Cell Culture, Transfection, and Immunofluorescence

The HeLa-derived cell line HLtat produces the HIV-1 Tat protein necessary for the activation of the long terminal repeat promoter of the gag reporter plasmid; human 293 and 293T cells were transiently transfected using Superfect (Qiagen) or calcium coprecipitation (20). Gag (HIV p24gag antigen capture assay; Zeptometrix), CAT activity, and GFP fluorescence were measured (20). For indirect immunofluorescence, the cells were fixed with 3.7% paraformaldehyde or cold methanol followed by permeabilization in 0.5% Triton X-100 (28). RBM15 polyclonal antibody (10587-1-AP; Proteintech), SC-35 mAb (SC-35; Sigma), HA epitope mAb (HA.11; Covance), and FLAG epitope mAb (M2; Sigma) were used as primary antibodies, followed by detection with Alexa-conjugated secondary antibodies (Molecular Probes, Eugene OR). In some experiments, the HA epitope was detected directly using fluorescein-conjugated HA antibodies (Roche Applied Science). The wide field epifluorescence images were acquired and processed as described (28). Image acquisition with ApoTome module was performed using Axio Observer microscope and AxioVision software (Carl Zeiss Microimaging).

Immunoprecipitation and in Vitro Binding Assays

For coimmunoprecipitation assays, 293 cells were transfected with 1–5 μg of expression constructs and harvested at day 2 post-transfection. The cells were extracted in 150 mm NaCl in the presence of 0.2% Triton X-100 in a buffer containing also 15 mm HEPES (pH 7.9), 0.1 mm EDTA, and 10% glycerol, treated with RNase A prior to immunoprecipitation with anti-FLAG-agarose (Sigma), and the complexes were eluted with 3×FLAG peptide (Sigma) or by boiling in SDS-PAGE loading buffer. For high stringency conditions, immunoprecipitations and subsequent washes were performed in the presence of 400 mm NaCl and 50 mm KCl, and the wash buffer was additionally supplemented with 2 m urea. In vitro protein binding assays were performed (28) using E. coli-expressed purified GST-tagged RBM15 and OTT3 and reticulocyte-produced NXF1 proteins.

Mice with Ott1^XK135 Allele

The mouse ES cell line XK135 (obtained from BayGenomics; funded by the NHLBI, National Institutes of Health), harbors a LacZ-neo fusion cassette inserted into the first coding exon of RBM15, 20 nucleotides upstream the initiator codon, generating a hypomorphic RBM15^XK135 allele. After injection into C57BL/6 blastocysts, the chimeric mice were bred for germline transmission. Genomic DNA from tail snips or yolk sacs was used for PCR genotyping. RBM15 alleles were detected by PCR using primers that span the cassette insertion site.

mRNA Microarrays

Total RNA from the genotyped mouse embryos was analyzed on GE Healthcare CodeLink^TM mouse whole genome bioarrays. Expression data and detailed protocols have been deposited in the NCBI Gene Expression Omnibus (29) and are accessible through GEO accession number GSE11785. Because the existing normalization approaches were not applicable for data sets reflecting biologically relevant intensity-dependent differences (30), the expression data were not subject to intensity normalization. The biologically relevant groups of transcripts in expression data sets were evaluated using PANTHER tools.

Statistical Analyses

The expression data were evaluated using the Mann-Whitney two-sample rank sum test, with α = 0.05. The embryonic genotype distributions were compared with the Mendelian distribution by using the χ² test. Data pruning was performed by averaging every 20 rows to produce one output row, using the GraphPad Prizm software.

