The antiobesity factor WDTC1 suppresses adipogenesis via the CRL4^{WDTC 1} E3 ligase

Abstract

WDTC1/Adp encodes an evolutionarily conserved suppressor of lipid accumulation. While reduced WDTC1 expression is associated with obesity in mice and humans, its cellular function is unknown. Here, we demonstrate that WDTC1 is a component of a DDB1‐CUL4‐ROC1 (CRL4) E3 ligase. Using 3T3‐L1 cell culture model of adipogenesis, we show that disrupting the interaction between WDTC1 and DDB1 leads to a loss of adipogenic suppression by WDTC1, increased triglyceride accumulation and adipogenic gene expression. We show that the CRL4^{WDTC 1} complex promotes histone H2AK119 monoubiquitylation, thus suggesting a role for this complex in transcriptional repression during adipogenesis. Our results identify a biochemical role for WDTC1 and extend the functional range of the CRL4 complex to the suppression of fat accumulation.

Introduction

Over 50 years ago, Dr. Winifred Doane isolated and extensively characterized a naturally derived D. melanogaster mutant, termed adipose (adp) 1, 2. The most obvious mutant phenotype observed was fat body hypertrophy due to excessive lipid storage. Adp is evolutionarily conserved from files to humans as a single gene and encodes a protein containing WD40 repeat domains and TPR motifs 3. The mammalian homolog of Adp is WDTC1 (WD40 and tetratricopeptide repeats 1). Loss of a single Wdtc1 allele results in obese mice with poor metabolic profiles, and conversely, transgenic expression of Wdtc1 in fat cells yields lean mice 4. Recently, population studies have linked reduced WDTC1 expression to human obesity 5, 6. Despite strong genetic evidence linking WDTC1 to antiadipogenic function, its molecular mechanism remains unknown.

Ubiquitin pathway plays a critical role in virtually all cellular processes. Ubiquitylation proceeds via an enzymatic cascade where E1 and E2 enzymes catalyze the activation and conjugation of ubiquitin, while E3s confer reaction specificity through substrate recruitment 7, 8. Substrates can be polyubiquitylated, which often leads to proteasome‐dependent degradation, or monoubiquitylated, which regulates the property and thus the function of the substrate. Comprising the largest family of E3 ligases is the cullin RING E3 ligase (CRL) complexes in which a cullin serves as the scaffold to bind small RING finger protein ROC1 (RBX1/HRT1) through a C‐terminal domain, and a linker‐substrate receptor dimer or a substrate receptor directly via a N‐terminal domain. Mammalian cells express two cullin 4 (CUL4) proteins, CUL4A and CUL4B, that bind damaged DNA binding (DDB1) protein. DDB1 acts as the linker to bridge the interaction between CUL4 and a subset of DDB1 binding WD40 repeat proteins (DWDs or DCAFs for DDB1 cullin‐associated factors) which function as substrate receptors to target specific substrates to the CRL4 E3 complexes 9, 10, 11, 12. The human genome encodes an estimated ~90 DWD proteins 9, but the functional interaction between CRL4 and DWDs remains unexplored for the vast majority. One of predicted DWD protein is WDTC1, raising the possibility that WDTC1 may function as a substrate receptor of CRL4 E3 ligase to inhibit fat accumulation. This study is aimed at determining this issue.

Results and Discussion

WDTC1 is a putative substrate receptor for CRL4 E3 ligase

Two structural features suggested that WDTC1 binds DDB1 (Fig 1A). First, WDTC1 contains two tandem DWD boxes with conserved WDXR submotifs, the signature motif present in nearly all WD40 proteins that bind DDB1 9. Second, the N‐terminal regions of vertebrate WDTC1 contain an α‐helical motif termed the H‐box which is shared by several DWD proteins and some viral proteins that bind DDB1 13. Although the H‐box of the DWD protein DDB2 makes a large contribution to DDB1 binding 11, 13, a mutation in the WDXR of DDB2 (R273) is found in human xeroderma pigmentosum patients and disrupts DDB2‐DDB1 binding 14. These observations underscore the importance and specificity of both DWD motif and H‐box in mediating DDB1 binding, thus offering a unique opportunity for investigating the WDTC1–CRL4 interaction.

Figure 1. WDTC1 is a substrate receptor of CRL4 E3 complexes.

1. Domain structure of human WDTC1 with locations of the H‐box and WDXR motifs indicated. Alignment of WDTC1 H‐box motifs in different species (boxed in gray) and bolded orange represents key residues contacting DDB1 13. Alignment of DWD boxes in different DWD proteins (bottom right) with WDXR submotif indicated; tandem DWD boxes are shown for WDTC1 only.
2. Endogenous CUL4A and CUL4B complexes were isolated from 293T cell lysates by immunoprecipitation (IP), and associated proteins were detected by immunoblotting (IB). Mock includes beads only control.
3. Flag‐tagged WT and mutant WDTC1 proteins were transiently expressed in 293T cells. Flag‐WDTC1 complexes were immunoprecipitated with anti‐FLAG, and their associated proteins were detected by immunoblotting as indicated; EV, empty vector control.
4. Top, the modeled structure of WDTC1 in complex with CRL4 (PDB 4A0K); WDTC1 superimposition was based on its H‐box (PDB 317N). WDTC1 domains are rendered in same colors as linear domain structure in (A); WDXR arginines were rendered in dark blue ball‐and‐stick. Bottom, schematic summarizing results presented in (C).
5. 293T cells were transfected with HA‐ubiquitin along with various combinations of plasmid and siRNA as indicated. Transfected cells were lysed under denaturing conditions to obtain whole‐cell extracts (WCE) and immunoprecipitated with anti‐FLAG. WDTC1 ubiquitylation was evaluated by immunoblotting as indicated.
6. 293T cells transiently expressing Flag‐WDTC1 were transfected with scramble (scrm) siRNA or siRNAs against DDB1, CUL4A, and CUL4B. Flag‐WDTC1 levels and knockdown efficiency were assessed by immunoblotting as indicated.

