mRNA Export from Mammalian Cell Nuclei Is Dependent on GANP

1Although the yeast mRNA export machinery is well characterized and is partly conserved in mammalian cells, there are important functional differences. First, the TREX (transcription-export) complex, which coordinates many steps in the gene expression pathway, is recruited cotranscriptionally in yeast but by the splicing machinery in mammals. Second, although Yra1 is essential for mRNA export in yeast cells, the metazoan homolog ALY is dispensable. This suggests that additional adaptor proteins mediate the interaction between the principal mRNA nuclear export factor, NXF1, and its cargoes in metazoa. Moreover, some active genes in yeast, such as GAL1, become tethered to nuclear pore complexes (NPCs) in a process known as “gene gating”. In contrast, the majority of active genes in mammalian cells lie in transcription foci or “factories” deep within the nucleus. These fundamental differences may reflect the much greater incidence of introns in metazoan genes. Here, we identify GANP (germinal center-associated nuclear protein) as an integral component of the mammalian mRNA export machinery. GANP is upregulated in germinal center B cells and a variety of lymphomas. However, these observations do not explain why GANP is expressed in essentially all mammalian cells, suggesting that it has a more general role. Confusion has arisen in databases and the literature because the MCM3AP (MCM3 acetylating protein) gene is contained entirely within the GANP gene. However, GANP residues 1–1259 have no counterpart in MCM3AP, and MCM3AP can be transcribed independently of GANP. We propose elsewhere that they should be referred to as independent but overlapping genes (V.O.W., P.I.A.M., A.D.M., Y.T., Y.A., S.M., J.M., and R.A.L., unpublished data).

2GANP contains regions of homology to two classes of protein involved in nuclear trafficking (Figure 1A). Residues 1–340 show 23%–32% identity to regions of several highly conserved NPC proteins (FG nucleoporins) including a cluster of six FG motifs (Figure 1A; see also Figure S1A available online), whereas residues 636–990 show 25% identity to Sac3p, a component of the yeast mRNA export machinery, and to Drosophila Xmas-2 (43% identity) (Figure 1A; Figures S1B and S1C). However, the Sac3 homology domain represents only 18% of GANP, and it is present in other proteins that are not involved in mRNA export.

3Immunoblotting with sheep antibodies raised against a unique region of GANP (residues 1050–1250) that is absent from MCM3AP recognized a 210 kDa band, which was abolished following small interfering RNA (siRNA) depletion (Figure 1D). Confocal immunofluorescence of intact HCT116 cells showed strong nuclear envelope staining and weaker nuclear interior staining. Both were abrogated following siRNA-mediated depletion of GANP (Figures 1C and 1E). Immunofluorescence of permeabilized human nuclei (Figure 1B) confirmed that this nuclear envelope staining colocalized with antibody mAb414 that recognizes four integral NPC components (Nups 62, 153, 214, and 358). Antibody access experiments showed that GANP is localized to the nuclear face of NPCs but is absent from the cytoplasmic face (Figure S1E).

4To ask whether GANP functions in mammalian mRNA export, we examined the effect of GANP depletion on poly(A)+RNA export via RNA fluorescence in situ hybridization (FISH). Nuclear accumulation of poly(A)+RNA was observed with two independent siRNAs directed against the unique region of GANP, but not with control siRNA that differed by two bases from that used to deplete GANP (Figures 2A and 2D). In control cells, most poly(A)+RNA was cytoplasmic, except for a few discrete foci in nuclei (Figure 2A), as observed previously. In contrast, GANP depletion caused nuclear accumulation of poly(A)+RNA (Figures 2A and 2B), and mean nuclear polyA(+)RNA levels were ∼50% higher in GANP-depleted cells compared to control cells (97 versus 63), even without correction for the large unstained nucleolar volume (Figure S2A). Because the siRNA used corresponded to the unique region of GANP, the effects on mRNA export were specific for depletion of GANP and not MCM3AP. Importantly, nuclear import and CRM1-dependent export of STAT2 proceeded in the absence of GANP, indicating that NPCs were functional for bidirectional transport of receptor-cargo complexes in these cells (Figure S2B). Thus, nuclear export of poly(A)+RNA is severely inhibited by GANP depletion.

