Degradation of the endoplasmic reticulum–anchored transcription factor MyRF by the ubiquitin ligase SCFFbxw7 in a manner dependent on the kinase GSK-3

The ubiquitin-proteasome system regulates the abundance of many cellular proteins by mediating their targeted degradation. We previously developed a method—differential proteomics–based identification of ubiquitylation substrates (DiPIUS)—for the comprehensive identification of substrates for a given F-box protein subunit of SCF-type ubiquitin ligases. We have now applied DiPIUS to the F-box protein Fbxw7 in three cell lines (mHepa, Neuro2A, and C2C12) and thereby identified myelin regulatory factor (MyRF), an endoplasmic reticulum–anchored transcription factor that is essential for myelination of nerves in the central nervous system, as a candidate substrate of Fbxw7 specifically in mHepa cells. Co-immunoprecipitation analysis confirmed that the NH2-terminal cytoplasmic domain of MyRF interacted with Fbxw7 in these cells. Furthermore, an in vitro ubiquitylation assay revealed that MyRF undergoes polyubiquitylation in the presence of purified recombinant SCFFbxw7. In addition, the stability of MyRF in mHepa cells was increased by mutation of a putative phosphodegron sequence or by exposure of the cells to an inhibitor of glycogen synthase kinase-3 (GSK-3). We found that MyRF mRNA is not restricted to the central nervous system but is instead distributed widely among mouse tissues. Furthermore, with the use of RNA sequencing in mHepa cells overexpressing or depleted of MyRF, we identified many novel potential target genes of MyRF. Our results thus suggest that Fbxw7 controls the transcription of MyRF target genes in various tissues through regulation of MyRF protein stability in a manner dependent on MyRF phosphorylation by GSK-3.

The ubiquitin-proteasome system mediates the degradation of many cellular proteins with high specificity and plays an essential role in most cellular processes (1). The ubiquitylation cascade is mediated by three enzymes: a ubiquitin-activating enzyme (E1), a ubiquitin-conjugating enzyme (E2), and a ubiq-uitin ligase (E3), with substrate specificity being determined by the E3 (2). The Skp1-Cul1-F-box protein (SCF) complex is one of the most well-characterized types of ubiquitin ligase. Each SCF complex is composed of four subunits: Skp1, Cul1, Rbx1, and an F-box protein (3)(4)(5). F-box proteins are responsible for substrate recognition by each complex and constitute a large family of Ͼ70 eukaryotic proteins (6). The relation between each F-box protein and its corresponding substrates has been largely unclear, however. We recently developed a new method, termed differential proteomics-based identification of ubiquitylation substrates (DiPIUS) (7), for the comprehensive identification of substrates for a given F-box protein. In the present study, we have applied DiPIUS to the F-box protein Fbxw7 (F-box and WD repeat domain-containing 7).
Fbxw7 was initially identified by genetic analysis as a negative regulator of LIN-12 (Notch)-mediated signaling in Caenorhabditis elegans (8,9). It was subsequently shown to play a pivotal role in the control of cell proliferation through ubiquitylation of cyclin E (10 -13), c-Myc (13)(14)(15), Notch (9,16,17), c-Jun (18,19), and Krüppel-like factor 5 (KLF5) (20,21). Furthermore, Fbxw7 has recently been shown to contribute to cell differentiation and development. The loss of Fbxw7 in neural stem cells thus impairs their differentiation into neurons as a result of the up-regulation of Notch (22,23). In addition, conditional ablation of Fbxw7 in the liver of mice results in a shift in the differentiation of liver stem cells from the hepatocyte lineage to cholangiocytes as well as in increased cell proliferation (24). Liver-specific loss of Fbxw7 also elicits lipid buildup in hepatocytes as a consequence of the accumulation of sterol response element-binding proteins (SREBPs), which are also substrates of Fbxw7 (24 -27). Depletion of Fbxw7 in C2C12 mesenchymal cells promotes their differentiation into osteoblasts in association with the accumulation of OASIS (28), whereas conditional ablation of Fbxw7 in primary mouse mesenchymal cells results in up-regulation of BBF2H7 and promotes chondrogenesis (28). SREBPs, OASIS, and BBF2H7 are all endoplasmic reticulum (ER) 2 -anchored transcription factors, with the NH 2 -terminal domain of these proteins undergoing cleavage and trans- cro ARTICLE location to the nucleus, where it triggers the transactivation of target genes and thereby promotes specific cellular differentiation. Fbxw7 is thus implicated in the control of such ER-anchored transcription factors in addition to plasma membraneanchored Notch.
Myelin regulatory factor (MyRF) is another ER membraneanchored transcription factor and undergoes intramolecular chaperone domain (ICD)-mediated autocleavage that results in the separation of its nucleus-targeted NH 2 -terminal region (MyRF(N)) from the COOH-terminal region containing the transmembrane domain (29). MyRF was first identified as the product of a gene expressed specifically by postmitotic oligodendrocytes in the CNS (30,31). MyRF(N) binds to the cisregulatory elements of multiple oligodendrocyte-specific genes related to myelination such as those for myelin basic protein, proteolipid protein 1 (PLP1), and myelin-associated glycoprotein. Conditional ablation of MyRF in the oligodendrocyte lineage of mice results in severe CNS dysmyelination and pronounced deficits in myelin gene expression (31). Inducible ablation of MyRF in mature oligodendrocytes of adult mice also results in a rapid down-regulation of myelin gene expression followed by a gradual degeneration of CNS myelin (32). These observations thus suggest that MyRF plays a central role in myelination in the CNS.
Unexpectedly, we found that MyRF is widely expressed in many tissues and cell types outside the CNS. Although many transcriptional targets of MyRF in the CNS have been identified and characterized, the roles of MyRF in other tissues are unknown. Here we show that Fbxw7 mediates the ubiquitylation of MyRF and controls its stability in mHepa mouse hepatocellular carcinoma cells. In addition, we identified many candidate target genes for MyRF by RNA sequencing in mHepa cells. Our results suggest that Fbxw7 regulates the abundance of MyRF and thereby plays unidentified roles outside the CNS.