Source data are available online for this figure.

To test whether WDTC1 and CRL4 interact in vivo, endogenous CUL4A or CUL4B complexes were immunoprecipitated from 293T cells, and the presence of WDTC1 in the immunoprecipitates was determined by immunoblotting. We found that WDTC1 interacts endogenously with both CUL4A and CUL4B (Fig 1B). We then generated two WDTC1 mutants, one deleted the N‐terminal 25 amino acid residues containing the H‐box (referred to as ΔH) and one substituted the arginine residues in the tandem WDXR motifs to alanines (referred to as RARA; Fig 1A). We found that wild‐type Flag‐WDTC1 coimmunoprecipitated all CRL4 subunits tested, including DDB1, CUL4A, CUL4B, and ROC1 (Fig 1C). In contrast, deletion of the H‐box (ΔH) completely ablated complex formation, while the RARA mutant showed a modest decrease in DDB1 binding and a marked reduction in CUL4 and ROC1 binding. A superimposed structure of WDTC1 is shown in complex with CRL4 (Figs 1D and EV1). As previously reported 13, WDTC1 is anchored through its H‐box into the BPC propeller domain of DDB1. Similar to DDB2‐DDB1 binding 15, the tandem WDXRs of WDTC1 are not predicted to make direct contact with DDB1 but are solvent‐exposed on the bottom surface of the β‐propeller. How these two signature arginine residues contribute to the binding of DWDs to DDB1 is presently unclear.

Figure EV1. WDTC1 structural model.

Modeled structure of the human WDTC1 depicting the unique combination of a classic seven‐bladed β‐propeller with protruding antiparallel α‐helices of the TPR motifs. The solvent‐exposed arginine residues of tandem WDXRs, rendered in dark blue ball‐and‐stick, are located on the bottom surface of the propeller fold. WDTC1 modeling was based on DDB2 (PDB 4A0K), Sgt2 (PDB 3SZ7), and the WDTC1 H‐box (PDB 3I7N) structural templates (see Materials and Methods for details). The structure was color‐coded according to the linear domain organization presented in Fig 1A (top).

We previously reported an example of a DWD protein, WDR5, that is ubiquitylated by its cognate CRL4B and targeted for proteolysis 16. We therefore tested whether WDTC1 is a CRL4 substrate by in vivo ubiquitylation assays. We found that WDTC1 was extensively ubiquitylated, but the levels were unchanged by either DDB1 knockdown or ΔH mutation (Fig 1E). Further, the steady‐state levels of Flag‐WDTC1 protein were largely unaltered by CUL4A or CUL4B knockdown, while their respective substrates p21 and cyclin E were clearly stabilized (Fig 1F), indicating that CRL4 does not regulate WDTC1 protein stability. We noted that DDB1 depletion resulted in a slight decrease in Flag‐WDTC1, possibly due to protein instability arising from the near loss of its primary binding partner. Together, these results demonstrate that WDTC1 is a component of DDB1‐CUL4A/B‐ROC1 complexes and likely functions as a substrate receptor of CRL4 E3 ligases.

The interaction of WDTC1 with CRL4 is critical for its function in adipogenic suppression

To determine the biological significance of the CRL4–WDTC1 interaction, we hypothesized that the fat‐suppressive role of WDTC1 is mediated through CRL4. We first examined the expression of WDTC1 and other CRL4 subunits in mouse tissues, including multiple adipose tissues. We found that DDB1, ROC1 and at least one CUL4 are expressed in all four fat tissues, along with readily detected expression of WDTC1 (Fig EV2A). We then assayed the adipogenic differentiation of 3T3‐L1 preadipocytes, the well‐characterized cell culture model to study adipogenesis (Fig EV2B). Thought to closely recapitulate adipogenesis in vivo, treating 3T3‐L1 cells to an adipogenic media triggers transcriptional activation of the terminal differentiation program and morphological changes following lipogenic accumulation of triglycerides 17.

Figure EV2. WDTC1 protein expression, schematic of 3T3‐L1 adipogenesis, and stable cell lines.

1. Expression of CRL4 components and WDTC1 proteins in a panel of adult C57BL/6 mouse tissues was evaluated by immunoblotting with the indicated antibodies; brown adipose tissue (BAT), inguinal white adipose tissue (IGWAT), gonadal white adipose tissue (GWAT), and mesenteric white adipose tissue (MWAT).
2. Schematic outline of experimental procedures performed with 3T3‐L1 cells in this study. Scale bar, 0.05 mm.
3. Lysate prepared from 3T3‐L1 preadipocytes was subjected to immunoprecipitation with the indicated antibodies, followed by immunoblot analyses as indicated to evaluate endogenous interactions in mouse 3T3‐L1 preadipocytes. WDTC1 antibodies α‐WDTC1‐1 and α‐WDTC1‐2 are commercially available from Abcam (ab174294) and Sigma (SAB1101463), respectively. Antibody SAB1101463 failed to immunoprecipitate endogenous WDTC1 and give a specific signal in this experiment. The CUL4A antibody cannot immunoprecipitate mouse CUL4A and to the best of our knowledge, a commercial source is presently unavailable.
4. 3T3‐L1 preadipocytes were retrovirally transduced to generate stable cell lines expressing vector control (EV) or Flag‐tagged wild‐type and mutant WDTC1 proteins. Flag‐WDTC1 protein expression and CRL4 binding were confirmed by immunoprecipitation followed by immunoblotting with antibodies as indicated.
5. To analyze cell cycle distribution, cells in (D) were plated at equal density and adipogenically induced. Following induction (24 h), cells were trypsinized and fixed overnight in 75% ethanol at 4°C. Fixed cells were stained with propidium iodide and then analyzed by flow cytometry to measure DNA content.
6. 3T3‐L1 stable cells expressing vector control (EV), Flag‐tagged wild‐type, or mutant WDTC1 proteins were adipogenically induced following standard induction procedure except insulin was replaced by 100 nM rosiglitazone (Cayman) in the induction cocktail. Adipogenesis was assessed by Oil Red O staining.
7. Lysates were prepared from uninduced (preadipocytes) or adipogenically induced 3T3‐L1 cells stably expressing vector control (EV), Flag‐tagged wild‐type, or mutant WDTC1 proteins. PPARγ protein levels were monitored to evaluate adipogenic marker protein induction.
8. 3T3‐L1 stable cells expressing vector control (EV), Flag‐tagged WT or mutant WDTC1 proteins were adipogenically induced. Expression of genes regulating lipogenesis (Acc1, Fasn, Scd1, Dgat1, and Dgat2), lipolysis (Atgl), and fatty acid oxidation (Cpt1b and Cpt2) was evaluated by RT–qPCR analysis in day 12 post‐induction cells (except Fasn; evaluated separately in day 4 post‐induction cells).

