RNA-binding Protein Quaking Stabilizes Sirt2 mRNA during Oligodendroglial Differentiation*

Myelination is controlled by timely expression of genes involved in the differentiation of oligodendrocyte precursor cells (OPCs) into myelinating oligodendrocytes (OLs). Sirtuin 2 (SIRT2), a NAD+-dependent deacetylase, plays a critical role in OL differentiation by promoting both arborization and downstream expression of myelin-specific genes. However, the mechanisms involved in regulating SIRT2 expression during OL development are largely unknown. The RNA-binding protein quaking (QKI) plays an important role in myelination by post-transcriptionally regulating the expression of several myelin specific genes. In quaking viable (qkv/qkv) mutant mice, SIRT2 protein is severely reduced; however, it is not known whether these genes interact to regulate OL differentiation. Here, we report for the first time that QKI directly binds to Sirt2 mRNA via a common quaking response element (QRE) located in the 3′ untranslated region (UTR) to control SIRT2 expression in OL lineage cells. This interaction is associated with increased stability and longer half-lives of Sirt2.1 and Sirt2.2 transcripts leading to increased accumulation of Sirt2 transcripts. Consistent with this, overexpression of qkI promoted the expression of Sirt2 mRNA and protein. However, overexpression of the nuclear isoform qkI-5 promoted the expression of Sirt2 mRNA, but not SIRT2 protein, and delayed OL differentiation. These results suggest that the balance in the subcellular distribution and temporal expression of QKI isoforms control the availability of Sirt2 mRNA for translation. Collectively, our study demonstrates that QKI directly plays a crucial role in the post-transcriptional regulation and expression of Sirt2 to facilitate OL differentiation.

Similar to myelin structural proteins (MBP and PLP), SIRT2 protein expression is also reduced in qk v /qk v mice (25). Mammalian SIRT2 is a class III NAD ϩ -dependent deacetylase (26), which is predicted to give rise to three isoforms (27). SIRT2 is enriched in brain and spinal cord tissue, predominately localized in the paranodal regions of the CNS myelin sheath (28 -30). During OL differentiation, SIRT2 is expressed early, prior to the expression of myelin-specific genes (25,30,31), promoting differentiation at both the cellular and molecular level (31). In addition to the qk v /qk v mice, SIRT2 expression is reduced in the Plp null mice (29) and Plp-ISEdel mutant mice (25). In Plp mutants, the loss of SIRT2 is observed primarily in the myelin sheath (25,29) but not in OPCs or OL cell bodies (25). Moreover, no putative QREs have previously been identified in Sirt2 mRNA. Thus, it was postulated that QKI indirectly regulates SIRT2 expression during CNS myelination through co-transport with PLP into the myelin sheath (25).
To further investigate the relationship between qkI and Sirt2 in OL development, we used mouse primary OLs and the CG4-OL cell line derived from neonatal rat forebrain O-2A progenitors that undergo defined stages of differentiation under controlled media conditions (31)(32)(33)(34)(35)(36). Coordinated expression of QKI and SIRT2 was observed during differentiation. Thus, we sought to delineate the molecular mechanisms that govern the direct or indirect interaction between the RNAbinding protein QKI and Sirt2 during OL development. We demonstrate a direct interaction between QKI and Sirt2 mRNA in OL progenitors and differentiating OLs. The binding site for QKI was mapped to the QRE ACUAAC at 1853-1858 bp in the 3Ј UTR of Sirt2 mRNA. Our findings indicate that Sirt2 is a direct target of QKI. Binding of QKI increases the post-transcriptional stability of Sirt2 mRNA and controls its availability for translation. Furthermore, our results show QKI-5 delays the transition from OPC through to post-mitotic, immature OL resulting in a delay in differentiation. The subcellular localization and coordinated temporal expression of specific QKI isoforms appear to govern Sirt2 expression for proper OL differentiation.

Results
Expression of QKI and SIRT2 Increase during OL Differentiation-QKI plays a critical role in OL differentiation by regulating the expression of several myelin specific genes, such as Mbp (6,7), Plp (9), and Mag (8,10,42). We have previously demonstrated that SIRT2 can promote OL differentiation (31). Hence, we sought to delineate the molecular interaction between qkI and Sirt2. In CG4-OL cells, the expression pattern of both qkI (Fig. 1) and Sirt2 (Fig. 2) increased over 6 days of differentiation. Detailed quantitative mRNA and protein analyzes revealed differential expression of specific isoforms of QKI and SIRT2 during OL differentiation in vitro. Expression of qkI mRNA (Fig. 1A) gradually increased with differentiation reaching a ϳ3-fold increase by day 6. There was a corresponding increase in expression of the QKI protein product QKI-6 isoform (ϳ2-fold increase at day 6 in differentiation medium (DM)) but not of QKI-5 or QKI-7, which co-migrate on the immunoblot (Fig. 1, B and C).
