Transfection analysis of expression of mRNA isoforms encoding the nuclear autoantigen La/SS-B.

Transcription of the gene encoding for the nuclear autoantigen La resulted in La mRNA isoforms. A promoter switching combined with an alternative splicing pathway replaced the exon 1 with the exon 1'. The exon 1' contained GC-rich regions and an oligo(U) tail of 23 uridine residues. Moreover, it encoded for three open reading frames upstream of the La protein reading frame. Despite this unusual structure, when exon 1' La mRNAs were expressed in transfected cells, both exon 1 and 1' La mRNAs were translated to La protein, whereas the upstream open reading frames of the exon 1' were not translated. In addition to full-length exon 1' La mRNAs 5'-shortened exon 1' La mRNAs were detected. The exon 1' 5'-starts varied in dependence on the analyzed tissues. Like the full-length exon 1' La mRNA a 5'-shortened exon 1' construct starting downstream of the oligo(U) tail but upstream of the open reading frames 2 and 3 was also well translated when transfected in mouse cells. Thus all La mRNA forms represent functional La mRNAs.

One of the target antigens of sera suffered from autoimmune patients with rheumatoid diseases such as systemic lupus erythematosus or primary Sjögren's syndrome (pSS) 1 is the nuclear autoantigen La (SS-B) (1). The La protein was described to associate at least transiently with all primary RNA polymerase III transcripts including precursor molecules of ribosomal 5 S RNA, tRNAs, and some 4.5 S RNAs, as well as a portion of the uridine-rich small nuclear RNAs U1 and U6 (2)(3)(4)(5). Common to all primary RNA polymerase III transcripts is their 3Ј-terminal oligo(U) tail, which is transcribed during the transcription termination step. These oligo(U) tails were shown to be a binding site for the La protein (6). In addition to the oligouridylated RNA polymerase III transcripts, an association of the La protein with some nonoligouridylated RNAs has been reported especially for some viral RNAs including the leader RNAs of the vesicular stomatitis virus and rabies virus (7,8).
The La protein is assumed to be involved in transcription termination of RNA polymerase III and internal initiation of translation of at least the poliovirus mRNA (9 -12).
Most recently five La cDNAs were isolated when a cDNA library made from peripheral blood lymphocytes (PBLs) of a patient with pSS was screened with her own anti-La serum (13). In two of these five La cDNAs the exon 1 was replaced with an alternative 5Ј-end. Genomic analysis revealed that these La cDNAs represented alternatively spliced transcripts of the La gene. An additional promoter site was identified in the intron between exons 1 and 2, which served as initiation site for transcription of the alternative exon 1Ј.
The exon 1Ј La mRNA form had an unusual 5Ј terminus. It contained GC-rich regions and an oligo(U) tail of 23 uridine residues and encoded for three upstream open reading frames (ORF1 to 3). The ORF1 encoded for a putative peptide of 5.4 kDa. It was interrupted from the La protein reading frame by two stop codons. The ORF2 and ORF3 were not in the reading frame of the La protein. Qualitative and quantitative analysis of expression of the exon 1 and exon 1Ј La mRNAs showed that both La mRNA forms represented finally processed abundant cytoplasmic mRNAs. Exon 1 to exon 1Ј La mRNAs were expressed at ratios between 1:1 and 1:5 (15,16).
Due to the unusual structure of the exon 1Ј La mRNA it still remained unclear whether the exon 1Ј La mRNA is a translatable mRNA and if so, which of the reading frames is used for translation. This was of special interest because one of the exon 1Ј La cDNAs contained a frameshift mutation in a recently detected hot spot region in the exon 7 of the La gene (14). The frameshift mutation caused a premature stop codon in the La protein reading frame. Thus the mutant exon 1Ј La mRNA could encode for a C-terminally truncated mutant La peptide of 29 kDa.
Here we present evidence that exon 1Ј La mRNAs can be translated to La protein, whereas the upstream ORFs are not used for translation.
Anti-La Antibodies and Sera-The anti-La mAb SW5, which was found to be directed to the N-terminal domain of La protein (17), was originally described by Smith et al. (18). The hybridoma supernatant was kindly provided by Prof. vanVenrooij (University of Nijmegen, The Netherlands). As anti-La serum we used the serum of the patient (Ma) (12).
Polymerase Chain Reaction-PCR was performed using a TC9600 Cycler (Perkin-Elmer, Ü berlingen, Germany). The 50-l assay in 1 ϫ Taq buffer contained 2 units Taq polymerase, 1.5 mM MgCl 2 , 5% (v/v) Me 2 So, 200 M dNTP, 20 pmol of each primer, and 1 ng of DNA. Cycling was started by heating for 3 min to 95°C. 40 cycles followed, each consisting of 15 s at 94°C, 15 s at 57°C, and 1 min at 72°C. Then the temperature was held for 10 min at 72°C and cooled down to 4°C.