Data information: In (H), bars represent the mean ± SEM from 3 independent experiments. *P < 0.05, significant change compared to EV control. Statistical significance was analyzed by paired two‐tailed Student's t‐tests.

We first confirmed that WDTC1 forms endogenous CRL4 complexes in 3T3‐L1 cells (Figs 2A and EV2C). We next evaluated the effects of disrupting the WDTC1–DDB1 interaction on WDTC1 function during adipogenesis. For this purpose, we generated stable 3T3‐L1 cells that express a vector control and flag‐tagged wild‐type (WT), ΔH, or RARA mutant WDTC1 (Fig EV2D) and verified that their expression did not significantly alter the cell cycle distribution (Fig EV2E). When these cells were adipogenically induced, WT WDTC1 suppressed adipogenesis as assessed by Oil Red O staining (ORO; Fig 2B). In striking contrast, WDTC1^ΔH expression promoted adipogenesis, suggesting that it may function as a dominant negative form of WDTC1. Further, the expression of WDTC1^RARA mutant also promoted adipogenesis, but at a lower efficiency than WDTC1^ΔH. While the WDTC1^ΔH mutant exhibits near‐complete loss of DDB1 binding, the WDTC1^RARA mutant is partially defective in DDB1 binding, suggesting that WDTC1^RARA mutant is functionally equivalent to a hypomorphic allele. This is consistent with genetic evidence indicating that WDTC1 exhibits dosage sensitivity 1, 4. Additionally, WDTC1 likely acts downstream of the insulin signaling pathway as replacing insulin with the PPARγ agonist rosiglitazone did not alter the observed ORO phenotypes (Fig EV2F). Validating the phenotypes evaluated by ORO, ectopic expression of WT WDTC1 suppressed triglyceride accumulation, while expression of the WDTC1^ΔH (3.5‐fold) and WDTC1^RARA (2.3‐fold) mutants resulted in significantly higher triglyceride levels (Fig 2C). Further, expression of WT WDTC1 decreased adipogenic marker expression but WDTC1^ΔH and WDTC1^RARA mutants increased the expression of these genes with WDTC1^ΔH again being more effective than WDTC1^RARA (Figs 2D and EV2G). We also assessed expression of genes regulating lipogenesis, lipolysis, and fatty acid oxidation, and found modest alterations that are likely due to altered adipogenesis in cells expressing WT or mutant WDTC1 proteins (Fig EV2H). The expression of WT WDTC1 resulted in a modest reduction in the expression of genes linked to lipogenesis (Fasn, Dgat1, and Dgat2), lipolysis (Agtl) and β‐oxidation of long‐chain fatty acids (Cpt2), while the expression of WDTC1^ΔH led to opposite changes in the expression of these genes. Together, these results support the idea that the CRL4 interaction is critically important for WDTC1 function in the negative regulation of adipogenesis.

Figure 2. WDTC1 suppresses adipogenesis in a CRL4‐dependent manner.

• A
  3T3‐L1 preadipocytes were adipogenically induced, and lysates were prepared from cells collected at indicated days. Endogenous WDTC1 interaction with CRL4 subunits was evaluated by immunoprecipitation with anti‐WDTC1 followed by immunoblot analyses with the indicated antibodies. Fatty acid synthase (FAS) levels were monitored to evaluate induction.
• B–D
  3T3‐L1 stable cells expressing vector control (EV), Flag‐tagged WT or mutant WDTC1 proteins were adipogenically induced. Adipogenesis was assessed by Oil Red O staining (ORO) (B); plate (top) and microscopic (bottom) views, 0.1 mm scale bar. Triglyceride levels were quantified by an enzymatic assay (C). Adipogenic gene expression was evaluated by RT–qPCR analysis (D).

Data information: In (C, D), bars represent the mean ± SEM from 3 independent experiments. *P < 0.05, **P < 0.005, significant change compared to EV control. Statistical significance was analyzed by paired two‐tailed Student's t‐tests. Source data are available online for this figure.

