Dysregulation of Ras guanyl nucleotide-releasing protein 1 (RasGRP1) in mice results in a systemic lupus erythematosus (SLE)-like disorder. We therefore looked for defective isoforms and/or diminished levels of human RasGRP1 in a cohort of SLE patients. PBMCs were collected from twenty healthy individuals and thirty-two patients with SLE. mRNA was isolated and five RasGRP1 cDNAs from each subject were sequenced. T cell lysates from healthy controls and SLE patients also were evaluated for their levels of RasGRP1 protein. The accumulated data led to the identification of 13 new splice variants of the human RasGRP1 gene. Not only did our SLE patients have increased levels and types of these defective transcripts relative to normal individuals, two SLE patients were identified whose PBMCs and T cells contained very little, if any, functional RasGRP1 mRNA and protein. The presence of aberrantly spliced RasGRP1 transcripts also was correlated with lower levels of RasGRP1 protein in the patients’ T cells. The lack of the normal isoform of RasGRP1 in some SLE patients and the increased prevalence of defective isoforms of RasGRP1 in others raise the possibility that dysregulation of this signaling protein contributes to the development of autoimmunity in a subset of SLE patients.
Systemic lupus erythematosus (SLE)3 is an autoimmune disease of unknown etiology characterized by the presence of high levels of autoantibodies. It has been proposed that a complex interaction of a number of undefined endogenous genes and their products with undefined pathogens and other factors in the environment somehow leads to dysregulation of adaptive immunity in SLE patients. Thus, a major effort has been made to segregate SLE patients into distinct subgroups based on the mechanism of their defects in adaptive immunity.
T cells use the TCR to distinguish self-Ags and foreign Ags and to undergo positive and negative selection in the thymus (1, 2). Positive selection occurs when TCR-derived signals of low intensity result in the activation of Erk-1, a rise in the intracellular levels of active Ras family members, and the activation of numerous transcription factors in the maturing lymphocytes (3). Ras cycles between an inactive GDP-bound form and active GTP-bound form. In T cells, Ras family members (e.g., N-Ras) are activated by ubiquitously expressed guanine nucleotide exchange factors (GEFs) such as Son of sevenless (4, 5) and Vav1 (6) and by the more restricted GEF Ras guanyl nucleotide-releasing protein (RasGRP) 1 (7, 8, 9, 10). RasGRP1 is an intracellular signaling protein that contains an N-terminal GEF domain and C-terminal calcium- and diacylglycerol (DAG)/phorbol ester-binding domains (11). Although RasGRP1 was initially identified and cloned from rat brain by Stone and coworkers (12), this intracellular protein is highly expressed in mouse and human T cells and to a lesser extent in B cells. The human RasGRP1 gene is 76.7 kb, contains 17 exons, and normally encodes an ∼95-kDa protein of 797 residues.
RasGRP1 resides in the cytoplasm of quiescent T cells. DAG is generated by phospholipase C when T cells are activated via their TCRs. Binding of DAG to the 50-mer protein kinase C1-like domain in the C-terminal half of RasGRP1 causes the transient translocation of the signaling protein to the inner leaflet of the lymphocyte’s plasma membrane. Eventually, RasGRP1 moves from the cell surface to the endoplasmic reticulum and Golgi by a down-regulation mechanism that remains to be determined (13, 14, 15, 16). Weakly selecting TCR signals depend on RasGRP1 to drive T cell development (17). Overexpression of RasGRP1 in Jurkat T cells enhances TCR-Ras-Erk signaling and results in increased IL-2 expression when these lymphocytes are exposed to calcium ionophore and PMA (7). RasGRP1−/− mice have a block in thymocyte development and diminished numbers of CD4+CD8− and CD4−CD8+ T cells (8). By contrast, the overexpression of RasGRP1 in mice results in increased numbers of CD4−CD8+ T cells (18). It therefore has been concluded that RasGRP1 participates in the final stages of T cell maturation (8). RasGRP1-null mice have high circulating levels of IgG and IgE and develop a late-onset lymphoproliferative autoimmune syndrome.
The lag mouse created by Layer et al. (19) also develops a SLE-like disorder. Young lag mice have reduced numbers of mature T cells due to a block in T cell development. Older lag mice have elevated levels of autoantibodies and also develop lymphadenopathy and glomerulonephritis. Positional mapping for a disease-associated gene in the lag mouse narrowed the responsible region to a 10-cM interval in the telomeric end of the long arm of chromosome 2, which contains the RasGRP1 gene among other genes. Even though the exact mutation is unknown, no functional RasGRP1 protein is present in the lag mouse’s splenocytes due to defective processing of its precursor transcript.
