G. Steiner, Division of Rheumatology, Department of Internal Medicine III, and Institute of Biochemistry, University of Vienna, Waehringer Guertel 18–20, A-1090 Vienna, Austria. Fax: + 431 40400 4306, Tel.: + 431 40400 ex. 4301 or 2121, E-mail: Guenter.Steiner@akh-wien.ac.at
Human Ro ribonucleoproteins (RNPs) are composed of one of the four small Y RNAs and at least two proteins, Ro60 and La; association of additional proteins including the Ro52 protein and calreticulin has been suggested, but clear-cut evidence is still lacking. Partial purification of Ro RNPs from HeLa S100 extracts allowed characterization of several subpopulations of Ro RNPs with estimated molecular masses of between 150 and 550 kDa. The majority of these complexes contained Ro60 and La, whereas only a small proportion of Ro52 appeared to be associated with Ro RNPs. To identify novel Y RNA-associated proteins in vitro, binding of cytoplasmic proteins to biotinylated Y RNAs was investigated. In these reconstitution experiments, several proteins with estimated molecular masses of 80, 68, 65, 62, 60 and 53 kDa, the latter two being immunologically distinct from Ro60 and Ro52, respectively, appeared to bind specifically to Y RNAs. Furthermore, autoantibodies to these proteins were found in sera from patients with systemic lupus erythematosus. The proteins bound preferentially to Y1 and Y3 RNA but, with the exception of the 53-kDa protein, only weakly to Y4 RNA and not at all to Y5 RNA. Coprecipitation of the 80, 68, 65, and 53-kDa proteins by antibodies to Ro60 and La was observed, suggesting that at least a proportion of the novel proteins may reside on the same particles as La and/or Ro60. Finally, the binding sites for these proteins on Y1 RNA were clearly distinct from the Ro60-binding site involving a portion of the large central loop 2, which was found to be indispensable for binding of the 80, 68, 65 and 53-kDa proteins, as well as the stem 3–loop 3 and stem 2–loop 1 regions. Interestingly, truncation of the La-binding site resulted in decreased binding of the novel proteins (but not of Ro60), indicating La to be required for efficient association. Taken together, these results suggest the existence of further subpopulations of Ro RNPs or Y RNPs, consistent with the heterogeneous characteristics observed for these particles in the biochemical fractionation experiments.
Ro ribonucleoproteins (RNPs), which may also be called Y RNPs, are composed of one of four small cytoplasmic RNAs (84–112 nucleotides) termed Y1–Y5 (Y2 is a degraded form of Y1 RNA) and at least two proteins: the 60-kDa protein, Ro60, and the 48-kDa phosphoprotein, La (reviewed in [1,2]). Ro60 and La belong to the family of RNA-binding proteins characterized by a conserved RNA-binding domain (also termed RNA-recognition motif) of ≈ 80 amino acids. A third protein of molecular mass 52 kDa (Ro52) has been reported to be associated with Ro RNPs, but this association has not been unequivocally proven so far [3–8]. Ro52 is completely unrelated to Ro60 and La and belongs to a group of proteins assumed to be involved in cell activation and transformation (reviewed in ). The four Y RNAs are closely related and are assumed to have similar predicted secondary structures. They are transcribed by RNA polymerase III but, in contrast with most other polymerase III transcripts, retain the oligo(U) stretch at their 3′ end. As this sequence is indispensable for binding of La, most if not all Y RNAs are presumably associated with La [10,11]. The binding site for Ro60 on Y RNAs was localized to the highly conserved stem structure formed by base-pairing of the 3′ and 5′ ends. In contrast, no binding site for Ro52 has been identified, and it was suggested that association of Ro52 with Ro RNPs is mediated via interaction with Ro60 .
For the La protein, which is about 50-fold more abundant than the two Ro proteins, several functions have been proposed. Almost a decade ago, Gottlieb and Steitz  demonstrated this protein to be essential for correct RNA polymerase III transcription, an observation recently confirmed and extended by other investigators [13,14]. In addition, a DNA–RNA unwinding activity was ascribed to La, and it was shown that La can inhibit the activation of the interferon-inducible protein kinase PKR by unwinding dsRNA [15–17]. Moreover, a possible involvement of La in translation of certain viral RNAs has been proposed [18–20], an assumption supported by the recently reported association of La with small ribosomal subunits . Taken together, these data indicate a (cytoplasmic) role for La in the regulation of eukaryotic translation in addition to its (nuclear) function as a regulatory transcription factor. In contrast, the role of the Ro proteins and Y RNAs, respectively, is largely obscure. They might be involved in similar biological processes as La, e.g. by acting as cofactors or even by regulating La functions.
The size of Ro RNPs was estimated to be between 200 and 350 kDa, which is larger than expected from the molecular masses of the known components [5,10]. Moreover, cDNAs encoding alternatively spliced variants of both Ro proteins and La have been described [22–24]. Thus, the detailed molecular structure of native Ro RNPs is largely unknown, and it is assumed that they may contain additional components yet to be identified. Although there is now strong evidence that Ro RNPs are localized in the cytoplasm of all higher eukaryotic cells, Ro proteins can also be detected in the nucleus. However, species located in the nucleus do not seem to be associated with Y RNAs, and it has been suggested that Ro proteins associate with Y RNAs immediately before leaving the nucleus [11,25,26] (reviewed in ).
As neither the role nor the molecular structure of Ro RNPs is known in detail, the aim of this study was to investigate the molecular composition of Ro RNPs using biochemical purification and in vitro reconstitution techniques. The results obtained indicate the existence of several proteins that, in addition to Ro60 and La, appear to bind specifically to Y RNAs.
