Functional Modules in Ribosomal Protein L5 for Ribonucleoprotein Complex Formation and Nucleocytoplasmic Transport*

Ribosomal protein L5 forms a small, extraribosomal complex with 5 S ribosomal RNA, referred to as the 5 S ribonucleoprotein complex, which shuttles between nucleus and cytoplasm in Xenopus oocytes. Mapping elements in L5 that mediate nuclear protein import defines three separate such activities (L5-nuclear localization sequence (NLS)-1, -2, and -3), which are functional in both oocytes and somatic cells. RNA binding activity involves N-terminal as well as C-terminal elements of L5. In contrast to the full-length protein, none of the individual NLSs carrying L5 fragments are able to allow for the predominating accumulation in the nucleoli that is observed with the full-length protein. The separate L5-NLSs differ in respect to two activities. Firstly, only L5-NLS-1 and -3, not L5-NLS-2, are capable of promoting the nuclear transfer of a heterologous, covalently attached ribonucleoprotein complex. Secondly, only L5-NLS-1 is able to bind strongly to a variety of different import receptors; those that recognize L5-NLS-2 and -3 have yet to be identified.

Large quantities of 5 S ribosomal RNA are synthesized in excess over other ribosomal components and stored in the cytoplasm during early stages of oogenesis in Xenopus (1). 5 S rRNA storage occurs in the context of small, nonribosomal ribonucleoprotein complexes (RNPs), 1 primarily in association with transcription factor IIIA (7 S RNP) (2)(3)(4), but also as part of a larger RNP, the 42 S RNP, which contains tRNA and the two proteins, p48 and p43, in addition to 5 S rRNA (5)(6)(7). Later in oogenesis, synthesis of the primary 5 S rRNA binding ribosomal protein L5 increases (8), resulting in the formation of the preribosomal 5 S RNP (9). Binding of newly synthesized 5 S rRNA to either L5 or TFIIIA is necessary for nuclear export of the corresponding RNPs in Xenopus oocytes (10). During later stages of oogenesis, concurrent with increased assembly of ribosomal subunits, the cytoplasmically stored 5 S rRNA is reimported into the nucleus and subsequently utilized for ribosome assembly in the nucleoli. This reimport of 5 S rRNA occurs in a complex with L5. In contrast, the complex with TFIIIA remains sequestered in the cytoplasm, which is due to masking of the NLS function in TFIIIA by the associated 5 S rRNA (11)(12)(13)(14).
The nucleocytoplasmic transport of proteins and RNA proceeds though the nuclear pore complex, a large, multiprotein complex embedded in the nuclear membrane. Ions, small metabolites, and small proteins can passively diffuse through the aqueous channel of the nuclear pore complex, whereas larger molecules are actively transported through the gated channel of the nuclear pore. Nucleocytoplasmic transport of macromolecules is a signal-mediated process and requires various transport factors, some of which interact with the nuclear pore complex (15)(16)(17)(18)(19). The directionality of the translocation process is controlled by the RanGTPase system (20 -22). RanGTP allows for the dissociation of import complexes in the nucleus and is also required for the formation of export complexes including those that serve to recycle import receptors. In the cytoplasm, hydrolysis of the Ran-bound GTP allows for the subsequent dissociation of such export complexes. All transport receptors identified to date are members of the superfamily of importin ␤-like, Ran-binding proteins (23,24).
The signal structures that target proteins to the nucleus are referred to as nuclear localization sequences (NLSs). The classical NLS contains either one cluster of basic amino acids (SV40 type) or two basic clusters, separated by a 10-amino acid spacer region (bipartite type) (25)(26)(27)(28). The first transport receptor identified, the importin ␣/␤ heterodimer, interacts with these classical, basic type NLSs (29 -37). Importin ␣ serves as an adaptor molecule that binds directly to the NLS of the substrate protein (37) and interacts via its importin ␤ binding domain with the ␤-subunit of the import complex (34,38). This heterotrimeric complex binds to the nuclear pore complex and is translocated into the nucleoplasm in an energy-dependent manner (34,39). Numerous other importin ␤-related proteins have subsequently been identified and found to be also involved in import and export pathways of proteins and RNA. One of these, termed transportin or karyopherin ␤2, has been shown to mediate the import of the heterogeneous nuclear RNP protein A1 by binding to a novel, nonbasic type NLS, named the M9 sequence (40 -44). An analogous protein, Kap104p, appears to be required for the import of various heterogeneous nuclear RNP proteins in yeast, although no matching M9-like yeast import signal has been identified to date (45).