The human RasGRP1 gene resides on chromosome 15q15 (20). Numerous gene-linkage studies have been conducted on SLE patients, and many candidates have been identified as possible disease-susceptibility genes in these patients. In agreement with the conclusion that SLE is a polygenic disorder influenced by undefined environmental factors, several sites in the human genome were identified by Rao et al. (21) in their multivariate linkage analysis of 101 SLE-affected sib pairs in the United States. One of the sites identified in that study was an undefined gene on chromosome 15q15. Despite the human gene-linkage data and the lag mouse data, abnormal isoforms of human RasGRP1 have not been described. Nevertheless, RasGRP1 acts downstream of the TCR and BCR in lymphocytes and upstream of N-Ras, and the gene that encodes the related signaling protein RasGRP4 is aberrantly spliced in some mast cells (22, 23, 24). We therefore looked for dysregulation of RasGRP1 in patients with different autoimmune diseases. We now describe a unique subset of SLE patients with a common abnormality in the expression of this important signaling protein.
Materials and Methods
Twenty apparently healthy Japanese individuals (four males and 16 females; 30.9 ± 6.2 years old, mean ± S.D.) and 32 unrelated Japanese patients with SLE (five males and 27 females; 32.0 ± 9.9 years old, mean ± S.D.) were studied. All patients in this cohort fulfilled the American College of Rheumatology criteria for SLE (25). Based on the British Isles Lupus Assessment Group scoring system (26), disease activity at the time of analysis of their RasGRP1 transcripts and proteins was 11.2 ± 7.3 (range = 0–28). Mean disease duration was 8.2 years (range = <1 month to 28 years). Twenty-three (72%) of our SLE patients were receiving prednisolone (PSL) when we measured the expression of RasGRP1 mRNA and protein in their PBMCs and T cells. The daily dose of PSL in this treated group was 16.0 ± 14.2 mg/day (mean ± S.D., range = 2.5–60) and was based primarily on the patient’s disease activity and overall weight. Six (26%) of our SLE patients on PSL therapy received the additional immunosuppressive agents cyclosporine (n = 2), methotrexate (n = 2), cyclophosphamide (n = 1), and mizoribine (n = 1). The remaining nine (28%) SLE patients were evaluated for the presence of aberrant RasGRP1 isoforms before the initiation of any therapy. Twelve patients with other autoimmune diseases served as autoimmune controls. We did not attempt to age or gender match this control group with our SLE patients. The patients in this autoimmune control group had rheumatoid arthritis (n = 5), polymyositis/dermatomyositis (n = 2), systemic sclerosis (n = 3), and Sjogren’s syndrome (n = 2). Our study was approved by the Human Ethics Committee of the Hokkaido University Graduate School of Medicine (Sapporo, Japan), and informed consent was obtained from each subject.
Isolation and characterization of novel human RasGRP1 transcripts
PBMCs were collected from our subjects using Ficoll-Paque PLUS (Amersham Biosciences), and RNA was isolated from the resulting cells using TRIzol (Invitrogen Life Technologies). The PBMC-derived transcripts were then converted into cDNA using reverse transcriptase (Toyobo) and an oligo(dT)12–18 primer (Invitrogen Life Technologies). The coding regions of the RasGRP1 cDNAs were amplified by a PCR method using the forward primer 5′-CGCGGCCATGGGCACCCTG-3′ and the reverse primer 5′-CTAAGAACAGTCACCCTGCTCCAT-3′, which correspond to sequences residing at the beginning and end of the coding domain of the normal human RasGRP1 transcript noted at GenBank accession no. NM_005739, respectively. After a heat denaturation step, each of the 30 cycles of the subsequent PCR steps consisted of a 15-s denaturing step at 94°C, a 15-s annealing step at 63°C, and a 2-min extension step at 72°C. The resulting PCR products were electrophoresed in 1.0% agarose gels and visualized. The PCR products were also subcloned into pcDNA3.1 V5-His-TOPO (Invitrogen Life Technologies). Five arbitrarily selected cDNAs from each individual were then sequenced using an ABI PRISM 3130 genetic analyzer (Applied Biosystems) and numerous internal forward and reverse sequencing primers.
The forward primer 5′-AAGGACCTCATCTCCCTGTA-3′ and the reverse primer 5′-AAGTAGGCTGTGATCTCATC-3′ also were used to evaluate the prevalence of the frequently found exon 11 deletion splice variants of the human RasGRP1 transcript in our SLE subjects. To serve as PCR controls in these transcript analyses, the forward primer 5′-GTCAGTGGTGGACCTGACCT-3′ and the reverse primer 5′-TCTTCAAGGGGTCTACATGG-3′ were used to detect and quantitate the GAPDH transcript. For amplification of GAPDH cDNAs, each of the 25 cycles of the subsequent PCR steps consisted of a 15-s denaturing step at 94°C, a 15-s annealing step at 55°C, and a 45-s extension step at 72°C. The RasGRP3-specific primers 5′-CCATGGGATCAAGTGGCCTTG-3′ and 5′-AGTCAGCCATCCTCACCATCCTG-3′ were used to isolate the nucleotide sequence of the transcripts that encode the entire coding domain of normal RasGRP3. The PCR conditions were identical with those used to evaluate RasGRP1 mRNA expression. We also evaluated the full-length CD4 transcripts in our patients’ PBMCs using the CD4-specific primers 5′-CACAATGAACCGGGGAG-3′ and 5′-GCCTCAAATGGGGCTAC-3′. In these CD4 transcript analyses, each of the 30 cycles of the PCR consisted of a 15-s denaturing step at 94°C, a 15-s annealing step at 55°C, and a 60-s extension step at 72°C.