Materials and methods
Partial purification of Ro RNPs
HeLa cells (1 × 1010; purchased from the Computer Cell Culture Center, Mons, Belgium) were washed twice in isotonic buffer (10 mm Tris/HCl, pH 7.9, 140 mm KCl, 1.5 mm MgCl2, 1 mm EDTA and 25% glycerol), resuspended in 2 pellet vol. buffer A (50 mm Tris/HCl, pH 7.4, 50 mm NaCl, 1.5 mm MgCl2, 1 mm EDTA and 25% glycerol) and disrupted by Dounce homogenization. Nuclei were separated by brief centrifugation (3 min at 3000 g), and the supernatant was first centrifuged for 20 min at 20 000 g and then for 1 h at 100 000 g. The resultant S100 extract (protein concentration 5–10 mg·mL−1) was loaded on a 220 × 30 mm DEAE-Sepharose column (Pharmacia Biotech, Uppsala, Sweden) equilibrated with buffer A. After extensive washing, bound material was eluted stepwise with buffer A containing successively 75, 130, 150, 175 and 250 mm NaCl. All these procedures were performed at 4 °C. The eluates were concentrated in Centricon 30 tubes (Amicon, Beverly, MA, USA) and analyzed by Western and Northern blotting using a human autoimmune serum containing strong reactivities to Ro and La proteins for detection of proteins, and antisense Y RNAs for detection of nucleic acids, respectively. Further separation was achieved by size-exclusion FPLC using either Superdex 200 (size separation range 10–600 kDa) or Superose 6 columns (range 50–5000 kDa) (both from Pharmacia Biotech). Columns were equilibrated in running buffer [20 mm Tris/HCl, pH 7.4, 140 mm NaCl, 1.5 mm MgCl2, 25% (v/v) glycerol]; the flow rate was 0.3 mL·min−1. The following protein standards (Sigma, St Louis, MO, USA) were used to calibrate the columns: thyroglobulin (669 kDa), apoferritin (443 kDa), β-amylase (200 kDa), alcohol dehydrogenase (150 kDa), BSA (66 kDa), and carbonic anhydrase (29 kDa).
Patients' sera and antibodies
In a previous study, sera from patients with systemic lupus erythematosus (SLE; diagnosed according to the established criteria of the American College of Rheumatology) or primary Sjögren's syndrome were investigated for the presence of autoantibodies to Ro and La proteins . Two sera (HM and LB) with pronounced reactivity to both Ro proteins and La were selected and subsequently used for immunoblot analysis of fractionated proteins. Some anti-Ro positive sera were found to contain antibodies to other Y RNA-associated proteins identified in the course of this investigation. Of these, one serum (BM) with pronounced reactivity to Ro60 but not to Ro52 was selected and used for immunoblot detection of Y RNA-associated proteins. The following monoclonal antibodies were used in immunoprecipitation studies: anti-La (SW5) , anti-Ro60 (2G10) , anti-Ro52 (2E7) . In some experiments, affinity-purified human antibodies were used. Purification from patients' sera using recombinant antigens immobilized on CNBr-activated Sepharose and control for monospecificity by immunoblotting, ELISA and immunoprecipitation was performed essentially as previously described [4,31]. Rabbit anti-(calreticulin fusion protein) serum was a gift from R. D. Sontheimer (University of Texas, Dallas, TX, USA) .
Protein samples (S100 extracts or aliquots of chromatographic fractions) were separated on SDS/12% polyacrylamide gels using an acrylamide to bisacrylamide ratio of 170 : 1, and subsequently blotted on to nitrocellulose as described . In a number of cases, poly(vinylidene difluoride) membrane (Immobilon-P; Millipore, Bedford, MA, USA) was employed when staining with Coomassie Blue R-250 was necessary. The nitrocellulose sheets were blocked for 1 h with 3% non-fat powdered milk in 10 mm Tris/HCl, pH 7.4, containing 140 mm NaCl and 0.1% Triton X-100 (Tris/NaCl/Triton) and subsequently incubated for 1 h with either human autoimmune sera diluted 1 : 50 or monoclonal or affinity-purified antibodies in Tris/NaCl/Triton/3% milk. After the nitrocellulose had been washed three times with Tris/NaCl/Triton, bound antibodies were detected with alkaline phosphatase conjugated goat anti-(human IgG) IgG or anti-(mouse IgG) IgG (Chemicon, Temecula, CA, USA). In some cases, immunoblots were developed by chemiluminescence detection using the ECL reagent from Amersham International (Amersham, Bucks, UK).
Monoclonal antibodies to Ro and La proteins were coupled to Protein A–Sepharose beads (Pharmacia Biotech) in high-salt buffer (25 mm Tris/borate, 3 m NaCl, pH 9) using 5 mg·mL−1 dimethyl pimelimidate as coupling agent ; affinity-purified human antibodies were coupled by the same procedure except that the NaCl concentration was 150 mm. For immunoprecipitation, 20 µL S100 cytoplasmic extract or aliquots of chromatographic fractions were diluted in 0.5 mL IPP-150 buffer (10 mm Tris/HCl, pH 7.5, 150 mm NaCl, 0.05% Nonidet P40) and incubated with 20 µL (packed bead volume) of Protein A–Sepharose coupled to the monoclonal antibodies of interest. After incubation for 1 h at room temperature, beads were washed three times with IPP-150, resuspended in 400 µL IPP-150 containing 0.5% SDS, and extracted with phenol/chloroform. RNA was precipitated with ethanol using 10 µg tRNA as carrier. Proteins were recovered from the phenol phase by adding 4 vol. acetone followed by precipitation for 1 h at −70 °C and centrifugation for 10 min at 15 000 g. For immunopurification on a larger scale, 0.5 mL affinity columns were prepared. Fractions or (reconstituted) S100 was applied using a peristaltic pump at low flow rate. After washing with at least 10 vol. IPP-150, bound complexes were eluted with 0.2 m glycine, pH 2.0, or 1.25 m NaSCN. Affinity columns were regenerated immediately afterwards with IPP-150.