OTT3 Promotes Export of the CAT Reporter mRNA via Its C-terminal Domain

We previously found that RBM15 acts at the post-transcriptional level and promotes export and expression of reporter transcripts (20). In this report, we tested OTT3 for such function using the DM128/B cat mRNA tethering assay (26) (Fig. 2). In this system, the reporter transcript contains the cat gene and the high affinity binding sites (Fig. 2A, box B) for the RNA-binding N-terminal domain of lambdoid phage P22 antiterminator protein N (Fig. 2A, N-peptide). This reporter RNA is designed to be retained in the nucleus and hence expresses only background levels of CAT protein. Fusion of the N-peptide to a protein with mRNA export activity leads to its tethering to the DM128/B transcript, resulting in stimulation of CAT expression. Because the major expression defect of DM128/B transcript is due to inefficient nuclear export, CAT expression activation provides a read-out of mRNA export activity of the tethered protein. Cotransfection of the cat reporter with the untagged proteins, which cannot interact with the box B-containing cat transcript, serves as specificity control.

Open in a separate window

FIGURE 2.

OTT3 stimulates cat mRNA expression when tethered to the transcript. A, tethering assay to detect nuclear export activity. CAT protein is only produced from DM128/B transcripts that are exported from the nucleus before splicing. Export can be activated when a N-tagged export factor is tethered to DM128/B via binding of the N-peptide moiety and its high affinity RNA ligand (box B) present within the intron of DM128/B. Stimulation of CAT production in the presence of a N-tagged protein provides a measure of its nuclear export activity. B, schematic of N-tagged OTT3 (aa 1–890) and of its mutants. N-OTT3 1–487 contains the RRM motifs and N-OTT3 488–890 includes the SPOC domain. To facilitate detection, these proteins also included HA epitope tags at the C terminus, and their expression was verified by Western blots using HA antibody. C, human 293 cells were transfected in duplicate with 0.02 μg of pDM128B reporter plasmid alone (−) or together with 0.3 μg of the indicated N-tagged OTT3 and RBM15 and, as positive control, the N-tagged NXF1 expression plasmids. As negative control, plasmids expressing untagged OTT3 and RBM15 were used. As internal control for transfection efficiency, all transfections included 0.5 μg of GFP expression plasmid pFRED25. Two days post-transfection, CAT activity and GFP fluorescence were measured and are plotted on the x axis (CAT, % conversion; GFP, firefly units (ffu)). The mean values and standard errors of two independently transfected plates of a representative experiment are shown. The lack of error bars in some of the columns is due to very small differences that cannot visualized in this scale.

N-peptide fusions with OTT3 and deletions thereof were generated (Fig. 2B), and their expression was verified using indirect immunofluorescence of HA-tagged proteins (data not shown). Coexpression of the N-tagged full-length OTT3 with DM128/B revealed its RNA export activity (Fig. 2C). In parallel transfections, we found that N-tagged RBM15 as well as the N-tagged NXF1 activated CAT expression as expected (20, 26). No activation was observed upon coexpression of OTT3 or RBM15 lacking the N-peptide, demonstrating that tethering to the transcript is essential for function and also shows that these proteins do not have transcriptional function. This assay further allows the dissection of the protein domains and allows a qualitative assessment of their activity. Fig. 2C shows that the C-terminal region spanning aa 488–890 of OTT3 is active, in contrast to the N-terminal region spanning aa 1–487, and this finding is in overall agreement with the reported properties of the domains of the related RBM15 (Fig. 2C) (20).

These data support the conclusion that OTT3 promotes the nuclear export of cat transcript, acting via its C-terminal region. Thus, OTT3 has mRNA export activity like previously reported RBM15, NXF1, and HIV-1 Rev (20, 26). However, OTT3 does not shuttle between the nucleus and cytoplasm, pointing to its function as export cofactor rather than export receptor, similar to the reported role of RBM15 (20).