To investigate the splice-donor and splice-acceptor sites of exon 11 in the human RasGRP1 gene, the relevant sequence was obtained using genomic DNA from the PBMCs of SLE patient 10, who had an exon 11 deletion in all five of his/her sequenced RasGRP1 cDNAs. The forward intron 10 primer was 5′-CAGATTGAGAATTTCAAATTAT-3′ and the reverse intron 11 primer was 5′-ATACAAAACTGAGCCAGG-3′. The forward primer resides 253 nucleotides from the 3′ end of intron 10 for the detection of the branch point sequence used to the remove this intron.
Generation and use of anti-human RasGRP1 Abs in SDS-PAGE immunoblot assays
Anti-human RasGRP1 Abs were generated against the 18-mer peptide MGTLGKAREAPRKPSHGC that corresponds to the normal protein’s N terminus. This peptide was chosen as the Ag because its sequence does not resemble that in human RasGRP2 (i.e., MGTQRLCGRGTOGWPGSS), RasGRP3 (i.e., MGSSGLGKAATLDELLCT), or RasGRP4 (i.e., MNRKDIKRKSHQECSGKA). A basic local alignment search tool (BLAST) protein search also revealed no similar sequence in any other known human protein. Finally, we concluded that Abs directed against the N terminus of RasGRP1 should be more valuable than the Abs directed the protein’s C terminus because the latter Abs would not allow us to identify truncated isoforms of this signaling protein in our patients’ lymphocytes. Five milligrams of the synthetic peptide were coupled to keyhole limpet hemocyanin (KLH) via its C-terminal cysteine. A New Zealand White rabbit was s.c. injected with the peptide-KLH complex in the presence of CFA for the primary immunization. This was followed by reimmunization of the animal with the peptide-KLH complex in IFA. The resulting Abs were affinity purified using the synthetic immunizing peptide coupled to Sulfo-Link gel (Pierce) via its C-terminal cysteine. Protein blots containing lysates of RasGRP1-expressing HEK-293 cells and a human thymus obtained from Clontech were probed with the Abs in the presence or absence of the immunizing peptide.
T cells were purified from ∼10 ml of peripheral blood from each subject using the RosetteSep human T cell enrichment cocktail (StemCell Technologies). The purities of these T cells were routinely >85% as assessed on a FACSCalibur flow cytometer (BD Biosciences) by using a FITC-labeled anti-CD3 Ab (BD Biosciences). A mammalian cell lysis kit (catalog no. MCL1; Sigma-Aldrich) was used to prepare lysates of each preparation of T cells. The protein concentrations of these samples were determined using the bicinchoninic acid protein assay (Pierce). Approximately 3 μg of each cell lysate was subjected to SDS-PAGE. The separated proteins were transferred to immune blot polyvinylidene difluoride membranes (Bio-Rad), and immunoblotting was performed using our rabbit anti-human RasGRP1 Abs and a mouse anti-human β-actin Ab (Sigma-Aldrich) followed by the relevant secondary Abs conjugated to HRP. The Ab-treated blots were developed by using the Immobilon Western chemiluminescent substrate (Millipore). The densities of the bands that correspond to β-actin and full-length RasGRP1 were then measured using the Quantity One software package (Bio-Rad).
Generation of recombinant human RasGRP1 isoforms found in SLE patients
Constructs encoding human RasGRP1 with or without a C-terminal V5 epitope tag were transfected into the epithelial cell line HEK-293 (line CRL-1573; American Type Culture Collection). This cell line was chosen rather than a T cell line because HEK-293 cells do not express RasGRP1, thereby allowing us to interpret data in an experimental situation without endogenous RasGRP1. The cDNAs that encode full-length RasGRP1 and three abnormal isoforms of the signaling protein identified in our SLE patients (splice variants A, B, and D) were subcloned into a pIRESpuro3 vector (Clontech). Transfections were performed using Lipofectamine 2000 reagent (Invitrogen Life Technologies) according to the manufacturer’s instructions. After a 24-h incubation, 5 μg/ml puromycin (Sigma-Aldrich) was added to the culture medium to select for RasGRP1-expressing transfectants. To evaluate the expression of recombinant RasGRP1 at the RNA level, the transfected cells were trypsinized, washed, and sonicated using a Microson ultrasonic cell disruptor (Wakenyaku). Total RNA was extracted from the lysates as described above. cDNAs were generated using the QuantiTect reverse transcription kit (Qiagen), which includes a DNase treatment step to minimize the possible contamination of the RNA samples with construct DNA. As an additional control for construct DNA contamination, PCRs were conducted on these RNA samples treated in the same manner but without reverse transcriptase.