RNA gel electrophoresis and Northern-blot hybridization
These procedures were carried out essentially as described . Briefly, RNAs were separated on 10% polyacrylamide/7 m urea gels, electroblotted on to nylon membranes (Zeta-probe; BioRad, Hercules, CA, USA) at 80 V in 0.5 × Tris/borate/EDTA buffer (45 mm Tris/borate, 1 mm EDTA, pH 8.2) and fixed by UV cross-linking. For detection of Y RNAs, digoxigenin or 32P-labeled antisense Y RNA transcripts were used. The probes were hybridized by overnight incubation at 65 °C in solution comprising 6 × NaCl/Cit (1 × NaCl/Cit is 0.15 m NaCl, 15 mm trisodium citrate, pH 7.0), 5 × Denhardt's solution, 0.1% SDS, containing 100 µg·mL−1 denatured salmon sperm DNA. Blots were washed twice with 2 × NaCl/Cit/0.2% SDS at room temperature and once with 0.1 × NaCl/Cit/0.1% SDS at 65 °C. Filters were developed by either autoradiography or using alkaline phosphatase-conjugated anti-digoxigenin Fab fragments in combination with a digoxigenin chemiluminescence detection kit (Boehringer-Mannheim, Mannheim, Germany).
In vitro transcription
In vitro transcription of antisense Y RNAs cloned into pTZ19 vectors by T7 RNA polymerase was carried out as follows: 1 µg EcoRI-linearized DNA template was incubated for 1 h at 37 °C in 40 mm Tris/HCl, pH 7.5, containing 10 mm MgCl2, 1 mm spermidine, 10 mm dithiothreitol, 1 mm ATP, CTP, GTP, 0.65 mm UTP, 0.35 mm digoxigenin-UTP (all nucleotides from Boehringer-Mannheim), 40 U RNasin (Promega, Madison, WI, USA), and 30 U T7 RNA polymerase (Promega). In vitro transcribed RNA was phenol extracted and recovered by ethanol precipitation in 0.6 m ammonium acetate using tRNA or glycogen as carrier. The precipitates were dissolved in hybridization buffer and used for detection of Y RNAs. For transcription of sense RNAs, DraI-linearized DNA templates from pUC18 vectors were excised from a 1% agarose gel and recovered using GenElute spin columns (Supelco, Bellefonte, PA, USA) followed by T7 run-off transcription as described above with the following differences: 0.5 mm UTP and 40 µCi [α-32P]UTP (NEN, Boston, MA, USA) for radiolabeled Y RNAs or 75 µm biotin 16-UTP (Boehringer-Mannheim) for production of biotinylated Y RNAs. Transcription of 3′-shortened forms of Y5 RNA followed linearization with AluI for Y5Δ-La RNA (deletion of 12 nucleotides from the 3′ end causing loss of the La-binding site) or with HinPI for Y5ΔLaδ60 RNA (deletion of 21 nucleotides from the 3′ end causing loss of the La-binding and Ro60-binding sites). The plasmids containing U5 RNA and 5S rRNA were gifts from A. Bindereif (Humboldt University, Berlin, Germany) and K. Nierhaus (Max Planck Institute for Molecular Genetics, Berlin, Germany), respectively.
In vitro reconstitution and purification of Y RNPs
To dissociate existing complexes, the salt concentration of the S100 extract was increased to 1 m KCl followed by 30 min incubation on ice as described by Granger et al. . Subsequently, 10 µg biotinylated Y RNA and 200 µg yeast tRNA (Boehringer-Mannheim) were added, KCl concentration was readjusted to 150 mm by diluting with reconstitution buffer [10 mm Tris/HCl, pH 7.9, 2 mm MgCl2, 1 mm dithiothreitol, 5% (v/v) glycerol], and the mixture was incubated for 20 min at 30 °C. Reconstituted biotinylated Y RNP complexes were purified by adding 20 µL ‘NeutrAvidin’ beads (Pierce, Rockford, IL, USA) and rotating for 1 h at 4 °C. Beads were then washed five times with 1 mL washing buffer (10 mm Tris/HCl, pH 7.9, 150 mm KCl, 0.1% Nonidet P40) and proteins were recovered with elution buffer (20 mm Tris/HCl, pH 7.9, 20 mm dithiothreitol, 2% SDS). Eluted proteins were heated at 65 °C for 5 min; 1 µL glycogen (20 mg·mL−1) and 4 vol. acetone were added. Samples were left at −70 °C for at least 30 min and subsequently centrifuged for 15 min at 15 000 g. Recovered proteins were dissolved immediately in SDS sample buffer, separated by SDS/PAGE, and analyzed by Coomassie Blue staining or immunoblotting.