The Sirt2.2 variant was the most abundant transcript in CG4-OL cells with ϳ4.5-fold and ϳ2700-fold greater expression than Sirt2.1 and Sirt2.3, respectively, under growth conditions ( Fig. 2A). Expression of the Sirt2.2 transcript increased ϳ8-fold within 24 h under differentiation conditions and was maintained through to day 6. Similarly, Sirt2.1 and Sirt2.3 mRNA expression increased throughout differentiation. SIRT2.2 was also the most abundant protein isoform in CG4-OL cells (Fig. 2B). Expression of SIRT2.2 protein increased ϳ2.5-fold by day 3 and was maintained through to day 6 ( Fig. 2C). Although Sirt2.1 and Sirt2.3 mRNA increased during differentiation, this did not translate to an increase in the SIRT2.1 or SIRT2.3 protein products. The extent of differentiation was analyzed by quantifying the expression of mature OL marker MBP. Expression of Mbp mRNA (supplemental Fig. S1A) increased during differentiation reaching a ϳ60-fold increase by day 6. Consistent with this, expression of MBP protein (supplemental Fig. S1B) also increased with differentiation. The coordinated expression patterns of qkI and Sirt2 during OL differentiation suggests that these two genes may interact, either directly or indirectly, to promote OL differentiation.
Overexpression of qkI Induces the Expression of Sirt2 mRNA and Protein-Quaking viable mutant mice (qk v /qk v ) exhibit reduced SIRT2 protein expression in myelin sheath (25), yet there is little known about the relationship between QKI and Sirt2 mRNA during CNS myelination. To investigate what role QKI may have in regulating the expression of Sirt2 during OL differentiation, qkI was overexpressed in CG4-OL cells by transfection with the common coding sequence of the RNA binding domain from QKI-5, QKI-6, and QKI-7. Up-regulation of qkI resulted in an increase in Sirt2 mRNA expression under both proliferating (growth media; GM) ( Fig. 3A) and differentiating (DM, day 6) conditions ( Fig. 3B). In GM, overexpression of qkI in proliferating CG4-OLs increased Sirt2.1 and Sirt2.2 mRNA levels to ϳ2and ϳ3-fold, respectively; in contrast, Sirt2.3 expression was not affected (Fig. 3A). In DM, expression of all three variants increased ϳ2-fold in differentiating CG4-OLs (Fig. 3B). Up-regulation of qkI also resulted in a corresponding increase in SIRT2.1 and SIRT2.2 protein expression under both proliferating (Fig. 3, C and D) and differentiating (Fig. 3, E and F) conditions in CG4-OLs. Furthermore, to examine the regulatory function of QKI in OLs, qkI was overexpressed in primary OLs isolated from the mouse brain (35,36). Similar to CG4-OLs, overexpression of qkI in mouse primary OLs increased Sirt2 mRNA expression under both proliferating (GM) (Fig. 4A) and differentiating (DM, day 6) conditions ( Fig. 4B). In proliferating primary OLs, up-regulation of qkI increased the Sirt2.1 and Sirt2.2 mRNA expression ϳ2-fold; whereas Sirt2.3 expression was not altered (Fig. 4A). In DM, expression of all three variants increased ϳ3-fold in differentiating primary OLs (Fig. 4B). There was a corresponding increase in SIRT2.1 and SIRT2.2 protein expression under both proliferating (Fig. 4, C and D) and differentiating (Fig. 4, E and F) conditions in primary OLs. Under these experimental conditions, expression of the SIRT2.3 protein isoform was not detectable either with or without qkI overexpression (Figs. 3, C and E, and 4, C and E). Thus, qkI appears to be a positive regulator of Sirt2.1 and Sirt2.2 mRNA and protein expression in developing OLs.
Presence of Putative QREs in Sirt2 Transcripts-SIRT2 protein expression is reduced in both qk v /qk v (25) and Plp (25,29) mutant mice, and the expression of PLP is critical for the transport of SIRT2.2 to the myelin sheath (25,29). Thus, it is not known whether QKI controls the expression of SIRT2 during CNS myelination through a direct interaction with Sirt2 mRNA or through an indirect pathway involving PLP. Using in silico analysis, we identified the presence of two putative QREs in Sirt2 mRNA in the 3Ј UTR (Fig. 5A). Both putative QREs, AUUAA(C/U) at 1639 -1643 bp (denoted as QRE-1) and ACUAA(C/U) at 1853-1858 bp (denoted as QRE-2) (Fig. 5A) in the Sirt2 mRNA, correspond to the predicted consensus sequences (8,18) and are highly conserved (Fig. 5A). Thus, the 3Ј UTR of Sirt2 mRNA has two putative interaction sites for QKI.
