Abstract

Introduction

In eukaryotic organisms, transcription is spatially separated from translation by a nuclear envelope. Consequently, gene expression requires nuclear export of mature mRNA. Although the distribution of individual mRNA export factors has been studied, as has that of several nuclear mRNAs, the use of bimolecular fluorescence complementation (BiFC) analysis makes it possible to study the in vivo formation of complexes between different export factors that evidence indicates are functionally associated with RNA. We have used this approach to study the distribution, dynamic behavior, and relationship of Y14–nuclear export factor 1 (NXF1) complexes to RNA synthesis. The assay relies on the reconstitution of fluorescent YFP by the association of two nonfluorescent YFP half-molecules, each linked to one of two proteins, whose interactions are of interest (Hu et al., 2002). Evidence indicates that many or all of the complexes visualized are associated with RNA. Thus, monitoring the interaction of Y14 and NXF1 by BiFC indirectly allows the observation of potentially export-competent mRNA. Y14 is known to bind mRNA as part of the exon–exon junction complex (EJC) at a late stage of splicing (Kataoka and Dreyfuss, 2004) and remains bound to mRNA until translation in the cytoplasm (Dostie and Dreyfuss, 2002). Bound to the EJC, NXF1 (also called TAP) promotes export of the mature mRNA (for reviews see Dreyfuss et al., 2002; Erkmann and Kutay, 2004).

We show that coexpression of the two modified proteins, YC-Y14 and YN-NXF1, carrying the COOH- and NH₂-terminal parts of YFP, respectively, allows observation of a characteristic BiFC pattern in cell nuclei. Unexpectedly, BiFC fluorescence accumulated in speckle-associated patches, suggesting an active role for speckles in mRNA processing, although they are otherwise considered mainly as storage sites for splicing and export factors (Reed and Hurt, 2002). Findings also provided insight into the idea that the nuclear retention of RNA is one way in which nature regulates gene expression. Concordantly, it had been found that only a small fraction of all transcribed RNA is exported to the cytoplasm, although most of nuclear polymerase II–derived RNA is maturely spliced and polyadenylated (Gondran et al., 1999; Jackson et al., 2000; Weil et al., 2000). Studies using BiFC to visualize Y14–NXF1 export complexes provide new evidence relating to the nuclear retention of mRNA in vivo.

Results

YC-Y14 and YN-NXF1 reconstitute YFP fluorescence with a characteristic nuclear distribution

Upon cotransfection of YC-Y14 and YN-NXF1, MCF7 cells emitted YFP fluorescence depending on BiFC maturation for 2 h at 30°C (Fig. 1 A). Fluorescence was observed in >90% of the cells. The signal was characterized by its nuclear localization and the composition of patchy accumulations embedded in a diffuse background. In nucleoli, the signal level was very low. Immunostaining of the YC epitope (the COOH-terminal area of YFP) essentially colocalized with the BiFC pattern (Fig. 1 A). Y14 tagged by full-length YFP displayed a similar pattern, except that it also stained nucleoli (Fig. 1 B, YFP-Y14). In contrast, patchy accumulations were less obvious with YFP-tagged NXF1, where focal accumulations aligned at the nuclear periphery appeared as a characteristic expression pattern (Fig. 1 B, YFP-NXF1).

Open in a separate window

Figure 1.

BiFC of YFP from YC-Y14 and YN-NXF1 depends on specific interaction of the NXF1 and Y14 moieties. (A) MCF7 cells transfected with YC-Y14 and YN-NXF1 were incubated for 2 h at 30°C for BiFC maturation. BiFC signals are shown in the top left image. Distribution of the YC epitope in the same cell was revealed by α-YC immunodetection (top right), and chromatin was stained by DAPI (bottom left). BiFC of YFP from YN-NXF1 and YC-Y14 was confined to nuclei, where it formed a patchy pattern in chromatin-poor regions, excluding nucleoli. YFP fluorescence by BiFC and immunodetection of YC-Y14 did correspond to each other (merged; bottom right). (B) YFP-Y14 (left) corresponded to the BiFC pattern, except that nucleoli were also labeled. In contrast, YFP-NXF1 (middle left) showed focal accumulations aligned at the nuclear periphery, whereas the accumulation in specklelike patches was less conspicuous because of a higher background of inhomogeneous signal throughout the nucleus. Deletion of part of the NH₂ terminus in NXF1 did not seriously alter the distribution of the protein. In particular, both deletion mutations (NXF1 Δ1–60 and NXF1 Δ1–371) localized to the nucleus and formed focal accumulations aligned at the nuclear periphery (middle right and right). (C and D) MCF7 cells, cotransfected with either YC-Y14–YN-NXF1 Δ1–371 (C) or YC-Y14–NXF1 Δ1–60 (D), were incubated for 2 h at 30°C for BiFC maturation. Without deletions, the BiFC pattern that is characteristic for full-length constructs was not observed. Merged images: red, α-YC; green, YFP; blue, DAPI. Bars, 5 μm.