Nuclear import of U snRNPs also appears to rely on the activity of importin ␤ (46). In Xenopus oocytes, the nuclear import signal of U snRNPs is defined by the trimethyl guanosine cap structure on the one hand and by the assembled Sm core proteins on the other (47)(48)(49)(50)(51)(52)(53). Snurportin, a protein containing an importin ␣-like importin ␤ binding domain, directly interacts with the trimethyl guansine cap structure of the RNA and stimulates U snRNP import in Xenopus oocytes. It therefore appears likely that Snurportin serves as an importin ␣-like adaptor molecule, mediating the binding of the transport substrate (U snRNP) to the import receptor (importin ␤). It is not yet clear, however, how the Sm core proteins carrying the second import signal are recognized. It seems possible that an additional, yet to be identified import factor binds to the Sm core proteins (54).
Ribosomal proteins are abundant, mostly small and basic proteins. The NLSs of several different ribosomal proteins have been defined. Although some of these resemble the basic NLSs of the SV40 type (55)(56)(57)(58), others seem to be more complex and do not fit any of the previously identified structural classes of import signals (59,60). Functional analyses in yeast revealed the importance of two importin ␤-related proteins, Yrb4p (Kap123p) and Pse1p (Kap121p), for the import of ribosomal proteins (61)(62)(63). Recently, it has been demonstrated that importin ␤, transportin, RanBP5, and RanBP7 are capable of binding to and mediating the import of different ribosomal proteins (including L5) in permeabilized HeLa cells (64). After translocation to the nucleus, ribosomal proteins must migrate to the nucleolus for ribosomal subunit assembly. Despite of the fact that some regions of nucleolarly localized proteins have been shown to be necessary for nucleolar accumulation, no consensus targeting sequence has been defined to date. Nucleolar localization is likely to occur by diffusion through the nucleoplasm, followed by nucleolar retention via interaction with other nucleolar components (65)(66)(67)(68)(69)(70)(71)(72)(73), even though it cannot as yet be excluded that this process might also be signaldependent and receptor-mediated. In some cellular proteins, as well as in a number of viral proteins, very small regions have been identified that were found to be sufficient to target a reporter protein to the nucleolus, although there is no direct evidence for signal/receptor-mediated transport (74 -84).
Here, we analyze aspects of the nucleocytoplasmic transport of ribosomal protein L5 and 5 S rRNA. L5 is found to contain three separate NLS elements that function in both somatic cells and oocytes, but none of these is sufficient to promote nucleolar accumulation at levels that are characteristic of the full-length protein. N-and C-terminal sequence elements in L5 are required for 5 S rRNA binding. Only two of the three NLSs, located at the N and C termini, respectively, are capable of mediating nuclear transfer of a covalently attached complex of TFIIIA and 5 S rRNA. Finally, we demonstrate that the ability of L5 to bind to a set of different known import receptors reflects a function of the N-terminal NLS but not of the other two NLSs.

MATERIALS AND METHODS
Plasmids and Cloning Procedures-All L5 proteins that were tested in the import or RNA binding assays were fused to ␤-galactosidase in the pAX4ϩ vector (85). For in vitro expression of the fusion proteins, L5 was fused to six N-terminal Myc epitopes in pCS2ϩMT (86). The control pm-␤-Gal was generated by inserting the Ecl136II/HindII (filled in) fragment from pAX4aϩ into pCS2ϩMT, which was cut with StuI.
The different deletion mutants utilized in the Xenopus oocyte microinjection experiments were generated either by use of the appropriate restriction sites or by PCR-based deletion mutagenesis (details available on request).
For transient transfection of HeLa cells with L5-␤-galactosidase fusion constructs, the full-length L5 and the different NLS fragments were cloned into a modified pSV ␤ expression vector via BglII/NheI (87). Primers that generate BglII and NheI restriction sites were used in a PCR amplification reaction.
For in vitro protein-protein interaction experiments transport factors were cloned into the pGEX 5X-1 vector (Amersham Pharmacia Biotech) to allow for protein expression and purification of glutathione S-transferase fusions. After a PCR-mediated amplification reaction, the obtained BamHI/XhoI-fragments were ligated into the BamHI/XhoI cut pGEX 5X-1 vector.
All L5 constructs that were used for in vitro protein-protein interactions contain six Myc epitope tags and were cloned via PCR fragments from the corresponding pm-␤-Gal-constructs into the pCS2ϩMT vector. To generate pm-nucleoplasmin, the nucleoplasmin gene was amplified by PCR from pTRB102 (88). For amplification, primers were used that introduced an EcoRI site at the 5Ј-end and a XbaI site at the 3Ј-end of the PCR fragment, which was inserted into pCS2ϩMT (EcoRI/XbaI).
pmTFIIIA and pmTFIIIA⌬5-6 were generated by fusing the TFIIIA wild-type and TFIIIA⌬5-6 DNA to the six N-terminal Myc tags within the pCS2ϩMT vector. These clones carry additional vector sequences (pBluescriptII KSϩ) downstream of the TFIIIA gene, including a BamHI site and an XbaI site.