Identification of substrates for SCF Fbxw7 by DiPIUS
To comprehensively identify the substrates targeted by a given F-box protein for degradation, we previously developed a differential proteomics approach termed DiPIUS (7). The DiPIUS system is based on a difference in binding affinity for substrates between WT and mutant F-box proteins. Whereas WT F-box proteins bind weakly to their substrates, with such binding being followed by substrate ubiquitylation and degradation, F-box proteins with an inactivating mutation in the F-box domain are expected to associate more stably with substrates and thereby to allow their accumulation in the cell (Fig. 1A).
Both WT Fbxw7␣ (an alternatively spliced isoform of Fbxw7) and a mutant form of the protein lacking the entire F-box domain (⌬F mutant) were tagged at the NH 2 terminus with the FLAG epitope and expressed separately in mHepa, Neuro2A (mouse neuroblastoma), and C2C12 (mouse myoblast) cells. Cell lysates were subjected to immunoprecipitation with antibodies to FLAG, and the immunoprecipitated proteins were analyzed by LC-MS/MS. The abundance of proteins that bound to the WT or mutant forms of Fbxw7␣ was compared by semi-quantitative spectral counting (DiPIUS-NL) (Fig. 1A). We identified MyRF as a candidate substrate of Fbxw7␣ with the DiPIUS-NL system in mHepa cells but not in Neuro2A or C2C12 cells (Fig. 1B). We also identified an additional 18 potential novel substrates for Fbxw7 in one or more of these three cell lines.

The NH 2 -terminal domain of MyRF interacts with Fbxw7
We verified the interaction between Fbxw7␣ and MyRF in mHepa cells with a co-immunoprecipitation assay. The fulllength form of MyRF localized in the ER undergoes ICD-mediated autocleavage (29). The released MyRF(N) fragment then undergoes translocation to the nucleus, where Fbxw7␣ is localized (33,34), and activates transcription of its target genes. Coimmunoprecipitation analysis revealed that the nuclear form of MyRF (MyRF(N)) was indeed associated to a greater extent with Fbxw7␣(⌬F) than with Fbxw7␣(WT) in mHepa cells, whereas the full-length form of MyRF (MyRF(FL)) did not immunoprecipitation. B, the abundance of proteins that bound to Fbxw7␣(WT) or Fbxw7␣(⌬F) in the indicated cell lines and in three independent experiments (Exp.) was compared by semiquantitative spectral counting (these data are the same as those presented previously (28)). Proteins with a mutant/WT ratio of Ͼ 2 in at least two of three independent experiments, including known substrates of SCF Fbxw7␣ as well as MyRF, are indicated. The ratios of the number of spectral counts for mutant and WT F-box proteins are indicated by the scale.