CRL4 binding mutants of WDTC1 cannot rescue WDTC1 function

To further validate the functional link between WDTC1 and CRL4 in the suppression of 3T3‐L1 adipogenesis, we sought to study the effect of CUL4 knockdown on 3T3‐L1 adipogenesis (Fig EV3A). The depletion of CUL4A resulted in a hyperproliferation phenotype, while CUL4B depletion resulted in a cell cycle defect and reduced cell proliferation (Fig EV3B and C). Oil Red O staining indicates that while 3T3‐L1 adipogenesis was enhanced by CUL4A depletion, it was substantially inhibited by CUL4B depletion (Fig EV3D). It is presently unclear to us why the knockdown of Cul4a and Cul4b has opposite effects on 3T3‐L1 adipogenesis. However, the function of CUL4A in promoting adipogenesis needs to be cautiously interpreted as it is likely that multiple distinct CRL4A complexes are assembled in 3T3‐L1 cells.

Figure EV3. CUL4A and CUL4B RNAi phenotypes cannot be clearly assessed in 3T3‐L1 adipogenesis.

1. 3T3‐L1 preadipocytes were stably transfected with either retroviral‐based shRNAs targeting GFP control, Cul4a, or Cul4b. Knockdown efficiency and the expression levels of various proteins were analyzed by immunoblotting as indicated. Elevated fatty acid synthase (FAS) in CUL4Ai cell line may reflect hyperproliferative activity of these cells.
2. To measure proliferation, cells in (A) were plated at equal density (1.5 × 10⁵). The cell number was counted every day for 7 days. An average from two separate plates is plotted for each cell line.
3. To analyze cell cycle distribution, cells in (A) were plated at equal density (1 × 10⁶). After 24 h, cells were trypsinized and fixed overnight in 75% ethanol at 4°C. Fixed cells were stained with propidium iodide and then analyzed by flow cytometry to measure DNA content.
4. 3T3‐L1 stable cells described in (A) were adipogenically induced. Their adipogenic potential was assessed by Oil Red O (ORO) staining; plate view (top) and microscopic view (bottom), scale bar 0.05 mm.

To more specifically demonstrate the function of CUL4–WDTC1 interaction in the suppression of 3T3‐L1 adipogenesis, we therefore carried out rescue experiments. Utilizing a lentiviral knockdown‐rescue vector strategy 18, we engineered 3T3‐L1 cells to simultaneously knockdown endogenous Wdtc1 by shRNA and ectopically re‐express shRNA‐resistant WDTC1 encoding either WT or DDB1‐binding mutants ΔH or RARA to levels comparable to endogenous WDTC1 (Fig 3A). We found that knocking down Wdtc1 dramatically enhanced 3T3‐L1 differentiation as assayed by ORO (Fig 3B), triglyceride accumulation (Fig 3C), and adipogenic gene expression (Fig 3D). While the re‐expression of WT WDTC1 almost completely rescued the loss of adipogenic suppression in all three assays, adding back either ΔH or RARA mutants failed to rescue the adipogenic defects resulting from the depletion of endogenous Wdtc1. Consistent with its partial disruption of DDB1 binding, the RARA mutant exhibited a weaker phenotype than the ΔH mutant, supporting the interpretation that the RARA mutant represents a hypomorphic form of WDTC1, while the ΔH mutant acts as a dominant negative. Collectively, these results demonstrate a strong parallel between CRL4^WDTC1 complex formation and WDTC1 function, underscoring the critical importance of the CRL4–WDTC1 interaction to the observed adipogenic suppression by WDTC1.

Figure 3. CRL4 binding mutants of WDTC1 cannot rescue WDTC1 knockdown phenotype.

• A
  Schematic of WDTC1 knockdown‐rescue lentiviral vector (top). 3T3‐L1 preadipocytes were lentivirally transduced to generate stable cells expressing a non‐specific (NS) shRNA control, shWDTC1 targeting endogenous Wdtc1 mRNA, or cells simultaneously expressing shWDTC1 and shRNA‐resistant WDTC1; bottom, confirmation of knockdown and WDTC1 ectopic expression.
• B–D
  3T3‐L1 stable cells described in (A) were adipogenically induced. Their adipogenic potential was assessed by ORO (B), triglyceride quantitation (C), and adipogenic gene expression (D).

Data information: In (C, D), bars represent the mean ± SEM from 3 independent experiments. *P < 0.05, **P < 0.005, significant change compared to NS control. Statistical significance was analyzed by paired two‐tailed Student's t‐tests. Source data are available online for this figure.

CRL4^WDTC1 promotes histone H2AK119 ubiquitylation

To gain insight into the mechanism by which CRL4^WDTC1 suppresses adipogenesis, we carried out two independent coupled immunoprecipitation and mass spectrometry (IP‐mass spec) analyses of WDTC1 complexes in 3T3‐L1 cells. Validating the IP‐mass spec analyses, several known CRL4^WDTC1 components were identified, including DDB1, CUL4B, CUL4A, ROC1, and all nine subunits of COP9 signalosome subunits (Fig EV4A). One potential novel WDTC1‐interacting protein is fatty acid synthase (FAS), a critical and positive regulator of lipid accumulation. We hypothesized that regulation of FAS by the CRL4^WDTC1 complex may contribute to the antiadipogenic action of WDTC1. We therefore characterized the interaction between WDTC1 and FAS and performed preliminary characterization of a few other candidates as well. However, our data showed that CRL4^WDTC1 E3 ligase does not appear to regulate FAS steady‐state protein levels and reached similar conclusions for other candidates we examined such as HDAC3 and PPARγ‐2 (Fig EV4B–D).

Figure EV4. WDTC1 interacting proteins.

