Elucidation of the 14-3-3ζ interactome reveals critical roles of RNA splicing factors during adipogenesis

Adipogenesis is facilitated by a complex signaling network requiring strict temporal and spatial organization of effector molecules. Molecular scaffolds, such as 14-3-3 proteins, coordinate such events, and we have previously identified 14-3-3ζ as an essential scaffold in adipocyte differentiation. The interactome of 14-3-3ζ is large and diverse, and it is possible that novel adipogenic factors may be present within it. Mouse embryonic fibroblasts from mice over-expressing a TAP-epitope-tagged 14-3-3ζ molecule were generated, and following the induction of adipogenesis, TAP-14-3-3ζ complexes were purified, followed by mass spectrometry analysis to determine the 14-3-3ζ interactome. Over 100 proteins were identified as being unique to adipocyte differentiation, of which 56 were novel interacting partners. Previously established regulators of adipogenesis (ie, Ptrf/Cavin1 and Phb2) were found within the 14-3-3ζ interactome, confirming the ability of this approach to identify regulators of adipocyte differentiation. An enrichment of proteins in the interactome related to RNA metabolism, processing, and splicing was identified, and analysis of transcriptomic data revealed that 14-3-3ζ depletion in 3T3-L1 cells affected the alternative splicing of mRNA during adipocyte differentiation. Of the RNA splicing factors within the 14-3-3ζ interactome, depletion of Hnrnpf, Hnrnpk, Ddx6, and Sfpq by siRNA revealed essential roles of these proteins in adipogenesis and their roles in the alternative splicing of Lpin1. In summary, novel adipogenic factors can be detected within the 14-3-3ζ interactome, and further characterization of additional proteins within the 14-3-3ζ interactome has the potential of identifying novel targets to block the expansion of adipose tissue mass that occurs in obesity.


Introduction
Central to the development of obesity are the increases in number and size of adipocytes according to nutrient availability (1,2). Despite various therapies to limit weight gain and promote weight loss, it is surprising that none specifically target the adipocyte to limit its expansion or growth (1,2). The complex transcriptional network and cellular processes that govern the differentiation of adipocyte progenitor cells contributes to the difficulty in targeting adipocytes therapeutically (1,2). Protein phosphorylation is a key post-translational modification that determines the activation state, subcellular localization, and stability of adipogenic regulators (3)(4)(5)(6)(7). Furthermore, phosphorylation status also determines their interactions with molecular scaffold proteins, which aid in the coordination of complex transcriptional networks (3,4) .
We previously identified the molecular scaffold, 14-3-3ζ, as a critical regulator of glucose homeostasis and adipogenesis (4,8,9). Specific to the adipocyte, systemic deletion of 14-3-3ζ in mice significantly reduced visceral adiposity and impaired adipocyte differentiation, whereas transgenic over-expression of 14-3-3ζ exacerbated high-fat diet induced obesity (4). The hedgehog transcription factor, Gli3, was identified as a critical downstream effector in 14-3-3ζ-mediated adipogenesis, but the diversity of proteins in the 14-3-3ζ interactome suggest the possibility that other interacting proteins or pathways parallel to Gli3 may be also involved.
Unbiased approaches, such as proteomics and transcriptomics, can lead to the discovery of novel factors that drive adipogenesis, in addition to providing insight into physiological pathways influenced by adipogenic regulators like 14-3-3ζ (4,10-15). All seven mammalian 14-3-3 isoforms have large, diverse interactomes (8,(16)(17)(18), and they are dynamic and change in response to various stimuli (11)(12)(13)19). Thus, inducing pre-adipocytes to differentiate may permit the identification of novel differentiation-specific factors within the 14-3-3ζ interactome and reveal pathways and biological processes that are essential to the development of a mature adipocyte.
To elucidate the 14-3-3ζ interactome during adipogenesis, we employed a proteomic-based discovery approach. Herein, we report that previously established factors required for adipogenesis (ie, Ptrf/Cavin1 and Phb2) can be detected in the interactome, and novel factors, such as those involved in RNA splicing, are also enriched in the interactome during differentiation. To test for their roles in adipogenesis, siRNA knockdown approaches were used and revealed the requirement for RNA splicing factors, such as Hnrnpf, Sfpq, and Ddx6.
Taken together these findings demonstrate the usefulness of examining the interactome of 14-3-3 proteins in the context of a physiological process, such as adipocyte differentiation, and highlight the ability to find novel functional regulators through this approach. Understanding how the interactome is influenced by disease states, such as obesity, may lead to the identification of novel proteins that contribute to disease pathogenesis.