For the cloning of the TFIIIA⌬C/L5-NLS fusion constructs, the L5 sequences coding for aa 1-25 (NLS-1), 17-253 (NLS-2), and 261-296(NLS-3) were amplified by PCR with primers that introduce a NsiI site at the 5Ј-end and a BamHI site and a stop codon at the 3Ј-end of the PCR fragments. After digestion with NsiI and BamHI, the fragments were inserted into pmTFIIIA, which was also cut with NsiI and BamHI. Because the L5 fragment aa 17-254 itself contains a NsiI site, this fragment was cloned in two steps. Due to the use of the NsiI site in TFIIIA, the sequences coding for the TFIIIA C terminus, including the nuclear export signal, were deleted. TFIIIA⌬C was constructed by the deletion of the XbaI fragment in pmTFIIIA. The plasmid was cut with XbaI, the ends were filled in, and the DNA was religated.
Expression of Proteins in Vitro-Proteins were expressed from cDNAs in the coupled transcription/translation (T N T) system from Promega. Reactions were performed following the Promega (T N T) protocol, and [ 35 S]methionine or [ 35 S]cysteine (Amersham Pharmacia Biotech) was used for radiolabeling. In vitro translation products were analyzed by SDS-polyacrylamide gel electrophoresis and phosphoimaging (Molecular Dynamics).
Microinjection Experiments and Immunoprecipitations-Preparation of oocytes and microinjection assays were performed as described in Ref. 13 with the exception that oocytes were separated by collagenase (Sigma, type II) treatment (collagenase-buffer contains 82.5 mM NaCl; 2 mM KCl; 1 mM MgCl 2 ; 5 mM Hepes, pH 7.5, 1 mg/ml collagenase).
Preparation of Radiolabeled RNAs-Xenopus laevis somatic type 5 S rRNA was transcribed in vitro using the RNA transcription kit from Stratagene according to the supplied standard reaction protocol. 1 g of DraI-digested T7-5SXls DNA (see Ref. 13) was used as template in this reaction, and [␣-32 P]UTP (Amersham Pharmacia Biotech) was added for the radiolabeling of the RNA. RNA was purified using UB-blue, a purification method for RNA in the presence of urea and methylene blue (12).
Coimmunoprecipitation Experiments-For coimmunoprecipitation experiments, L5-␤-Gal-fusion proteins were in vitro translated in the coupled T N T system of Promega in the presence of 32 P-labeled 5 S rRNA. The samples were incubated for 90 min at 30°C and subsequently added to protein G-Sepharose-Myc antibody pellets. The immunoprecipitation was performed in a final volume of 500 l of NET-2 (50 mM Tris-HCl, pH 7.4, 150 mM NaCl, 0,05% Nonidet P-40, supplemented with 1 mM phenylmethylsulfonyl fluoride and 1 g/ml pepstatin A) for 1 h at 4°C. The immunoprecipitates were washed three times with NET-2 before 32 P-labeled 5 S rRNA was phenol-extracted and ethanolprecipitated. Unbound 32 P-labeled 5 S rRNA from the supernatant was also extracted, and all fractions were analyzed on a 8% polyacrylamide-7 M urea gel followed by phosphoimaging. As a control for the efficiency of the T N T protein synthesis, all reactions were additionally performed in the presence of [ 35 S]methionine. After immunoprecipitation, the bound fractions of labeled fusion proteins were analyzed by SDS-polyacrylamide gel electrophoresis and phosphoimaging as described above.
Nuclear Import of in Vitro Reconstituted RNPs-TFIIIA wild-type and mutant proteins were translated in vitro with the T N T system (Promega) in the presence of a mixture of the 32 P-labeled 5 S RNA mutants Maxi and CA67 (13). As a control, one reaction was carried out without DNA. The samples were injected into the cytoplasm of oocytes and 18 h after injection, the RNPs were immunoprecipitated from cytoplasmic and nuclear fractions via the Myc-tagged proteins. The 5 S RNAs were extracted from the immunopellet with phenol, precipitated with ethanol, and analyzed by polyacrylamide gel electrophoresis (7% urea). Subsequently, the gels were exposed to a PhosphoImager screen.
HeLa Cell Culture, Transfection, and Immunofluorescence-HeLa Cell culture, transfection experiments, and immunofluorescence were done as described in Ref. 87.