The PCR products were further analyzed by agarose gel electrophoresis. PCR fragments were visualized by ethidium bromide staining. Isolated PCR fragments were either directly sequenced or sequenced after subcloning into pGEM-T (13).
DNA Sequence Analysis-DNA was prepared using the QIAprepspin kit and concentrated by ethanol precipitation to a final concentration of about 1 g/l in Tris-EDTA buffer (10 mM Tris, 1 mM EDTA, pH 8.3). DNA sequencing was performed as described previously (13).
5Ј-Rapid Amplification of cDNA Ends Analysis-For an analysis of the 5Ј-ends of the La mRNAs we chose the 5Ј-rapid amplification of cDNA ends system (5Ј-RACE) supplied by Life Technologies, Inc. It includes a first strand cDNA synthesis and a PCR amplification step. During reverse transcription 40 units of RNase inhibitor were added to each 1-g RNA sample. The RNA samples were isolated from either (adult) liver (L) of the tumor patient, human fetal spleen (20th week), PBLs of the patient with pSS, and PBLs of a (adult) control person, and mouse LTA cells, which were untransfected or transfected with the human either La gene or La exon 1 La cDNA (19). Although the liver tissue of the tumor patient did not obviously show metastasis, the tissue displayed an abundant expression of the c-myc oncogene (data not shown).
The RNA was isolated as described earlier (13). The reactions and the following 5Ј-RACE steps were performed according to the instructions of the supplier. For the first strand synthesis the primer P1 locating within the exon 4 was used (P1, CACTGATTTCCATGAGT-TCTGCCTTGG). This primer was extracted with phenol/chloroform and diluted in diethyl pyrocarbonate-treated bidistilled water. The first amplification was performed using the Anchor primer of the supplier as upstream primer and the primer P2 locating in the exon 3 (P2, TGTC-CCGTGGCAAATTGAAGTCGCC) as downstream primer. The cDNA synthesis and the first amplification step is schematically summarized in Fig. 2 The cDNAs were further amplified using nested primer pairs. The nested primer pairs consisted of the common downstream exon 2 primer P3 (P3, GATGACAGATTTTGGCCTCCAG) in combination with either the exon 1 specific upstream primer P4 (P4, GAGTCGTTGCTGTTGCT-GTTTG) or the exon 1Ј specific upstream primer P5 (P5, TTCTAGTCT-CACCGAAGGCTTGTG) (see also Fig. 2 (A1). Alternatively the cDNAs were amplified using a combination of the universal amplification primer (UAP) of the supplier with either the exon 1Ј specific downstream primer P6 (P6, CTGAAACCTGATGTGAGCGATG) or the exon 1 specific downstream primer P7 (P7, CCACAGGCTCACAAACAGCAAC) as schematically summarized in Figs. 2 (B1) and 3 (S).
Preparation of Exon 1 and Exon 1Ј Constructs for Transfection Studies-All exon 1 and 1Ј constructs were cloned in the two transfection vectors pCI-neo and pCI according to the following strategy. Cloning started from the La cDNA La23 in pBluescript SK(Ϫ). The La insert in La23 started at the 5Ј-site with an oligo(dT) tail and represented a 5Ј-shortened exon 1Ј La mRNA derivative. Therefore, in a first step an exon 1Ј full-length La cDNA had to be constructed. In parallel an exon 1 full-length La cDNA was constructed. Because La23 started at the 5Ј terminus with 51 dT residues, this irregular oligo(dT) tail was shortened to 5 dT residues to obtain the 5Ј-shortened exon 1Ј construct. During cloning we learned that La23 contained a frameshift mutation within the coding region (Ref. 14; see also Fig. 6,ORF1 La(N)). Because all constructs were derivatives of La23, in a further cloning step the mutated reading frame was replaced with the correct reading frame, which was obtained from the La cDNA La19.