Ro RNP complexes were reconstituted as described above except that 32P-labeled Y RNA was used. The reaction mixtures were transferred to a microtiter plate, put on ice and irradiated with a UV lamp (‘Stratalinker’; Stratagene) at 254 nm at a dose of 10 mJ·mm−2. Immunoprecipitations of cross-linked complexes were carried out by incubating Protein A–agarose beads with the appropriate affinity-purified antibodies by rotation for at least 1 h in Tris/NaCl/NP40 (10 mm Tris/HCl, pH 7.5, 500 mm NaCl, 0.05% Nonidet P40) followed by washing twice with Tris/NaCl/NP40 and twice with Tris/KCl/dithioerythritol/EDTA (10 mm Tris/HCl, pH 8.0, 100 mm KCl, 1 mm dithioerythritol, 1 mm EDTA, 0.05% Nonidet P40). After incubating the coated beads with the UV-cross-linked extracts in Tris/KCl/dithioerythritol/EDTA for 2 h at 4 °C, the beads were washed three times with Tris/KCl/dithioerythritol/EDTA and incubated with 2 μg RNase A for 45 min at 37 °C. Finally, radiolabeled proteins were separated on an SDS/10% polyacrylamide gel and visualized by autoradiography.
Y1 RNA-deletion mutants
The plasmid for Y1wt RNA transcription has been described . The mutant Y1 RNA constructs were derived from this Y1 RNA construct using PCR-based strategies with 5′ primers containing the T7 RNA polymerase promoter sequence. Using the appropriate primers, the Y1 RNA mutants Y1ΔS1L1, Y1ΔS1L1sty, and Y1ΔL1S2 were amplified and cloned into the EcoRI–HincII sites of pUC19. The mutants Y1ΔS3L3, Y1ΔS4L4, and Y1ΔL2b were created using the approach described . This procedure requires three oligonucleotide primers: two flanking primers complementary to regions upstream and downstream of the mutation site, respectively, and a mutagenic primer. As flanking primers, the 5′ and 3′ primers of Y1wt RNA were used. The resultant PCR products were digested with EcoRI and HindIII and cloned into the corresponding sites of pUC19. For transcription, the clones were linearized with DraI. All PCR products were sequenced to control for unwanted PCR mutations.
Fractionation of Ro RNPs by anion-exchange chromatography and gel filtration
To investigate Ro RNP composition, a HeLa S100 extract was subjected to anion-exchange chromatography using a protocol similar to that described by Boire et al. [6,10]. Resultant fractions were analyzed by Northern and Western blotting as appropriate (Fig. 1). The 50 mm NaCl flow-through and the 75 mm NaCl fractions contained La and Ro52 but neither Ro60 nor Y RNAs (Fig. 1, lanes 1 and 2). Fractions obtained at 130 mm NaCl were heterogeneous: peak fractions contained Ro60, Ro52, La, Y3, Y4 and Y5 RNA, with Y5 RNA predominating (lanes 3 and 4), whereas later fractions contained Ro60, La and Y5 RNA but only little or no Ro52 (lanes 5 and 6). The 150 mm NaCl eluate showed a similar composition, being relatively enriched for Ro60, La and Y5 RNA (lanes 7 and 8). Compared with the previous fractions, the 175 mm eluate had much less protein and RNA, containing relatively small amounts of Ro60 and Y RNA in addition to La; remarkably, this was the only fraction in which all four Y RNAs were present in comparable concentrations (lane 9). In contrast, fractions obtained at 250 mm NaCl had abundant Ro60 and Y1, Y3 and Y4 RNA, but lacked Y5 RNA (lane 10). In immunoprecipitation experiments, coprecipitation of Ro60, La and Y RNAs by both anti-Ro60 and anti-La IgGs was seen in all fractions investigated, demonstrating that the interaction of Ro60 and La with Y RNAs was maintained under the chromatographic conditions employed. In contrast, only a small portion of Ro52 (< 5%) was coprecipitated by the two antibodies, and the monoclonal anti-Ro52 IgG precipitated only Ro52 (not shown here; see Fig. 2).
To estimate the size of Ro RNPs, the major fractions from anion-exchange chromatography (obtained at 130 mm, 150 mm, 175 mm and 250 mm NaCl) were further separated by gel filtration. The 130-mm peak fraction was of particular interest because it was the only fraction that contained all known Ro RNP components including Ro60, Ro52, La, and Y RNAs, with Y5 RNA predominating and Y1 RNA being almost undetectable. In Superdex-200 gel filtration, a major portion of Ro52 and Ro60 copurified together with La and Y5 RNA in one fraction with an estimated molecular mass of 150–250 kDa (Fig. 2, panels I–III, lane 4). Interestingly, a part of Ro52 was eluted at ≈ 450–700 kDa (panel I, lanes 1 and 2); these fractions contained only very small amounts of Ro60 and La, with Y RNAs being undetectable. In immunoprecipitation experiments (panels II and III), anti-La efficiently coprecipitated Ro60 and Y5 RNA but only very little if any Ro52 (II A, lane 4); the monoclonal anti-Ro60 IgG gave a similar result, precipitating La, Y5 RNA, and also a small amount of a 52-kDa protein indistinguishable from Ro52 (II B, lanes 3 and 4). Moreover, this antibody precipitated in addition to Y5 RNA small amounts of Y3 and Y4 RNA from the 300–400-kDa fraction (III B, lane 3). In contrast, the monoclonal anti-Ro52 IgG precipitated only Ro52 (II C), being unable to recognize the particle-associated protein as described previously .
Gel filtration of the 150-mm NaCl fraction (containing Y5 RNA, Ro60 and La, but no Ro52) showed Y5 RNA to copurify with Ro60 and La in two fractions between 200 and 400 kDa, whereas Y1, Y3 and Y4 RNAs derived from the 175-mm and 250-mm fractions were eluted (together with Ro60 and La) between 300 and 550 kDa (not shown). Taken together, the gel-filtration experiments allowed further biochemical characterization of Ro (and La) particles as summarized in Table 1.