Do YC-Y14 and YN-NXF1 functionally interact with RNA?

The aforementioned experiments support the hypothesis that functional interaction of the proteins YC-Y14 and YN-NXF1 is a prerequisite for the reconstitution of fluorescent YFP. As such, BiFC should depend on active mRNA transcription. Accordingly, BiFC was compromised upon RNA polymerase II (pol II) inhibition with either 5,6-dichloro-1-β-d-ribofuranosylbenzimidazole (DRB) or α-amanitin at 37°C, even though expression of the YC epitope still appeared unperturbed (Fig. 2 A; α-amanitin not depicted). In contrast, pol II inhibition after BiFC maturation at 30°C did not abolish the signal, which remained stable within 5% for a period of at least 6 h (Fig. 2 B). Immunoprecipitation of YC-Y14 in nuclear extracts of BiFC-matured cells confirmed the expected interactions. Targeted against the YC epitope of YC-Y14, immunoprecipitation coprecipitated YN-NXF1, NXF1, Y14, and radiolabeled RNA (Fig. 3).

Open in a separate window

Figure 2.

BiFC of YFP from YC-Y14 and YN-NXF1 depends on pol II activity during BiFC maturation. (A) Pol II inhibition by DRB at 37°C compromises subsequent BiFC maturation at 30°C (left), even though the expression level of proteins appears unperturbed (middle, YC immunodetection in identical cells; right, DAPI image). (B) Pol II inhibition by DRB at 30°C after maturation of BiFC did not abolish the already established BiFC signal. DRB treatment caused BiFC patches to gather and fuse into larger particles over time (duration of DRB incubation as indicated), which is a behavior already described for SC35 domains (Spector et al., 1991). Bars, 5 μm.

Open in a separate window

Figure 3.

YC-Y14 forms complexes with Y14, YN-NXF1, NXF1, and RNA. (A) Immunoprecipitation of YC-Y14 from the nucleoplasmic fraction of BiFC-matured MCF7 cells coprecipitated YN-NXF1, as well as endogenous NXF1 and Y14. Addition of ATP to the buffer did not dissociate the complexes. (bottom) Prolonged exposure of the same blot. (B) Immunoprecipitation of YC-Y14 from the nucleoplasmic fraction of BiFC-matured cells coprecipitated radiolabeled RNA. In contrast to proteins, the beads used for immunoprecipitation bind RNA without antibody coating. As shown by the mock control (beads alone), this false positive background was low compared with the amount of coprecipitated RNA. ATP administration did not dissociate RNA from YC-Y14.

BiFC patches colocalize with SC35 domains and the perispeckle region

Signal patches of BiFC from YC-Y14 and YN-NXF1 resembled speckles in respect to size, shape, and number. Speckles are nuclear bodies enriched in splicing and export factors (Spector, 1990; Zhou et al., 2000; Rodrigues et al., 2001). Colocalization of the BiFC patches with speckles was investigated by immunodetection of SC35, a marker protein for speckles (Fig. 4). All BiFC patches colocalized with SC35 domains, and vice versa. However, the BiFC patches were larger than their corresponding SC35 domain, surmounting the speckle perimeter by up to 500 nm. In ~10% of the cells, the perispeckle signal was even higher than the SC35-positive center.

Open in a separate window

Figure 4.