A confirmatory real-time quantitative PCR (qPCR) approach was then used to monitor the levels of the different RasGRP1 transcripts in the transfectants. In these instances, the level of the RasGRP1 transcript was normalized to that of the GAPDH transcript using an ABI Prism 7000 sequence detection system and TaqMan MGB probes specific for RasGRP1 (assay identifier Hs00996723_m1) and GAPDH (assay identifier Hs00266705_1) (Applied Biosystems). We chose a RasGRP1-specific primer set in these qPCRs that generates a DNA fragment derived from exons 12 and 13, because this sequence is present in all of our expression constructs. Thus, the primer set can be used to measure the levels of the transcripts that encode normal RasGRP1 and its three splice variants in the transfectants. Relative quantification was performed using the comparable cycle threshold (CT) method in which ΔCT is the level of the RasGRP1 transcript in the RNA sample relative to that of the GAPDH transcript. The difference in the expression of the transcripts that encode normal RasGRP1 and its splice variants was defined in each instance as ΔΔCT and the changes in mRNA levels were defined by 2−ΔΔCT.
For protein expression analysis, the cells were washed with PBS and then lysed using the mammalian cell lysis kit (Sigma-Aldrich). The presence of the different isoforms of RasGRP1 in the transfectants at the protein level was evaluated by the SDS-PAGE immunoblot method described above, using our anti-RasGRP1 Abs. In some experiments, the transfectants were placed in Opti-MEM I medium (Invitrogen Life Technologies) supplemented with the proteasome inhibitor MG-132 at 25 μM (Calbiochem). The 10 mM stock solution of MG-132 was dissolved in 100% DMSO. Thus, the final DMSO concentration in the culture medium was 0.25%. After a 15-h incubation, the treated cells were washed with PBS and cell lysates were prepared and analyzed as described above. Transfected HEK293 cells also were exposed for 14 h to 2 μM monensin sodium sulfate (Sigma-Aldrich) in Opti-MEM I medium before lysates of these cells were analyzed for their RasGRP1 protein levels.
A cell-free in vitro transcription:translation assay was performed using the PROTEINscript II T7 kit (Ambion) according to the manufacturer’s instruction. Briefly, our pIRESpuro3 vectors encoding the V5-labeled normal full-length hRasGRP1 and its defective splice variants A, B, and D were subjected to the transcription step. The rabbit reticulocyte lysates in the kit were then used to translate the transcripts, which were evaluated by the SDS-PAGE immunoblot assay using anti-V5 Ab (Invitrogen Life Technologies).
Expression of IL-2 in stimulated CD4+ T cells
CD4+ T cells were purified from 10 ml of the peripheral blood of normal individuals and SLE patients using the RosetteSep human CD4+ T cell enrichment cocktail (StemCell Technologies). In each instance, 500,000 purified T cells were suspended in 0.5 ml of RPMI 1640 medium containing 10% FCS and then stimulated by plate-bound anti-CD3 (1 μg/ml) and anti-CD28 (2.5 μg/ml; BD PharMingen) Abs. Purified T cells that were not stimulated with the anti-CD3 and anti-CD28 Abs served as negative controls. After a 3.5-h incubation at 37 °C in a humid chamber, total RNA was prepared from each sample and subjected to real-time qPCR for quantitation of its IL-2 mRNA levels. The TaqMan MGB primer set specific for IL-2 (assay identifier Hs00174114_m1) was used in the qPCR protocol. This experiment was performed two to three times for each individual in a triplicate manner.
The χ2 test with Yates’ correction was used to compare the frequencies of the identified RasGRP1 variants in SLE patients vs healthy individuals. Clinical features or other parameters between control individuals and patients or between patients with and without splice variants were compared using Mann-Whitney’s U test. p values below 0.05 were considered to be significant.