Table 1. Biochemical characterization of Ro and La particles based on anion-exchange chromatography, gel filtration and immunoprecipitation data. The concentration of NaCl in the anion-exchange elution buffer is shown. Some Ro RNPs (population 1 and 2) may contain Ro52, although immunoprecipitation did not provide an unambiguous result.
Molecular mass (kDa)
Ro60, La (Ro52)
Y 3, 4, 5
Ro60, La (Ro52)
Y1, 3, 4, 5
Y1, 3, 4
In vitro reconstitution of Y RNPs reveals novel Y RNA-binding proteins
The molecular masses of Ro RNPs or Y RNPs indicated the presence of additional proteins in these complexes. Although immunopurification of Ro RNPs (using anti-Ro60 or anti-La immunoaffinity columns) yielded some preliminary data, no clear result could be obtained, as the low abundance of Ro RNPs did not allow us to isolate sufficient amounts of such proteins in a reproducible manner. Therefore, as an alternative approach, binding of cytoplasmic proteins to in vitro transcribed biotinylated Y RNAs was investigated. Thus, S100 extracts were incubated with biotinylated Y1 RNA (or biotinylated U5 snRNA as a control) to allow reconstitution of (biotinylated) Y RNPs, which were subsequently purified via avidin–agarose and analyzed by SDS/PAGE. Coomassie Blue staining revealed the presence of several proteins with molecular masses between 40 and 80 kDa in reconstituted Y1 RNP complexes (Fig. 3). The strong 48-kDa and 60-kDa bands in the Y1 RNA lane corresponded to La and Ro60, respectively, as they were recognized by patients' sera and monoclonal antibodies (not shown here; see Fig. 5). In contrast, two proteins of similar molecular mass visible in the U5 snRNA lane were not recognized by anti-Ro/La positive sera or monoclonal antibodies and were therefore apparently different from these two proteins (see Fig. 5). Among the other proteins, five proteins with molecular masses of 53, 62, 65, 68 and 80 kDa appeared to be specifically bound by Y1 RNA, as they interacted only very weakly or not at all with U5 snRNA. The 68-kDa protein was almost as efficiently bound as Ro60, whereas the 80-kDa band was rather weak and sometimes difficult to recognize.
Autoantibodies to Y RNA-binding proteins in patients' sera
As it is well known that autoimmune sera often contain antibodies directed to antigens (proteins as well as nucleic acids), which are biochemically associated in multimeric complexes [36,37], we investigated the possibility that some of the novel Y RNA-binding proteins may be recognized by anti-Ro positive patients' sera. To this end, 50 sera from SLE patients, among them 30 anti-Ro positive sera, were probed by immunoblotting with Y1 RNA-binding proteins isolated by in vitro reconstitution as described above. These analyses revealed several additional reactivities in 14 of the anti-Ro60 positive sera which were mainly directed to proteins of molecular mass 53 kDa and 65 kDa, apparently corresponding to two of the proteins identified in the reconstitution assay. The 53-kDa protein was obviously distinct from Ro52 because it was neither recognized by (monoclonal or affinity-purified) anti-Ro52 IgG (not shown) nor by sera strongly reactive with native or recombinant Ro52 (Fig. 4, lanes 1 and 2). Four of the 14 sera contained reactivities directed to proteins of 62, 68 and 80 kDa (Fig. 4, lanes 3–6). The serum shown in lane 6 (patient BM) contained antibodies to all Y1 RNA-associated proteins identified in the reconstitution assay (including Ro60 and La) and was therefore used for immunodetection in subsequent studies on the interaction of the novel proteins with Y RNAs described in the following sections.
As the Ca2+-binding protein calreticulin, which has been reported to bind to Y RNA , migrates on SDS/PAGE between 60 and 65 kDa, we investigated the possibility that one of the proteins in this molecular-mass range may be related to calreticulin. However, none of the novel Y RNA-binding proteins was recognized by a rabbit anti-calreticulin IgG, although calreticulin was clearly detectable in the S100 extract (not shown).
The novel Y RNA-binding proteins bind preferentially to Y1 and Y3 RNA
The presence of autoantibodies in patients' sera allowed the use of immunoblot analysis in further investigations on the interaction of the novel proteins with Y RNAs. To this end, affinity-purified (biotinylated) Y1–Y5 RNPs were separated by SDS/PAGE and subsequently transferred to a poly(vinylidene difluoride) membrane which was first stained with Coomassie Blue (Fig. 5A) and then, after destaining, incubated with patient serum BM (Fig. 4, lane 6) containing autoantibodies to Ro60, La and all five novel Y RNA-binding proteins (Fig. 5B). In a second experiment, proteins were transferred to a nitrocellulose membrane avoiding the staining/destaining procedure (Fig. 5C). These experiments confirmed and extended the data described above (Fig. 3). Apart from Ro60 and La, four proteins with apparent molecular masses of 68, 65, 62 and 53 kDa were efficiently bound by Y1 and Y3 RNAs (Fig. 5A, lanes 1 and 2), whereas the 80-kDa protein appeared to associate only with Y1 RNA. With exception of the 53-kDa protein, these proteins bound only weakly to Y4 RNA (lane 3), and none of them appeared to bind specifically to Y5 RNA (lane 4) or two Y5 deletion mutants lacking the binding sites for La or for both La and Ro60 (lanes 5 and 6), nor did they show significant interaction with U5 snRNA (lane 7). The two proteins visible in control lanes (6–8) at the positions of La and Ro60 bound non-specifically to avidin–agarose (lane 8) and were different from La and Ro60, as revealed by immunostaining of the same membrane (Fig. 5B). Thus, pronounced staining of the Ro60 band can be seen in lanes 1–5, and staining of the La band in lanes 1–4, whereas no staining of these two bands is seen in lanes 6–8. With respect to the novel proteins, the 53-kDa band is visible in lanes 1–3, the 65-kDa band as well as the 68-kDa band in lanes 1 and 2, although staining intensity is weaker than in lane 2 (in other experiments the differences between Y1 and Y3 RNPs were less pronounced; see Fig. 5C); finally, the 80-kDa band can be seen in lane 1 and, although rather weakly, also in lane 2. The 62-kDa protein was not recognized in this experiment, whereas it can be clearly seen in Fig. 5C where a nitrocellulose blot was developed using the same serum as in Fig. 5B. Therefore, we assumed that the staining/destaining procedure of the poly(vinylidene difluoride) membrane had affected the epitope structure of the 62-kDa protein. Interestingly, a weak 58-kDa double band was revealed on the nitrocellulose membrane which is also not visible on the poly(vinylidene difluoride) membrane.