Fbxw7 promotes ubiquitin-dependent degradation of MyRF
interact with Fbxw7␣ ( Fig. 2A). These results suggested that the NH 2 -terminal domain of MyRF encounters Fbxw7␣ in the nucleus, and that Fbxw7␣ might attenuate the function of MyRF to prevent excessive transactivation of its target genes.

Fbxw7 controls the stability of MyRF
We found that the ectopic expression of Fbxw7␣(WT) reduced the amount of MyRF(N), but not that of MyRF(FL), in mHepa cells ( Fig. 2A). Fbxw7␣(⌬F), which lacks ubiquitylation activity, did not show such an effect, suggesting that Fbxw7␣ regulates the abundance of MyRF(N) in a ubiquitin-dependent manner. Cycloheximide chase analysis revealed that the halflife of MyRF(N) was indeed reduced by forced expression of Fbxw7␣(WT) (Fig. 2, B and C).
We also examined whether MyRF is a direct substrate of SCF Fbxw7␣ in vitro. MyRF(N) tagged with His 6 and HA epitopes was expressed in and purified from Sf21 insect cells with nickelnitrilotriacetic acid (Ni-NTA) beads. Skp1 and Cul1 each tagged with the His 6 and Myc epitopes, Myc-tagged Rbx1, and His 6 -FLAG-tagged Fbxw7␣ were also expressed in Sf21 cells and purified as the assembled SCF Fbxw7␣ complex with antibodies to FLAG. An in vitro ubiquitylation assay with these recombinant proteins revealed that the ubiquitylation of MyRF(N) was apparent only in the presence of all reaction components, including Uba1 (E1), UbcH5C (E2), SCF Fbxw7␣ (E3), and ubiquitin (Fig. 2D). Collectively, these results suggested that SCF Fbxw7␣ is a bona fide ubiquitin ligase for MyRF(N).

CPD phosphorylation is required for recognition of MyRF by Fbxw7
Most substrates of Fbxw7 contain a conserved amino acid sequence for phosphorylation, termed the Cdc4 phosphodegron (CPD) (5). MyRF contains four potential CPD sequences (Fig. 3A). Given that three of these sequences are located in the nuclear translocation region of MyRF, we introduced mutations into these three motifs to examine which one is essential for the Fbxw7-mediated degradation of MyRF(N). Co-immunoprecipitation analysis revealed that a mutant (S138A/S142A) form of MyRF in which both Ser 138 and Ser 142 in the second CPD were replaced with alanine did not interact with Fbxw7␣(⌬F) (Fig. 3B). Furthermore, cycloheximide chase analysis revealed that this mutant was more stable than WT or other CPD mutant proteins in mHepa cells (Fig. 3, C and D). Given that many CPDs of Fbxw7 substrates are phosphorylated by glycogen synthase kinase-3 (GSK-3) (5), we examined the stability of MyRF(N) in the presence of the GSK-3 inhibitor 6-bromoindirubin-3Ј-oxime (BIO). Cycloheximide chase analysis revealed that BIO treatment prolonged the half-life of MyRF(N) from 53 to 116 min (Fig. 3, E and F). In addition, the S138A/ S142A mutant form of MyRF(N) did not undergo ubiquitylation in vitro (Fig. 3G). Together, these results thus indicated that the GSK-3-dependent phosphorylation of Ser 138 and Ser 142 in MyRF(N) is required for recognition and subsequent ubiquitylation by SCF Fbxw7 .

Fbxw7 restrains oligodendrocyte maturation
We found that depletion of Fbxw7 in mouse oligodendrocyte progenitor cells (OPCs) by shRNA-mediated RNAi promoted the differentiation of these cells into mature oligodendrocytes. Expression of the genes for the oligodendrocyte marker proteins PLP1 and 2Ј,3Ј-cyclic nucleotide-3Ј-phosphodiesterase (CNP), both of which are target genes of MyRF, was thus upregulated in OPCs depleted of Fbxw7 compared with those expressing a control shRNA (Fig. 4). In contrast, expression of the gene for the OPC marker platelet-derived growth factor receptor ␣ (PDGFR␣) (31) was attenuated by Fbxw7 depletion. These results suggested that Fbxw7 depletion leads to the accu-mulation of MyRF(N) in OPCs and thereby promotes their maturation into oligodendrocytes.