QKI Binds to Sirt2 mRNA to Regulate Its Expression-To determine whether QKI binds to Sirt2 mRNA at the putative binding sites, RNA co-immunoprecipitation (RNA co-IP) was carried out using whole cell lysates from CG4-OL under both FIGURE 1. Expression of qkI mRNA and protein increases during differentiation. CG4-OL progenitor cells undergo high rates of proliferation in GM followed by differentiation into mature, pre-myelinating OLs when transferred to DM. Whole cell lysates were collected at 24-h intervals during differentiation (days 0, 1, 2, 3, 4, 5, and 6). A, quantitative real time-PCR was carried out to analyze the changes in the mRNA levels of qkI (n ϭ 3 biological replicates). A gradual increase in qkI, expression was observed over the 6-day experimental timeline. Data were normalized to ␤-actin and represented relative to day 0 (mean Ϯ S.E.; one-way ANOVA, *, p Ͻ 0.05; **, p Ͻ 0.01). B, representative immunoblot showing a corresponding increase in QKI protein expression over the course of a 6-day differentiation. C, densitometric analysis of immunoblots (n ϭ 3 biological replicates) shows QKI-6 is the predominate isoform expressed during OL differentiation. Data were normalized to ␤-tubulin and represented relative to the respective isoform at day 0 (mean Ϯ S.E.; two-way ANOVA, *, p Ͻ 0.05; **, p Ͻ 0.01; ***, p Ͻ 0.001).
growth (GM) and differentiating (DM day 6) conditions. QKI was found to bind Sirt2 mRNA in both proliferating and differentiating CG4-OLs (Fig. 5, B and C). In comparison, QKI was found to bind Plp mRNA only under growth conditions (supplemental Fig. S3A). In both GM and DM, QKI did not bind with Gapdh mRNA (Fig. 5B, bottom panel). Further investigation determined that under growth conditions QKI bound to Sirt2.1 and Sirt2.2 only in CG4-OLs (Fig. 5D), whereas under differentiating conditions QKI bound to all the three variants of Sirt2 mRNA (Fig. 5E). In addition, interaction of QKI and Sirt2 mRNA in primary OLs was also examined. QKI was found to interact with Sirt2 mRNA in both proliferating and differentiating primary OLs (Fig. 6A). In proliferating primary OLs under growth conditions, QKI bound to Sirt2.1 and Sirt2.2 mRNA FIGURE 2. Expression of Sirt2 mRNA and protein increases during differentiation. CG4-OL progenitor cells in GM were transferred to DM and whole cell lysates were collected at 24-h intervals (days 0, 1, 2, 3, 4, 5, and 6). A, Sirt2 mRNA expression increases under differentiation conditions. Sirt2.2 mRNA is the most abundant transcript expressed in CG4-OL cells and increases by day 1 of differentiation. qRT-PCR data (n ϭ 3 biological replicates) were normalized to ␤-actin and represented relative to Sirt2.3 variant at day 0 (mean Ϯ S.E.; two-way ANOVA). ‡ denotes Sirt2.1 mRNA with p Ͻ 0.001; ␣ denotes Sirt2.2 mRNA with p Ͻ 0.001; † denotes Sirt2.3 mRNA with p Ͻ 0.001 relative to their respective variants at day 0. B, representative immunoblot showing a corresponding increase in SIRT2 protein expression over the course of differentiation. C, densitometric analysis of immunoblots (n ϭ 3 biological replicates) reveal that SIRT2.2 is the predominate isoform expressed during OL differentiation. Data were normalized to ␤-tubulin and represented relative to the respective isoform at day 0 (mean Ϯ S.E.; two-way ANOVA, ***, p Ͻ 0.001).
( Fig. 6B), whereas QKI bound to all the three variants of Sirt2 in differentiating primary OLs (Fig. 6C). Collectively, these data demonstrate that QKI interacts with all three variants of Sirt2 mRNA, presumably via the putative QRE-1 and/or QRE-2 in the 3Ј UTR.