Mass spectrometry
Equal amounts of cell lysates from undifferentiated and differentiated TAP-14-3-3ζ MEFs were subjected to an overnight incubation with IgG coupled to protein-G beads (ThermoFisher Scientific) in RIPA buffer. Bound proteins from each pull down were eluted with 1X SDS sample buffer without reducing agents and separated by SDS-PAGE prior to in-gel digestion (20). For each sample, peptides from three fractions (<50KDa, >50KDa, IgG bands) were then purified on C-18 stage tips (21) and analyzed using a LTQ-Orbitrap Velos (ThermoFisher Scientific) as previously described (22). Data were processed with Proteome Discoverer v. 1.2 (ThermoFisher Scientific) followed by a Mascot analysis (2.3.0, Matrix Science, Boston, MA) using the Uniprot-Swissprot_mouse protein database (05302013, 540261 protein sequences). Only proteins with at least two peptides (false positive discovery rate <=1%) in one of the two samples were retained. Two independent pulldowns were used for mass spectrometry and proteomic analysis. Proteins were analyzed with DAVID and String-Db to analyze proteins based their biological processes (23,24).

Analysis of differential exon usage
To understand how adipocyte differentiation and depletion of 14-3-3ζ affected alternative splicing of mRNA, differential exon usage via DEXSeq was used as a surrogate measurement (25). Our previous transcriptomic data [GSE60745] (26) were aligned to the mouse genome (Ensembl NCBIM37) via Tophat (v. 2.1.1), and the number of reads mapping to a particular exon were compared to the total number of exons in a given gene (25). A false discovery rate (FDR) of 0.05 was used to filter results. This dataset was also analyzed to examined how depletion of 14-3-3ζ or differentiation affects the expression profile of target genes. Genes identified by DEXSeq were subjected to gene ontology analysis to categorize genes by biological function (27).
Alternatively, analysis of Lpin1 splicing was performed by RT-PCR, as described previously (28). PCR products were resolved on an agarose gel, followed by densitometric analysis of splice variants by ImageJ (29).

siRNA-mediated knockdown, RNA isolation and quantitative PCR
3T3-L1 cells were seeded at a density of 75,000 per well prior to transfection with control siRNA or target-specific Silencer Select siRNAs (ThermoFisher Scientific). Transfection was performed using

Regulation of mRNA processing by 14-3-3ζ
Using our previous transcriptomic analysis of differentiating 3T3-L1 cells (26), we re-analyzed the effects of differentiation and 14-3-3ζ depletion on RNA processing. Differential exon usage (DEXSeq) was used as a surrogate measure of alternative splicing of mRNA ( Figure 2A) (25). Any changes in splice variant levels were not due to global effects of 14-3-3ζ depletion on RNA transcription, as no gross differences in the incorporation of a uracil analog were detected ( Figure 2B). Comparison of genes that displayed differential exon usage at 24 and 48 hours post differentiation revealed that 163 and 172 genes, respectively, that were unique to each time point ( Figure 2C). Gene ontology analysis revealed that at each time point, distinct groups of genes were alternatively spliced ( Table 4). The use of this approach to detect genes with differential exon usage was validated by the ability to detect Pparg variants after 48 hours of differentiation ( Figure S1) (35). The effect of 14-3-3ζ depletion was assessed at each time point, and 78, 37, and 36 genes were affected following 14-3-3ζ knockdown at 0, 24, and 48 hours, respectively, after the induction of differentiation ( Figure 2D). However, only in undifferentiated 3T3-L1 cells could enrichments in genes associated with macromolecular complex assembly (GO:0065003, p=3.44 x 10 -3 ), macromolecular complex subunit organization (GO:0043933, p= 7.56 x 10 -4 ), and regulation of biological quality (GO:0065008, p=9.51 x 10 -3 ) be detected by gene ontology analysis. Collectively, these data demonstrate that adipogenesis promotes the alternative splicing of genes and this process can be influenced by 14-3-3ζ.