Identification of novel candidate target genes of MyRF
It was unexpected that we identified MyRF as a target of Fbxw7 in mHepa cells, given that most previous studies of MyRF have focused on its role in the CNS. To investigate the distribution of MyRF mRNA, we performed RT and realtime PCR analysis of mouse tissues. MyRF mRNA was detected not only in the CNS but in all tissues examined with the exception of quadriceps and bone marrow (Fig. 5), suggesting that MyRF might have other unrecognized roles in

Fbxw7 promotes ubiquitin-dependent degradation of MyRF
addition to transactivation of myelin-associated genes in the CNS.
To investigate possible roles of MyRF outside of the CNS, we performed RNA sequencing in mHepa cells to identify novel genes that might be targeted by MyRF. RNA samples from mHepa cells engineered to overexpress or to be depleted of MyRF were purified and used to prepare RNA libraries for analysis with the HiSeq 1500 system (Illumina) and calculation of FPKM (fragments per kilobase of exon per million mapped reads) values. We identified 39 genes as positively regulated by MyRF (overlap of 1321 genes that were up-regulated by MyRF overexpression and 174 genes that were down-regulated by MyRF knockdown) (Fig. 6A). Conversely, 33 genes were negatively regulated by MyRF (overlap of 1737 genes that were down-regulated by MyRF overexpression and 113 genes that were up-regulated by MyRF knockdown) (Fig. 6B). We confirmed by RT and real-time PCR analysis that the amounts of mRNAs for 4930486L24Rik, Gm5549, and Hpgds, the genes revealed by RNA sequencing to be most up-regulated by MyRF overexpression, were indeed increased in mHepa cells overexpressing MyRF (Fig. 6C). We also verified by RT and real-time PCR analysis that the expression levels of Fam46a, Col14a1, and Akr1c18, the genes most down-regulated by MyRF overexpression, were reduced in mHepa cells overexpressing MyRF (Fig. 6D). Together, these data suggested that MyRF is expressed and plays unidentified roles in tissues outside of the CNS.

Discussion
With the use of the DiPIUS system, we have identified MyRF as a novel substrate of Fbxw7␣. A series of validation assays confirmed that the nucleus-targeted NH 2 -terminal region of MyRF is a bona fide substrate of Fbxw7␣. Consistent with our findings, a mutation that disrupts the Fbxw7 ortholog in zebrafish was shown to elicit hypermyelination (35), possibly as a result of the accumulation of MyRF in oligodendrocytes because of its impaired degradation. Further studies will be required to confirm a role for Fbxw7 in myelination in mammals.
Fbxw7␣ mediates the nuclear degradation of many transcription factors, some of which are localized at the ER membrane under basal conditions but undergo cleavage and translocation to the nucleus in response to specific cues. SREBPs are such ER-anchored transcription factors that are cleaved in response to a reduction in the intracellular level of cholesterol, with the mature form of these proteins being translocated to the nucleus to mediate activation of lipogenesis. OASIS and BBF2H7 are also ER-anchored transcription factors that are cleaved in response to ER stress and promote osteogenesis and chondrogenesis, respectively. MyRF undergoes autocleavage and promotes myelination. Fbxw7␣ targets the nuclear forms of all of these transcription factors (SREBPs, OASIS, BBF2H7, and MyRF). Fbxw7 was originally discovered as a regulator of Notch (8,9), which is also a membrane-anchored transcription factor that undergoes cleavage, with the COOH-terminal intracellular domain of this protein being translocated to the nucleus and there activating downstream target genes to induce cellular differentiation. Fbxw7 might thus function to extinguish the transactivation activity of these various transcription factors in the nucleus, thereby contributing to the strict regulation of transcription by membrane-anchored transcription factors (Fig. 7).
We found that MyRF mRNA is not restricted to the CNS but is instead widely distributed among mouse tissues. We identified many candidates for target genes of MyRF in mHepa cells by RNA sequencing. Among the candidate genes positively regulated by MyRF, Ngef (encoding Ephexin-1) and Ctgf (which encodes connective tissue growth factor and is also known as Ccn2, Fisp12, or Hcs24) contribute to maturation of the neuromuscular junction (36,37) and to angiogenesis (38,39), chondrogenesis (38), and pancreas development (40), respectively. On the other hand, Apela (also known as Ela, Ende, Gm10664, or Tdl) and Traf5, two of the genes negatively regulated by MyRF, have been shown to contribute to heart development (41) and mesendodermal cell movement during gastrulation (42) and to osteoclastogenesis (43), respectively. Collectively, these observations suggest that MyRF might function to orchestrate developmental programs outside the CNS. Fbxw7 might thus contribute to the control of developmental processes as an upstream regulator of the stability of MyRF as well as that of other ER-anchored transcription factors including SREBPs, OASIS, and BBF2H7.

