Transcription-dependent nucleocytoplasmic distribution of hnRNP A1 protein in early mouse embryos

The heterogeneous nuclear RNA-binding proteins (hnRNP proteins) constitute a family of over 20 proteins, designated A-U, which bind to nascent pre-mRNA molecules immediately upon their emergence from the transcription apparatus (reviewed in Dreyfuss et al., 1993; Kiledjian et al., 1994). The hnRNP proteins are thought to be involved in the processing reactions required to generate mature mRNA and they also play an important role in export of mRNA from the nucleus (Michael et al., 1995a; Izaurralde et al., 1997). Among hnRNPs, the A1 and C1 proteins have been extensively characterised. Both are very abundant proteins (there are 7-10×10⁷ molecules of each in the nucleus; Kiledjian et al., 1994) that differ in their intracellular transport properties. Although both proteins appear to be localised to the nucleus by immunofluorescence, A1 continuously shuttles between the nucleus and cytoplasm, whereas C1 is always retained in the nucleus. The signal within hnRNP A1 that mediates its nucleocytoplasmic shuttling is a C-terminal 38 amino acid domain, termed M9 (Siomi and Dreyfuss, 1995; Weighardt et al., 1995; Michael et al., 1995a). By contrast, hnRNP C proteins contain a nuclear retention signal that is capable of retaining proteins in the nucleus that would normally be exported (Nakielny and Dreyfuss, 1996).

Import of A1 to the nucleus occurs by a receptor-mediated pathway that relies on a novel importin-related mediator, transportin (Pollard et al., 1996). A1 binds to poly(A)⁺ RNA in both the nucleus and the cytoplasm, and electron microscopic observations show that it remains associated with Balbiani ring mRNAs as they traverse the nuclear pores (Visa et al., 1996).

An intriguing feature of some hnRNP proteins, including A1, is that their accumulation in the nucleus is blocked in the presence of RNA polymerase II transcription inhibitors, implying a link between transport and transcriptional activity (Piñol-Roma and Dreyfuss, 1991; Piñol-Roma and Dreyfuss, 1992; Piñol-Roma and Dreyfuss, 1993; Michael et al., 1995b).

In the present work we have analysed the nucleocytoplasmic distribution of hnRNP proteins in early mouse embryos because these cells contain two haploid pronuclei that remain transcriptionally inactive for a period of several hours. Thus, this system provides an important advantage over the more commonly studied somatic cells in culture, where transcription starts within minutes after completion of mitosis and assembly of daughter nuclei (Ferreira et al., 1994).

In the mouse embryo activation of zygotic gene expression is controlled by a biological clock (Schultz, 1993; Majumder and DePamphilis, 1994) and occurs by the two-cell stage, at approx. 40 hours after fertilisation (Majumder and DePamphilis, 1995; Nothias et al., 1995). However, RNA polymerase II-dependent genes can be transcribed in the one-cell embryo (Clegg and Pikó, 1982; Vernet et al., 1992; Latham et al., 1992; Ram and Schultz, 1993; Matsumoto et al., 1994; Christians et al., 1995) and it was recently discovered that fertilised mouse eggs can delay expression of zygotic genes by uncoupling translation from transcription (Nothias et al., 1996). A direct demonstration that endogenous transcription is actually taking place in mouse embryos as early as 14-17 hours after fertilisation was presented by Bouniol et al. (Bouniol et al., 1995), who succeeded in detecting bromouridine incorporated into nascent RNA transcripts following microinjection of 5-bromouridine 5′-triphosphate precursors (BrUTP). This observation was later extended by Aoki et al. (Aoki et al., 1997), who showed that BrUTP incorporation starts by mid-S phase in the one-cell mouse embryo and increases thereafter progressively.