Protein Expression and in Vitro Protein-Protein Binding Experiments-All transport factors fused to glutathione S-transferase were purified from Escherichia coli as follows. 50 ml of LB medium were inoculated with a 5-ml starter culture, and cells were grown at 37°C until they reached an A 550 of 0.7. Protein expression was induced after addition of 1 mM isopropyl-1-thio-␤-D-galactopyranoside for 3 h at 30°C. The bacterial pellet was shock-frozen and then dissolved in 5 ml of cold Z-buffer (25 mM Tris-HCl, pH 7.7, 100 mM NaCl, 12.5 mM MgCl 2 , 20% glycerol, 0, 1% Nonidet P-40, protease inhibitors), sonified, and centri-fuged at 12000 ϫ g. The crude extract was aliquoted and frozen in liquid nitrogen. For protein-protein interaction experiments, the bacterial crude protein extracts were loaded on glutathione-agarose pellets (Sigma). After washing two times with binding buffer (50 mM Tris, pH 7.5, 400 mM NaCl, 2.5 mM MgAc, 0,05% Nonidet P-40, 2 mM mercaptoethanol, protease inhibitors), 10 l of 35 S-labeled in vitro translated L5 proteins or NLS fragments were added to the preloaded pellets in 300 l of binding buffer plus 2 mg/ml bovine serum albumin. After 1 h incubation, the pellets were washed three times with binding buffer (without bovine serum albumin), and bound proteins were eluted with SDS loading buffer and analyzed by SDS-polyacrylamide gel electrophoresis. RanQ69L was purified as described (32), including 0.1 mM GTP during purification steps.

RESULTS
The Ribosomal Protein L5 Contains Three Separate Nuclear Localization Signals-For the analysis of signal structure(s) responsible for nuclear import of L5, an extensive series of Nor C-terminal and internal fragments of L5 was fused to the C-terminal third of ␤-galactosidase in order to generate cargo molecules for nucleocytoplasmic transport that are above the critical molecular weight for nuclear transfer by passive diffusion. All constructs were Myc epitope-tagged in order to allow for purification by immunoprecipitation. The ability of these L5 derived fragments to mediate the nuclear import of the corresponding fusion proteins was investigated by microinjection of in vitro translated, 35 S-labeled proteins into the cytoplasm of Xenopus oocytes. 20 h after microinjection, the ␤-Gal-L5 fulllength fusion protein is efficiently transported into the nucleus. An N-terminal fragment (aa 1-25), which contains basic amino acid residues that resemble the classical basic NLS of the nucleoplasmin type, mediates import into the nucleus at a rate comparable to full-length L5 (Fig. 1A). Sequence elements within the first 16 amino acids of this NLS bearing fragment are necessary for nuclear import function, as shown by injection of an aa 17-227-containing fragment, which is retained in the cytoplasmic fraction of the oocytes.
A second NLS function that localizes to the central portion of L5 appears to require amino acids that are widely separated within the primary sequence. The L5 fragment aa 17-253 carries this second nuclear import function (Fig. 1B). When amino acids were deleted from the C terminus of this fragment, import into the nucleus was lost, and this was similarly found to be the case upon N-terminal truncation. The L5 fragment aa 32-296⌬254 -281 is efficiently transported into the nucleus (a third C-terminal NLS is deleted in this fragment; see below). The deletion of an additional 10 N-terminal amino acids leads to cytoplasmic retention. Therefore, a second nuclear import function in L5 is contained in the region defined by amino acids 32-253. Deletion of internal sequences within this NLS, as well as the replacement of these amino acids with a spacer peptide, leads to loss of nuclear transport activity (data not shown), indicating either a role of tertiary protein folding or involvement of extended sequence elements for NLS function.
A second basic cluster of amino acids, which again resembles the classical, basic NLS of the nucleoplasmin type, is located in the C terminus of the protein and is also found to carry nuclear import function (Fig. 1C). The C-terminal fragment (aa 217-289) contains this third NLS and is transported into the nucleus at a rate comparable to full-length L5. Internal deletion of the basic cluster within the C-terminal, NLS-containing fragment abolishes nuclear import activity (aa 217-289⌬254 -281). For a further delineation of the third NLS, additional mutational analyses were performed. C-terminal amino acids outside of the basic cluster are not required for NLS function (aa 217-285). Basic amino acids outside of the nucleoplasmin NLS homology region cooperate in NLS activity but are not essential (aa 217-280), whereas a partial removal of the nucleoplasmin homology region leads to a loss of NLS activity (aa 217-273).
N-terminal amino acids adjacent to the basic cluster and including some additional basic residues are not required for NLS function (aa 261-296). In contrast, the deletion of the N-terminal basic amino acids of the nucleoplasmin NLS homology region results in the sequestration of the protein to the cytoplasm (aa 268 -296). Therefore, the sequence responsible for the C-terminal NLS function of L5 is defined by amino acids 261-285 (Fig. 1, C and D).