1. List of WDTC1‐WT interacting proteins identified by mass spectrometry. CRL4 subunits and associated proteins are highlighted in bold; FAS and histone H2A are highlighted in bold blue. Complete data sets are presented in the source data.
2. Lysates were prepared from 293T cells transiently expressing empty vector control (−), Flag‐tagged wild‐type, or ΔH mutant WDTC1 proteins. Flag‐WDTC1 complexes were immunoprecipitated with anti‐FLAG, and their associated proteins were detected by immunoblotting as indicated.
3. Lysates were prepared from 293T cells transiently coexpressing Flag‐WDTC1 wild‐type or ΔH, and either empty vector control (−) or HA‐tagged PPARγ‐2 proteins. HA‐PPARγ‐2 complexes were immunoprecipitated with anti‐HA and their associated proteins were detected by immunoblotting as indicated.
4. 293T cells were transfected with HA‐ubiquitin along with empty vector control (−) or Flag‐PPARγ‐2, and siRNA scramble control (−) or siDDB1 as indicated. Transfected cells were lysed under denaturing conditions to obtain whole‐cell extracts (WCE) and immunoprecipitated with anti‐FLAG. PPARγ‐2 ubiquitylation was evaluated by immunoblotting as indicated.
5. Nucleosomal proteins were released from insoluble chromatin preparation from induced 3T3‐L1 cells stably expressing control (EV) or Flag‐WDTC1 proteins by deoxyribonuclease (DNase) digestion, and the resulting extract was used for immunoprecipitation by anti‐FLAG and immunoblotted as indicated.
6. Flag‐WDTC1 complexes were affinity‐purified from 293T cells and used as the source of E3 ligase in in vitro ubiquitylation reactions with recombinant H2A as substrate.

Source data are available online for this figure.

We then investigated the possibility that histone H2A is a CRL4^WDTC1 substrate. This was inspired in part by the established role of H2AK119 monoubiquitylation (referred to as H2AK119ub1 hereafter) as a key transcription repressive mechanism in the control of development and differentiation, and also because it has been demonstrated that the antiadipogenic function of WDTC1 is lost upon nuclear exclusion and WDTC1 physically interacts with ectopically expressed histones 4. Further, ectopic WDTC1 expression is associated with reduced adipogenic marker expression (Fig 2D; 4).

We first examined the subcellular distribution of Flag‐WDTC1 and found that while it is abundantly recovered in the cytosolic fraction, it is also clearly present in the nuclear fractions, and in the MNase‐digested chromatin fraction (Fig 4A). We then performed immunoprecipitation experiments with nuclear extracts and demonstrated that WDTC1 associated with H2A, as well as H2B and H3 (Fig 4B). In this experiment, we noted that WDTC1 interacted with H2A, but not the presumably monoubiquitylated form of H2A, an observation that is consistent with the possibility that CRL4^WDTC1 catalyzes H2AK119ub1. Chromatin coimmunoprecipitation assay confirmed the binding between WDTC1 and nucleosomal histones in vivo at physiologic levels (Fig EV4E). Notably, the DDB1 binding mutant, WDTC1^ΔH, retained histone binding activity, indicating that histone binding is independent of CRL4 binding. Because WDTC1^ΔH mutant disrupts DDB1 binding and thus disrupts probable ubiquitylation activity, and appears to functionally antagonize WDTC1 during adipogenesis, we took advantage of this mutant to investigate the effect of WDTC1 expression on histone H2AK119ub1 levels. We examined the global level of H2AK119ub1 in preadipocyte and induced 3T3‐L1 cells stably expressing either a vector control, WT WDTC1 or WDTC1^ΔH mutant (Fig 4C). We did not detect a notable change in H2AK119ub1 levels among these cell lines in the preadipocyte state. In contrast, expression of WT WDTC1 resulted in a modest, but significant and reproducible, increase in H2AK119ub1 while WDTC1^ΔH expression caused a slight decrease of H2AK119ub1. This result suggests that WDTC1‐promoted H2AK119ub1 is linked to adipogenesis and is dependent on the association with the CRL4 E3 complex.

Figure 4. CRL4^{WDTC 1} E3 ligase promotes H2AK119 ubiquitylation.

1. Induced 3T3‐L1 cells stably expressing Flag‐WDTC1 were subjected to subcellular fractionation. Aliquots of total cell extract (TCE) and subcellular fractions were analyzed by immunoblot as indicated; fraction purity was assessed by α‐tubulin and histone H3 antibodies.
2. Chromatin‐enriched nuclear extracts from 293T cells cotransfected with Flag‐WDTC1 and HA‐tagged core histones were immunoprecipitated (IP) with anti‐FLAG and immunoblotted (IB) as indicated. The probable H2AK119ub1 band detected in the nuclear extract is indicated by an asterisk.
3. Soluble chromatin extracts were prepared from preadipocyte or induced 3T3‐L1 cells stably expressing control (EV) and Flag‐WDTC1 proteins. H2A monoubiquitylation was detected by H2AK119ub1‐specific antibody and total H2A served as a loading control. Relative intensity represents ratio of H2AK119ub1 over total H2A signal normalized to EV control.
4. Affinity‐purified Flag‐WDTC1 wild‐type and mutant ΔH complexes from 293T cells were the source of E3 ligase in in vitro ubiquitylation assays. Copurification of CRL4 E3 ligase proteins was confirmed by immunoblotting and Coomassie Blue staining (left panels). In vitro ubiquitylation assays of recombinant H2A were performed in the presence of the indicated proteins. Reaction products were resolved by SDS–PAGE, and ubiquitylated H2A was detected by anti‐H2AK119ub1 (right panel).
5. In vitro ubiquitylation assays were performed with purified wild‐type and mutant histone proteins as substrates. H2A ubiquitylation was assessed by immunoblotting with anti‐ubiquitin and various antibodies as indicated.

Data information: In (C), bars represent the mean ± SD from 2 (preadipocyte) or 3 (induced) independent experiments. *P < 0.05, significant change compared to EV control. Statistical significance (induced only) was analyzed by paired two‐tailed Student's t‐tests. Source data are available online for this figure.