5Because GANP contains local homology both to nucleoporins and to Sac3p, we compared its depletion phenotype to those of either Nup153 or mRNA export factor NXF1. Figures 2C and 2D show that the punctate accumulation of poly(A)+RNA in the nucleus following GANP depletion closely resembles that seen following NXF1 depletion but differs from the pattern observed following Nup153 depletion. Poly(A)+RNA did not accumulate at the nuclear envelope of GANP-depleted cells but instead accumulated in a distinct punctate focal pattern throughout the nucleus, excluding nucleoli (Figure 2D).

6To test whether GANP is associated with nuclear messenger ribonucleoproteins (mRNPs), we harvested nuclear poly(A)+RNPs from mammalian nuclei with oligo(dT) cellulose. Endogenous GANP specifically copurified with the poly(A) fraction, as did NXF1, a known component of nuclear mRNPs (Figure 3A). Conversely, MCM2, a component of the DNA prereplication complex, did not copurify with the poly(A) fraction (Figure 3A). These results indicate that GANP associates with nuclear mRNPs, either directly or indirectly through other proteins.

7NXF1 is a major metazoan mRNA export factor and has a similar depletion phenotype to GANP (Figure 2D). We asked whether GANP interacts directly with NXF1 to facilitate efficient mRNA export in mammalian cells. Figures 3B and 3C show that endogenous NXF1 was coimmunoprecipitated from cell extracts by antibodies against GANP and that endogenous GANP was coimmunoprecipitated by antibodies against NXF1. NXF1 still coimmunoprecipitated with GANP in the presence of RNase A, confirming that this interaction was not RNA mediated (Figure 3B). Because NXF1 binds the FG-repeat region of Nup214, which is 29% identical to the N-terminal FG motif-containing region of GANP (Figure 1A), we investigated whether this region of GANP could mediate its interaction with NXF1. Figure 3D shows that the fragment generated by the fusion of cyan fluorescent protein (CFP) to the first 313 residues of GANP did indeed interact in vivo with endogenous NXF1, but that CFP alone did not. In vitro binding assays with recombinant GANP(1-313) and the FG-repeat binding region of NXF1(371-621) showed that this GANP fragment binds directly to NXF1 (Figure 3E), confirming that the first 313 residues of GANP are sufficient for interaction with NXF1. When overexpressed in HCT116 cells, CFP-GANP(1-313) was located throughout the nucleus, with no detectable NPC preference (Figure S3A). However, this was accompanied by an increased intensity of NXF1 staining within the nucleus coupled with a concomitant reduction at NPCs (Figure S3C) and nuclear poly(A)+RNA accumulation in a pattern identical to that of CFP-GANP(1-313) (Figure S3D). This effect was not due to increased NXF1 expression following CFP-GANP(1-313) expression (Figure S3B). These results suggest that the N-terminal GANP(1-313) fragment retained NXF1 and its bound mRNA within the nucleus, inhibiting their recruitment to NPCs and export (Figure S3).

8The observations that GANP interacts directly with NXF1, that it is present in both the nuclear envelope and the nuclear interior, and that it is required for efficient export of poly(A)+RNA from the nucleus suggest a role for GANP in mammalian mRNA nuclear export whereby GANP recruits NXF1-containing mRNPs in the nuclear interior and delivers them to the NPC. One prediction of this hypothesis is that GANP depletion should reduce the association of NXF1 with NPCs. Figure 3G shows that GANP depletion indeed decreased the amount of NXF1 at the NPC, with a corresponding increase in the nuclear interior, whereas the distribution of GANP was unaltered by NXF1 depletion. These results suggest that GANP contributes to the recruitment of NXF1 from the nuclear interior to NPCs but that GANP recruitment to the NPC is independent of NXF1. However, because NXF1 can itself bind directly to NPCs through transient interactions with FG nucleoporins, GANP may contribute to the efficient NPC recruitment of NXF1.