OTT3 Promotes Expression of the RTE-containing Reporter Transcript

We previously found that RBM15 recognizes the RNA export element RTE and activates RTE-mediated reporter gene expression (20), and here we tested OTT3 for similar function. The reporter NLgag mRNA encodes the HIV p55gag protein and produces only low levels of Gag protein, whereas the insertion of RNA export elements (i.e. CTE, RTE, and RRE) 3′ to the gag gene promotes the export and expression of the unspliced gag reporter RNA (20, 22, 33–35). We expressed the empty NLgag and the RTE-containing NLgag-RTE (Fig. 3A) in human HeLa-derived HLtat cells, in the absence or presence of coexpressed SPEN proteins. We found that none of the SPEN proteins had an effect on empty NLgag, confirming the absence of nonspecific effects. These data further support the notion that the export factors need to interact specifically with the target transcript (see also Fig. 2). As expected, the presence of RTE in the gag-RTE transcript resulted in induction of Gag expression (20, 23). Interestingly, coexpression of OTT3 led to a further strong activation of expression of NLgag-RTE (Fig. 3A). Similarly, RBM15 promoted Gag expression from NLgag-RTE (20), whereas SHARP had no effect (Fig. 3A). These data demonstrated that, in contrast to SHARP, both OTT3 and RBM15 act as post-transcriptional activators (Figs. 2 and ​and3)3) and specifically activate the RTE-containing reporter RNA (Fig. 3). Together, OTT3 acts at post-transcriptional steps by (i) promoting expression of the RTE-containing reporter transcripts (Fig. 3), (ii) tethering the cat reporter to the export machinery (Fig. 2 and see below), and (iii) acting as splicing suppressor of a β-thalassemia transcript (19). Because SHARP is a transcription factor with no reported role in post-transcriptional regulation, we speculate that the precursors of SPEN proteins split into the large and small structural classes (Fig. 1B, node 1) because of functional specialization. It is plausible that the large proteins retained their original role in transcription, whereas the small proteins, like OTT3 and RBM15, evolved as post-transcriptional regulators. In support of these findings, the SPOC domain of SHARP was reported to have a much stronger transcriptional repression activity than those of RBM15 and OTT3 (19), which is consistent with the loss of transcription coregulator activity concomitant with the acquisition of post-transcriptional function.

Open in a separate window

FIGURE 3.

OTT3 activity is enhanced by the mRNA receptor NXF1. A, HeLa cells were transfected in triplicate with 1 μg of reporter NLgag or NLgag-RTE plasmids alone (−) or together with 0.5 μg of OTT3, RBM15, or SHARP expressing plasmids, as shown to the left of the panel. All transfections included 0.5 μg of GFP expression plasmid pFRED25 that served as internal control of transfection efficiency. p24^gag antigen and GFP fluorescence were measured at day 2 post-transfection, and raw expression values (p24^gag, ng/ml; GFP, firefly units (ffu)) are plotted on the x axis. The mean values and standard deviation of three independently transfected plates of a typical experiment are shown. The lack of error bars in some of the columns is due to very small differences that cannot visualized in this scale. B, human 293T cells were transfected in triplicate with 0.5 μg of reporter NLgag or NLgag-RTE plasmids in the presence or absence of NXF1 (0.5 μg) and p15/NXT1 (0.1 μg) expression plasmids. All of the transfections included 0.1 μg of HIV-1 Tat-expressing plasmid pBstat, necessary to activate expression from long terminal repeat (LTR) promoter. p24^gag antigen was measured at day 2 post-transfection, and raw expression values (p24^gag, ng/ml) are plotted on the x axis. The means and standard deviation of three independently transfected plates of a representative experiment are shown. C, Western blot analysis of OTT3-FLAG and RBM15-FLAG proteins from cells transfected as described for A.