Immunoprecipitation of native and reconstituted Y RNPs
These data demonstrated the presence of proteins in S100 extracts that had the capacity to bind in vitro specifically to Y RNAs. To address whether these proteins resided on the same particles as Ro60 and/or La, native complexes were immunoprecipitated with monoclonal anti-Ro60 and anti-La antibodies, and the precipitates analyzed by immunoblotting using serum BM (Fig. 6A). As can be seen, the anti-Ro60 IgG precipitated only a 53-kDa protein in addition to Ro60 and La, whereas the anti-La IgG also precipitated the 65-kDa and 68-kDa proteins. To investigate whether these proteins make direct contact with Y RNAs, S100 extracts were incubated with 32P-labeled Y1 RNA, cross-linked by UV irradiation and subsequently immunoprecipitated with affinity-purified anti-La and anti-Ro60 IgGs (the monoclonal anti-Ro60 IgG did not recognize the cross-linked antigen). Precipitated complexes were treated with RNase and analyzed by SDS/PAGE and autoradiography. Remarkably, under these conditions both antibodies precipitated the same set of proteins (Fig. 6B): thus, apart from La and Ro60, three (labeled) proteins with apparent molecular masses of 65, 68 and 80 kDa could be clearly distinguished, which presumably corresponded to three proteins of comparable molecular mass identified in the reconstitution experiments. The diffuse band visible below the Ro60 band in lane 1 may correspond to the 58-kDa double band seen in Fig. 5C (see also next section; Fig. 7). Cross-linking of the 53-kDa and 62-kDa proteins could not be observed, which may indicate that these proteins were bound via protein–protein interactions or that the photocross-linking reaction had been inefficient. These results indicated that, under the experimental conditions used, at least some of the novel proteins can bind directly to Y1 RNA and may, at least partially, reside on the same particles as Ro60 and/or La. A summary of the data obtained for the novel Y RNA-binding proteins is given in Table 2. Remarkably, in both experiments, coprecipitation of the novel proteins by anti-La was stronger than by anti-Ro60 even though the anti-Ro60 IgG efficiently coprecipitated the La protein. Possible explanations include impaired recognition of Ro60 by antibodies when additional proteins are present in the complex, destabilization of complexes on antibody binding or preferential association of these proteins with Y RNPs that do not contain Ro60.
Table 2. In vitro binding of cytoplasmic proteins to Y RNAs. The data shown are primarily based on binding studies employing biotinylated or 32P-labeled Y RNAs in reconstitution experiments. For immunoprecipitation from in vitro UV-cross-linked or native S100 extracts monoclonal or affinity-purified antibodies to Ro60 and La were used. NO, not observed.
a Autoantibodies to these proteins were found in several sera from patients with SLE, most of which also contained anti-Ro60 IgGs. b Precipitation of 32P-labeled complexes after UV cross-linking. c Precipitation of native complexes. d The 60-kDa protein comigrated with Ro60; binding of this protein and the presence of autoantibodies against it was evident from the anti-60 kDa immune reactivity observed in the absence of Ro60 (Fig. 7, lanes 2 and 3).
The central loop of Y1 RNA is essential for binding of the novel Y RNA-associated proteins
To investigate the interaction of the newly identified proteins further with Y RNAs a series of (biotinylated) Y1 RNA-deletion mutants was constructed (Fig. 7A) and employed in reconstitution/immunoblotting assays as described above. A semiquantitative summary of the data is given in Table 3. These experiments identified a region within the central loop of Y1 RNA as being indispensable for interaction with the novel proteins (Fig. 7B): binding of all five proteins (and also of the proteins corresponding to the 58-kDa doublett) was greatly reduced or almost undetectable with mutant Y1ΔL2b lacking nucleotides 71–86 (loop 2b), whereas binding of Ro60 and La remained unaffected (lane 4). Removal of the stem 2–loop 1 region (mutant Y1ΔS2L1) reduced binding of the proteins, except for the 62-kDa (and also the 58-kDa) protein by ≈ 30–50% (lane 5). With mutant Y1ΔS3L3 lacking the stem 3–loop 3 region, binding of the 68-kDa and 65-kDa protein was about 60%, binding of the 53-kDa protein was about 40%, and binding of the 80-kDa protein was about 20% as compared with wild-type Y1 RNA, whereas binding of the 62-kDa protein did not decrease (lane 6). Finally, removal of either stem 1–loop 1 (mutant Y1ΔS1L1; lane 2) or stem 4–loop 4 (mutant Y1ΔS4L4; lane 7) had no strong effects on binding of any of the novel proteins. Mutant Y1ΔS1L1 lacking the binding site for Ro60 still bound a 60-kDa protein (lane 2), which, however, was distinct from Ro60 as it was not recognized by the monoclonal anti-Ro60 IgG (Fig. 7C); interestingly, the binding characteristics of the 60-kDa protein appeared to be similar to those of the 62-kDa protein, suggesting that the two proteins may be related. Surprisingly, an intact La-binding site seemed to be required for efficient binding of all novel proteins, as mutant Y1ΔS1L1sty lacking both the Ro60 and the La binding site bound the proteins only rather weakly (lane 3). A similar result was obtained with a mutant lacking the La-binding site only (not shown).