BiFC-labeled Y14–NXF1 complexes accumulate in speckles and a perispeckle region. Immunolocalization of speckles (anti-SC35) in MCF7 cells, which express BiFC-YFP from YC-Y14 and YN-NXF1. (A) All BiFC patches always colocalize with speckles, and vice versa. All BiFC patches surmount their corresponding speckles in size (green border in merged image). Arrows indicate the position of a line scan across three speckles. (B) In ~10% of BiFC-expressing cells, the perispeckle BiFC signal was even stronger than the signal over the speckle. Line scan across two speckles. Red, anti-SC35; green, BiFC-YFP; blue, DAPI-labeled chromatin. Bars, 5 μm.

Export of BiFC complexes partially depends on ATP

Export capability of the BiFC complexes was investigated in digitonin-permeabilized cells. Permeabilization allows observation of unidirectional nuclear export (Kehlenbach, 2003) because soluble factors necessary for reimport become depleted. Upon permeabilization, the whole nuclear level of BiFC fluorescence decreased by two thirds within 5 min (Fig. 5 A) and remained stable for at least 5 min further. However, the remaining YFP fluorescence was immediately extinguished upon administration of 5 mM ATP. To further verify the role of ATP in releasing the Y14–NXF1 complexes, cells were pretreated with sodium azide to deplete cellular levels of ATP. Sodium azide treatment reduced overall BiFC intensity. In particular, the diffuse background appeared extracted and a diffuse rim emerged at the nuclear periphery. Upon digitonin permeabilization, the signal level decreased by 50%. Again, the administration of 5 mM ATP caused the complete loss of fluorescence (Fig. 5 B).

Open in a separate window

Figure 5.

Depletion of BiFC-YFP upon ATP administration. (A) Upon selective permeabilization of the cell membrane by digitonin, the initial BiFC fluorescence level (left) decreased by roughly 65% (middle). The remaining fluorescence vanished within seconds upon administration of 5 mM ATP (right). (B) Sodium azide treatment reduced overall BiFC intensity, and a diffuse stain at the nuclear periphery became prominent (left). In these cells, the initial signal level diminished by 50% upon digitonin permeabilization without ATP (middle). Complete loss of fluorescence could also be triggered by the administration of 5 mM ATP (right). Bar, 10 μm.

Short-term immobile fractions of BiFC complexes slowly turn over in living cells

To measure immobile fractions and the recovery speed of BiFC complexes in speckles and the nonspeckle nucleoplasm, FRAP was performed within 2-μm-diam circular regions of interest (Fig. 6). Immobile fractions measured by FRAP were confirmed by fluorescence loss in photobleaching (FLIP; Table I). Because temperature dependence was of special concern in this study, FRAP was also measured for YFP-Y14 to observe a possible shift from 37 to 30°C and RT (Table I). Besides a marginal trend of decreased immobile fraction in the cold, temperature had no significant effect within the error of individual measurements. For the BiFC signal, immobile fractions were much higher in speckles (46%) than in the nonspeckle nucleoplasm (21%). In contrast, recovery time of the mobile fraction was similar in both compartments (1.6 s). Sodium azide treatment significantly increased the immobile fractions (74 and 55%) and also decreased the recovery times. In particular, fluorescence recovery was much slower in speckles compared with the nonspeckle nucleoplasm (Table I). Performed in a timescale of 35 s, FRAP measured the mobility of fluorochromes already established at the time of bleaching. Bleaching all fluorescence of the complete nucleus, in contrast, allowed observation of de novo BiFC formation (Fig. 6). It took roughly 2 h for a steady state of restoration, which resembled the initial state. Separate analysis of the time course of recovery within speckles and the nonspeckle nucleoplasm revealed that speckles reached only 80% of their initial value.

Open in a separate window

Figure 6.

A substantial amount of BiFC complexes are immobile in vivo. The immobile fraction and half-life recovery upon photobleaching was measured within 2-μm-diam circular regions of interest that were centered over speckles and the nonspeckle nucleoplasm. (A) The FRAP curves shown were cumulated over n = 10–15 single measurements. Differences between measurements at 30°C and RT were not significant (Table I). Sodium azide treatment significantly increased the immobile fractions (74 and 55%) and slowed down recovery time. (B) Recovery of fluorescence by de novo maturation of BiFC was observed after bleaching all fluorescence of the complete nucleus (left, maximum projection of nuclei before and immediately after total bleach). 2 h after total bleach, the initial BiFC signal was almost restored (middle, images of the same nucleus before total bleach and 120 min after bleach). The time course of fluorescence intensity measured separately over speckles and the nonspeckle nucleoplasm reveals that recovery reached only 80% of its initial value (right). Control measurement in fixed cells demonstrates 5% curing of the bleached fluorochromes. Bars, 5 μm.