All traffic of RNA and protein molecules between the nucleus and the cytoplasm is routed through nuclear pore complexes. Although small molecules (<60-70 kDa) may cross the pores by simple diffusion, eukaryotic cells have evolved complex transport pathways mediated by saturable receptors or carriers (for recent reviews see Ohno et al., 1998; Görlich and Kutay, 1999). As expected from a receptor-mediated process, nucleocytoplasmic transport is energy-dependent and substrate-specific. Accordingly, distinct pathways involved in nuclear import and export of different classes of proteins and RNAs have been identified (reviewed by Nigg, 1997; Nakielny and Dreyfuss, 1997; Ohno et al., 1998; Stutz and Rosbash, 1998; Görlich and Kutay, 1999).

The results reported here indicate that in transcriptionally inactive one-cell embryos the hnRNP A1 protein diffuses passively through the nuclear pores and equilibrates between nucleus and cytoplasm. Following transcriptional activation the protein concentrates in the pronuclei, making use of a carrier-mediated transport pathway. Thus, the receptor-mediated nucleocytoplasmic transport of hnRNP A1 appears to be coupled to a transcription-dependent modification of the protein. This modification most likely takes place in the nucleus, since A1 failed to concentrate in transcriptionally inactive pronuclei that were transplanted into the cytoplasm of transcriptionally active embryos.

Random-bred Swiss albino mice (4-10 weeks old; Charles River Breeding Laboratories) were used. Females were injected with human chorionic gonadotrophin (hCG) to induce superovulation and mated overnight, as described (Hogan et al., 1994). Under these conditions, fertilisation is assumed to occur approximately 12 hours post hCG injection. Fertilised eggs were recovered from ampullae, briefly treated with hyaluronidase (1 mg/ml) to remove cumulus cells and kept in culture in either Whitten’s or M2 medium (Hogan et al., 1994). For drug treatment, embryos were cultured in the presence of 5 μg/ml actinomycin D (Sigma Chemical Co., St Louis, MO, USA) or 75 μM 5,6-dichloro-1β-D-ribofuranosylbenzimidazole (DRB, Sigma). As the stock solutions of these drugs were prepared in ethanol, mock-treated embryos were cultured in Whitten’s medium in the presence of ethanol (final concentration 0.6-1%).

Microinjections were performed using a Eppendorf microinjector 5242 and a Narishige holder (IM-5, IM-6). Micropipettes obtained from Clark Electromedical Instruments (Pangbourne, UK) were freshly prepared on a P-87 puller (Sutter Instruments, USA).

For detection of transcription, embryos were microinjected in the cytoplasm with 100 mM 5-bromo-2′-uridine-5′-triphosphate (BrUTP, Sigma) as described (Bouniol et al., 1995). The incorporated bromo-uridine was detected using a monoclonal antibody directed against bromodeoxyuridine (Boehringer Mannheim, Germany) and a secondary antibody labelled with fluorescein.

Wheat germ agglutinin (WGA) (Sigma) was microinjected at a concentration of 5 mg/ml WGA in phosphate-buffered saline (PBS). Bovine serum albumin (BSA, fraction V, Sigma) was conjugated to lissamine rhodamine B-sulphonyl chloride (Polysciences).

Female F₁ hybrid mice (C57BL/6×CBA) were superovulated and mated with F₁ males overnight. One- and two-cell embryos were recovered at 16 or 39 hours post-hCG injection and kept in culture. Fusion of the two blastomeres was performed between two electrodes in a non-conductive 0.3 M mannitol solution (Sigma), using 2 DC pulses of 1.5 kV cm⁻¹ field strength for 100 milliseconds (Grass stimulator, Quincy, USA). All micromanipulations were performed at room temperature on an inverted microscope (Olympus IMT2) equipped with De Fonbrune micromanipulators (CIT-Alcatel, Annecy, France). After incubation in PB1 medium containing 5 mg ml⁻¹ cytochalasin B and 0.01 mg ml⁻¹ nocodazole (Sigma) for 15-20 minutes, a karyoplast containing pronuclei from a 18-20 hours post-hCG one-cell embryo was introduced under the zona pellucida of a 41-43 hours post-hCG two-cell embryo (McGrath and Solter, 1983). Membrane electrofusion was performed as above.