OTT3 Acts as NXF1 Cofactor to Activate mRNA Expression

Because OTT3 is primarily a nuclear protein (19) (see Figs. 6 and ​and7),7), we next tested the possible cooperation between OTT3 with the nuclear export receptor NXF1. OTT3 and NXF1 were coexpressed with the gag reporter transcripts in the human 293T cell line that best allows study of the effects of NXF1 and its cofactor p15 (36, 37). Fig. 3B shows that coexpression of NXF1/p15 with OTT3 or RBM15 had no effect on the empty gag reporter. In contrast, NXF1/p15 further augmented the OTT3-mediated expression from NLgag-RTE, suggesting cooperativity between OTT3 and NXF1. Similarly, NXF1/p15 activated the function of RBM15 (Fig. 3B), as reported previously (20). Notably, in 293T cells, there was a pronounced difference in the extent of gag reporter gene activation comparing OTT3 (5-fold) and RBM15 (10-fold), and dose dependence studies confirmed that it was due to the intrinsic properties of RBM15 and OTT3 rather than their expression levels (data not shown). In addition, coactivation of OTT3 by NXF1/p15 (2-fold) was reproducibly slightly weaker than that of RBM15 (~3-fold). To exclude the possibility that the difference was due to lower OTT3 expression, we performed Western blot analysis using the extracts from the experiments shown in Fig. 3B, which revealed that OTT3 was expressed to even higher levels than RBM15 (Fig. 3C). Taken together, these results demonstrate the ability of OTT3 to cooperate with NXF1 in the activation of the RTE-mediated reporter gene expression and further indicate that in some cell types OTT3 function is intrinsically weaker than that of RBM15.

Open in a separate window

FIGURE 6.

OTT3 and RBM15 interact with Aly/REF. A, schematic depicts the domain structure of REF1-II and REF2-II proteins. B, association of OTT3 and RBM15 with REF2-II in vivo. Coimmunoprecipitation assays were performed in RNase A-treated extracts under using 150 mm NaCl as described for Fig. 4. C, the GFP-REF2-II was expressed in HeLa cells alone or in the presence of HA-tagged RBM15 or OTT3. After paraformaldehyde fixation and Triton X-100 permeabilization, the SPEN proteins were visualized by indirect immunofluorescence with HA antibodies, and REF2-II was detected by GFP fluorescence.

Open in a separate window

FIGURE 7.

OTT3 and RBM15 associate with the nuclear envelope and SFC and form complexes in vivo. HeLa cells expressing the indicated epitope-tagged proteins were subject to indirect immunofluorescence. A, dual-color detection of the exogenously expressed HA-tagged SPEN proteins and endogenous SC35 antigen. The cells were counterstained with nuclear dye 4′,6-diamidino-2-phenylindole (DAPI). B, dual-color detection of the exogenously expressed HA-tagged OTT3 and RBM15 proteins and FLAG-tagged NXF1. The images were acquired using ApoTome module (Carl Zeiss Microimaging). C, single-color detection of FLAG-tagged OTT3 and RBM15 proteins and their deletion mutants. D, dual-color detection of the exogenously expressed HA-tagged SPEN proteins OTT3 and SHARP and the endogenous RBM15. Raw, unfiltered images of representative fields are shown, depicting equatorial sections of the nuclei. The cells were counterstained with nuclear dye 4′,6-diamidino-2-phenylindole. E, coimmunoprecipitation of OTT3 with RBM15. Human 293 cells were transfected as indicated, and proteins were detected on Western blots in 1% aliquots of raw extracts (Input) and in OTT3-FLAG and RBM15-FLAG complexes that were immunoprecipitated with FLAG antibodies (IP: α-FLAG). Control, untransfected cells. Representative experiments are shown, and similar data were obtained in three independent experiments.