To test whether H2AK119 is a direct target of CRL4^WDTC1 E3 ligase, we performed in vitro ubiquitylation assays. We affinity‐purified Flag‐tagged WT WDTC1 or WDTC1^ΔH complexes from 293T cells and confirmed the presence of CRL4 components in the WT, but not ΔH, WDTC1 eluted fraction (Fig 4D, left). Recombinant H2A was incubated with purified WDTC1 complexes and reaction products were resolved by SDS–PAGE, followed by immunoblotting with H2AK119ub1 antibodies. The results show that H2A is robustly ubiquitylated by WT WDTC1, but not WDTC1^ΔH complexes (Figs 4D, right, and EV4F), demonstrating that H2A was a specific substrate of CRL4^WDTC1 in this in vitro system. We noted that multiple ubiquitylated H2A forms were detected, perhaps due to the high efficiency nature of in vitro ubiquitylation reactions. To confirm that K119 residue is the primary target of CRL4^WDTC1, wild‐type H2A and its mutants, H2AK118R and H2AK119R, were tested in in vitro ubiquitylation assays (Fig 4E). While disrupting the K118 residue had very little effect, K119R mutation resulted in a substantial loss of H2A ubiquitylation by the CRL4^WDTC1 E3 ligase. These results collectively suggest that CRL4^WDTC1 promotes H2AK119ub1 in vitro and in vivo in an adipocyte lineage‐specific manner.

The most extensively characterized H2A E3 ligase is the RING1A/B of PRC1 complex which plays a critical and evolutionarily conserved role in developmental control in flies and mammals. It has been unclear whether and how a single E3 ligase regulates different biological processes that involve epigenetic silencing, especially when considering the abundance of H2AK119ub1 in cells (5–15% of total H2A). Two reports support the latter possibility, 2A‐HUB (DZIP3/hRUL138) that mediates selective repression of a specific set of chemokine genes in macrophages 19 and CRL4B^RBBP4/7 that collaborates with PRC2 to repress various genes involved in cell growth and migration 20. Our study supports the possibility that in addition to the historic PRC1, there may exist additional H2AK119 E3 ligases that function in epigenetic repression.

Multiple lines of genetic evidence from Drosophila, C. elegans, and mouse, and human epidemiology, have linked the function of WDTC1/Adp to the negative regulation of adipogenesis and fat metabolism. In this study, we demonstrate that the regulation of adipogenesis by WDTC1 requires its assembly into a CRL4 E3 ligase complex. A recent RNAi screen in Drosophila identified 47 genes which, when silenced, produced flies with high body fat (so‐called antiobesity genes) 21. Among them are Cul4 and Csn4, which encodes a subunit of the COP9 signalosome that functions to activate CRL E3 ligases. These new findings provide additional support for the function of CRL4^WDCT1 as a negative regulator of adipogenesis and fat formation.

Materials and Methods

Plasmids

The human WDTC1 (NM_015023.4) coding sequence was PCR‐amplified from a HepG2 cDNA library and cloned into pENTR^™/D‐TOPO (Invitrogen) to generate a WDTC1 entry vector. WDTC1^ΔH was subcloned from pENTR‐WDTC1 and WDTC1^RARA was generated by sequential QuikChange site‐directed mutagenesis (Stratagene) to pENTR‐WDTC1. WDTC1 entry vectors were recombined with Gateway‐adapted destination vectors (kind gift of Dr. K. I. Nakayama) p3XFLAG for transient and pMX‐FLAG‐puro for stable expressions. To generate WDTC1 knockdown‐rescue vectors, shRNAs targeting mouse Wdtc1 and shRNA refractory human WDTC1 coding sequences were cloned into pLL‐5.5‐IRES‐EGFP (kind gift of Dr. J. Bear), as described previously 18, 22. Mouse Cul4a and Cul4b shRNA sequences were designed using BLOCK‐IT RNAi tool (Invitrogen) and cloned into pMKO.1‐puro retroviral vector for stable knockdown of respective mRNAs in 3T3‐L1 cells. All plasmids were sequence‐verified, and shRNA oligos are listed in Table EV1.

Cell transfections, retroviral and lentiviral infections

293T cells were transfected at 1:3 plasmid‐to‐reagent ratio using Fugene 6 transfection reagent (Promega) for transient overexpression or transfected with 50 nM of siRNA for knockdown using Lipofectamine 2000 reagent (Invitrogen) according to manufacturers' protocols. Cells were incubated for 72 h for knockdown by siRNA; siRNA sequences were described previously 23 and listed in Table EV1. For retroviral or lentiviral production, 293T cells were cotransfected with various plasmids for viral packaging as described 24. For WDTC1 knockdown‐rescue transfections, rescue vector plasmids were cotransfected with shRNA alone plasmids (2:1) for more complete knockdown of endogenous Wdtc1 mRNA. Viral media was collected 48 h post‐transfection at two 12‐h intervals, syringe‐filtered through 0.45‐μm filter, and polybrene (8 μg/ml)‐supplemented. Preconfluent 3T3‐L1 preadipocytes were infected with the viral media twice within a 24‐h interval. Cells were split to maintain preconfluency. Retrovirally transduced cells were selected with 4 μg/ml puromycin and maintained in selection media until adipogenic induction. Lentivirally transduced pLL‐5.5‐IRES‐EGFP cells were visually selected for > 90% of cells with EGFP expression.