Embryos were rinsed twice in PBS, twice in HPEM buffer (60 mM Pipes, 25 mM Hepes, 10 mM EGTA, 2 mM MgCl₂, pH 6.9) and fixed/permeabilised in 3.7% paraformaldehyde in HPEM buffer containing 0.5% Triton X-100, for 40 minutes at room temperature with gentle agitation. The embryos were then washed in PBS containing 0.05% Tween 20 (PBS-Tw) for 15 minutes, incubated with 5% normal goat serum in PBS-Tw for 15 minutes, incubated with primary antibodies diluted in PBS-Tw for 1 hour at room temperature, washed twice for 15 minutes in PBS, and incubated for 1 hour with the appropriate secondary antibodies conjugated to either fluorescein or TexasRed (Vector Laboratories, Peterborough, UK). Finally, the embryos were washed twice for 15 minutes, rinsed in PBS-Tw containing 1 mg/ml DAPI (as a DNA stain) and mounted in VectaShield (Vector). Alternatively, embryos were counterstained for DNA with Propidium Iodide (PI).

The following antibodies were used: monoclonal antibodies 9H10, 4F4, 4D11 and 2A6 directed against hnRNP A1, C, L and M proteins, respectively (Choi and Dreyfuss, 1984; Piñol-Roma et al., 1989; Burd and Dreyfuss, 1994); monoclonal antibody 4G3 directed against the U2snRNP protein B′′ (Habets et al., 1989); rabbit anti-hTBP KD 55/1 polyclonal antibodies (Metz et al., 1994); and a monoclonal antibody directed against bromodeoxyuridine (Boehringer Mannheim, Mannheim, Germany).

Samples were examined with a Zeiss LSM 410 microscope. Confocal microscopy was performed using argon ion (488 nm) and HeNe (543 nm) lasers. For double-labelling experiments, images from the same focal plane were sequentially recorded and superimposed. In order to obtain a precise alignment of superimposed images the equipment was calibrated using multicolor fluorescent beads (Molecular Probes, Eugene, USA), and a dual-band filter that allows simultaneous visualisation of red and green fluorescence. UV fluorescence was acquired using a SIT-camera and an ARGUS-10 image processor (Hamamatsu Photonics, Japan).

Approximately 5×10⁶ murine ES cells were lysed in 62.5 mM Tris-HCl, Ph 6.8, 2% SDS, 5% β-mercaptoethanol, 10% glycerol, 200 U/ml benzonase, as previously described (Almeida et al., 1998). Proteins were separated on a SDS/12% polyacrylamide gel. After transfer to nitrocellulose filters, the hnRNP A1 and C proteins were detected using antibodies 9H10 and 4F4, and the ECL detection kit (Amersham) according to the manufacturer’s instructions.

At the time of ovulation the mouse oocyte is arrested at second meiotic metaphase. Fertilisation induces the completion of meiosis and the appearance of two pronuclei, one originating from the sperm and one from the oocyte germinal vesicle (Fig. 1). The first mitotic cell cycle lasts 16-18 hours and onset of transcriptional activity is detected by mid-S phase, approximately 5-8 hours after pronuclear formation (Bouniol et al., 1995; Aoki et al., 1997).

In order to induce superovulation, female mice were injected with human chorionic gonadotrophin (hCG) and the time of injection was taken as reference (Fig. 1). The nucleocytoplasmic distribution of hnRNP proteins was analysed in transcriptionally inactive early embryos (collected at 21-23 hours post-hCG) and transcriptionally active late embryos (collected at 29-30 hours post-hCG).

Previous studies on nuclear transport of hnRNP proteins have made use of cultured somatic cells treated with transcription inhibitors such as actinomycin D (Piñol-Roma and Dreyfuss, 1991; Piñol-Roma and Dreyfuss, 1992; Piñol-Roma and Dreyfuss, 1993; Michael et al., 1995b). In non-treated somatic cells, all hnRNP proteins are predominantly localised in the nucleus. Following exposure to actinomycin D, hnRNP A1 protein accumulates in the cytoplasm of interphase cells and is not detected in the nucleus of post-mitotic cells (Piñol-Roma and Dreyfuss, 1991; Fig. 2A). In contrast, the intranuclear localisation of hnRNP C, K, L and U proteins as well as of splicing factors is not affected by the drug (Piñol-Roma and Dreyfuss, 1991; Ferreira et al., 1994).