In summary, we have identified three separate sequence elements in L5 that exhibit nuclear import function. The Nterminal L5-NLS-1 (aa 1-25) and the C-terminal L5-NLS-3 (aa 261-285) resemble the basic NLS of the nucleoplasmin type. The third NLS (L5-NLS-2) requires sequence elements widely separated within the primary amino acid sequence (aa 31-253) and therefore does not fall into any class of NLS consensus sequences that have been defined previously.
N-and C-terminal Sequences of the L5 Protein Are Required for Specific 5 S rRNA Binding-The complex of ribosomal protein L5 and 5 S rRNA not only serves as a precursor for ribosomal subunit assembly but also plays an essential role in reimporting the cytoplasmically stored 5 S rRNA into the nucleus of Xenopus oocytes. In order to be able to correlate RNA binding and NLS activities, a series of N-and C-terminally truncated L5 mutants was tested in an in vitro RNA binding assay. For this purpose, rat and Xenopus L5 fragments fused to ␤-galactosidase and Myc epitope tags were in vitro translated in the presence of 32 P-labeled Xenopus 5 S rRNA in order to allow for RNP formation. These preformed RNPs were then subjected to immunoprecipitation with Myc antibodies, and coprecipitated 5 S rRNA was extracted and analyzed by gel electrophoresis. Full-length Xenopus L5 and an N-terminally truncated fragment (aa 17-296) bind efficiently to radiolabeled 5 S rRNA, whereas the ␤-Gal fusion component alone has no detectable 5 S rRNA binding activity (Fig. 2). Two other N-and C-terminally truncated L5 variants (aa 17-253 and 47-296) were also found to bind to 5 S rRNA, albeit at a reduced efficiency. Shorter N-terminal fragments of the protein, containing amino acids 17-95, 17-164, and 1-227, respectively, had no detectable 5 S rRNA binding activity. Qualitatively corresponding results were obtained with series of N-and C-terminal deleted fragments of the rat L5 protein (Fig. 2). Full-length rat L5 bound well to Xenopus 5 S rRNA. Neither N-terminally truncated rat L5 fragments nor the majority of C-terminally deleted fragments bound to Xenopus 5 S rRNA. A very weak 5 S rRNA binding activity could be obtained with a protein containing amino acids 1-251. The importance of Nand C-terminal sequence elements in Xenopus L5 for 5 S rRNA binding was also demonstrated using electrophoretic mobility shift assays. A systematic series of N-and C-terminally truncated Xenopus L5 deletion mutants was equally translated in vitro in the presence of labeled 5 S rRNA, and such preformed RNPs were then analyzed via nondenaturing gel electrophoresis. Results obtained were qualitatively similar to those from the coimmunoprecipitation assay (data not shown). Even though the binding activities of the rat L5 variants are comparatively low, which may be a consequence of using heterologous 5 S rRNA substrate, results obtained with L5 from the two different species are qualitatively comparable. It seems noteworthy that the primary sequence of L5 has been highly con-served in evolution; the rat and Xenopus protein are 88.5% identical. In summary, we have found that sequence elements located in the N-terminal as well as in the C-terminal portion of L5 and overlapping with the NLS-2 function (see above) are required for the 5 S rRNA specific binding activity.
L5 Nuclear Import Signals and Subnuclear Localization in Somatic Cells-In order to investigate the subcellular localization of the different import competent L5 variants, and also in order to test whether the three different NLSs in L5 that were defined in the oocyte injection system also function in somatic cells, transient transfection experiments were performed. The full-length L5 and the different NLS bearing fragments were fused to ␤-galactosidase and transiently expressed in HeLa cells. The nucleocytoplasmic distribution of the corresponding fusion proteins was monitored by use of indirect immunofluorescence methods. As control proteins we used the linker histone H1 0 , which is efficiently imported into the nucleus but does not accumulate in the nucleoli, as well as an N-terminal fragment of H1 0 , which contains no functional NLS and is therefore not transported into the nucleus (87). L5-NLS-1, -2, and -3 were found to be equally competent to mediate the nuclear import of the corresponding ␤-galactosidase fusion proteins in HeLa cells (Fig. 3). An almost exclusive nucleolar accumulation is observed with the full-length L5 protein. In contrast, the individual NLSs are also strongly detected within the nucleoplasm, even though they also stain the nucleolar area. It is interesting to note that the L5-NLS-2 bearing fragment (aa 17-253), in contrast to the L5-NLS-1 and -3 fragments, also exhibits specific, albeit somewhat weakened 5 S rRNA binding activity in vitro (see above and Fig. 2). It therefore appears that nuclear transport and the ability to form RNP are not sufficient for nucleolar accumulation and assembly into ribosomal subunits.