Identification of known and novel regulators of adipocyte differentiation
Within the 14-3-3ζ interactome, we were able to detect proteins with known roles in adipogenesis, such as Ptrf/Cavin1 and prohibitin-2 (Phb2) (36-40) and confirmed their roles in adipocyte differentiation ( Figure 3).
This confirmed that known regulators of adipogenesis can be detected within the 14-3-3ζ interactome and suggested the possibility that novel factors could be identified. Additional proteins in the 14-3-3ζ interactome, such as Fragile-X mental retardation protein-1 (Fmr1) and Rpn2, were also examined for their roles in adipogenesis, as they have previously been shown to be associated with obesity or weight gain (41,42).
However, siRNA-mediated knockdown of either protein had no effect on 3T3-L1 differentiation, indicating that these proteins are not required for adipogenesis ( Figure 3A-D), at least in this in vitro model system.
As proteins associated with RNA processing and splicing were highly enriched during differentiation (Table 3), we sought to examine contribution of RNA splicing factors to adipogenesis. Using siRNA in 3T3-L1 pre-adipocytes, 8 splicing factors, which were identified in our proteomic analysis of the 14-3-3ζ interactome (Table 1), were screened for their roles in 3T3-L1 adipogenesis. They were chosen by the number of connections exhibited within each cluster of proteins ( Figure 1D) (24). Of note, mRNA levels of the chosen splicing factors were generally unaffected by knockdown of 14-3-3ζ; however, some splicing factors were influenced by differentiation ( Figure S2) (26). Transient knockdown of Ddx6, Sfpq, Hnrnpf, or Hnrnpk was sufficient to impair 3T3-L1 differentiation, as assessed by Oil Red-O incorporation ( Figure 4). Closely related proteins with similar roles, such as Ddx1, Nono, Hnrnpm, and Syncrip/Hnrnpq were not required for 3T3-L1 adipogenesis ( Figure 4B, C). Knockdown of Ddx6 or Hnrnpk by siRNA did not have an effect of Pparg, which suggests that these factors act downstream of the mRNA expression of this master transcription factor ( Figure   4D). Other pro-or anti-adipogenic genes are alternatively spliced during adipocyte differentiation. For example, Lpin1 mRNA is spliced to generate Lipin-1α and Lipin-1β, which have differential roles on adipogenesis (28). To examine the effect of depletion of 14-3-3ζ, Hnrnpf, Ddx6, Hnrnpk, and Sfpq on Lpin1 splicing, 3T3-L1 cells were transiently transfected with siRNA, followed by the induction of differentiation. Gene silencing of all target genes was found to prevent the generation of the Lpin-1α variant during differentiation ( Figure 4E, F). Collectively, these findings demonstrate that novel regulators of adipogenesis can be identified within the interactome of 14-3-3ζ and highlight the involvement of 14-3-3ζ in regulating the alternative splicing of mRNA.