Identification of 13 novel RasGRP1 transcripts in SLE patients caused by defective splicing of its precursor transcript
RasGRP1 cDNAs were amplified from each patient’s PBMCs by using a RT-PCR approach and primers that correspond to the beginning and end of the coding domain of normal RasGRP1 in human T cells (Fig. 1,A). For this initial screening analysis, a limited number of patients with other rheumatic diseases also were evaluated for their expression of normal and defective RasGRP1 transcripts (Fig. 1,B). The 2.4-kb cDNA that encodes the normal isoform of RasGRP1 was found in all healthy individuals, as well as in those patients with non-SLE autoimmune disorders. No abnormal RasGRP1 transcripts were found in the PBMCs of SLE patients 1, 4, or 5. In contrast, SLE patients 2, 6, and 7 contained both normal and abnormal RasGRP1 transcripts. Surprisingly, the normal 2.4-kb RasGRP1 RT-PCR product was below detection in SLE patients 3, 8, and 9. When SLE patient 3 was reanalyzed 16 mo later, his/her PBMCs had predominantly normal RasGRP1 transcripts (Fig. 1,C). The defect in RasGRP1 expression therefore was variable in SLE patient 3. In contrast, analysis of the PBMCs taken from SLE patients 8 and 9 ∼8 mo after the initial analysis of these patients revealed a similar defect in RasGRP1 expression (Fig. 1,C). Of the two SLE patients who reproducibly had virtually no normal RasGRP1 mRNA in their PBMCs, SLE patient 9 had been treated with PSL before the analysis of his/her PBMCs. However, SLE patient 8 had not been given any therapeutic drug. Thus, the failure to detect appreciable amounts of the normal RasGRP1 isoform in SLE patient 8 was not a therapy-induced abnormality. Disease status was significantly different in SLE patients 3 and 8 between the first and the second evaluations of their PBMCs for RasGRP1 expression. SLE patient 3 was treated with cyclosporine to alleviate his/her thrombocytopenia, whereas SLE patient 8 was treated with prednisolone and cyclophosphamide to alleviate his/her fever, cytopenia, and glomerulonephritis. In contrast, neither disease activity nor treatment had changed in SLE patient 9. As noted in Fig. 1 D, SLE patients 8 and 9 had only full-length CD4 and RasGRP3 transcripts in their PBMCs. Similar CD4 and RasGRP3 transcript findings were obtained when two normal individuals, two rheumatoid arthritis patients, and SLE patient 4 were analyzed. Thus, the T cells in SLE patients 8 and 9 did not have a global defect in mRNA processing, including the posttranscriptional processing of another RasGRP family member.
Different sized PCR products were detected in many of our SLE patients. Nevertheless, as noted below, RasGRP1 transcripts lacking exon 11 were frequently observed. We therefore next used a PCR approach to amplify exons 10, 11, and 12 in the RasGRP1 transcripts from the same cDNA samples noted in Fig. 1. Nucleotide sequence analysis confirmed that the generated ∼500-bp product noted in the middle panels of Fig. 1 corresponds to that in normal RasGRP1 transcripts with its three exons, whereas the ∼400-bp product corresponds to abnormal RasGRP1 transcripts lacking exon 11. The ∼500-bp fragment was prevalent in every control subject even though lesser amounts of the ∼400-bp fragment were also found in these samples. In contrast, the ∼400-bp product was prevalent in the same three SLE patients in whom we failed to detect the normal RasGRP1 transcript with the first primer set. Because the data with the exon 10/exon 12 primer set essentially confirmed the data with the exon 1/exon 17 primer set, the RasGRP1 gene is transcribed in all of our SLE patients but its precursor transcript is sometimes aberrantly spliced.
To identify the exons that are preferentially lost in the defective RasGRP1 transcripts in our SLE patients, we next collected samples from age- and sex-matched SLE patients and healthy controls. We then attempted to sequence five arbitrarily subcloned RasGRP1 cDNAs from each individual. Although we were not able to generate any RasGRP1 cDNAs from SLE patients 8 or 9 (Fig. 1,A), RasGRP1 cDNAs from our other 30 SLE patients were obtained. Thus, we sequenced 100 RasGRP1 cDNAs from 20 healthy individuals and 150 RasGRP1 cDNAs from 30 SLE patients. More than 1,000 sequencing reactions were conducted in this study to understand at the molecular level how the RasGRP1 transcript is aberrantly processed in our cohort of SLE patients. The accumulated data resulted in the identification of 13 new isoforms of human RasGRP1 due to alternate splicing of exons 5–17 in its precursor transcript (Fig. 2). The novel splice variants are as follows deletion of exon 11 (splice variant A, GenBank accession no. AY954625); deletion of exons 11 and 16 (splice variant B, GenBank accession no. AY858556); deletion of exons 11 and 13 (splice variant C, GenBank accession no. AY966005); deletion of exons 11, 15, and 16 (splice variant D, GenBank accession no. AY634315); deletion of exon 11 with insertion of a genomic fragment from intron 16 (splice variant E); deletion of exon 16 (splice variant F); deletion of exons 15 and 16 (splice variant G); insertion of 40 nucleotides from intron 12 between exons 11 and 12 (splice variant H); deletion of exons 11 and 15 (splice variant I); deletion of exons 10 and 11 (splice variant J); deletion of exon 13 (splice variant K); deletion of exon 15 (splice variant L); and deletion of the last 44 nucleotides in exon 5 (splice variant M).
The most prominent abnormal RasGRP1 isoform identified in our SLE patients was splice variant A, which lacks only exon 11. Loss of this exon does not cause a frame-shift abnormality or a premature translation termination codon in the transcript. Nevertheless, many of the other RasGRP1 splice variants identified in our study result in a premature translation termination codon. If translated, splice variants A to M should give rise to C-terminal truncated isoforms of this intracellular signaling protein that contain 173–762 residues rather than the 797 residues of normal RasGRP1. An improperly processed transcript that contains one of the gene’s first nine introns also should lead to a nonfunctional protein if translated.