FIG. 3. Analysis of L5-NLS-1, -2, and -3 import activities in somatic cells.
HeLa cells were transiently transfected with plasmids driving the expression of various ␤-galactosidase fusion proteins, as indicated. The intracellular localization of these proteins was detected by immunofluorescence analysis (Tritc), and the nuclei were visualized by DNA staining (DAPI). H1 0 was utilized as a positive control for nuclear import and nucleoplasmic accumulation; H1 0 -N is a deletion mutant that lacks NLS activity and serves as a negative control. The full-length L5 sequence (L5-WT) mediates pronounced nucleolar accumulation; the individual L5-NLS fragments allow nuclear transport but are more equally distributed between nucleoplasm and nucleoli.
Only Two of the Three NLSs in L5 Mediate the Import of a Heterologous RNP-The apparent redundancy of nuclear import signals as defined in L5 raises the question of whether these three NLSs are functionally equivalent, in particular in respect to mediating the nuclear import of the 5 S RNP. We therefore tested which one of these three different NLSs is by itself able to mediate the nuclear import of a heterologous 5 S rRNA containing RNP (the 7 S RNP) that is covalently attached. The 7 S RNP, consisting of TFIIIA and 5 S RNA, is normally not imported into the nucleus of Xenopus oocytes, which is due to masking of the TFIIIA NLS function upon 5 S RNA binding (13). The three different L5 fragments bearing NLS activity were fused to a TFIIIA variant (TFIIIA⌬C) that lacks the C-terminal, leucine-rich nuclear export signal, in order to avoid loss of nuclear accumulation due to a dominant export activity. The different TFIIIA⌬C/L5-NLS fusion proteins were translated in vitro and injected into the cytoplasm of Xenopus oocytes for the analysis of RNP import function. TFIIIA⌬C served as a control for cytoplasmic RNP retention and the corresponding ␤-Gal/L5-NLS fusion proteins were coinjected as a control for protein import (Fig. 4A). Only two of the three L5-NLSs were found to be able to mediate the nuclear import of the corresponding heterologous RNPs, namely L5-NLS-1 and L5-NLS-3. In contrast, the TFIIIA⌬C/L5-NLS-2 RNP was retained in the cytoplasm, even though the same NLS mediates efficient import when fused to ␤-galactosidase (␤-Gal-L5-NLS-2). We note that the import rate obtained with TFIIIA⌬C/L5-NLS-1 and TFIIIA⌬C/L5-NLS-3 appears to be reduced when compared with the import rate of the corresponding ␤-Gal-L5-NLS-1 and -3 fusion proteins.
As a further control, we also tested the ability of the different TFIIIA/L5-NLS fusion proteins to bind to 5 S rRNA in vitro (Fig. 4B). With the exception of TFIIIA⌬5-6, which is known not to be capable of 5 S rRNA binding and was included as a negative control, all TFIIIA fusion proteins analyzed in the import assay bind to 5 S rRNA, which is compatible with the idea that the nuclear import of TFIIIA⌬C/L5-NLS-1 and TFIIIA⌬C/L5-NLS-3 should indeed reflect RNP import rather than import of the free proteins. To test this latter interpretation more directly, we examined the nuclear import of RNPs reconstituted in vitro by translation of the appropriate Myctagged TFIIIA protein variants in the presence of radiolabeled 5 S RNA binding partners. The RNA preparation utilized was a mixture of the two different 5 S RNA mutants, Maxi and CA67. Maxi binds only to TFIIIA, not to L5, and is normally not imported after cytoplasmic injection (13). The CA67 mutant binds to L5 but not to TFIIIA and was included as a control for the specificity of RNP formation. After injection, RNPs were immunoprecipitated from cytoplasmic and nuclear fractions via the Myc tag. Coprecipitated RNA was extracted and analyzed by gel electrophoresis (Fig. 4C). As expected, in complex with wild-type TFIIIA, Maxi was only detected in the cytoplasmic fraction. The same result was obtained for Maxi in complex with TFIIIA⌬C/L5-NLS-2. In contrast, Maxi was also recovered with the nuclear fraction in complex with TFIIIA⌬C/L5-NLS-1 or with TFIIIA⌬C/L5-NLS-3. A significant amount of CA67 was only co-precipitated with TFIIIA⌬C/L5-NLS-2 from the cytoplasmic fraction. This is consistent with the observation that the L5 fragment contained in this fusion protein carries 5 S RNA binding activity (as discussed above and as illustrated in Fig. 2) and is therefore also able to recognize CA67. Co-precipitation of CA67, which only binds to L5, not to TFIIIA, with TFIIIA⌬C/L5-NLS-2 also indicates that the structural integrity of the L5-NLS-2 fragment is maintained in the context of this TFIIIA fusion protein, albeit being unable to mediate RNP import.