For the first step of the cloning procedure 5Ј-exon 1Ј and 5Ј-exon 1 fragments were required and prepared by PCR using the proofreading Pfu polymerase. In the case of the exon 1Ј fragment, PCR was performed using DNA of a genomic subclone as substrate. The subclone was prepared from the charon phage Lambda 2.1 (19), which was a gift by Prof. J. D. Keene (Duke University, Durham, NC). Restriction of Lambda 2.1 DNA with EcoRI resulted in a 4.4-and a 4.6-kilobase fragment. The 4.6-kilobase EcoRI fragment was isolated and subcloned into pBluescript SK(Ϫ). This subclone contained besides the exons 1 and 2 the intron between exons 1 and 2 including the exon 1Ј. PCR was performed using as upstream primer P8 (P8, CGCTTTACTAGTGCGC-GACTGCGCGTTTCC; the artificial SpeI site is underlined) and as downstream primer P6. The resulting exon 1Ј fragment started at the predicted 5Ј-start of exon 1Ј (13) and ended downstream of an EcoO109 site, which located downstream of the oligo(dT) stretch in the La sequence. The exon 1Ј fragments were cleaved with SpeI and EcoO109 and cloned into the respective sites of La23. For this purpose La23 was linearized with SpeI, and the isolated DNA was partially digested with EcoO109. Despite using Pfu polymerase a series of clones had to be characterized, because the vaste majority of clones had an incorrect length of dT residues in the oligo(dT) tail ranging from 14 to 28 residues. Finally a single clone was isolated with the correct length of 23 dT residues. The 5Ј-exon 1 fragment was obtained as follows. As substrate we used the La cDNA M13-3, which was originally described by Chan et al. (20) and which was a gift of Dr. E. K. L. Chan (Scripps Clinic, La Jolla, CA). Upstream exon 1 primer served the primer P9 (P9, CGCTT-TACTAGTCGGTCCCCATCTTCTTGG; the artificial SpeI site is underlined). The downstream primer P1 located in the exon 4 downstream of the KpnI site in the La sequence. The exon 1 PCR fragments were cleaved with SpeI and KpnI and cloned into the respective sites of La23. For this purpose La23 was linearized with SpeI, and the isolated linearized DNA was partially digested with KpnI. The exon 1 and 1Ј La inserts were isolated from the pBluescript SK(Ϫ) constructs by cleavage with SpeI and XhoI and cloned into pCI-neo and pCI, which were restricted with NheI and SalI. The exon 1Ј construct in pCI was termed as ORF1 La(N) (see also Fig. 6, I).
In the case of the 5Ј-shortened exon 1Ј construct, a PCR fragment was prepared using as upstream primer P10 (P10, CGCTTTACTAGTTTT-TACCTCCACCGCCTTC; the artificial SpeI site is underlined) in combination with P1 as downstream primer and La23 as substrate. The resulting fragment was cleaved with SpeI and KpnI and cloned in the respective sites of La23.
Finally the reading frame of the exon 1 and 1Ј constructs was corrected as follows. La19 cDNA containing the correct La coding sequence was restricted with BstEII, which cleaved in the exon 9 of the La sequence, and BglII, which cleaved in the exon 5 of the La sequence. The exon 1 and 1Ј pCI-neo and pCI constructs were linearized with BstEII and after isolation of the linearized DNA partially digested with BglII. Then the BglII-BstEII fragment of La19 was cloned in the respective sites of the exon 1 and 1Ј pCI-neo or pCI construct. The final constructs were sequenced.
Transfection, Cell Culture, and Staining Procedures-The human Raji and XPTA cell lines were grown in RPMI 1640 medium containing 10% FCS in a humidified CO 2 atmosphere. Mouse LTA and NIH 3T3 cells and human HeLa cells were grown in DMEM containing 10% FCS in a humidified CO 2 atmosphere either in culture flasks for preparation of extracts or on coverslips for epifluorescence microscopy. The mouse cells were transfected transiently according to the following protocol. Transfection was performed in 6-well tissue culture plates (35 mm) containing coverslips. Cells were grown to confluency of 70 -80% in 4 ml of DMEM containing 10% FCS. Prior to transfection the serum/medium was removed, and the cells were washed for 30 min with 2 ml of DMEM without FCS and antibiotics. In parallel 1.5 g of plasmid DNA (see below) was dissolved in 100 l of Opti-MEM medium and combined with 100 l of DMEM (without antibiotics and FCS) containing 6 l of LipofectAMINE. After removal of the 2 ml of DMEM, 0.8 ml of DMEM lacking FCS and antibiotics were added to the cells, and the DNA mixture followed. After an 5-h incubation 1 ml of DMEM containing 20% FCS and antibiotics was added. 20 h after the beginning of transfection the medium was replaced by 2 ml of DMEM containing 10% FCS and antibiotics. 44 h after transfection the cells were either harvested for preparation of total extracts or fixed with methanol/EGTA for immunofluorescence microscopy. In general we observed that even after optimization of the transfection conditions the expression level of the pCI-neo constructs was only 25% compared with the pCI constructs.
In addition to transiently transfected cells, permanently transfected LTA cell lines were used. These lines were kindly provided by Dr. K. Keech of the group of Prof. J. McCluskey and Prof. T. Gordon (Flinders Medical Center, Bedfort, South Australia). The cells were transfected with either the human La gene or the human exon 1 La cDNA (19). The permanently transfected LTA cells were grown in the presence of 0.02% geneticin.