Experimental procedures
Mice WT C57BL/6J mice were obtained from CLEA Japan (Tokyo). Mouse experiments were approved by the Animal Ethics Committee of Kyushu University.

Antibodies
Antibodies to the FLAG epitope (M2) were obtained from Sigma; those to the HA epitope (HA.11 or Y-11) from Covance (Princeton, NJ) or Santa Cruz Biotechnology (Dallas, TX), respectively; those to Skp1, to Hsp90, and to CD140a (PDGFR␣) from BD Biosciences; those to Cul1 from Cell Signaling Technology (Danvers, MA); and those to c-Myc from Abcam. HRP-conjugated goat antibodies to mouse or rabbit IgG were obtained from Promega (Madison, WI).

Plasmids
Complementary DNAs encoding mouse Fbxw7␣ or its ⌬F mutant were subcloned into p3ϫFLAG-CMV 7.1 (Sigma), and those for HA epitope-tagged mouse MyRF or its mutants were subcloned into pcDNA3 (Invitrogen). The resulting vectors were introduced into mHepa cells with the use of the Lipofectamine 2000 transfection reagent (Invitrogen) for transient expression. For retroviral expression, cDNAs for FLAG epitope-tagged mouse Fbxw7␣ or for HA epitope-tagged mouse MyRF, or shRNA expression cassettes (mouse U6 gene promoter and each target sequence for mouse MyRF or mouse Fbxw7) were subcloned into pMX-puro (kindly provided by T. Kitamura). The resulting vectors were introduced into Plat E cells with the use of the X-tremeGENE 9 DNA transfection reagent (Roche), and recombinant retroviruses released into culture supernatants were used to infect mHepa cells or primary OPCs. The infected cells were subjected to selection in medium containing puromycin (6 g/ml). For baculoviral expression, cDNAs for FLAG-tagged mouse Fbxw7␣, HA epitope-tagged mouse MyRF(N), and Myc epitope-tagged mouse Cul1 and Skp1 were subcloned into pFASTBac HT (encoding a His 6 tag), whereas that for human Rbx1 containing an NH 2 -terminal Myc epitope tag was subcloned into pFASTBac1 (Invitrogen).

Cycloheximide chase analysis, GSK-3 inhibitor treatment, immunoprecipitation, and immunoblot analysis
For cycloheximide chase experiments, cycloheximide (100 g/ml) was added to culture medium and the cells were harvested at the indicated times thereafter. For GSK-3 inhibition, cells were cultured in the presence of 1 M BIO for 1 h before the addition of cycloheximide. Cells were lysed as described previously (46). Immunoprecipitation and immunoblot analysis were also performed as described previously (47), with Hsp90 as a loading control and immunoblot intensity quantitated with ImageJ software.

RNAi
Construction of shRNA vectors was performed as described previously (50). The sequence targeted for mouse MyRF was ER-anchored transcription factors (MyRF, BBF2H7, OASIS, and SREBPs) and plasma membrane-anchored transcription factors (Notch proteins) are cleaved to release fragments that undergo translocation to the nucleus, where they activate target genes. The nuclear forms of these proteins are phosphorylated, ubiquitylated by SCF Fbxw7 , and thereby targeted for degradation by the proteasome, thus preventing excessive transactivation of target genes. U, ubiquitin; P, phosphate.

RNA sequencing and data analysis
RNA sequencing was performed as described previously (51). Total RNA was extracted from retrovirus-infected mHepa cells. Illumina libraries were constructed with the use of a NEB-Next Ultra Directional RNA Library Prep Kit for Illumina. Single-end reads were mapped to mm10 (UCSC) with TopHat (v2.1.1) and the mm10 GTF file. Data were processed with the Linux system for calculation of FPKM values with the use of Cuffdiff (v2.2.1) and the option "-N." Sequence data have been deposited in the DDBJ Sequence Read Archive under the accession number DRA005955.

Statistical analysis
Where indicated, quantitative data are presented as mean Ϯ S.D. and were analyzed with Student's t test. A p value of Ͻ 0.05 was considered statistically significant.