Table I.

Mobility of BiFC complexes and YFP-Y14 in MCF7 cell nuclei

	Half-life recoveries		Immobile fraction
	Speckle	Nucleoplasm	Speckle	Nucleoplasm
	s	s	%	%
FRAP
YFP-Y14 37°C	0.26 (0.26/0.3)	0.25 (0.28/0.24)	36 (27/40)	12 (9/19)
YFP-Y14 30°C	0.28 (0.26/0.3)	0.25 (0.24/0.29)	33 (29/37)	13 (5/21)
YFP-Y14 RT	0.26 (0.26/0.26)	0.25 (0.23/0.26)	29 (27/33)	17 (14/20)
BiFC 30°C	1.6 (1.3/2.1)	1.3 (1.0/2.1)	47 (43/49)	22 (17/31)
BiFC RT	1.9 (1.1/2.1)	1.5 (1.2/1.8)	46 (42/50)	20 (12/27)
BiFC + sodium azide	3.7 (3.2/4.5)	2.5 (2.3/4.1)	74 (72/75)	55 (43/66)
FLIP
BiFC 30°C	-	-	43 (31/53)	23 (17/30)
BiFC RT	-	-	34 (30/44)	20 (17/25)

Open in a separate window

Of n = 10–15 acquired values, the ranked mean is shown together with the third/ninth value in parentheses.

Discussion

Recently, in vivo analysis of nuclear mRNA greatly contributed to our current understanding of RNA dynamics. However, measuring the dynamics of processed endogenous mRNA in vivo has been challenging. Studies that visualize mRNA export factors individually cannot distinguish functionally associated reporters from unbound reporters. We used the BiFC approach to visualize the interaction of Y14 and NXF1. Furthermore, evidence indicates that these complexes are associated with nuclear RNA. Thus, monitoring the interaction of target molecules instead of labeling single molecules by fluorescent protein tags or by FISH probes allows discrimination of false positive backgrounds from noninteracting entities. Compared with fluorescence resonance energy transfer, which is another method used to visualize protein–protein interaction in situ, the signal-to-noise ratio is higher in the BiFC experiments because the signal is measured directly. In contrast, fluorescence resonance energy transfer analysis requires cross talk compensation between donor and acceptor channels, which intrinsically increases statistical noise.

As NXF1 binds to processed mRNA and mediates the interactions of the export complex with the nuclear pore complex, it is considered a mRNA export receptor (Katahira et al., 1999; Bachi et al., 2000; Herold et al., 2001). The combination of the splicing-associated protein Y14 with NXF1 was thought to label spliced mRNA with putative export capacity. BiFC from YC-Y14 and YN-NXF1 reproducibly demonstrated a specific nuclear distribution of the interacting proteins, revealing spatial association with the speckle compartment in particular.

The BiFC approach is particularly well suited for in vivo observations. A peculiar requirement is the maturation at a low temperature (30°C). Because cells still grow and divide at this temperature, mRNA processing and export go on at physiologically acceptable rates. Accordingly, even though we observed a trend in reduction of the immobile fraction of YFP-Y14 in speckle-associated patches, temperature dependence of immobile fractions and recovery times upon photobleaching were not significant. Thus, we consider the shift to 30°C of marginal influence on the analysis of RNA maintenance by the BiFC assay.