For immunofluorescence microscopy the cells were fixed with meth-anol containing 0.02% EGTA at Ϫ20°C for 1 h. Indirect immunofluorescence of cells with the anti-La mAb SW5 was performed by incubating the fixed cells, which had been rehydrated for 5 min with PBS, with cell culture supernatant of hybridomas secreting the anti-La mAb SW5 for at least 15 min. The cells were washed with PBS (5 min), and the bound anti-La mAb was detected using Cy3-conjugated anti-mouse antibody developed in goat. The incubation time for the secondary antibody was 15 min. The staining occurred at room temperature and the unbound secondary antibodies were removed by washing with PBS (twice, 5 min). The stained cells were mounted using PBS/glycerol (1:1 v/v). Confocal laser scanning microscopy (cLSM) was performed using a Zeiss LSM 10. The stained specimens were cut automatically into horizontal sections (512 ϫ 512 pixels/8 bit, objective lenses Plan-Neofluar 40 ϫ/1.3 oil). Evaluation of the stored stacks of the horizontal optical sections was performed with the LSM 10 image processing unit. Preparation of Cell Extracts-Total extracts of transiently transfected cell lines were prepared by incubation with 350 l of SDS-PAGE sample solution (95°C). Total extracts from XPTA cells or permanently transfected LTA cells were harvested by adding 100 l of hot cell lysis buffer (100 mM Na 2 HPO 4 , pH 8.3, 200 mM dithiothreitol, 1% SDS, 10% glycerol (v/v)) per cm 2 . Raji cells were harvested by centrifugation (250 ϫ g, 10 min). The lysed cells were heated for 5 min to 95°C and centrifuged at 14,000 ϫ g for 5 min at 4°C. 5-l aliquots were mixed with SDS-PAGE sample solution and used for SDS-PAGE.
SDS-Polyacrylamide Gel Electrophoresis and Immunoblotting-SDS-PAGE and immunoblotting was performed as described (22,23). After blocking and washing the blots were incubated with cell culture supernatant of hybridoma cells secreting the anti-La mAb SW5. Formed immune complexes were visualized using the enhanced chemiluminescence-Western blotting detection reagents. Then the anti-La SW5 immune complexes were eluted, and the blot was processed a second time using either the patient's anti-La antibody or the anti-ORF1 rabbit serum. The formed immune complexes were detected using anti-human or anti-rabbit antibodies conjugated with alkaline phosphatase and 5-bromo-4-chloro-3-indolyl phosphate/4-nitro blue tetrazolium chloride as substrate.
Preparation of ORF1 Constructs-To obtain antibodies against the ORF1, a fusion protein consisting of glutathione S-transferase (GST) and a portion of ORF1 was constructed. For this purpose an exon 1Ј fragment spanning nucleotides 171-260 (6) was amplified by PCR using the upstream primer P11 (P11, CCCGGGATCCATGCGAGATCCTGG; the artificial BamHI site is underlined; A indicates a point mutation introduced in the native BamHI site) and the downstream primer P12 (P12, AGGATCCAAGACGCAATGGGGATGAG; the artificial BamHI site is underlined) subcloned into pGEM-T and after restriction with BamHI inserted into the BamHI site of pGEX-2T. To proof the specificity of the anti-ORF1 serum and to look for or to rule out an expression of the ORF1 further constructs were prepared. In a first step the complete exon 1Ј was cloned in the transfection vector pCI. Because the translation initiation site of the ORF1 in the exon 1Ј is not optimal according to Kozak (24), a further construct was prepared in which the translation initiation site was optimized according to the Kozak rules. This construct was termed ORF1 Kozak construct. Finally, the ORF1 was cloned downstream of a 3Ј-terminally truncated exon 1Ј-La mRNA. This clone was termed as ORF1 La(N)-ORF1 (see also Fig. 6, I).
The ORF1 in the pCI transfection vector was obtained as follows. The pCI vector containing the full-length exon 1Ј La cDNA was cut with SmaI. Then the cleaved DNA was partially digested with BalI and religated. The resulting clone contained the 5Ј leader sequence of the exon 1Ј La mRNA (nucleotides 1-137), the complete ORF1 (starting at nucleotide 138 and ending at nucleotide 276 including the oligo(dT) tail). It ended at nucleotide 293 in the exon 1Ј.
The ORF1 Kozak construct was prepared as follows. The full-length exon 1Ј La cDNA served as substrate for PCR using as upstream primer the primer P13 (P13, TGCTAGCGCCACCATGGGGGATCCTGGGG-TTC; the artificial NheI site is underlined, the mutated optimized translational initiation site according to Kozak is given in italics). The primer P14 (P14, TCTAGACGCTATGGGGATGAGG; the artificial XbaI site is underlined) served the downstream primer. The resulting fragment was subcloned in pGEM-T, isolated by cleavage with NheI and XbaI, and cloned into the respective sites of pCI. The resulting ORF1 construct contained the same portion of the ORF1, which was also cloned in the GST-ORF1 construct.
The ORF1 La(N)-ORF1 construct was prepared as follows. The exon 1Ј full-length construct in pCI was cleaved with XbaI and dephosphorylated. The resulting deletion La mutant encoded for amino acids 1-228 of La protein. The ORF1 insert was amplified from the above mentioned ORF1 subcloned in pGEM-T using as upstream primer P15 (P15, TGTAAAACGACGGCCAGTG) and as downstream primer P16 (P16, ATGAAATCACTAGTGATTGCTAGCGCCACC; the SpeI site is underlined). The resulting PCR fragment was cleaved with SpeI and XbaI and cloned in the XbaI site of the N-terminal truncated La deletion construct.