In an attempt to detect transcription-dependent changes in the nucleocytoplasmic distribution of hnRNP proteins in a drug-free system, early and late one-cell embryos were immunolabelled with the monoclonal antibody 9H10, which specifically reacts with the mammalian hnRNP A1 protein (Burd and Dreyfuss, 1994; Yang et al., 1994; Fig. 2D). During the transcriptionally inactive stage, A1 appears equally distributed between the cytoplasm and the nucleoplasm of both paternal and maternal pronuclei (Figs 2B,C, 3A). In contrast, after the onset of transcription the A1 protein is predominantly concentrated within both pronuclei (Fig. 3B). Immunolabelling experiments using monoclonal antibodies specific for hnRNP C and M proteins reveal similar changes in the nucleo-cytoplasmic distribution pattern between early and late one-cell embryos (Fig. 3C-F). A similar nucleocytoplasmic redistribution of proteins during early embryogenesis was previously reported in Drosophila (Dequin et al., 1984).

In contrast to hnRNP A1, the hnRNP L protein (Fig. 4A), the U2 snRNP protein B′′ (Fig. 4B) and the TATA-binding protein (Fig. 4C) are all concentrated in the pronuclei of transcriptionally inactive early embryos.

In summary the data show that hnRNP A1, C and M proteins fail to concentrate in the pronuclei before activation of transcription, while the hnRNP L, the U2 snRNP protein B′′, and the TATA-binding protein are exclusively detected in the nucleus before the onset of transcription. As activation of zygotic gene expression is delayed until the two-cell stage at approx. 40 hours after fertilisation (Majumder and

DePamphilis, 1995; Nothias et al., 1995; Nothias et al., 1996), these results most likely represent changes in nucleocytoplasmic distribution of maternally derived proteins.

To unambiguously demonstrate that the observed redistribution of hnRNP A1 is induced by transcriptional activation, early one-cell embryos (at 23 hours phCG) were treated with either actinomycin D or dichloro-ribofuranosylbenzimodazole (DRB) and incubated for 6 hours to allow progression of the embryos to the late stage. Both actinomycin D and DRB have the advantage of acting very rapidly in vivo, in contrast to α-amanitin, which requires hours to inhibit transcription in cultured cells (Nguyen et al., 1996). DRB has the further advantage of producing a reversible inhibition of transcription. Actinomycin D exerts its effects by binding to the DNA template, thereby interfering with the elongation of the growing RNA chain, whereas DRB is a specific inhibitor of processive transcription that blocks phosphorylation of the carboxy-terminal domain of RNA polymerase II large subunit (reviewed by Bentley, 1995).

The results show that in mock-treated embryos, the A1 protein is concentrated in the pronuclei, whereas in actinomycin-treated embryos the protein is uniformly distributed in the cytoplasm and in the nucleoplasm (Fig. 5A,B). A similar block on the import of A1 protein is observed when early embryos are treated with DRB (Fig. 5C). This effect is completely reverted within 2 hours of removal of the drug from the culture medium (Fig. 5D).

Similarly, actinomycin D and DRB prevent the accumulation of hnRNP C and M proteins in the pronuclei of late one-cell embryos (Fig. 6 and data not shown). In contrast, treatment of one-cell embryos with actinomycin D between 23 and 29 hours phCG does not affect the pronuclear localisation of hnRNP L, U2snRNP protein B′′ or TBP (data not shown).

Thus, the concentration of hnRNP A1, C and M proteins observed in the pronuclei of late one-cell embryos depends upon transcriptional activation. Surprisingly, in somatic cells transport of hnRNP C to the nucleus is transcription-independent (Piñol-Roma and Dreyfuss, 1991) and the reason for this discrepancy is unknown.

A striking difference between the intracellular distribution of hnRNP A1 in early embryos and in somatic cells is that in postmitotic cells treated with transcription inhibitors the A1 protein is excluded from the daughter nuclei, whereas in embryos the protein is uniformly present in the cytoplasm and in the nucleoplasm (cf. Fig. 1A,B).