Table 3. Semiquantitative analysis of the interaction of the novel Y RNA-binding proteins with (biotinylated) Y1 RNA-deletion mutants. Binding was quantitated by densitometric evaluation of the immunoblots using the ImageQuant program. Data were obtained in three independent experiments. ++++, > 80–100% binding [100% is wild-type (WT)]; +++, > 60–80%; ++, > 40–60%; +, > 20–40%; –, < 10%.
Little is known about the detailed molecular structure of the small cytoplasmic Ro RNPs, and their function is still largely obscure. The presence of four different Y RNAs suggests the existence of at least four subpopulations of Ro RNPs or Y RNPs. There is now substantial evidence that most Y RNAs are complexed with La and Ro60 [4–6,10,11]. Thus, Ro60 and La may be considered the ‘core’ proteins of Y RNPs, in analogy with the Sm proteins of snRNPs, whereas additional proteins may be more or less stably bound to individual Y RNAs. The experiments described strongly indicate the existence of such proteins. Evidence for this is based on: (a) size of Ro particles as determined by gel filtration; (b) specific binding of several cytoplasmic proteins to Y1 and Y3 RNAs in reconstitution assays; (c) binding sites for these proteins on Y1 RNA (and presumably also Y3 RNA) distinct from the binding sites for Ro60; (d) coprecipitation of some of these proteins by anti-La and anti-Ro60 IgG; (e) the presence of autoantibodies to several of the Y RNA-binding proteins in anti-Ro positive autoimmune sera.
Association of a 52-kDa protein (Ro52) with Ro RNPs was suggested several years ago . Although no direct binding of this protein to Y RNAs could be demonstrated, data from reconstitution experiments suggested association of Ro52 by protein–protein interaction with Ro60 . Moreover, human anti-Ro52 autoantibodies and rabbit sera raised against synthetic peptides of Ro52 were shown to precipitate Y RNAs, although less efficiently than anti-Ro60 IgGs [7,8,31]. Furthermore, experiments in which immunization of animals with the Ro60 protein led to the development of an antibody response not only to Ro60 but also to Ro52 and La, presumably because of ‘intermolecular spreading’ of the anti-Ro60 immune response, may be considered additional although circumstantial evidence for biochemical association [38,39] (reviewed in ). However, a stable interaction of Ro52 with Ro RNPs has never been clearly demonstrated [5,6]. In our experiments, a portion of Ro52 consistently copurified with Ro60, La and Y RNAs in the 150–250-kDa region indicative of Y RNP association. Although monoclonal antibodies to Ro60 and La appeared to coprecipitate small amounts of a protein indistinguishable from Ro52, these experiments also did not provide conclusive evidence for stable association of Ro52 with Ro RNPs. Thus, the role of this protein in Ro RNP biochemistry is still unclear. Remarkably, most Ro52 was present in the cytoplasm apparently not as a free protein, as it was eluted on gel filtration with an apparent molecular mass between 150 and 700 kDa. The molecular composition of high-molecular-mass Ro52 complexes has not yet been addressed in detail. So far, our data do not indicate the presence of RNA in such complexes (data not shown). However, association of Ro52 with other proteins in vivo is not unlikely as this protein contains Zn finger domains, which are known to mediate protein–protein interactions (reviewed in ).
The molecular masses of Ro RNPs as estimated in the gel-filtration experiments (using two different types of columns) were comparable with or somewhat greater than those reported by others [5,10]. Thus, Y5 RNPs appeared to be of smaller size than Y1, Y3 and Y4 RNPs, which would be compatible with the data obtained in the reconstitution experiments in which no binding of additional proteins to Y5 RNA could be observed. Interestingly however, Y5 RNPs separated into four distinct subpopulations differing both by charge (i.e. binding to the anion exchanger) and size. The smallest Y5 RNPs may indeed contain only Ro60 and La and thus would have a calculated molecular mass of ≈ 150 kDa whereas larger Y5 RNPs may also contain other proteins (including Ro52) possibly bound via protein–protein interactions.