RESULTS
In a recent study Keech et al. (21) describe a mouse LTA cell line transfected with the human La gene. This cell line expressed human La protein in addition to the endogenous mouse La protein. As shown in Fig. 1 a differentiation between the human La protein and the endogenous mouse La protein in this mouse LTA cell line was possible by two techniques, including immunofluorescence microscopy (Fig. 1A) and SDS-PAGE/immunoblotting (Fig. 1B). As shown in Fig. 1A (a) LTA cells transfected with the human La gene were stained with the anti-La mAb SW5, whereas untransfected cells were not stained (Fig. 1A, b and c). After SDS-PAGE/immunoblotting, the anti-La mAb SW5 reacted with a single protein band according to a molecular mass of 50 kDa from the extract of the LTA cell line transfected with the human La gene (Fig. 1B, lane  2). The same band was also recognized by the patient's anti-La antibody (Fig. 1B, lane 4). The anti-La mAb SW5 did not react with the total extract of the untransfected LTA cells (Fig. 1B,  lane 1). In contrast, the patient's antibody reacted with a further protein band according to a molecular mass of 45 kDa in the extract of both the transfected (Fig. 1B, lane 4) and the untransfected cells (Fig. 1B, lane 3). These data allowed the following conclusions: (i) the human and the mouse La protein can be separated by SDS-PAGE, (ii) the protein band with a molecular mass of about 50 kDa represented the human La protein, (iii) the protein band with a molecular mass of 45 kDa  1 and 3) and an extract of mouse LTA cells permanently transfected with the human La gene (lanes 2 and 4) was analyzed by SDS-PAGE/immunoblotting with the anti-La mAb SW5 (lanes 1 and 2). After detection the anti-La mAb SW5 immune complexes were eluted from the blot, and the same blot was processed a second time with the monospecific patient's (La(P)) anti-La antibody (lanes 3 and 4) represented the endogenous mouse La protein, (iv) the patient's anti-La antibody reacted with both the human and the mouse La protein, and (v) the anti-La mAb SW5 reacted only with the human but not with the endogenous mouse La protein.
Consequently, transfection of mouse cells with a human exon 1Ј La construct should allow the decision if the exon 1Ј La mRNA is a translatable mRNA. Moreover, a mouse cell line transfected with an exon 1Ј La construct but also the LTA cell line transfected with the human La gene should be useful to look for translation products of the upstream ORFs of the exon 1Ј La mRNA.
One prerequisite to use the LTA cell line transfected with the human La gene for such a purpose was that the exon 1Ј La mRNA was made and expressed similarly to human cells. Therefore, in a first step we analyzed if the mouse LTA cells allowed the expression of the exon 1Ј La mRNA type from the human La gene.
Detection of Exon 1Ј La mRNAs in Mouse LTA Cells Transfected with the Human La Gene-For this purpose 5Ј-RACE experiments were performed as schematically summarized in Fig. 2 (S). The PCR products obtained after the first amplification step were further amplified using a primer combination consisting either of the common exon 2 downstream primer together with an exon 1 (Fig. 2, A1, PCR 2a) or exon 1Ј (Fig. 2, A1, PCR 2b) specific upstream primer or of the common UAP upstream primer in combination with an exon 1Ј specific downstream primer (Fig. 2, B1, and also Fig. 3, S, B).
As shown in Fig. 2 (A2, lanes 1), both the exon 1Ј and exon 1 human La mRNA forms could be detected in the mouse LTA cells transfected with the human La gene, whereas they were not detectable in the untransfected cell line (Fig. 2, A2, lanes 3). The human exon 1Ј La mRNAs were also not detectable in a LTA cell line permanently transfected with the human exon 1 La cDNA (Fig. 2, A2, exon 1Ј, lane 2), whereas the human exon 1 La mRNA was detectable in this LTA cell line (Fig. 2, A2, exon  1, lane 2). Finally, when the exon 1 and 1Ј bands were excised, subcloned, and sequenced, the PCR products could be unequivocally characterized as human exon 1 or exon 1Ј products.