QKI Stabilizes and Protects Sirt2 mRNA from Degradation-QKI regulates myelin gene expression during development, in part by modulating the stability of myelin gene transcripts. To determine whether this is also the case with Sirt2, a standard mRNA stability assay (20 -22) was used to assess the ability of QKI to protect or degrade Sirt2 transcripts. Following transfection with pcDNA or pcDNA-qkI (48 h), actinomycin D treatment was used to inhibit transcription under GM conditions and subsequent degradation of Sirt2 mRNA was monitored using qRT-PCR. QKI was found to regulate the stability of Sirt2.1 and Sirt2.2 variants (Fig. 8, A-C). Overexpression of qkI delayed the degradation of Sirt2.1 (Fig. 8A) and Sirt2.2 (Fig. 8B), but not Sirt2.3 (Fig. 8C). The half-life (t1 ⁄ 2 ) of each of the Sirt2 . qRT-PCR data (n ϭ 3 biological replicates) was normalized to ␤-actin and represented relative to pcDNA control vector (mean Ϯ S.E.; two-way ANOVA, *, p Ͻ 0.05; ***, p Ͻ 0.001). Representative immunoblots (C and E) and densitometric analysis (D and F) shows a corresponding increase in SIRT2.1 and SIRT2.2 protein with the overexpression of qkI. SIRT2.3 protein was below the limits of detection. Densitometry data (n ϭ 3 biological replicates) were normalized to ␤-tubulin and represented relative to pcDNA control vector (mean Ϯ S.E.; two-way ANOVA, *, p Ͻ 0.05; ***, p Ͻ 0.001).
Overexpression of qkI-5 Induces the Expression of Sirt2 mRNA but Not SIRT2 Protein-There is a wealth of data supporting the role of QKI-6 in promoting OL differentiation and myelination (5, 9, 10, 13, 20 -22), yet the role of QKI-5 is not well defined (5,6). Although qkI expression increased gradually with differentiation ( Fig. 1), this was found to reflect an increase in qkI-5 and qkI-6 transcripts but no change in qkI-7 expression (Fig. 9A). qkI-5 mRNA is expressed early during differentiation, with a ϳ1.5-fold increase by day 3 and ϳ2-fold increase by day 6, whereas qkI-6 mRNA expression increased ϳ1.5-fold by day 6 in DM. QKI-5 protein expression did not show a significant increase during differentiation most likely due to the co-migration of QKI-5 and QKI-7 (Fig. 1, B and C).  As Sirt2 mRNA expression increases early in OL differentiation (Fig. 2), we chose to further investigate the specific role of QKI-5 in regulating Sirt2 expression. Overexpression of qkI-5 induced expression of qkI-5 transcript only and did not alter qkI-6 or qkI-7 mRNA levels in differentiating OLs (Fig. 9B, supplemental Fig. S2, A-C). Up-regulation of qkI-5 resulted in an increase in the expression of Sirt2 transcripts. A ϳ5-fold increase in Sirt2 mRNA compared with cells transfected with control vector was observed by day 6 (Fig. 9C); however, upregulation of qkI-5 did not increase SIRT2 protein expression (Fig. 9D). Similarly, up-regulation of qkI-5 increased Plp mRNA expression but not PLP protein expression (supplemental Fig.  S3, B-D). Although QKI-5 is able to modulate transcript levels of Sirt2 and Plp in differentiating OLs, this does not result in a corresponding change in expression of the protein product.
QKI-5 Delays CG4-OL Differentiation-Given that QKI-5 regulates Sirt2 only at the level of mRNA expression, we sought to better understand its functional role during OL differentiation. To assess differentiation, CG4-OL cell cultures were analyzed for phenotypic markers of specific OL developmental stages at 24-h intervals (Fig. 10). In the OL lineage, A2B5 ϩve cells (OPCs) actively proliferate and have a bipolar morphology, FIGURE 7. QKI interacts with Sirt2 via the QRE (ACUAAC) at 1853 bp in the 3 UTR. A 593-bp fragment of the Sirt2 3Ј UTR containing putative QKI binding sites was amplified and cloned into pGL3-promoter vector. A, site-directed mutagenesis of the putative QKI response elements, QRE-1 at 1639 bp (Mut1), QRE-2 at 1853 bp (Mut2), or both QREs at 1639 and 1853 bp (Mut1*Mut2) were carried out using the QuikChange Lightning kit. The underlined and bold sequences represent putative QREs and the mutated sequences represented in red. B, HEK293 cells were co-transfected with pcDNA or pcDNA-qkI (common coding region of qkI-5, qkI-6, and qkI-7) along with pGL3 vector harboring wild-type or mutated QREs in the Sirt2 3Ј UTR, as indicated. RNA co-immunoprecipitation with His tag antibody in these cell extracts reveals high affinity binding of QKI to wild-type or mutated QRE-1 of Sirt2 3Ј UTR but not to mutated QRE-2. qRT-PCR data (top panel) (n ϭ 3 biological replicates) was normalized to ␤-actin and represented relative to pcDNA control vector (mean Ϯ S.E.; two-way ANOVA, ***, p Ͻ 0.001). RT-PCR products (bottom panel) were detected in a RedSafe-stained agarose gel; lanes are as indicated. C, for luciferase reporter assay, CG4-OL cells were co-transfected with pcDNA or pcDNA-qkI (common coding region of qkI-5, qkI-6, and qkI-7) along with pGL3 luciferase vector containing wild-type or mutated QREs in the Sirt2 3Ј UTR. Overexpression of qkI increased luciferase reporter activity of Sirt2 3Ј UTR and Sirt2 3Ј UTR-Mut1 but not Sirt2 3Ј UTR-Mut2 or Sirt2 3Ј UTR-Mut1*Mut2. This indicates that QRE-2 at 1853 bp is critical for the interaction of QKI with Sirt2. Relative luciferase activity (n ϭ 4 biological replicates) was normalized to pcDNA control vector co-transfected with pGL3 luciferase vector and represented as percent activity (mean Ϯ S.E.; two-way ANOVA, **, p Ͻ 0.01).