Antibodies, immunoprecipitation, and immunoblotting

A rabbit polyclonal antibody (Abgent, AP4944b) and a rabbit monoclonal antibody (Abcam, ab174294, 25) were used to detect WDTC1. DDB1, CUL4A, and ROC1 antibodies were generated by our laboratory as described previously 23, and all other antibodies used in this study are commercially available and listed in Table EV2. Cells were lysed on ice with a NP‐40 lysis buffer [50 mM Tris (pH 8.0), 150 mM NaCl, 10% glycerol, 1 mM EDTA, and 0.1% NP‐40] supplemented with Halt^™ (Thermo Sci.) protease/phosphatase inhibitor cocktail (PPIC). For immunoprecipitation experiments, clarified lysates prepared from Flag‐WDTC1 expressing cells were immunoprecipitated by anti‐FLAG M2 affinity gel (Sigma) overnight at 4°C. Immunoprecipitates were washed 3–5 times in lysis buffer with rotation, eluted by boiling in Laemmli buffer, resolved by SDS–PAGE, transferred to PVDF membrane (Millipore), and detected by immunoblotting with indicated antibodies. Immunoblotting was performed following standard protocols; blots were cut into strips to allow probing of the same gel with multiple antibodies when possible. Blots were developed using an ECL reagent (GE Amersham). Protein band densitometry analyses were performed using ImageJ software (U.S. National Institutes of Health).

3T3‐L1 differentiation, Oil Red O staining, and triglyceride assay

3T3‐L1 cells were cultured and differentiated as described previously 26. 3T3‐L1 preadipocytes were cultured in DMEM supplemented with 1× penicillin–streptomycin solution (Corning) and 10% (v/v) fetal calf serum (Colorado Serum Co.). To induce differentiation, 2 day post‐confluent 3T3‐L1 cells (day 0) were treated with an induction media containing 1 mM dexamethasone, 0.5 mM isobutylmethylxanthine (IBMX), and 1 μg/ml insulin (all from Sigma) in 10% fetal bovine serum supplemented DMEM. Two days later, induction media was replaced with 1 μg/ml insulin only for the duration of the experiment and media was changed every 2 days. A schematic summarizing 3T3‐L1 differentiation with time points indicated for different experimental procedures is shown (Fig EV2B). To stain lipid droplets, cells were fixed in 3.7% buffered formaldehyde and stained with 0.3% Oil Red O. Triglyceride levels were measured with the Triglyceride Quantitation Kit and manufacturer's protocol (BioVision), and levels were normalized to sample protein concentrations by BCA assay (Thermo Sci.).

Real‐time quantitative PCR

Total RNA was isolated from cells using Trizol (Invitrogen) and purified to remove residual phenol/chloroform using RNAeasy Mini cleanup (Qiagen). First‐strand cDNA was synthesized with 800 ng of RNA using Superscript II reverse transcriptase kit (Invitrogen). Quantitative PCR (qPCR) was performed in duplicate using 1 μl of cDNA and SYBR Green PCR master mix (Applied Biosystems) in an Applied Biosystems 7900HT Fast Real‐Time PCR system. Gene expression was normalized to TPB levels and analyzed by ΔΔCt method. qPCR primers for Wdtc1 and SREBP1c were generated using Primer‐BLAST (NCBI), and all other primers were reported previously 27, 28 and listed in Table EV1.

Subcellular fractionation and chromatin extraction

Cell fractionation and soluble chromatin extraction by MNase digestion was performed as described previously 29. Pelleted 3T3‐L1 cells (8 × 10⁶) were resuspended and incubated on ice for 5 min in 200 μl buffer A [10 mM HEPES (pH 7.9), 10 mM KCl, 1.5 mM MgCl₂, 0.34 M sucrose, 10% glycerol, 1 mM DTT and Halt^™ PPIC] plus 0.1% Triton X‐100. Nuclei were recovered in pellet 1 (P1) after centrifugation (4 min, 1,300 g, 4°C), and the supernatant (S1) was further clarified by centrifugation to obtain cytosolic fraction (S2). For nucleoplasmic fraction, P1 pellet was washed once in buffer A and lysed in 100 μl buffer B (3 mM EDTA, 0.2 mM EGTA, 1 mM DTT, and Halt^™ PPIC), and centrifuged and the supernatant was recovered (S3). For chromatin‐enriched fraction, P1 pellet was resuspended in buffer A and treated with 2 U MNase (1 min, 37°C) and centrifuged; treated nuclei were then lysed in 100 μl buffer B and the chromatin fraction was recovered in the supernatant (S3) of MNase treated nuclei. Immunoprecipitation with soluble chromatin extracts prepared by DNase treatment was performed as described previously 30.

In vivo and in vitro ubiquitylation assays

In vivo ubiquitination assays were performed as described previously 31. For the detection of ubiquitylated WDTC1 in vivo, 293T cells were first transfected with HA‐ubiquitin, split ~10 h later, transfected again with various combinations of plasmids and siRNAs at 24 h post‐HA‐ubiquitin transfection to minimize variations in HA‐ubiquitin expression across plates. At 67 h post‐siRNA transfection, cells were treated with MG132 (10 μM) and collected 5 h later. Cells were lysed under denaturing conditions in 1% SDS buffer [50 mM Tris (pH 7.5), 0.5 mM EDTA, 1% SDS, 1 mM DTT] by boiling for 10 min, and clarified lysate was immunoprecipitated in 0.1% SDS by 10‐fold dilution with NP‐40 buffer. Flag‐tagged WDTC1 was immunoprecipitated by anti‐FLAG M2 agarose beads, resolved by SDS–PAGE, and ubiquitylated WDTC1 was detected by immunoblotting with anti‐HA antibodies. For in vitro ubiquitylation assays, WDTC1 immunocomplexes were purified from transiently transfected 293T cells expressing either Flag‐WDTC1‐WT or Flag‐WDTC1‐ΔH. Flag‐tagged WDTC1 was immunoprecipitated with anti‐FLAG M2 agarose beads overnight on a rotator at 4°C. Immunocomplexes were washed three times in lysis buffer and twice in TBS followed by elution with molar excess of Flag peptide (Sigma). Ubiquitylation reactions were performed according to manufacturer's protocol using an ubiquitin conjugation kit (Enzo Life Sci.). In 50 μl reaction volume, unless noted otherwise in figure, 100 nM of either WT or ΔH immunocomplexes as the source of E3 and 200 nM of human recombinant histone H2A substrate (New England Biolabs) were combined with a ubiquitylation buffer containing 100 nM E1, 1 μM E2‐UbcH5c, 1 μM human recombinant WT ubiquitin or ubiquitin Mutant No K (Boston Biochem), 1 U inorganic pyrophosphatase, 1 mM DTT, and 5 mM Mg‐ATP. Reactions were incubated at 37°C for 30 min and terminated by addition of SDS sample buffer. Reaction products were resolved by SDS–PAGE and detected by immunoblotting with either anti‐H2AK119ub1 or anti‐ubiquitin antibodies.