Identification of Varying 5Ј-Starts of Exon 1Ј La mRNAs-Already during determination of the 5Ј-start of the exon 1Ј La mRNA type a series of 5Ј-shortened exon 1Ј La  3. Identification of varying 5-starts of the exon 1 La mRNA isoform. S, 5Ј-RACE products were prepared from different mRNA preparations according to the procedure that is schematically summarized in Fig. 2 (S). Then the PCR products were further amplified using the UAP as upstream primer in combination with either the exon 1 (P7) specific downstream primer ([A]) or the exon 1Ј (P6) specific downstream primer ([B]). A and B, the PCR products were prepared from different human tissues including fetal spleen (FS, lanes a), peripheral blood lymphocytes of the patient with pSS (P p , lanes b) or a healthy donor (P h , lanes c), and human liver (L, lanes d). All the exon 1 bands (A) and the four regions of the separated exon 1Ј products as indicated in B were excised, eluted, subcloned, and sequenced. The 5Ј-start of the exon 1 (A1) and exon 1Ј (B1-B4) fragments were schematically summarized. The exon 1Ј fragments started upstream of the ORF1 (B1), upstream of the ORF2 but downstream of the ORF1 (B2), upstream of the ORF3 but downstream of the ORF2 (B3), or downstream of the ORF3 (B4). The ORF1 was in the La protein reading frame but interrupted by two stop codons (*). ORF2 was in the ϩ1 reading frame, and ORF3 was in the Ϫ1 reading frame. Due to the upstream ORFs cDNAs were isolated. The decision of the 5Ј-start of the exon 1Ј La mRNA was made on the longest 5Ј-exon 1Ј La cDNA fragment obtained by the 5Ј-RACE technique. However, when the 5Ј-RACE products obtained for mRNAs isolated from the LTA cells transfected with the human La gene were separated on a NuSieve-agarose gel (Fig. 2, B2, lane 1), several bands were obtained. Therefore, further 5Ј-RACE studies were performed, and the PCR products were characterized. For this purpose exon 1 and exon 1Ј 5Ј-RACE products were prepared in parallel from different mRNA preparations according to the procedure that is schematically summarized in Fig. 2 (S). After the first amplification (Fig. 2, S, PCR(1)) the PCR products were further amplified as schematically summarized in Fig. 3 (S). The second amplification was performed using the common UAP primer in combination with either an exon 1 (Fig. 3, S, [A]) or an exon 1Ј (Fig. 3, S, [B]) specific downstream primer. The mRNA preparations used for the 5Ј-RACE experiments were obtained from different human tissues including liver (Fig. 3, A  and B, lanes L), PBLs of a patient with pSS (Fig. 3, A and B, lanes P p ) and a healthy donor (Fig. 3, A and B, lanes P h ), and fetal spleen (Fig. 3, A and B, lanes FS).
As shown in Fig. 3 (A, lanes a-d), the exon 1 5Ј-RACE products gave a single band when separated on an agarose gel. The bands obtained from the mRNAs of the different tissues were excised and subcloned, and a representative amount of clones were sequenced. All exon 1 inserts started around the predicted 5Ј-start of the exon 1 La mRNA as schematically summarized in Fig. 3 (A1). No difference was observed between the different mRNA preparations. In contrast, when the exon 1Ј 5Ј-RACE products were separated on an agarose gel (Fig. 3B,  lanes a-d) up to at least five bands were obtained. Moreover, the exon 1Ј 5Ј-RACE patterns differed between the liver tissue (Fig. 3B, lane d) and the patterns for the PBL preparations (Fig. 3B, lanes b and c) and the fetal spleen (Fig. 3B, lane a). Four regions of the agarose gel as indicated by the bars in Fig.  3 (B, 1-4) were excised. The extracted DNAs were subcloned, and representative amounts of clones were sequenced. The results of these experiments were schematically summarized in Fig. 3 (B1, B2, B3, and B4). The characterized exon 1Ј La cDNAs started either upstream of the ORF1 ATG at the predicted 5Ј-start of the exon 1Ј (Fig. 3, B1), around the oligo(dT) stretch (Fig. 3, B2), downstream of the ORF2 ATG but upstream of the ORF3 ATG (Fig. 3, B3), or downstream of the ORF3 ATG (Fig. 3, B4).
In consequence, these data increased the complexity, and we had to look for translational products not only of the exon 1Ј full-length La mRNA but also of the 5Ј-shortened exon 1Ј La mRNAs.
Translation of Exon 1Ј La mRNAs in Mouse LTA Cells-In a first step, full-length exon 1, full-length exon 1Ј, and a 5Јshortened exon 1Ј construct starting downstream of the oligo(U) tail but upstream of the ORF2 AUG were prepared (see "Experimental Procedures") and mouse LTA cells were transiently transfected. The transfected cell lines were analyzed for expression of human La protein by cLSM (Fig. 4) and SDS-PAGE/immunoblotting (Fig. 5). As shown in Fig. 4 (a-c) the anti-La mAb SW5 gave the same nuclear staining pattern, which was also obtained for the LTA cell line permanently transfected with the human La gene (see Fig. 1A, a), on the LTA cells being transiently transfected with either the fulllength human exon 1, full-length exon 1Ј, or the 5Ј-shortened exon 1Ј La pCI-neo construct. No staining was found on cells transfected with a pCI-neo control lacking a La specific insert (Fig. 4d). The staining intensities of the three constructs slightly differed. It appeared highest in the case of the 5Јshortened exon 1Ј derivative, followed by the exon 1 and the exon 1Ј full-length construct. Comparable results were obtained when instead of the pCI-neo construct the corresponding pCI construct was used. In all cases the pCI construct gave a higher protein expression than the pCI-neo construct.