whereas GalC ϩve cells (pre-myelinating OLs) are post-mitotic and develop extensive branching of cellular processes (43)(44)(45). In differentiating CG4-OL cultures, ϳ80% of cells expressed A2B5 at day 1 under all treatment conditions. This was followed by a steady decline in A2B5 ϩve cells corresponding to an increase in GalC ϩve cells over the 6 days of differentiation (Fig.  10, E and F); no difference was observed between wild-type and blank pIRES2 vector-transfected cells. Overexpression of qkI-5 resulted in an increase in the percentage of A2B5 ϩve cells from day 3 to 6 ( Fig. 10, B and E). By day 6, the number of A2B5 ϩve cells was 44.6 Ϯ 2.7% in cultures transfected with qkI-5 compared with 26.4 Ϯ 1.9% in cultures transfected with control vector. This was paralleled by a decrease in the percentage of GalC ϩve cells (Fig. 10, D and F). At day 6, the number of GalC ϩve cells was 43.2 Ϯ 3.7% in cultures transfected with qkI-5 compared with 58.5 Ϯ 4.5% in cultures transfected with control vector. These data indicate that QKI-5 delays the transition from OPC through to post-mitotic, immature OL resulting in a delay in differentiation.

Discussion
Sirt2 is a key regulator of OL development (31). In myelinating OLs, SIRT2 is predominately cytoplasmic, localizing to the paranodal loops (28 -30), but under certain cellular conditions, SIRT2 has also been shown to shuttle between the cytoplasm and nucleus (26,46). As a NAD ϩ -dependent deactylase, it has a multitude of cellular targets involved in gene transcription (47,48), proliferation (46,47,49,50), cell polarity (51), and cytoskeletal remodeling (26,28,30). We have previously shown expression of Sirt2 drives OL differentiation at the cellular level by enhancing process outgrowth and arborization (31). QKI facilitates proper CNS myelination via post-transcriptional regulation of several transcripts during OL development (5-10, 20, 22, 23). Interestingly, in qk v /qk v mutant mice there is reduced FIGURE 8. QKI preferentially stabilizes and protects Sirt2.2 mRNA. A-C, the degradation of Sirt2 mRNA was monitored in CG4-OL cells transfected with pcDNA and pcDNA-qkI (common coding region of qkI-5, qkI-6, and qkI-7) in growth conditions. After 48 h following transfection, transcription was blocked with actinomycin D and mRNA levels of all the three Sirt2 variants were determined by qRT-PCR at the indicated time points (0, 15, 30, 60, and 120 min). Overexpression of qkI significantly increased the quantity of Sirt2.2 mRNA remaining 15 to 120 min after actinomycin D treatment (B) and that of Sirt2.1 mRNA remaining at 120 min after actinomycin D treatment (A). The stability of Sirt2.3 mRNA was not significantly impacted (C). A-C, qRT-PCR data (n ϭ 3 biological replicates) were normalized to 18S rRNA and are represented as the percentage of Sirt2 mRNA variant levels measured at time 0 min (mean Ϯ S.E.; one-phase decay). The half-lives were calculated as the time necessary for each Sirt2 mRNA variant to reduce to 50% of its initial amount at time 0 min (dashed lines). D, half-lives of Sirt2.1, Sirt2.2, and Sirt2.3 mRNA are represented in minutes (mean Ϯ S.E.; two-way ANOVA, ***, p Ͻ 0.001).
protein expression of SIRT2, whereas the expression of Sirt2 mRNA is not altered in the brain tissue (25). In addition, it has been reported that QKI indirectly controls the expression of SIRT2 during myelination through co-transport with PLP to the myelin sheath (25,29). Although transport of the SIRT2 protein into the myelin sheath is dependent on PLP (25, 29), the developmental expression of Sirt2 mRNA and protein is most likely regulated by an alternate mechanism(s) because Sirt2 is expressed prior to myelin structural proteins both in vitro (25,31) and in vivo (25, 28 -30). In the present study we have investigated the precise molecular interactions governing the expression of qkI and Sirt2 during OL differentiation. show that qkI-5 mRNA levels were significantly increased by day 3 and qkI-6 mRNA levels were significantly increased by day 6. To further evaluate the role of QKI-5 in regulating Sirt2 expression, CG4-OL cells were transfected with a pIRES2 vector containing a qkI-5 cDNA (see "Materials and Methods") under differentiation conditions. Transfection with pIRES2-qkI-5 increased expression of qkI-5 mRNA (B). Overexpression of qkI-5 increased the expression of Sirt2 mRNA (C) on days 3 and 6 of differentiation but did not affect the expression of SIRT2 protein (D). RT-PCR (n ϭ 3 biological replicates) and immunoblot (n ϭ 3 biological replicates) data were normalized to ␤-actin and represented relative to day 0 (mean Ϯ S.D.; two-way ANOVA, *, p Ͻ 0.05).