We considered three possibile explanations for the presence of A1 in early pronuclei. Either A1 is imported because of a very low level of transcription occurring immediately after pronuclear assembly, or A1 is ‘trapped’ in the nucleoplasm during nuclear envelope assembly, or A1 (molecular size 36 kDa) enters the pronuclei by passive diffusion through the nuclear pores.

To address the first idea, early embryos were treated with actinomycin D at 14 hours phCG, before the formation of pronuclei (Fig. 7A,B). When these cells reach the late one-cell stage (at 30 hours phCG), the pronuclei have an abnormal chromatin configuration (not shown) and contain multiple nucleolus-like bodies (Fig. 7B). The level of hnRNP A1 protein present in these pronuclei is similar to that observed in the cytoplasm (Fig. 7B). This excludes the possibility that the presence of A1 in pronuclei is due to a very low level of transcription.

The second hypothesis implies that some cytoplasmic proteins may leak into the nucleoplasm during pronuclear formation, and to address this point we made use of bovine serum albumin (molecular size 66 kDa) coupled to rhodamine (BSA-rhodamine) as a marker protein. First, BSA-rhodamine (0.9 mg/ml) was microinjected in the cytoplasm of interphase and mitotic rat kangaroo kidney epithelial cells (PtK2). As expected, the results confirmed that when this protein is injected in the cytoplasm of interphase cells it does not enter the nucleus and when it is injected into mitotic cells it becomes excluded from the daughter nuclei (data not shown). The BSA-rhodamine conjugate was then injected into early one-cell embryos at 15 hours phCG (before formation of the pronuclei), and the cells were examined approximately 2 hours later. In embryos that have not yet assembled the pronuclei, BSA-rhodamine closely surrounds the condensed chromosomes (Fig. 8A,B, arrows), but as soon as the pronuclei form, BSA-rhodamine is excluded from the nucleoplasm (Fig. 8C,D, arrows). Thus, zygotic pronuclei are efficiently assembled as compartments distinct from the cytoplasm, similar to what has been observed in post-mitotic somatic cells (Benavente and Khrone, 1986; Swanson and McNeil, 1987).

Finally, the mechanism of translocation of A1 through the nuclear pores was studied by microinjection of WGA, because this lectin binds to proteins of the nuclear pore complex and selectively blocks carrier-mediated transport without interfering with passive diffusion (Davis, 1995). Importantly, the inhibitory activity of WGA is reversible, and therefore the cells must be analysed within 1-2 hours after microinjection (Yoneda et al., 1987). When early embryos are injected with WGA before pronuclear formation and examined 2 hours later, the pronuclei remain very small and have highly condensed chromatin (compare Fig. 9A,C,E with B,D,E). This reflects the general block in import of nuclear proteins, as previously shown after microinjection of WGA into mitotic PtK2 cells (Benavente et al., 1989). Although pronuclei are condensed, the A1 protein equilibrates between cytoplasm and nucleoplasm of embryos injected with WGA (Fig. 9D). In contrast, TATA-binding protein (TBP), which crosses the pores by a carrier-mediated pathway (Bustamante et al., 1995; Pemberton et al., 1999; Morehouse et al., 1999), is completely excluded from the pronuclei (Fig. 9F). This indicates that hnRNP A1 may diffuse through the nuclear pores and therefore equilibrate between the nucleus and cytoplasm.

Having established that A1 may enter the pronuclei of early embryos by passive diffusion, we next studied the mechanism of A1 concentration in the pronuclei of late embryos. DRB was added to the culture medium of embryos at 23 hours phCG, removed at 28 hours phCG, and cells observed 1 hour later. Under these conditions, the A1 protein is either concentrated in both pronuclei or appears predominantly localised in the paternal (larger) pronucleus (Fig. 10A). However, when one-cell embryos are treated with DRB and then allowed to recover from the drug for 2 hours, A1 appears concentrated in both paternal and maternal pronuclei (Fig. 5D). The observation that at 1 hour after drug removal, some embryos contain A1 predominantly in the paternal pronucleus is consistent with previous reports indicating that the paternal pronucleus starts to transcribe first and supports a higher level of transcription than the maternal (Ram and Schultz, 1993; Worrad et al., 1994; Henery et al., 1995; Aoki et al., 1997).