OTT3 Interacts with NXF1 in Vivo

The activation of OTT3 by NXF1 suggests interaction of the factors, and thus, we tested OTT3 for its ability to bind to NXF1. We performed coimmunoprecipitations (Fig. 4) from extracts of human 293 cells that expressed FLAG-tagged OTT3 together with HA-tagged NXF1. As control, we also transfected FLAG-tagged RBM15, which we previously reported to interact with NXF1 (20). To ensure that only RNA-independent interactions were analyzed, the extracts were pretreated with RNase A. Western blots verified the expression of HA-NXF1 as well as of OTT3 and RBM15 (Fig. 4, A and B, Input). Immunoprecipitations (Fig. 4A) using anti-FLAG antibody revealed that NXF1 coprecipitated efficiently and specifically with FLAG-OTT3 at physiological ionic strength (150 mm NaCl). Interestingly, under high stringency conditions (400 mm NaCl, 50 mm KCl, 2 m urea), coimmunoprecipitation of OTT3 with NXF1 was less efficient, whereas the RBM15-NXF1 association was not changed significantly (Fig. 4B). These data provided evidence of a differential interaction of NXF1 with OTT3 and RBM15, respectively, which likely reflects the intrinsic molecular properties of proteins rather than their availability to interact (e.g. overall expression levels). These data suggested that the reduced activity of OTT3 could be due in part to its weaker interactions with NXF1 (Fig. 3).

Open in a separate window

FIGURE 4.

OTT3 associates with NXF1 in vivo. Human 293 cells were transfected with HA-tagged NXF1 in the absence or presence of plasmids expressing FLAG-tagged intact OTT3 or RBM15 (A and B) or N- and C-terminal mutants of RBM15 and OTT3 (C). Coimmunoprecipitations were performed using anti-FLAG agarose in the presence of physiological conditions using 150 mm NaCl (A and C) or high salt using 400 mm NaCl, 50 mm KCl, and 2 m urea (B). The epitope-tagged proteins were analyzed in 1% aliquots of raw extracts (Input). The FLAG-immunoprecipitated (IP) complexes were visualized using Western immunoblot analysis. OTT3 and RBM15 proteins were detected with α-FLAG antibody, and the pulled down NXF1 was detected with α-HA antibodies. control, untransfected cells. The lower band of OTT3, pulled down on the FLAG-beads under physiological conditions (A, right panel), likely represents a degradation product.

Our previous work showed that the C-terminal region of RBM15 binds to NXF1 and suggested that this interaction is key to RBM15 activity (20). We therefore tested FLAG-tagged OTT3 deletion mutants for their association with HA-tagged NXF1 (Fig. 4C, mutants containing the C-terminal regions are indicated in bold type). These experiments revealed that OTT3 interacted with NXF1 via its C-terminal region (aa 488–890), similarly to RBM15. We noted that, in contrast to the full-length protein, the isolated C-terminal region of OTT3 was able to pull-down NXF1 more efficiently than that of RBM15 (Fig. 4C). Collectively, these data demonstrated that NXF1 binds both OTT3 and RBM15 via their C-terminal regions and that NXF1 is able to discriminate between these two proteins.

OTT3 Interacts with NXF1 in Vitro

To further address the specificity of the interactions of OTT3 and RBM15 with NXF1, we performed in vitro binding studies to map their interacting regions within NXF1. The full-length NXF1 and its deletion mutants were produced as outlined in Fig. 5A and metabolically labeled using reticulocyte extracts. These proteins were used in GST pull-down assays together with recombinant OTT3 and RBM15 proteins spanning the C-terminal regions (Fig. 5B). To exclude the RNA-mediated interactions, binding reactions were performed in the presence of RNase A. We found that OTT3 interacted specifically with the full-length NXF1, as reported for RBM15 (20), but not luciferase, which served as a negative control, and OTT3 bound more efficiently than RBM15 (Fig. 5C), consistent with the in vivo data (Fig. 4C). Fig. 5D shows the mapping of the binding domain within NXF1 by testing regions spanning aa 1–370 and 370–619 (left panel) and aa 1–505, 505–619, and 330–551 (right panel), and the data are summarized in Fig. 5A. No interaction was found with NXF1 aa 1–370, whereas aa 370–619 interacted efficiently (left panel). Analysis of additional NXF1 deletion mutants showed that the C-terminal aa 505–619 (right panel) did not interact, whereas the N-terminal region of NXF1, aa 1–505, interacted with OTT3 and RBM15, indicating that the region spanning aa 370–551, which contains the NXF1 NTF2-like domain (Fig. 5A), harbors the domain necessary for binding. Indeed, testing of the NXF1 region (aa 370–551) confirmed interaction with both OTT3 and RBM15 (Fig. 5D, right panel).