The results obtained with cLSM were confirmed by SDS-PAGE/immunoblotting. As shown in Fig. 5 (lane 1) the patient anti-La antibody reacted only with the endogenous mouse La protein in LTA cells transiently transfected with the pCI-neo control construct lacking a La insert. The endogenous mouse La protein was also detected in all the other extracts obtained from the transiently transfected LTA cells (Fig. 5, lanes 2-4). In addition to the endogenous mouse La protein, the patient's anti-La antibody also detected the human La protein in the mouse LTA cells transfected with full-length exon 1 (Fig. 5,  lane 2), full-length exon 1Ј (Fig. 5, lane 3), and the 5Ј-shortened exon 1Ј (Fig. 5, lane 4) pCI-neo construct. There was no obvious difference between the molecular mass of La protein translated from the exon 1 or the exon 1Ј La mRNAs.
From the epifluorescence and the SDS-PAGE/immunoblotting results, we concluded that all La mRNA forms can be translated to the same La protein. The shortening of the 5Ј-end of the full-length exon 1Ј La mRNA might increase the translational efficiency. None of the upstream ORFs should be fused to La protein when translated from an exon 1Ј La mRNA.
Analysis of an Expression of the ORF1 Reading Frame-Finally, we analyzed whether the ORF1 of the exon 1Ј is translated. For this purpose a recombinant GST-ORF1 fusion protein was prepared and used for immunization of a rabbit. After adsorption of the rabbit serum to GST, the anti-serum did no more react with the GST-carrier protein but still reacted with the recombinant GST-ORF1 fusion protein (data not shown). No reactivity was detected in extracts of several human cell lines including HeLa (see below), Raji, or XPTA cells and mouse cells transfected with the human La gene or full-length exon 1Ј La mRNA (data not shown).
To proof the specificity of the ORF1 serum and to look for a translational product of the ORF1 further exon 1Ј La mRNA, derivatives were constructed as schematically summarized in  (14). Because the epitope recognized by the anti-La mAb SW5 is located upstream of the frameshift mutation, it still reacted with the truncated La protein (see below). The remaining N-terminal La coding region contained a XbaI site upstream of the premature stop codon. This site was used to fuse a second ORF1 sequence to the C terminus of the N-terminal La peptide. Thus the resulting construct ORF1 La(N)-ORF1 (Fig. 6, I) contained a first ORF1 locating upstream of but interrupted by two stop codons from the La coding region and a second ORF1 fused to the C terminus of the truncated N-terminal La peptide.
As shown in Fig. 6 (II, A, lane b) only the ORF1 La(N)-ORF1 La extract reacted with the anti-ORF1 serum. No reactivity was found for the ORF1 La(N) construct irrespective of whether the construct was transfected into human HeLa (Fig.  6, II, A, lane d) or mouse 3T3 (Fig. 6, II, A, lane f). Furthermore, no reactivity was found in the control HeLa and 3T3 cell extracts. The anti-ORF1 immune complexes were eluted from the blot. Then the same blot was processed a second time using the anti-La mAb SW5. As shown in Fig. 6 (II, B, lanes b, d, and h) the anti-La mAb SW5 reacted with the endogenous human La protein in all HeLa cell extracts. In the case of the transfected cell lines the anti-La mAb SW5 detected equal amounts of either the N-terminal truncated La peptide lacking the ORF1 (Fig. 6, II, B, lanes d and f) or the N-terminal La-ORF1 fusion peptide (Fig. 6, II, B, lane b). As expected the mobility of the N-terminal La-ORF1 fusion peptide was less than the mobility of the N-terminal truncated La peptide lacking the ORF1.
These data show that the ORF1 fused to the C terminus of the N-terminal La peptide was efficiently made and detectable by the anti-ORF1 serum. In contrast, the ORF1 encoded from the same exon 1Ј La mRNA but upstream of the La protein reading frame was not made at an equivalent amount or was very rapidly degraded. This interpretation was supported by the results obtained with two further ORF1 constructs. One construct contained only the complete exon 1Ј sequence cloned into pCI. In the case of the other construct we tried to increase the probability of a translation of the ORF1 reading frame by optimization of the translation initiation site of the ORF1 (see "Experimental Procedures"). However, in both cases no ORF1 translation product was detected when these constructs were transfected into human HeLa or mouse 3T3 cells (data not shown). All the results that were obtained by SDS-PAGE/im-munoblotting for all the ORF1 constructs were confirmed by using enzyme-linked immunosorbent assay and indirect immunofluorescence technique. In none of these studies was an ORF1 peptide detected (data not shown).