Clone numbers and the percentages of splice variants are indicated in Fig. 2 and Table I. The splice variants A, B, D, and F were frequent abnormal isoforms in our SLE patients; these were even detected in some healthy subjects. In contrast, splice variants C, E, G, H, I, J, K, L, and M were only found in the SLE patients. Moreover, the frequency of any aberrant RasGRP1 isoform was significantly greater in the patient group than the healthy group, even if one does not take into account in this statistical analysis our failure to obtain RasGRP1 cDNAs from SLE patients 8 and 9. We compared the clinical features of those patients who have no abnormal RasGRP1 isoform with those who have at least one splice variant (Table II). The SLE patients who had at least one abnormal RasGRP1 cDNA of five sequenced cDNAs tended to be older and carried the diagnosis longer than those SLE patients that lacked an aberrant RasGRP1 isoform. However, there was no significant correlation between disease activity and the presence of aberrant RasGRP1 isoforms. To evaluate the effect of PSL therapy on the generation of aberrant isoforms of RasGRP1, the presence of at least one abnormal splice variant cDNA of five sequenced cDNAs were evaluated in patients who were analyzed before or during therapy. There was no statistical difference in the frequency of having at least one defective RasGRP1 isoform between the patient groups before and during therapy (Table II). Moreover, SLE patient 8, who had no normal RasGRP1 transcripts (Fig. 1), had not been on any therapy before his/her PBMCs were evaluated.
|.||Percentage (%) .||Odds Ratio .||95% CI .||p Value .|
|Loss of exon 11|
|Lost of any exon|
|.||Percentage (%) .||Odds Ratio .||95% CI .||p Value .|
|Loss of exon 11|
|Lost of any exon|
Odds ratio, confidential interval (CI), and p values were calculated by a χ 2 test with Yates’ correction to evaluate the significance of the RasGRP1 clones lacking exon 11 or any exon in the cohort of healthy subjects and SLE patients. The percentage (%) of sequenced clones that contained an aberrant RasGRP1 isoform is also indicated. Every healthy subject contained at least one normal RasGRP1 cDNA. In contrast, only abnormal RasGRP1 cDNAs were detected in our analyses of four SLE patients, including patients 8 and 9.
|.||No Splice Variant .||Any Splice Variant .||p Value .|
|Age (years)||26.0 ± 6.0||36.0 ± 11.0||0.024*|
|BILAG score||9.8 ± 5.3||11.4 ± 7.0||0.494*|
|Disease duration (mo)||53.8 ± 72.9||15.2 ± 11.6||0.020*|
|PSL dose (mg/day)||11.1 ± 15.1||12.1 ± 14.2||0.790*|
|Patients on therapy||4||3|
|Patients on no therapy||7||16||0.403†|
|.||No Splice Variant .||Any Splice Variant .||p Value .|
|Age (years)||26.0 ± 6.0||36.0 ± 11.0||0.024*|
|BILAG score||9.8 ± 5.3||11.4 ± 7.0||0.494*|
|Disease duration (mo)||53.8 ± 72.9||15.2 ± 11.6||0.020*|
|PSL dose (mg/day)||11.1 ± 15.1||12.1 ± 14.2||0.790*|
|Patients on therapy||4||3|
|Patients on no therapy||7||16||0.403†|
The p values in the first four rows (∗) were calculated using Mann-Whitney’s U-test and the p value in the last row (†) was calculated by χ 2 test with Yates’ correction. BILAG, British Isles Lupus Assessment Group.
Sequence analysis of a RasGRP1 genomic fragment from a patient who was preferentially expressing RasGRP1 transcripts in his/her PBMCs that lacked exon 11 (e.g., splice variant A) revealed no point mutation from the 3′ end of intron 10 to the 5′ end of intron 11 of his/her RasGRP1 gene. Thus, there is no germline mutation in the intron 10/exon 11 and exon 11/intron 11 splice-sites in either allele of this patient’s RasGRP1 gene. The branch point sequence in intron 10, which is used to remove this U2-dependent intron, also was found to be intact in this SLE patient.
Analysis of RasGRP1 protein levels in T cells
Rabbit polyclonal Abs were generated against an 18-mer synthetic peptide that corresponds to the unique N terminus of human RasGRP1. The resulting affinity-purified Abs were then used to evaluate RasGRP1 protein levels in the T cells in the peripheral blood of 12 of our 32 SLE patients. The specificity of the preimmune sera, postimmune sera, and affinity-purified anti-RasGRP1 Abs were first evaluated using HEK-293 cells that had been transfected with an expression vector lacking or containing cDNAs that encoded normal RasGRP1 or its abnormal splice variant A. As noted in Fig. 3, no immunoreactive band was detected in lysates of the control transfectants in our SDS-PAGE immunoblot analysis. In contrast, the affinity-purified anti-RasGRP1 Abs recognized the ∼95-kDa normal RasGRP1 isoform and its abnormal ∼90-kDa splice variant in the transfectants. A nonspecific band of ∼64 kDa that could not be blocked by the immunizing peptide was occasionally observed in the T cell lysates from some individuals (data not shown). Nevertheless, our Abs consistently recognized an ∼95-kDa protein in lysates of the human thymus and T cells purified from normal peripheral blood (Fig. 3). In all instances, reactivity of the latter band was greatly diminished if the protein blots were reprobed with the anti-RasGRP1 Abs in the presence of the immunizing peptide.