Open in a separate window

FIGURE 5.

In vitro binding of NXF1 to OTT3 and RBM15. A, schematic of the NXF1 deletion mutants and summary of the in vitro binding data. Proteins representing functional domains of NXF1 were expressed and metabolically labeled in reticulocyte extracts and tested for binding to E. coli-produced purified GST-OTT3 and GST-RBM15 proteins representing their C-terminal regions. B, expression of the recombinant proteins GST-OTT3 (488–890) and GST-RBM15 (530–977). Arrowheads, full-length; asterisks, truncated forms. C and D, GST pulldowns of ³⁵S-radiolabeled reticulocyte-produced NXF1 and its mutants. In C, firefly luciferase protein (³⁵S-Luciferase) is also included, as binding specificity control. The ³⁵S-radiolabeled proteins (10% aliquots of input fractions) and the GST pull-downs were analyzed by SDS-PAGE and visualized by radiofluorography.

Taken together, these results established that NXF1 bound in vitro to the C-terminal regions of RBM15 and OTT3 via its NTF2-like domain (aa 370–551), whereas our in vivo studies suggest that the C-terminal region of OTT3 has an intrinsically reduced interaction with NXF1 compared with RBM15, which may explain its lower functional activity observed in transfected human 293T cells (Fig. 3B).

RBM15 and OTT3 Associate with NXF1 Cofactor Aly/REF in Vivo

By screening known NXF1 cofactors using coimmunoprecipitation assays, we found that OTT3 and RBM15 also associated with Aly/REF, a RNA-binding factor that was implicated in the assembly of NXF1 with export-ready mRNP complexes, serving as a link between pre-mRNA splicing and mRNA export (38–42). We tested the interaction of OTT3 and RBM15 with the REF variants REF1-II and REF2-II, which differ by the presence of an N-terminal variable region (Fig. 6A). Coprecipitation of HA-tagged REF2-II (Fig. 6B) with FLAG-tagged OTT3 and RBM15 revealed an interaction of REF2-II with both RBM15 and OTT3. Pull-down experiments using recombinant proteins further revealed that REF2-II interacts with the C-terminal portions of OTT3 and RBM15 (data not shown), which also contain the interaction sites with NXF1 (Fig. 5). In contrast, REF1-II did not coprecipitate efficiently with RBM15 or OTT3 (data not shown).

We next compared the subcellular localization of GFP-tagged REF2-II in the absence or presence of cotransfected HA-tagged OTT3 and RBM15 in HeLa cells (Fig. 6C). REF2-II was found in the nucleoplasm and was enriched in nuclear speckles reported to represent SFC as expected (38, 43). REF2-II also accumulated in the nucleoli, which could be attributed to its nonspecific association with RNA in this compartment, and these effects were not further addressed. Coexpression of REF2-II with OTT3 and RBM15, respectively, showed extensive colocalization (Fig. 6C). REF2-II, OTT3, and RBM15 proteins were enriched in SFC, in agreement with our data in Fig. 7A (staining with anti-SC35 antibody). In contrast, REF1-II did not colocalize with OTT3 and RBM15 (data not shown), supporting our biochemical data that showed the selective binding of OTT3 and RBM15 to REF2-II but not to REF1-II. Because REF2-II mostly differs from REF1-II by the presence of a N-terminal variable region (Fig. 6A), it is plausible that this region contributed to both the localization of REF2-II with SFC and its affinity to OTT3 and RBM15. Together, these data revealed highly selective interactions of both small SPEN proteins with the mRNA export factor REF2-II that acts upstream of NXF1, further supporting our model that OTT3, like RBM15, is an integral part of the NXF1 pathway.