In summary, these results allow the following conclusions: (i) the ORF1 of an exon 1Ј La mRNA is not translated to either a separate ORF1 peptide or a ORF1 La fusion protein and (ii) the La protein reading frame in an exon 1Ј La mRNA is efficiently translated.

DISCUSSION
Frequently sera of patients with pSS or systemic lupus erythematosus contain self-reacting antibodies directed to nuclear antigens. One of the targets of anti-nuclear antibodies is the nuclear autoantigen La/SS-B (1). The La protein was described as a housekeeping protein (19). It was proposed to be involved  (N)). The N-terminal domain contained a XbaI site upstream of the frameshift mutation. In this site a second ORF1 was cloned. Thus the resulting ORF1 La(N)-ORF1 construct contained the ORF1 twice, one ORF1 upstream of the La peptide coding region but interupted by two stop codons and a second ORF1 fused in frame to the C terminus of the N-terminal La peptide. II, SDS-PAGE/ immunoblot analysis of extracts obtained from mouse 3T3 or human HeLa cells transfected with the ORF1 La(N) or ORF1 La(N)-ORF1 pCI construct. In a first step the immunoblot was analyzed with the anti-ORF1 rabbit serum (A). After detection the anti-ORF1 immune complexes were eluted, and the same blot was processed a second time with the anti-La mAb SW5 (B). Lanes a, c, e, g, and i were not used. in transcription/termination of RNA polymerase III and in internal initiation of translation especially of (polio)virus mRNAs (9 -12).
In a recent study we identified alternative La mRNAs (13). Both La mRNA forms were characterized as abundant finally processed cytoplasmic mRNAs (see the Introduction). However, due to the unusual 5Ј terminus of the exon 1Ј La mRNA isoform its function still remained obscure. The exon 1Ј was GC-rich, contained an oligo(U) tail of 23 uridine residues, and encoded for three ORFs locating upstream of the La protein reading frame. mRNAs with GC-rich 5Ј-termini and/or upstream ORFs had also been described in the case of other housekeeping proteins and also in the case of (proto)oncogenes, growth factor receptors, and proteins involved in immune response and inflammation (24). In most of these cases it was suggested that the alternative mRNA forms may play a regulatory role. It was assumed that the mRNA isoforms are either expressed in parallel to throttle the protein production of critical protooncogene products or represent nonfunctional mRNAs, which are alternatively expressed instead of the functional gene product. Only in few cases were the upstream ORFs used for translation.
Based on this background, we asked which if any of the reading frames encoded by the exon 1Ј La mRNA can be translated into protein. Looking to the upstream ORF1 to ORF3 it appeared likely that only the ORF1 could be translated to a 5.4-kDa peptide, whereas the ORF2 and ORF3 reading frames are (i) most likely too short and (ii) not in the La protein frame. Thus, we looked for expression of a peptide from either the ORF1 or the La protein reading frame of the exon 1Ј La mRNA. To look for an expression of the ORF1 or the human La protein reading frame a cell system and antibodies had to be established that allowed a specific detection of the human La protein and an ORF1 peptide when made. During these studies we looked to find out whether the selected mouse cell system was able to express the human La gene similarly to a human cell. Thereby it became evident that the 5Ј-start of the human exon 1Ј La mRNAs differed. However, similar results had also been observed when analyzing mRNA preparations that were obtained from a series of other human tissues. The observed 5Ј-start variations might not be the result of a 5Ј-terminal nonspecific RNA degradation because the exon 1 La mRNAs analyzed in parallel from the same RNA preparations did not show 5Ј-terminal variations, and the longest 5Ј-starts of exon 1Ј La mRNAs were found in human liver tissue, which certainly contained the highest RNase concentration during preparation of the mRNAs. It is also unlikely that the distinct 5Ј-terminal start regions of the exon 1Ј La cDNAs are the result of premature stops during reversed transcriptase due to secondary structures in the exon 1Ј La mRNAs because the pattern of a certain mRNA preparation could be repeated (data not shown).
Because exon 1Ј La mRNAs lacking the ORF1 could have different properties if compared with the full-length exon 1Ј La mRNAs, we included a 5Ј-shortened exon 1Ј La mRNA derivative in our expression studies. The selected 5Ј-shortened exon 1Ј La mRNA form started like the two exon 1Ј La cDNAs isolated from the patient's cDNA library at the oligo(U) tail in the exon 1Ј.
The two independent techniques, immunofluorescence microscopy and SDS-PAGE/immunoblotting, showed that the La protein reading frame in full-length exon 1 and exon 1Ј and the 5Ј-shortened exon 1Ј La mRNAs can be translated in transiently transfected cells, whereas the ORF1 reading was not used. The translation of the full-length exon 1Ј product appeared to be less efficient if compared with the exon 1 mRNA, whereas the 5Ј-shortened exon 1Ј La mRNA appeared to be more efficient despite the still existing two upstream ORFs 2 and 3. However, future studies using reporter gene constructs will be required to determine the translational efficiency of the different La mRNA forms.