After having established the specificity of our anti-RasGRP1 Abs, we next used them to evaluate the protein levels of varied isoforms of RasGRP1 in T cells purified from the peripheral blood of five healthy individuals and 12 SLE patients (Fig. 4). The RasGRP1 and β-actin protein blot data are shown in Fig. 4, A and B, respectively. In each instance, the ratio of the density of the RasGRP1 immunoreactive band relative to that of the β-actin immunoreactive band was determined. Density ratios in all healthy controls and in all SLE patients were 0.87 ± 0.05 and 0.64 ± 0.24, respectively, without statistical difference (p = 0.082). However, some SLE patients had low levels of RasGRP1 protein, thereby confirming the mRNA data. RasGRP1:β-actin density ratios of individuals without splice variants were significantly higher than those of individuals with any splice variants or whose sequence was not available (0.83 ± 0.16 vs 0.59 ± 0.21, p = 0.027) (Fig. 4,C). Thus, the presence of defective RasGRP1 transcripts in SLE patients generally correlates with the lower levels of normal RasGRP1 protein. Clinical status and information on the sequenced RasGRP1 cDNAs are illustrated in Fig. 4 D for the individuals whose T cell lysates and RasGRP1 cDNAs were available.
The ability of CD4+ T cells to produce IL-2 after anti-CD3 and anti-CD28 Ab stimulation was evaluated using a real-time qPCR approach (Fig. 4 E). As previously found by others, the amount of IL-2 mRNA generally increased >10-fold when the peripheral blood T cells from our cohort were stimulated ex vivo via their surface CD3 and CD28 proteins. Although diminished responses were observed in healthy individual 11, the CD4+ T cells isolated from RasGRP1-defective SLE patient 8 barely increased their levels of IL-2 mRNA (n = 2–3).
Despite the presence of transcripts for splice variants A to M in SLE patients, we could not detect appreciable amounts of low m.w. immunoreactive proteins in the T cells of SLE patients using our anti-RasGRP1 Abs. The absence of proteins slightly smaller than normal RasGRP in our patients’ T cells that are recognized by our anti-RasGRP1 raised the possibility that splice variants A to M are translated but that many of these newly expressed proteins are rapidly catabolized by an undefined proteolytic pathway to ensure that defective isoforms of the signaling protein do not accumulate in the patient’s lymphocytes. An alternate explanation is that some aberrant RasGRP1 transcripts simply are not translated in T cells. To address those possibilities, we next evaluated what happens when cultured HEK-293 cells are transfected with expression constructs that encode the RasGRP1 splice variants A, B, and D. Because an exogenous peptide added at the protein’s C terminus might affect the targeting and/or metabolism of these three isoforms of RasGRP1 in the transfectants, additional expression constructs were created to encode recombinant RasGRP1 isoforms that lack epitope tags. We focused on splice variants A, B, and D in this aspect of our study because these abnormal RasGRP1 transcripts are abundant in SLE patients (Fig. 2). Another reason for selecting splice variant D, which lacks exon 15, is because defective splicing of exon 15 in the RasGRP4 transcript occurs in the mast cells of the C3H/HeJ mouse (24).
As assessed by semiquantitative RT-PCR gel analysis (Fig. 5,A) and by real-time qPCR analysis (Fig. 5,D), all four stable transfectants contained abundant amounts of the transcript that encodes the relevant RasGRP1 isoform. Due to the loss of exon 11, the RT-PCR products present in the transfectants that had been induced to express splice variants A, B, and D were smaller than the RT-PCR products present in the transfectant that had been induced to express normal RasGRP1 (Fig. 5,A). At the protein level, the positive control transfectant contained substantial amounts of ∼95-kDa RasGRP1. As expected, the transfectants that had been induced to express splice variant A produced a slightly smaller immunoreactive product due to loss of the 35-mer peptide that links the GEF domain to the calcium- and phorbol ester/DAG-binding domains. However, the amount of immunoreactive protein in the latter transfectant was noticeably less than that in the transfectant, which expressed full-length RasGRP1. Surprisingly, no immunoreactive RasGRP1-related protein products were detected in the transfectants that had been induced to express splice variants B or D (Fig. 5,C), even though their transcripts were abundant (Figs. 5, A and D). Similar results were obtained when cells were transfected with constructs that encode V5-labeled proteins (data not shown). Thus, our failure to the detect splice variants B and D at the protein level was not due to a technical problem with the anti-RasGRP1 Abs.