Our findings reveal that during OL differentiation, there is coordinated expression of qkI and Sirt2 (Figs. 1 and 2) and overexpression of qkI was found to impact the expression of Sirt2 (Figs. 3 and 4) raising the possibility that QKI may directly modulate Sirt2 mRNA expression. In silico analysis revealed two putative quaking response elements, QRE-1 (AUUAAC) at 1639 bp and QRE-2 (ACUAAC) at 1853 bp in the 3Ј UTR of Sirt2 (Fig. 5A), which conform to the predicted core binding sequence ACUAAY (18) or its variant AUUAAY (8) and binding of QKI with Sirt2 transcripts was confirmed by RNA co-IP (Figs. 5, B and C, and 6A). Binding of QKI to Sirt2 mRNA was observed in both proliferating and differentiating OLs, suggesting that the interaction between QKI and Sirt2 mRNA occurs throughout OL development (Figs. 5, B and C, and 6A). More-over, we have previously shown expression of Sirt2 drives differentiation at the molecular level by enhancing the expression of MBP (31). Thus, the interaction of QKI with Sirt2 mRNA may be critical early in OL differentiation prior to the onset of myelination. Further analysis revealed that the QRE sequence ACUAAC at 1853-1858 bp was essential for binding of QKI to Sirt2 mRNA (Fig. 7, B and C). This reflected the predicted sequence for high affinity binding (18) and was highly conserved across mammalian Sirt2 transcripts (Fig. 5A). The ability of QKI to bind Sirt2 mRNA indicates that QKI controls SIRT2 protein expression during OL development through an interaction with Sirt2 transcripts.
A reduction in myelin-specific gene transcripts in qk v /qk v mice has been shown to be due to post-transcriptional regula- tion by QKI as transcription rates are not altered (7). QKI has been implicated in regulating the stability (20 -22), splicing (8,10), transport (6,7), and translation (52,53) of mRNA in developing OLs. Although QKI was found to bind all three variants of Sirt2 (Figs. 5, D and E, and 6, B and C), our data show QKI primarily modulates the post-transcriptional stability of the Sirt2.1 and Sirt2.2 mRNA (Fig. 8, A-D). In addition, up-regulation of qkI protects Sirt2.1 and Sirt2.2 mRNA from degradation (Fig. 8). Thus, similar to regulating major myelin transcripts such as Mbp, Plp, and Mag, QKI also controls the stability of Sirt2 during OL development.
In this study we demonstrate QKI was able to, in some capacity, promote the expression of all three Sirt2 variants; however, the functional interaction between QKI and Sirt2.2 appears to be important for proper OL development. Indeed, SIRT2.2 is the most abundant isoform in OLs in vitro (Fig. 2) and in vivo (25), and the main isoform incorporated into the myelin sheath (29). Although there is efficient binding of QKI to both Sirt2.1 and Sirt2.2 transcripts, the interaction between QKI and Sirt2.2 mRNA is associated with greater stabilization of this transcript (Fig. 8, B and D). These data, along with an increase in the endogenous expression of both Sirt2.2 mRNA and protein at early stages of differentiation (Fig. 2), are indicative that the regulation of Sirt2.2 by binding with QKI is critical for proper OL development. As all QKI isoforms share a common RNA binding domain (12) and the Sirt2 QRE is located in the common 3Ј UTR (Figs. 5 and 7), it is still unclear how preferential binding and protection with Sirt2.2 is achieved. Because Sirt2.2 is the most abundant transcript, possibly due to preferential usage of transcriptional start site (27) and/or alternative splicing, this increased availability could be the reason for a more prominent functional interaction with QKI. Hence, a combination of the temporal and spatial expression pattern of QKI isoforms, along with the corresponding abundance of specific target transcripts, may dictate their molecular interactions at any given stage of OL differentiation.