Several experiments were performed to demonstrate that the BiFC assay labels functional interaction of the two proteins with RNA. First, the cross exchanged pair YN-Y14–YC-NXF1 did not reconstitute YFP, indicating that BiFC depended on the sterical configuration of the linked proteins, Y14 and NXF1, and that the YC and YN parts did not interact in solution to reconstitute fluorescence. Second, the characteristic BiFC pattern was not observed if the NH₂-terminal 60 or 371 amino acids of NXF1 were deleted, even though topological expression of the proteins was not affected. The NH₂-terminal part of NXF1 is responsible for RNA binding and interaction with the EJC (Braun et al., 1999; Bachi et al., 2000; Liker et al., 2000). According to Bachi et al. (2000), deletion of the first 371 amino acids blocks nuclear export and impairs viability of the cells. This may explain the bad condition of many cells transfected with YN-NXF1 Δ1–371. BiFC fluorescence, with YN-NXF1 Δ1–371 exclusively seen in shrunken nuclei, may have evolved because of aggregation of the BiFC chimeras. The first 60 amino acids are involved in interaction of NXF1 with the EJC protein E1B-AP5 (Bachi et al., 2000). Because NXF1 Δ1–60 still binds to RNA (Braun et al., 1999, 2001), interaction of YN-NXF1 with YC-Y14 may occur with low efficiency, explaining the few cases where BiFC fluorescence could be monitored, although at very low intensity. Third, immunoprecipitation of YC-Y14 from nuclear extracts not only coprecipitated YN-NXF1 but also Y14, NXF1, and radiolabeled RNA, demonstrating that all expected interactions of YC-Y14 did occur. In particular, this experiment shows directly that the YC tag does not prohibit interaction with NXF1. Fourth, BiFC from YC-Y14 and YN-NXF1 requires pol II activity. Inhibition of pol II by DRB or α-amanitin before BiFC maturation prohibited fluorescence reconstitution, whereas already established fluorescence was not affected by pol II inhibition. All together, these experiments suggest that BiFC of YFP from YC-Y14 and YN-NXF1 monitors endogenous mRNA at a terminal state of maturation.

Because BiFC-YFP represents the interaction of YC-Y14 and YN-NXF1, the characteristic BiFC pattern should be part of both YFP-Y14 and YFP-NXF1 distributions. This was the case, as both constructs showed nuclear location with accumulation in specklelike patches. Differences from the characteristic BiFC pattern may be attributable to entities that are not functionally associated with RNA, such as the nucleolar localization of YFP-Y14. The focal accumulations of YFP-NXF1 at the nuclear edge may represent cytoplasmic entities that become docked to nuclear pores before import (Forler et al., 2004).

The BiFC complexes accumulated within speckles and the perispeckle region, suggesting a functional role of the speckle compartment in mRNA processing, including maturation and/or transport. Furthermore, the pronounced enrichment of BiFC complexes in the perispeckle space indicates that speckles associate with a specific nuclear environment.

The accumulation of RNA in speckles had already been observed for both microinjected and endogenous RNA that were labeled by FISH with either an oligo d(T) probe or sequence-specific probes (Carter et al., 1991; Zirbel et al., 1993; Melcak et al., 2001; Shopland et al., 2002). Because BiFC has a maturation time that is >2 h (Hu and Kerppola, 2003), the observed accumulation in speckles derives from complexes of rather long nuclear residence time. The poor mobility of BiFC complexes was experimentally revealed by FRAP and FLIP and was further confirmed in cell permeabilization assays. Upon cell permeabilization, roughly two-thirds of the BiFC signal was eluted. The remaining fluorescence of an apparently immobile fraction completely disappeared upon administration of ATP. One might argue that ATP destroys the BiFC complexes. However, YN-NXF1 still coprecipitates with YC-Y14 upon ATP administration, indicating that the effect of ATP is the release of the complexes. ATP-dependent mobilization of splicing- and export-related factors were described for the alternate splicing factor (Misteli et al., 1997) and the EJC protein SRm160 (Wagner et al., 2004). ATP-dependent export was also observed for in vitro–transcribed rab11 mRNA (Ossareh-Nazari et al., 2000).

In cells treated with sodium azide, the diffuse background of BiFC complexes strongly decreased. The speckle-associated accumulations became more prominent, and a rim emerged at the nuclear periphery. This reorganization may be caused by the extraction of a soluble part of BiFC complexes or by the reduced release of bound complexes. Accordingly, a lower fraction of fluorescence was eluted upon digitonin permeabilization. The peripheral rim was broad and not sharply delimited, unlike the focal accumulations of NXF1. It may represent a zone of weak binding for mRNA, which only becomes visible when the diffuse background of presumptively mobile BiFC complexes disappeared.