Injection of bromo-UTP into embryos incubated with DRB confirms that these cells lack transcriptional activity (data not shown), whereas removal of DRB for 1 hour is sufficient to activate transcription (Fig. 10C). Injection of WGA before DRB removal does not affect the onset of transcriptional activity (Fig. 10D), but completely prevents concentration of A1 in the pronuclei (Fig. 10B). This shows that the transcription-dependent concentration of A1 in late pronuclei involves a carrier-mediated transport mechanism.

How is transcription coupled to nucleocytoplasmic transport of hnRNP A1? It was previously proposed that either synthesis of an RNA molecule was required, or an RNA-dependent modification of the protein had to occur in the cytoplasm (Piñol-Roma and Dreyfuss, 1991). According to either of these hypothesis, hnRNP A1 should be transported into transcriptionally inactive nuclei placed in a transcriptionally active cell. To investigate this idea, we have made use of embryonic nuclear transfer, a technique that has been extensively used for the cloning of cattle and other mammals (see Heyman et al., 1998; Le Bourhis et al., 1998). Most importantly, it has been previously shown that nuclei transplanted into early embryos follow a programme of transcriptional activity similar to that observed during normal embryogenesis (Kanka et al., 1996). Early (transcriptionally inactive) pronuclei were transferred to 2-cell embryos and analysed either 5 or 12 hours later. At 5 hours after the transplant the pronuclei do not concentrate A1 despite the fact that they are immersed in a ‘competent’ cytoplasm (Fig. 11A,B), whereas at 12 hours A1 is clearly accumulated in the transplanted pronuclei (Fig. 11D,E). As a control, one-cell embryos from the same group as those used for transplants were maintained in culture for 5 or 12 hours. At 5 hours, these embryos are still in the early stage and do not concentrate A1 in the pronuclei (Fig. 11C), whereas at 12 hours the embryos have reached the late stage and accumulate A1 in the pronuclei (Fig. 11F). These results show that the presence of newly synthesised RNAs in the cytoplasm is not sufficient to induce the accumulation of hnRNP A1 in the nucleus.

In this work we show that the onset of transcription induces a nucleo-cytoplasmic redistribution of hnRNP A1 protein, which becomes accumulated in the pronuclei of one-cell mouse embryos. This redistribution involves a carrier-mediated transport pathway and most likely results from the appearance of nascent transcripts in the nucleus.

All macromolecules in transit between nucleus and cytoplasm are routed through the nuclear pore complexes, which form aqueous channels penetrating and fusing the double bilayer membrane of the nuclear envelope. Nuclear pore complexes contain a diffusion channel of 9 nm in diameter, which allows passive movement of small molecules in and out of the nucleus, whereas globular molecules greater then 60-70 kDa fail to diffuse across the pores at a significant rate. However, macromolecular complexes as large as 25-50 MDa can be translocated through the pores by active or facilitated transport mechanisms. Moreover, small molecules are also selectively transported by carrier-mediated pathways when they contain a nuclear localisation signal. In order to allow the translocation of receptor-cargo complexes the nuclear pore complex undergoes a change in conformation and expands the channel to a maximum of approximately 25 nm in diameter (Nigg, 1997; Ohno et al., 1998; Görlich and Kutay, 1999).

Several studies performed on cultured mammalian somatic cells have identified signals within the shuttling protein hnRNP A1 that mediate both its nuclear import and export. A so-called M9 domain of 38 amino acids near the carboxyl terminus of A1 has been shown to be necessary to localise the protein to the nucleus. Furthermore, M9 is sufficient to localise normally cytoplasmic proteins to the nucleus when they are fused with this domain (Siomi and Dreyfuss, 1995; Weighardt et al., 1995), and an M9-interacting protein, termed transportin, was identified as a novel receptor for nuclear protein import (Pollard et al., 1996).