Our observations using primary OLs further provide evidence of physiological interaction of QKI and Sirt2 mRNA during OL development in vivo. Interestingly, up-regulation of both qkI (Figs. 3, A and B, and 4, A and B) and qkI-5 (Fig. 9C) increased expression levels of Sirt2 mRNA, but only up-regulation of qkI was associated with an increase in expression levels of SIRT2 protein (Figs. 3, C-F, and 4, C-F). The C terminus of QKI-5 is known to harbor a nuclear localization signal not found in QKI-6 and QKI-7 (17), which supports the differential subcellular localization of the QKI isoforms (4,13). This suggests that similar to Mbp (6, 7), QKI-5 may bind and retain Sirt2 transcripts in the nucleus. Although the mechanism of nuclear export involving QKI is unclear, the ratio between the nuclear isoform QKI-5 and the cytoplasmic isoforms QKI-6/QKI-7 appears to be critical (6). All QKI isoforms contain identical RNA binding and dimerization (QUA1-KH-QUA2) domains (12,15), and therefore have the ability to bind and stabilize Sirt2 transcripts. Higher expression levels of QKI-5 at early stages of OL development (i.e. glial progenitors and OPCs) would confine the mRNA in the nucleus increasing transcript levels but limiting its availability for translation. In contrast, higher expression levels of cytoplasmic QKI isoforms at later stages (i.e. pre-myelinating and myelinating OLs) would stabilize mRNA increasing the number of copies available for translation and promoting the expression of SIRT2 protein. This would serve to tightly control SIRT2 expression during OL differentiation (Fig. 11). Further complexity is added by the potential that dimerization (16,54) and the phosphorylation status (55) of the different QKI isoforms may regulate their activity at various stages of OL development.
The embryonic and early postnatal expression of qkI-5 is suggestive of a role in early OL differentiation rather than later stages of myelination (12,13,15,56). It is the predominant isoform expressed by glial progenitors in the VZ/SVZ (4), and expression peaks during early postnatal development (13,15) when OPCs undergo active proliferation and migration in the cortex (1). This is followed by a decline in qkI-5 and a corresponding increase in qkI-6 and qkI-7 during the active myelination period (13,15). There have been contradictory reports in the literature of qkI-5 both promoting (5) and impeding (6) OL differentiation. We find that selective overexpression of qkI-5 in CG4-OL cells results in a delay in differentiation (Fig. 10). This became apparent at day 3, which corresponds to the transition from proliferative OPC to the post-mitotic, immature OL phenotype denoted by the transition between expression of A2B5 and GalC, and persisted throughout day 6 under differentiating conditions. This is consistent with findings from Larocque et al. (6) showing overexpression qkI-5 delayed differentiation of OLs both in vitro and in vivo. In contrast to the dominant role of QKI-6 during peak myelination (5, 9, 10, 13, 20 -23), this suggests the role of QKI-5 may be a more subtle and complex.
Unraveling the molecular interactions between qkI and Sirt2 begins to shed some light on this complexity. Thus, it can be speculated that early in OL development the nuclear isoform QKI-5 would increase the stability of Sirt2 transcripts, limiting their availability for translation (Fig. 11). This would delay transition of OPCs to mature OLs thus allowing OPCs to proliferate and migrate from VZ/SVZ to the rest of the CNS. As the OLs reach their final destination they begin differentiating into mature, myelinating OLs where SIRT2 expression in OLs promotes downstream expression of myelin proteins (31). The phenotypes associated with the loss of qkI-5 in the qk v /qk v (13) and qk v /qk l-1 mutants (56) suggests that ability of OLs to sequester mRNAs and limit their availability for translation at early stages is a critical step for proper myelination. In differentiating OLs, cytoplasmic QKI isoforms (e.g. QKI-6) would rapidly increase quantity of Sirt2 transcripts available for translation. The competition between the differentially localized QKI isoforms would control the availability of target mRNAs for translation and allow OLs to synchronize the timing of myelin protein expression during proliferation, migration, and differentiation. Our findings on the interaction between two key regulators, qkI and Sirt2, during oligodendroglial differentiation sheds further light into this vital aspect of OL development and myelination.