Distinct Subcellular Localization of OTT3

To examine possible differences between the human SPEN proteins, we first compared their subcellular localization upon transfection of HA-tagged proteins into HeLa cells. We found that HA-OTT3 and HA-RBM15 accumulated strongly in nuclear speckles (Fig. 7A). Staining with anti-SC35 antibody showed that these speckles were positive for SC35 antigen and hence represented the SFC (Fig. 7A). For RBM15, this location is expected based on the previous characterization of RBM15 as a spliceosome-associated protein (44). In contrast, HA-SHARP formed a finely punctate pattern in the nucleoplasm as previously published (10) and was not enriched in the SFC or at the nuclear envelope (NE) (Fig. 7A), further supporting a distinct biological function of the large member of the SPEN family.

For a more detailed analysis, we compared the localization of HA-OTT3 and HA-RBM15 with that of NXF1, a known NE-associated protein (24, 45, 46). To facilitate detection, NXF1 was tagged with the FLAG epitope. We found that FLAG-NXF1 was present at the NE as expected (24, 45) (Fig. 7B) and that it was also found in nuclear speckles that represent the SFC (data not shown). Both HA-RBM15 and HA-OTT3 were also prominently enriched in SFC in agreement with the data shown in Fig. 7A. In addition, HA-RBM15 accumulated at the NE, like FLAG-NXF1, whereas HA-OTT3 was mostly present in the nucleoplasm and was barely detectable at the NE (Fig. 7B).

To understand whether these differences were due to intrinsic properties of the OTT3 NE localization signal, we tested the FLAG-tagged deletion mutants of OTT3 and RBM15 for NE association (Fig. 7C). We found that the isolated N-terminal portions of RBM15 (aa 1–530) and OTT3 (aa 1–487) contain NE localization signals, whereas their C-terminal portions (RBM15, aa 530–977; OTT3, aa 488–890) localized to the nucleoplasm. Notably, in contrast to the full-length proteins (Fig. 7C, top row; see also Fig. 7, A and B), the isolated N-terminal portions of both OTT3 and RBM15 associated with the nuclear envelope with comparable efficiencies (Fig. 7C, middle row). These analyses revealed that the N-terminal regions of both proteins contain NE localization signals of similar strength but that this signal is less accessible within the full-length OTT3 protein. Collectively, these data led us to conclude that OTT3 and RBM15 localized similarly within the cell (nucleus; SFC association; NXF1 association) but that RBM15 accumulated more strongly at the nuclear envelope.

OTT3 and RBM15 Are Present in the Same Complexes in Vivo

Because OTT3 and RBM15 colocalize at the SFC (Fig. 7A), we asked whether they can interact with each other. We tested the colocalization of endogenous RBM15 and HA-tagged OTT3 and used as negative control HA-tagged SHARP. We found that the endogenous RBM15 efficiently colocalized with OTT3 (Fig. 7D) and was found enriched in the SFC compartment as identified in Fig. 7A. HA-tagged SHARP did not colocalize with the endogenous RBM15 (Fig. 8A). These data strongly suggest that RBM15 and OTT3 could be part of the same complexes in vivo. We tested this possibility by performing coimmunoprecipitation assays from extracts of cells that expressed HA-RBM15 alone or in combination with FLAG-OTT3 or FLAG-RBM15 (Fig. 7E). To exclude the RNA-mediated interactions, extracts were pretreated with RNase A. Fig. 7E shows that precipitation with the anti-FLAG antibody revealed both the RBM15-RBM15 association as well as the RBM15-OTT3 association. These data support the existence of mixed complexes in vivo.

Open in a separate window

FIGURE 8.

OTT3 cannot replace RBM15 function. A, mouse embryos at E11.5. B, levels of transcripts encoding RBM15 (red) and its homologue OTT3 (blue) were analyzed using mRNA microarrays (CodeLink, mouse whole genome bioarray, 36K), and the raw values for individual embryos (levels) are plotted on the y axis as intensity units. WT, wild type.