The novel Y RNA-associated proteins identified in the course of this study may indeed contribute to the relatively high molecular masses observed for Ro RNPs and further support the assumption that these particles may have a more complex structure than initially assumed. Thus, four proteins (68, 65, 62, and 58 kDa) were found to bind to Y1 and Y3 RNA, a 53-kDa protein bound to Y1, Y3 and Y4 RNA and a 80-kDa protein associated preferentially with Y1 RNA. The observed coprecipitation of the 80, 68, 65 and 53-kDa proteins by anti-La and anti-Ro60 IgGs suggests that these proteins reside on the same particles as La and Ro60, although efficient coprecipitation of the 80, 68 and 65-kDa proteins was seen only after UV cross-linking, indicating instable or transient interaction with Y RNAs. On the other hand, this experiment provided good evidence for a direct interaction of at least these three proteins with Y1 RNA. Remarkably, the proteins were more efficiently precipitated by anti-La IgG than by anti-Ro60 IgG. This may indicate that at least some of the novel proteins are associated with Y RNPs that contain La but no Ro60. An alternative explanation for the weaker coprecipitation by anti-Ro60 IgG would be destabilization of the complexes on antibody binding or that antibody binding was partially impaired by sterical hinderance or conformational changes induced by interaction(s) with other components of the complex. The existence of Y RNPs lacking Ro60 is also suggested by data from glycerol density gradient fractionation experiments in which a portion of Y1 and Y3 RNA was found to sediment free from Ro60 (but not from La) with apparent molecular masses of 300–400 kDa . Thus, we would have to add an as yet unknown number of Y RNP subpopulations to the various Ro60 RNP subpopulations characterized by us and others. On the basis of our data, we hypothesize a dynamic interaction between Ro60 RNPs and Y RNPs, leading to the formation of numerous complexes with different composition (Fig. 8).
Remarkably, the internal pyrimidine-rich central loop of Y1 (and presumably also Y3) RNA was found to be essential for binding of all novel proteins: thus, binding of the 80, 68, 65 and 53-kDa proteins (and also the 58-kDa protein) was almost completely abolished, and binding of the 62 and 60-kDa proteins was reduced by more than 50% when this region was deleted. Loop 2b, which is not required for binding of Ro60 (and La), has been suspected to be involved in interactions with other proteins or RNAs . Furthermore, because of its inaccessibility to chemical modifications and to enzymatic probing, this region may also be involved in tertiary interactions with other parts of the RNA . Thus, the significantly decreased binding of the 80, 68, 65, 53-kDa proteins (but not of the 62, 60 and 58-kDa proteins) to deletion mutants lacking stem 3–loop 3 or stem 2–loop 1, respectively, suggests that these regions together with loop 2b form the binding sites for these four proteins. This would also be compatible with the generally weaker binding capacity of Y3 RNA, which lacks stem 3–loop 3. In line with this interpretation, Y4 RNA lacking both the pyrimidine-rich section and stem 3–loop 3 bound only the 53-kDa protein, but less efficiently than Y1 or Y3 RNA, and Y5 RNA was obviously unable to associate with any of these proteins. Although these data demonstrate the importance of loop 2b for binding, they do not allow us strictly to conclude that there is direct interaction of this part of the RNA with the novel proteins. Removal of this section may gravely affect secondary and tertiary structure of the RNA, leading to disruption of protein-binding sites.
In summary, the 80, 68, 65 and 53-kDa proteins showed comparable binding properties, which somewhat differed from those observed for the 62-kDa and 60-kDa proteins, which were less dependent on loop 2b and the binding of which was not affected by removal of stem 3–loop 3. The finding that truncation of the La-binding site significantly decreased binding of all novel proteins was unexpected and indicates that La is required for optimum binding. As presumably all Y RNAs are associated with La in vivo, this protein may be required for correct RNA folding and consequently for efficient binding of other proteins (except Ro60). It has recently been proposed that La may function as an RNA chaperone in the biosynthesis of several RNAs and their assembly into RNPs [14,43,44]. This leads us to speculate that La may have some chaperone-like function in Ro RNP assembly in vivo or that the association of these additional proteins may occur in a co-ordinated fashion requiring La to be bound first.
The presence of autoantibodies to all six proteins in patients' sera was remarkable given that autoantibodies to nucleic acid-binding proteins, a hallmark of systemic autoimmune diseases, are often directed to biochemically associated antigens. This has led to the formulation of the ‘particle hypothesis’ of autoimmunity, suggesting that loss of tolerance to one component of a multimeric complex will eventually lead to loss of tolerance to other components because of the phenomenon of intermolecular epitope spreading (reviewed in [36,37,40]). Thus, antibodies to the whole set of Sm proteins and U1 snRNP-associated antigens are often found in the same serum, as are autoantibodies to Ro60, Ro52 and La. Therefore, the concurrence of autoantibodies to Y RNA-associated proteins in sera containing anti-Ro IgGs may be considered indirect evidence for a biochemical interaction. Finally, it should be mentioned that this study addressed only cytoplasmic Y RNPs. As there is active transport of Ro RNP components into and from the nucleus, the various proteins identified may also have a role in nuclear Ro RNP biochemistry (e.g. export), which has to be addressed in future studies.
To date the function of the mainly cytoplasmic Ro RNPs is not known and it is also not clear why there exist so many subpopulations. There is now sufficient evidence that cytoplasmic La may be involved in regulation of translation [16,18–21], and it is conceivable that the function of Ro RNPs/Y RNPs may be related to that. Work is in progress to purify the novel Y RNA-binding proteins in order to obtain data on their primary sequence. Identification of novel proteins associated (stably or transiently) with Y RNAs may not only provide new information on the structure of these complexes but also yield clues to their role in the biochemistry of the eukaryotic cell.
The authors wish to thank Elisabeth Höfler and José P. H. Thijssen for expert technical assistance, Ron Peek for providing affinity-purified antibodies, Richard D. Sontheimer for providing the anti-calreticulin IgG, and Albrecht Bindereif for providing a plasmid encoding U5 snRNA. We are grateful to Andrea Barta and Walther van Venrooij for their continuous interest in our experiments and for stimulating discussions, and Joseph S. Smolen for encouraging this work. The work was supported in part by a grant from the Austrian Ministry of Science and by the Netherlands Foundation for Chemical Research (SON) with financial aid from the Netherlands Organization for Scientific Research (NWO).