Primary Oligodendrocyte Cell Culture and Electroporation-Primary OLs were prepared from C57BL/6 mice at postnatal day 1, as previously described (35,36). Briefly, mixed glial cells were prepared from the meninges-free cortices isolated from the neonatal pups. The dissected cortices were digested with Accumax solution (Sigma) and passed through a 70-m nylon cell strainer. The filtered cell suspensions were plated on a poly-D-lysine (Sigma)-coated T75 tissue culture flask and maintained as mixed glial cells in mixed cell medium containing DMEM/F-12, 10% FBS, and penicillin-streptomycin solution. Half of the culture medium was changed every other day. After 7 days, to enrich OLs in the mixed glial cells, mixed cell medium was replaced with OL growth medium (OGM) containing 10 ng/ml of biotin, 5 g/ml of insulin, 50 g/ml of transferrin, 2 mM glutamine, 30 nM sodium selenite, 0.1% BSA, 10 nM hydrocortisone, 1% penicillin-streptomycin solution in DMEM/F-12 and 30% of B104-conditioned medium. After 14 days, primary OLs were shaken off (at 200 rpm; 37°C) from the mixed glial cultures and plated in the OGM supplemented with 10 ng/ml of human recombinant platelet-derived growth factor (PDGF-AA) and 10 ng/ml of basic fibroblast growth factor (bFGF). To induce differentiation, purified primary OLs were plated in OL differentiation medium containing 10 ng/ml of biotin, 5 g/ml of insulin, 50 g/ml of transferrin, 2 mM glutamine, 30 nM sodium selenite, 0.1% BSA, 10 nM hydrocortisone, 1% penicillin-streptomycin solution in DMEM/F-12, 5 g/ml of N-acetyl-L-cysteine and 1% FBS. Transfections were carried out by electroporation (Amaxa nucleofection apparatus, Lonza) in 3-5 ϫ 10 6 primary OLs using Ingenio electroporation solution (Mirus FIGURE 11. Schematic diagram illustrating how interaction of QKI and Sirt2 regulates OL differentiation and can be critical for proper CNS myelination. QKI-5 is the predominant isoform expressed by glial progenitors in the VZ/SVZ, and expression peaks during early postnatal development when OPCs undergo active proliferation and migration in the cortex. Interaction between QKI-5 and Sirt2 in the nucleus would stabilize Sirt2 mRNA but not allow for translation. As OPCs reach their final destination in the CNS interaction between cytoplasmic QKI and Sirt2 would stabilize the mRNA and promote the translation of SIRT2 protein facilitating differentiation. This would allow OL cells to accumulate sufficient copies of Sirt2 mRNA but tightly regulate their availability for translation. Hence, competition between the differentially localized QKI isoforms during OL development would control the availability of Sirt2 mRNAs for translation and allow OLs to synchronize the timing of myelin protein expression during proliferation, migration, and differentiation. Thus, an interaction between QKI and Sirt2 would serve to regulate the timing of OL differentiation for proper CNS myelination. Bio LLC). Primary OLs were resuspended in OGM (supplemented with PDGF-AA ϩ bFGF) or OL differentiation medium and harvested 6 days after electroporation for protein and RNA extraction.
Luciferase Reporter Assay-CG4-OL cells maintained in GM were seeded in a white opaque 96-well plate (Costar) and were co-transfected with pcDNA vector or pcDNA-qkI expression vector along with Sirt2 3Ј UTR luciferase vector (pGL3-Sirt2 3Ј UTR) or putative QREs mutated Sirt2 3Ј UTR luciferase vector (pGL3-Sirt2 3Ј UTR-Mut1 or pGL3-Sirt2 3Ј UTR-Mut2 or pGL3-Sirt2 3Ј UTR-Mut1*Mut2). pRL-CMV Renilla luciferase vector was used as a control for normalizing transfection. The Dual-Glo Luciferase Assay System (Promega) was used to measure luciferase activity of cells co-transfected with qkI overexpression vector and wild-type or mutated Sirt2 3Ј UTR luciferase vectors. Luciferase assay was performed according to the manufacturer's instruction. Briefly, 48 h after transfection, an equal volume of Dual-Glo reagent was added to the cells in culture medium to measure the firefly luminescence. Subsequently, Dual-Glo Stop & Glo Reagent was added to inactivate firefly luciferase activity and the Renilla luciferase activity was measured. Relative luciferase activity was calculated by normalizing the firefly luciferase values to its respective Renilla luciferase values. Data were normalized to the Sirt2 3Ј UTR cotransfected with pcDNA control vector and represented as percentage luciferase activity.
RNA Co-immunoprecipitation-Immunoprecipitation of mRNAprotein complexes was performed according to Peritz et al., (38). The primary OLs and CG4-OLs were harvested, either in GM or after DM day 6, and lysed with polysome lysis buffer. The mixture was precleared with protein A-agarose beads before immunoprecipitation to remove nonspecific binding to the beads. Subsequently, the supernatant was incubated with 10 g of anti-QKI antibody or normal rabbit IgG overnight at 4°C. The following day, fresh protein A-agarose beads were added and incubated for 4 h at 4°C. Beads were collected after and washed four times with polysome lysis buffer and finally with polysome lysis buffer containing 1 M urea. Next, 0.1% SDS and 30 g of proteinase K were added to the beads and heated at 50°C for 30 min to retrieve the protein-RNA complex. RNA isolation and qRT-PCR was performed, as described above. HEK293 cells were co-transfected with wild-type or mutated putative QREs of the Sirt2 3Ј UTR along with pcDNA or pcDNA-qkI, to access the critical QRE site(s) involved in QKI-