Despite having a nuclear localization signal, the hnRNP A1 protein is a small molecule of 36 kDa. Therefore, A1 may passively diffuse along the 9 nm channel of the resting nuclear pore complex. This is in fact observed at the early stages of mouse cell-embryos, when A1 equilibrates between cytoplasm and nucleoplasm. In contrast, when mitotic somatic cells are treated with the transcription inhibitor actinomycin D, A1 appears excluded from the nucleus. Most likely, A1 enters all nuclei by passive diffusion, but in somatic cells it is rapidly re-exported to the cytoplasm. Consistent with this view, in somatic cells the A1 protein shuttles rapidly between the nucleus and the cytoplasm, and export to the cytoplasm is not affected by the presence of transcription inhibitors (Piñol-Roma and Dreyfuss, 1992).

Following the onset of transcriptional activity, the hnRNP A1 protein becomes concentrated in the pronuclei by a mechanism that is blocked by WGA. This plant lectin recognises the O-linked N-acetylglucosamine residues present in a subset of the proteins that constitute the nuclear pore complex (Davis, 1995). As a consequence, WGA arrests translocation of receptor-cargo complexes through the pore, similar to the effect observed after ATP depletion. The result that hnRNP A1 fails to accumulate in transcriptionally active pronuclei treated with WGA (Fig. 10B) argues against the possibility that A1 enters the nucleus by diffusion and becomes concentrated there due to retention by binding to nascent transcripts. Rather, the effect produced by WGA injection strongly suggests that the selective accumulation of A1 in transcriptionally active pronuclei involves a receptor-mediated transport pathway.

In addition to hnRNP A1, the nuclear transport receptor transportin mediates the nuclear import of other RNA-binding proteins, such as the poly(A) binding protein PABP2 (Calado et al., 2000). However, the facilitated import of PABP2 to the nucleus is independent of ongoing RNA synthesis (Calado and Carmo-Fonseca, 2000), suggesting that transcription-dependent nuclear import is determined by the cargo and not by the transport receptor. Accordingly, the nucleo-cytoplasmic distribution of transportin is not affected by transcription inhibitors (Siomi et al., 1997). These data imply that an RNA-dependent modification of hnRNP A1 is required to trigger its selective accumulation in the nucleus.

As the A1 protein shuttles, an increase in its steady state nuclear concentration could be the result of either faster nuclear import or slower export to the cytoplasm. The result that A1 fails to accumulate in a transcriptionally inactive pronucleus transplanted into a transcriptionally active embryo (Fig. 11A,B) indicates that the presence of newly synthesised RNAs in the cytoplasm is not sufficient to induce the accumulation of hnRNP A1 in the nucleus. The accumulation of hnRNP A1 in the transplanted pronucleus was shown to be delayed until its own transcription is activated (Fig. 11D,E), suggesting that the appearance of nascent transcripts in the nucleus play a crucial role in this process. Possibly, binding of A1 to nascent transcripts retards its export to the cytoplasm and therefore contributes to concentrate the protein in the nucleus. Considering that A1 is a major hnRNA-binding protein that associates with nascent RNAs co-transcriptionally, it is most plausible that, in the presence of nascent transcripts, the A1 protein becomes anchored to chromatin via RNA and, consequently, is retarded in the nucleus.

We wish to acknowledge Prof. David-Ferreira for support, Dr João Ferreira and Dr Peter Jordan for critical comments, Dr Pierre Adenot for help in some experiments, and Mr. Romão for animal care facilities. We are also grateful to Prof. G. Dreyfuss and Dr Maurice Swanson, Prof. W. van Venrooij and Dr R. Bravo for generously providing anti-hnRNP, anti-B′′ and anti-TBP antibodies, respectively. This study was supported by grants from Junta Nacional de Investigação Científica e Tecnológica/Programme PRAXIS XXI and from the European Union (Human Capital and Mobility Programme).

D.V. was the recipient of a postdoctoral fellowship from the Human Capital and Mobility Programme (CT930251); C.C. and A.C. were supported by PRAXIS XXI fellowships.