Discussion
In the present study, affinity proteomics was used to determine how adipogenesis influences the interactome of 14-3-3ζ. Surprisingly, the interactome was dynamic, as differentiation altered the landscape of proteins that interact with 14-3-3ζ. This approach also permitted the identification of known adipogenic factors within the 14-3-3ζ interactome and revealed novel proteins that are required for adipocyte differentiation. An enrichment of proteins associated with RNA processing and splicing were detected, and the novel contributions of RNA splicing factors, such as Hnrnpf, Ddx6, and Sfpq, in adipogenesis were identified. The usefulness of this approach was also evident in the ability to identify process that may be regulated by 14-3-3ζ during adipocyte differentiation.
We previously identified an essential function of the hedgehog signaling effector Gli3 in 14-3-3ζregulated adipocyte differentiation (4). However, due to the large, diverse interactome of 14-3-3 proteins (10,13,16,17), we hypothesized that it is unlikely that one protein would be solely responsible for 14-3-3ζmediated adipogenesis. It is known that the interactomes of 14-3-3 proteins are dynamic and change in response to various stimuli (11)(12)(13)19). The functional significance of such changes in the interactome is not clear, but it suggests that 14-3-3 proteins may be regulating biological processes critical for adipocyte development through their interactions. Using a gene onotology-based approach, we found that the 14-3-3ζ interactome is enriched with proteins involved in RNA binding and splicing during differentiation and confirms its contribution to the alternative splicing of mRNAs. As over 100 proteins were found to be unique to the14-3-3ζ interactome during adipocyte differentiation, it suggests that 14-3-3ζ could also regulate other cellular processes required for adipocyte development. For example, we detected an interaction of 14-3-3ζ with the mitochondrial regulator, Prohibitin-2 (Phb2), which others have shown to be essential for the expansion of mitochondria mass and mitochondrial function during adipogenesis (36-38). Further in-depth studies are required to assess whether 14-3-3ζ has regulatory roles in mitochondrial dynamics, but when taken together, it demonstrates the possibility of examining the contributions of interacting partners to reveal novel biological processes required for adipocyte differentiation.
Through the use of a functional siRNA screen, we identified novel roles of various RNA splicing factors, Alternative splicing of mRNA is critical for maintaining genetic diversity and cell identity, in addition to the expression of key factors required for differentiation (49,50). Specific to adipogenesis, differential promoter usage and alternative splicing are required for the expression of the canonical adipogenic transcription factor Pparγ (51-53). Other regulatory factors are also formed from alternative splicing, including nCOR1 and Lipin1 (54,55). Future studies are required to determine whether 14-3-3ζ directly binds to these splicing factors and how it regulates their splicing activity to generate essential adipogenic factors.
Protein abundance of 14-3-3ζ and other isoforms is increased in visceral adipose tissue from obese individuals (56,57), and we have previously reported that systemic over-expression of 14-3-3ζ in mice is sufficient to potentiate weight gain and fat mass in mice fed a high-fat diet (4). With respect to the pancreatic βcell, single cell transcriptomic analysis revealed higher mRNA expression of YWHAZ in β-cells from subjects with type 2 diabetes (58), and we have found that systemic over-expression of 14-3-3ζ was sufficient to reduce β-cell secretory function in mice (9). The exact mechanisms owing to how changes in 14-3-3ζ function affects the development of obesity or β-cell dysfunction are not known, but In-depth examination of the interactome in the context of both conditions may yield novel biological insight as to how 14-3-3ζ influences their development.
This approach has already been useful in understanding how changes in 14-3-3ε or 14-3-3σ expression can lead to the development of various forms of cancer and the identification of novel therapeutic targets (19,59-61).
In conclusion, this study provides compelling evidence demonstrating the usefulness of elucidating the interactome of 14-3-3ζ as a means to identify novel factors required for adipogenesis. Additionally, a systematic investigation of interacting partners may also provide insight as to which physiological processes are essential for 14-3-3ζ-mediated adipocyte differentiation. Lastly, deciphering how various disease states influence the interactome of 14-3-3 proteins may also aid in the discovery of novel therapeutic targets for the treatment of chronic diseases, such as obesity and type 2 diabetes.