variants with Cnot1. The Western blot showed that IMP2-DD has stronger interaction with Cnot1 relative to wild-type IMP2 and IMP2-AA (Fig. 6F). Therefore, the acidic charges at IMP2 Ser162Ser164 promote degradation of Fzd8 mRNA through regulating the recruitment of deadenylase complex CCR4-NOT to Imp2-Fzd8 mRNA ribonucleoprotein (RNP).

Finally, we treated Imp2 null ADSC with NSC654259, a small molecule that specifically binds to the Wnt binding site on the FZD8 cysteine-rich domain. The treatment led to reduced phosphorylation of Lrp6, a Wnt coreceptor (Fig. 7A), decreased β-catenin abundance (Fig. 7B), elevated expression of preadipocyte markers (Fig. 7C), and significant induction of adipogenesis (Fig. 7D and Supplementary Fig. 5A and B). This experiment demonstrated that inhibition of FZD8 function can rescue the reduced adipogenesis of Imp2-deficient ADSC. Together, our data showed that Imp2 is required for the commitment of MSC to adipocyte lineage by regulating Fzd8 mRNA degradation (Fig. 7E).

Worldwide obesity has nearly tripled since 1975. A consequence of this is the increased susceptibility of many pathologies such as T2D. Excessive adiposity is a major cause of obesity and insulin resistance, a main risk of T2D. Therefore, it is critical to understand the root mechanisms responsible for the development and maintenance of normal WAT.

In 2007, genome-wide association studies identified the association of SNPs within the human IMP2 gene with increased risk for T2D. This was unexpected, as no role for IMP2 in nutrient metabolism had previously been reported. Subsequently, we found that mice with global deficiency of Imp2 exhibit a complex metabolic phenotype dominated by a reduced white fat mass with a marked resistance to diet-induced obesity and fatty liver (15). Seeking to understand the contribution of IMP2 to T2D susceptibility and its function in major metabolic tissues, we further generated and characterized mice with tissue-specific Imp2 deficiency. Mice with pancreatic β-cell–specific Imp2 deletion showed hyperglycemia due to impaired β-cell proliferation and insulin secretion on HFD feeding (17). This discovery is consistent with findings that people with causal variants in IMP2 gene have reduced IMP2 expression and decreased insulin secretion in islets (13,14). Studies of muscle-specific Imp2 null mice revealed that Imp2 promotes protein synthesis in skeletal muscle mass and optimizes muscle metabolism during voluntary exercise (18). Mice with hepatocyte-specific IMP2 deficiency showed diet-induced fatty liver due to reduced fatty acid oxidation (19). Reported here, the ADSC-specific loss of Imp2 impairs white fat accumulation and protects mice from diet-induced obesity. However, mice bearing Imp2 deletion in muscle, liver, and adipocyte tissues all have normal glucose tolerance and insulin sensitivity until 30 weeks old. Therefore, the collected data to date demonstrate that Imp2 contributes to T2D susceptibility primarily through β-cell insulin secretion; any strong evidence for IMP2 association with insulin sensitivity is lacking.

The timing, dosage, and location of gene expression are fundamental determinants of different cell types such as MSC and ADSC. IMPs regulate RNA life cycle in a transcriptome-specific manner (4). The abundance of IMP1–3 polypeptides and of their client RNAs both contribute to the regulation.

White adipocytes are derived from MSC. When MSC undergo adipogenesis, they first differentiate into preadipocytes, a proliferative adipocyte precursor cell, after which they undergo terminal differentiation into mature adipocytes. During this biological process, the abundancy of IMP2 polypeptides is most enriched in isolated mouse MSC (>10×) compared with preadipocytes and mature adipocytes (Fig. 1A). Little or no Imp2 expression was observed days 4–5, and virtually no Imp2 expression remains in fully differentiated cells day 7 (Ning Dai, unpublished data). This is consistent with the general expression pattern of IMPs: highly abundant in stemness/undifferentiated state and very low in mature/differentiated state (31,32). As IMPs are considered “onco-fetal” gene to stimulate cell growth and proliferation, it is not surprising to see that Imp2 null preadipocytes proliferated and differentiated normally in vitro. This result could be due to the relative low expression of Imp2 in preadipocytes, a more differentiated cell type compared with MSC, in which Imp2 is not critical enough for their growth and differentiation. Alternatively, it could be that in vitro experimental conditions may not accurately mimic the microenvironmental milieu or stem cell niche necessary to support proliferation and differentiation. Another limitation of the study is that we could not rule out the contribution of extra-adipose to the phenotype of the PIMP2-KO mice, since PDGFRα is expressed in perivascular stromal cells within a variety of tissues. Future in vivo lineage tracing experiments will help to justify the observation based on in vitro differentiation assays.