Taken together, these observations provide a strong argument for the notion that L5-NLS-1 and L5-NLS-3, but not L5-NLS-2, are competent to mediate 5 S RNP import into the nucleus. This situation thus defines a first functional difference in a comparison of the activities of the three separate NLS elements in ribosomal protein L5.
Interaction of Various Known Import Receptors with L5-NLS-1 in Vitro-It has been demonstrated more recently that a number of different import receptors, namely RanBP5, RanBP7, transportin and, to some extent, also importin ␤, can mediate import of human L5 in vitro; this is in apparent contradiction to the ability of immobilized human L5 to select importin ␤, RanBP5, and RanBP7 but not transportin from total egg extract (64). In extension of these studies, we have analyzed the ability of Xenopus L5, as well as of the isolated L5-NLS-1, -2, and -3, to interact with recombinant importin ␣, importin ␤, transportin, and RanBP7 (Fig. 5A); note that the binding experiments by Jä kel and Görlich (64) were performed out of a competitive situation with a complex mixture of the different import receptors, whereas our analysis assays the binding ability with individual recombinant proteins, i.e. out of a noncompetitive situation. However, we cannot exclude the possibility that the interactions that we detect are indirect, because the radioactively labeled L5 variants employed are produced by in vitro translation in rabbit reticulocyte lysate that contains multiple compounds of the import machinery. The different import receptors were expressed as glutathione S-transferase fusion proteins; if in vitro translated nucleoplasmin was used as a binding substrate, it was found to interact exclusively with importin ␣, thus serving as a positive control for binding specificity. The binding activity under conditions of elevated ionic strength, which reflects optimized assay conditions for the interaction of import receptors with L5, is rather weak; it is equally specific but much stronger at reduced ionic strength (data not shown). Full-length Xenopus L5 interacts relatively weakly but specifically with all four import receptors tested. The same pattern of binding activities was observed with the L5-NLS-1 fragment; in contrast, neither L5-NLS-2 nor L5-NLS-3 exhibits significant binding to any of the import receptors analyzed in this study. The relatively low binding observed with L5-NLS-2 prefers smaller protein fragments over the full-length protein and is therefore interpreted as an unspecific binding activity. The presence of 12 M RanQ69L strongly inhibited complex formation of importin ␤, transportin and RanBP7 with Xenopus ribosomal protein L5 (Fig. 5B), establishing that these interactions are highly specific. The binding of importin ␣ is not reduced to the same extent, suggesting that the interaction of L5 and importin ␣ is direct and not necessarily achieved via formation of the importin ␣/␤ heterodimer. We conclude that, as already described for human L5 (64), Xenopus L5 is able to interact with multiple import receptors. We further find that the full spectrum of binding activities is reproduced by the N-terminal 25 amino acids of Xenopus L5. The inability of hL5 to select importin ␣ and transportin from total egg extract (64), which is in apparent contrast to our in vitro binding studies with Xenopus L5, may be explained by use of a different assay system. DISCUSSION Our analysis of RNA binding and nuclear import activities in the ribosomal protein L5 from X. laevis allows for several different conclusions. L5 contains three separate nuclear localization signals, which are functionally distinct. Whereas L5-NLS-1 and -3 function in RNP import, L5-NLS-2 does not. L5-NLS-1 binds to a number of different known NLS receptors in vitro, and L5-NLS-2 and -3 do not. Complex formation with 5 S ribosomal RNA requires both N-and C-terminal sequence elements of L5. Finally, we also report that NLS function and RNA binding activity are not sufficient to allow wild-type levels of nucleolar accumulation.
The existence of multiple nuclear localization signals in ribosomal proteins has been reported previously, such as for yeast L25 and L29, as well as for human S6, S7, and L7 (55, 56, 58 -60). Many of these NLSs consist of basic amino acid stretches, which show similarities to nuclear localization signals of the SV40 or nucleoplasmin type; however, others do not fit into any of these types of consensus sequences. Of the three different NLSs in L5, the two located at the N and C termini resemble the canonical NLS of the nucleoplasmin type. The third one, referred to as L5-NLS-2 and located within the central portion of the protein, is structurally distinct and includes sequence elements that are widely separated within the primary structure of L5. L5-NLS-2 is also functionally distinct from L5-NLS-1 and -3; in contrast to these latter NLSs, it is not capable of mediating the nuclear import of a covalently attached, heterologous RNP. Because the NLS-2 bearing L5 fragment retains 5 S rRNA binding activity and also due to its complex structure, one could imagine that L5-NLS-2 is masked in the 5 S RNP and thus only visible in the free L5 protein.