Mass spectrometry

WDTC1‐associated proteins were isolated from 3T3‐L1 preadipocytes and induced adipocytes stably expressing Flag‐WDTC1‐WT and Flag‐WDTC1‐ΔH proteins by FLAG affinity purification, and cells transfected with an empty vector (EV) served as contamination controls for Flag immunoprecipitation. For analyses of interacting proteins in preadipocytes, WDTC1‐WT and WDTC1‐ΔH complexes were purified from ten 15‐cm plates at day 0. For analyses of interacting proteins in induced adipocytes, day 0 preadipocytes were induced to differentiate and 2 days post‐induction, WDTC1 complexes were purified from five (WT) or ten (ΔH) 15‐cm plates; twice the number of cells were used in the purification of WDTC1‐ΔH complexes due to its lower stable expression compared to WDTC1‐WT. Cell pellets were lysed for 30 min on a rotating platform at 4°C in NP‐40 lysis buffer [50 mM Tris (pH 8.0), 150 mM NaCl, 10% glycerol, 1 mM EDTA, and 0.1% NP‐40] supplemented with Halt^™ PPIC. Cell extracts were clarified by centrifugation, and the soluble fractions were incubated with anti‐FLAG M2 agarose beads (Sigma) overnight on a rotating platform at 4°C. Immunoprecipitated proteins were washed ten times by inverting tubes in 10‐ml wash volumes; five times in NP‐40 buffer followed by five times in PBS. Protein complexes were eluted with 50 μl of 3XFLAG peptide (200 μg/ml) for 5 min by shaking (800 rpm) on a Thermomixer (Eppendorf) at 37°C; each sample was eluted three times and fractions were combined in a single tube on ice. A small amount of total purified protein (~2–5%) was resolved by SDS–PAGE and visualized by either silver staining or Coomassie Blue staining for confirmation. Eluted fractions were either in solution digested (preadipocytes) or in‐gel digested (induced cells) at 37°C using trypsin (Promega). Digested peptides were desalted and purified by C₁₈ resin (C₁₈ columns or C₁₈ Zip‐Tip from Pierce or Millipore, respectively) following manufacturers' instructions. The identities of eluted proteins were determined by mass spectrometry analyses carried out at the UNC Proteomics Technology Development Core Facility using a LTQ‐Velos‐Orbitrap (Thermo Sci.) mass spectrometer coupled to a 2D nano‐ultra‐liquid chromatography system (Eksigent). The data were processed using MaxQuant software, and the MS/MS data sets were searched against UniProt mouse database. Processed data for Flag‐WDTC1‐WT and Flag‐WDTC1‐ΔH interacting proteins were manually sorted against contaminating proteins identified in EV control samples.

Sequence alignment

WDTC1 protein sequences from Homo sapiens (Hs), Mus musculus (Ms), Xenopus laevis (Xl), Danio rerio (Dr), Drosophila melanogaster (Dm), Caenorhabditis elegans (Ce), and Arabidopsis thaliana (At) were aligned to compare H‐box sequence similarity in different species. Sequences were aligned using CLUSTAL 2.1 multiple sequence alignment tool.

Molecular modeling of WDTC1 and CRL4^WDTC1

Modeling was performed at the University of North Carolina R. L. Juliano Structural Bioinformatics Core. For modeling WDTC1, suitable templates for the H‐box, WD40, and TPR domains of WDTC1 were identified using HHpred 32. Three template structures were utilized as follows: WDTC1 H‐box in complex with DDB1 (PDB 3I7N; 13), DDB2 in complex with CRL4 (PDB 4A0K; 15), and TPR domains of Sgt2 (PDB 3SZ7; 33). The Hbox and WD40 domains were modeled simultaneously using the Modeller software package 34 and two template structures: PDB ID 3I7N and PDB: 4A0K. The TPR domains were modeled separately based on the template structure of Sgt2, a TPR structure from Aspergillus fumigatus (PDB ID 3SZ7). The model of the WDTC1‐CRL4 complex was based on the structure of the DDB2‐CRL4 complex (PDB ID 4A0K) with WDTC1 superimposed on and then replacing DDB2. The TPR domains were manually moved to a location near their insertion within the beta‐propeller. Models were rendered using Pymol.

Author contributions

BSG and YX designed research; BSG and FY performed experiments; MDS provided experimental support; YY and XC performed mass spectrometry analyses; and BSG and YX analyzed data and wrote the manuscript.

Conflict of interest

The authors declare that they have no conflict of interest.

Supporting information

Expanded View Figures PDF

Table EV1

Table EV2

Source Data for Expanded View

Review Process File

Source Data for Figure 1

Source Data for Figure 2

Source Data for Figure 3

Source Data for Figure 4

EMBO Reports (2016) 17: 638–647