mTOR pathway is the major nutrient-sensitive regulator of growth and is the key link between the availability of nutrients in the environment and the control of anabolic and catabolic processes in the body (33). mTOR is a serine/threonine protein kinase. In mammals, it constitutes the catalytic subunit of two distinct complexes known as mTOR complex 1 (mTORC1) and mTORC2 (33). These complexes are distinguished by their accessory proteins and their differential sensitivity to rapamycin, as well as by their unique substrates and functions. mTORC1 is the major nutrient sensor complex (34–36). The components of mTORC1 such as Raptor, S6K, and 4E-BPs are essential for the development of adipocyte tissues (37–39). As WAT is the largest repository of energy, previous mTOR studies mostly focus on its function in adipocyte differentiation: increasing the synthesis and deposition of triglycerides (40).

Wnt/β-catenin signaling is well-known to regulate MSC fate by inhibiting adipogenesis and promoting osteoblastogenesis (41). The signaling initiates on binding of Wnt ligands to the seven-pass transmembrane receptor FZDs and the single-pass low-density lipoprotein receptor–related protein 5 or 6 (LRP5/6) on the cell surface. This event further triggers phosphorylation of LRP5/6 and inactivation of β-catenin destruction complex. As the core receptor of Wnt proteins, the strength of the Wnt/β-catenin signaling pathway is exquisitely sensitive to the level of FZDs on the plasma membrane (42). Regulation of FZDs has emerged as a key regulatory node of the Wnt signaling pathway (43,44).

Disturbed mTOR or Wnt signaling has been implicated in numerous diseases. Both signalings play important roles in adipose tissue development: the former by acting to induce adipogenesis and the latter acting to inhibit it. Cross talk between the mTOR and Wnt signaling pathways has been reported to occur at different levels and in different cellular contexts (45–47). Whether these two fundamental pathways couple to exert physiological functions during adipogenesis remains mysterious. The data presented here demonstrate that mTOR and Wnt signaling pathways exhibit crosstalk at the level of FZDs during the early commitment of MSCs into ADSCs into preadipocytes: mTOR induces phosphorylation of IMP2, promotes the binding of IMP2 with CCR4-NOT deadenylase complex, stimulates the Fzd8 mRNA degradation and destruct β-catenin. Thus, the IMP2/FZD8 mRNA interaction acts as a mechanistic bridge between the mTOR and Wnt pathways to regulate the commitment of MSC into adipocytes, which could play a role in the pathophysiology of metabolic diseases.

In summary, our study demonstrated that mice with Imp2 MSC-specific deletion have reduced WAT and are resistant to diet-induced obesity, although to an extent that does not alter glucose metabolism and insulin sensitivity. Imp2 plays a key role in WAT development by stimulating the early commitment of MSC into adipogenic lineage, but it is dispensable for preadipocytes proliferation and terminal differentiation. At the molecular level, Imp2 destabilized Fzd8 mRNA by recruiting CCR4-NOT complex in an mTOR-dependent way during MSC commitment. The data highlight the importance of posttranscriptional regulation of gene expression in maintaining WAT homeostasis. This study expands our knowledge of the cross talk between mTOR and Wnt signaling during adipogenesis and may help with discovery of new treatment and prevention of obesity, and the subsequent metabolic dysregulation.

Acknowledgments. The authors thank the MGH microscopy core of the program in membrane biology for imaging and Dr. Peroni at Brighgam and Woman’s Hospital for adipocyte size determination.

Funding. This work was supported by National Institutes of Health grants DK17776 and DK057521 and Massachusetts General Hospital (MGH) institutional funds.

Duality of Interest. No potential conflicts of interest relevant to this article were reported.

Author Contributions. J.A., H.W., and N.D. conceptualized the studies; L.R., W.W., F.J., H.W., F.J., and N.D. conducted research; J.A. and N.D. wrote the original draft; W.W., H.W., J.A., and N.D. reviewed and edited the manuscript; H.W. and N.D. supervised the studies; and H.W. and N.D. administrated the whole project. N.D. 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.