9The proportion of GANP located at NPCs was higher than that seen for NXF1 (Figure 4A), consistent with GANP's contributing to the NPC recruitment of NXF1, but not vice versa (Figure 3G), and suggesting that GANP binds to NPCs. Consistent with this hypothesis, GANP coimmunoprecipitated with a fraction containing Nup153 and Nup358 but not Nup62 (Figure 3F), and also with either mAb414 or Nup358 antibodies (Figure 3F). These results indicate that the proportion of GANP localized to NPCs interacts either directly or indirectly with a subset of nucleoporins. However, it is unlikely that GANP is an integral NPC protein, because GANP was not identified in a large-scale screen for mammalian NPC components and a proportion of GANP is inside the nucleus. Therefore, GANP interacts with both NXF1 and NPCs, supporting the hypothesis that it functions at a late stage in mRNA export.

10Nuclear mRNP export is a dynamic process, and most actively transcribed genes localize at focal concentrations of RNA polymerase II (RNA Pol II) deep in the nucleus called transcription factories. We have shown that a proportion of GANP was localized in nuclear foci and that depletion of GANP resulted in retention of mRNPs in punctate nuclear foci. The role for GANP in mRNA export indicated by our data suggests that GANP is actively involved in recruitment and transport of mRNPs through the nucleus to NPCs. If GANP is shuttling and carrying mRNPs from sites of transcription and processing to NPCs, then inhibition of transcription should perturb the distribution of GANP along this pathway. Therefore, we examined the effect of inhibiting transcription on the localization of GANP. In untreated intact cells, GANP was concentrated at the NPC, with a lower concentration in the nucleus (Figure 4B). Strikingly, treatment with DRB, a specific inhibitor of RNA Pol II elongation, redistributed GANP into foci throughout the nucleus, with less at the NPC (Figure 4B). A similar result was observed following treatment with the more general transcription inhibitor actinomycin D (Figure S4). These results suggest that GANP is mobile and that it shuttles between transcription factories and the NPC.

11Our results are consistent with a model for GANP function in mammalian mRNA nuclear export whereby GANP is recruited to mature mRNPs after NXF1 is attached, enhancing the delivery of mRNPs to the NPC (Figure 4C). In contrast to yeast, where the GANP homolog Sac3 functions with Sus1 and Cdc31 in the TREX-2 complex to target active genes such as GAL1 to NPCs, we propose that in mammalian cells, GANP facilitates targeting of NXF1-containing mRNPs to NPCs. We propose that when these mRNPs reach NPCs, the higher concentrations of FG motifs in FG nucleoporins displace GANP from NXF1 in the mRNP, freeing it to pass through the pores, facilitated by interactions between NXF1 and the FG repeats that line the transport channel. This hypothesis is supported by several lines of evidence: (1) GANP associates with nuclear mRNPs, and treatment with transcription inhibitors redistributes GANP in foci throughout the nucleus, with a reduction at NPCs, consistent with a shuttling function between transcription factories and the NPC; (2) GANP interacts with the same domain of NXF1 that interacts with FG-repeat nucleoporins; (3) GANP depletion results in retention of mRNPs in nuclear foci and not at NPCs, similar to the pattern seen following NXF1 depletion but not that observed following Nup depletion; (4) overexpression of the GANP(1-313) fragment that localizes in the nucleus and not at NPCs retains NXF1 within the nucleus, preventing its efficient recruitment to NPCs and subsequent nuclear export of NXF1-bound mRNPs; and (5) GANP depletion reduces the amount of NXF1 at NPCs with a corresponding increase in the nuclear interior, suggesting that GANP contributes to the recruitment of NXF1 from the nuclear interior to NPCs.

12Our data establish a role for GANP distinct from those proposed previously, which focused on cells of the immune system in which GANP was discovered but failed to account for GANP's being expressed universally rather than only in cells of the immune system. By contrast, our data instead suggest that other phenotypic effects of GANP depletion could be secondary consequences of the function of GANP in mRNA export.