Candidate nuclear import receptors for L5-NLS-2 remain to be identified, because results obtained in in vitro binding assays with a set of different known import receptors were negative. However, the fact that L5-NLS-2 can mediate nuclear protein import but not RNP import seems to indicate that these processes may occur via separate, perhaps only partially overlapping pathways.
The redundancy of nuclear import signals in L5 may also directly reflect a redundancy of nuclear import pathways. Very recently, Jä kel and Görlich (64) were able to demonstrate that four different importin ␤-like transport factors, namely importin ␤, transportin, Ran BP5, and Ran BP7, are able to mediate the nuclear transport of ribosomal proteins, including L5, in vitro. This raises the question of whether the three different NLSs in L5 may represent binding sites for different transport factors. Interestingly, L5-NLS-1 reproduces the spectrum of import receptor binding that is observed with the full-length protein. Thus, it seems likely that this basic, N-terminal NLS in L5 can utilize different import receptors or pathways. This situation is reminiscent of what has been described for ribosomal protein L23a. A relatively small sequence element in the N-terminal region of L23a, which is equally enriched in basic amino acids, serves as a binding site for four different import receptors, and it has been proposed that this sequence element might be considered as an archetypal import signal that evolved before the various import receptors diverged in evolution (64). Our observations with NLS-1 in L5 appear to reinforce the same notion, even though L5 NLS-1 does not seem to be closely related to the corresponding sequence element in L23a, referred to as the BIB domain (64), except for the notion FIG. 5. L5-NLS interactions with various import receptors. A, the bacterially expressed import receptors importin ␣, importin ␤, transportin, and Ran BP7, as well as their fusion domain glutathione S-transferase, were immobilized and incubated with in vitro translated nucleoplasmin (Nuc), Xenopus ribosomal protein L5, or the NLS containing L5 fragments (L5-NLS-1, -2 and-3). The protein fraction that was bound to the different immobilized recombinant import receptors was analyzed by gel electrophoresis. WT, wild-type. B, the interaction of Xenopus ribosomal protein L5 with different import receptors was analyzed in the presence and absence of 12 M RanQ69L (Ran-GTP). that both sequences appear relatively enriched in basic amino acids.
Our RNA binding studies with ribosomal protein L5 demonstrate that sequence elements located in the N-terminal as well as in the C-terminal portion of the protein are required for 5 S rRNA recognition and binding. Previous coprecipitation analyses by Michael and Dreyfuss (89) using biotinylated 5 S rRNA had suggested that an N-terminal fragment of rat L5 containing amino acids 1-93, as well as several other fragments that contain the same N-terminal portion of the protein, are sufficient to bind specifically to human 5 S rRNA. Using the same type of binding assay, we have been able to reproduce those results, and we have also obtained qualitatively similar results with fragments derived from Xenopus L5. However, using biotinylated U6 snRNA as a control revealed a high background of unspecific binding with the same set of proteins (data not shown). We conclude that the stringency of our coimmunoprecipitation analyses with antibodies directed against the protein moiety of the RNP complex and with an unmodified RNA substrate is significantly higher than in this latter type of assay. The importance of N-and C-terminal amino acid residues in the yeast L5 homolog, referred to as YL3, for binding to yeast 5 S rRNA has similarly been reported previously. By in vitro and in vivo binding assays, it was demonstrated that point mutations in the N-and C-terminal part of the protein, as well as C-terminal deletions, interfere with 5 S rRNA binding activity (90).
After translocation into the nucleus, ribosomal proteins must travel to the nucleolus in order to be able to participate in ribosomal subunit assembly. Despite the fact that some regions of nucleolar proteins, including some ribosomal proteins, have been shown to be necessary for nucleolar targeting, no consensus targeting sequence has been defined to date. Nucleolar localization could either occur by a combination of diffusion through the nucleoplasm and nucleolar retention by interaction with other nucleolar components (65)(66)(67)(68)(69)(70)(71)(72)(73), or it could be a signal-dependent, receptor mediated process. Sequence elements in some cellular and in a number of viral proteins have indeed been identified that are sufficient to target a reporter protein to the nucleolus, although there is no direct evidence for a receptor-mediated transport (74 -84). In transient transfection experiments, we were able to show that only the fulllength L5 protein is able to mediate efficient nucleolar accumulation. This observation is compatible with the idea that nucleolar accumulation of L5 requires the complex activities of the protein, i.e. its ability to interact with ribosomal RNA and other ribosomal proteins in the context of ribosomal subunit assembly. However, our experiments do not exclude the existence of a specific nucleolar localization signal that would have a complex architecture and would not be maintained in the different L5 fragments that we have used.