In vivo analysis of mobility by FRAP was kept simple. Half-life recovery and immobile fractions were read out directly from the recovery plots of bleach-corrected raw data. Neither curve fitting to decipher the complexity of presumptively superposed events nor modeling of diffusion constants wer performed. Thus, the data presented are valuable for direct comparison only. Recovery of YFP-Y14 was faster, compared with BiFC-YFP. Because the YFP-Y14 signal does not represent a pure population of complexed molecules, a fraction of unbound molecules will account for this apparent faster mobility. One might also argue that BiFC formation, by itself, impedes mobility of the complexes.

Measured by FRAP or FLIP, roughly half of the BiFC-forming complexes in the speckle-associated patches were immobile, and the immobile fraction was lower in the nonspeckle nucleoplasm, suggesting a functional association of BiFC complexes with speckles. Further evidence for functional association with speckles compared with haphazard immobilization caused by the artificial BiFC tag is provided by the efficient release of immobile BiFC complexes upon ATP administration in permeabilized cells.

Upon incubation of cells in sodium azide, most of the BiFC signal became immobile. According to the ATP-dependent release in permeabilized cells, ATP depletion may have impaired the release of bound complexes rather than decreased the nuclear export rate. FRAP of diffusible molecules is a fast process. In our study, an approximately steady state was reached within a period of 35 s. In contrast, bleaching the entire pool of fluorescence in a cell allows the observation of fluorescence turnover instead of diffusion. Upon total bleach of a nucleus, the initial BiFC pattern recovered within a period of ~2 h, demonstrating its turnover even within BiFC patches.

Previous studies reported that speckles contain poly(A) RNA (Carter et al., 1991; Huang et al., 1994). As shown by Johnson et al. (2000) and Shopland et al. (2002), at least part of this speckle RNA comes from transcription of protein-coding genes. Although possibly all actively transcribed genes associate with the surface of speckles (Huang and Spector, 1991; Johnson et al., 2000), some genes' interaction with speckles is particularly intimate, as their transcription products pervade the associated speckle (“typeI genes;” Smith et al., 1999). Transcripts can even move over and accumulate in the entire speckle compartment of the nucleus (Hattinger et al., 2002). Enrichment of messengers in speckles, however, appears to depend on the particular conditions of either the cell state or the transcript. Thus, observable accumulation throughout the entire speckle compartment of the induced gene p21 transcripts occurred in <20% of the induced cells (Hattinger et al., 2002). Comparing the distribution of collagen 1α1 and 1α2 transcripts in cultured fibroblast nuclei, Shopland et al. (2002) found significant different distribution patterns. The collagen 1α1 product pervades the entire associated speckle. Because transcripts of a splice mutation of collagen 1α1 that never leave the nucleus also pervade the entire associated speckle (Johnson et al., 2000), one may argue that speckle-associated mRNA is retained before export. This also explains why their prevalence remains practically stable upon transcripition inhibition (Huang et al., 1994; Shopland et al., 2002). Long residence time is also a characteristic feature of the observed BiFC complexes because of the duration of fluorescence maturation. There may be complexes that exit the nucleus quicker. These, however, would not be visible by the assay, as the complex is supposed to decay in the cytoplasm. Although we cannot exclude the possibility that the BiFC complexes are artificially bound within speckles because of their tag, they may equally represent a fraction of normally retained RNA. In any case, the assay reveals a possible involvement of the speckle-associated nuclear space in retention of processed mRNA.

BiFC of YFP from YC-Y14 and YN-NXF1 indicates that mammalian cell nuclei harbor NXF1-associated mRNA with very long residence times, implying that association of RNA with NXF1 is insufficient for immediate export. Because the BiFC signal can be released by ATP, an ATP-dependent mechanism may be involved in the export prohibition of NXF1-associated RNA.

Materials and methods

Acknowledgments

Notes

Abbreviations used in this paper: BiFC, bimolecular fluorescence complementation; DRB, 5,6-dichloro-1-β-d-ribofuranosylbenzimidazole; EJC, exon-exon junction complex; FLIP, fluorescence loss in photobleaching; NXF1, nuclear export factor 1; pol II, RNA polymerase II.

References