To exclude the possibility that the truncated splice variants B and D were preferentially secreted into the condition medium, we next placed cultures of these transfectants overnight in medium supplemented with 2 μM monensin and then performed SDS-PAGE immunoblot analysis on the resulting cell lysates. Because no immunoreactive protein was detected in the lysates of the monensin-treated cells (data not shown), our failure to detect splice variants B and D inside the transfectants apparently was not due to exocytosis of these defective RasGRP1 isoforms into the conditioned medium. In support of these data, we could not detect immunoreactive RasGRP1 isoforms in the conditioned medium of the non-monensin-treated cells that were expressing the splice B and D variants. The levels of immunoreactive RasGRP1 protein in the splice D transfectant also did not increase when these cells were cultured 15 h in the presence of the proteasome inhibitor MG132 (25 μM) (data not shown). The results obtained from an in vitro translation assay using rabbit reticulocyte lysate are shown in Fig. 5 E. The normal RasGRP1 transcript and its splice variant A resulted in translated proteins that could be detected with anti-V5 Ab. In contrast, no translated protein was obtained with the splice variant B and D transcripts. The accumulated data suggest that defective splice variants B and D cannot be translated in HEK-293 cells.
RasGRP1 is an intracellular signaling protein essential for the final stages of T cell development (11, 20, 27). Along with RasGRP3 (28, 29, 30), RasGRP1 also has been implicated in B cell proliferation and development by facilitating BCR signaling (12). This intracellular GEF functions downstream of the TCR in T cells, downstream of the BCR and Lyn in B cells, and upstream of Ras family members in both cell types. Lyn-null mice develop a SLE-like disorder (31), and decreased expression of Lyn has been reported in some SLE patients (32, 33). A Gly→Asp mutation at residue 13 in N-Ras also causes a human autoimmune lymphoproliferative disorder (34). The accumulated data have led to the conclusion that dysregulation of numerous intracellular proteins that participate in BCR- and/or TCR-dependent signaling pathways in lymphocytes can lead to autoimmune disorders. In regard to our study, loss of RasGRP1 due to targeted ablation of its gene (8) or defective splicing of exon 4 from its precursor transcript (19) causes a lupus-like disorder in mice. Despite the impressive mouse data on the role of RasGRP1 in lymphocyte development and function, no human has been identified with a defect in RasGRP1 expression. Thus, the relevance of the mouse RasGRP1 data to SLE patients remained to be determined. We now describe a subset of SLE patients who preferentially express defective isoforms of RasGRP1 due to aberrant splicing of its precursor transcript.
The 76.7-kb human RasGRP1 gene contains 17 exons, and the proper splicing of its precursor transcript results in an ∼5-kb mRNA of which 2.4 kb represents its coding domain. Because microarray assays that simply measure RasGRP1 mRNA levels in a person’s lymphocytes cannot be used to identify improperly spliced RasGRP1 transcripts, we used a RT-PCR approach in the initial phase of our study to evaluate the global status of the coding domain of the RasGRP1 transcript in the noncloned PBMCs of healthy individuals, patients with SLE, and patients with other autoimmune disorders. The normal full-length RasGRP1 transcript was abundant in the PBMCs of healthy individuals and our autoimmune control group (Fig. 1). In contrast, two of these initially analyzed SLE patients reproducibly had greatly diminished levels of the normal RasGRP1 transcript in their PBMCs. Additional SLE patients were identified that contained an unusually high level of abnormal RasGRP1 transcripts in addition to the normal RasGRP1 transcript. Analysis of our cDNA sequence data resulted in the identification of 13 new isoforms of human RasGRP1 due to alternate splicing of exons 5–17 in its precursor transcript (Fig. 2). As noted in Fig. 1,A, SLE patients sometimes had RasGRP1-specific PCR products in their PBMCs that were >2.4 kb in size, suggesting a failure to remove at least one intron in the processed transcript. Introns 1, 2, and 3 in the human RasGRP1 gene are 4.6, 33.6, and 6.7 kb, respectively. Because RasGRP1 cDNAs that contain one or more large introns most likely would be missed in our cDNA selection and sequencing approach to identify new isoforms of RasGRP1, the number of abnormal isoforms of RasGRP1 in SLE patients almost certainly is greater than that noted in Fig. 2.
We failed to identify defective RasGRP1 transcripts that contained introns 1–10 (Fig. 2). Failure to remove any of these introns in the processed RasGRP1 transcript would result in a very early translation termination codon. Defective transcripts are rapidly degraded by the nonsense posttranscriptional pathway in yeast if the premature translation termination codon resides in the initial half of the coding domain of the transcript (35), as occurs in the defective RasGRP4 transcript in HMC-1 cells (22). Thus, our failure to identify improperly processed RasGRP1 transcripts in our SLE patients that contain introns 1 to 7 probably is the result of rapid catabolism of those defective RasGRP1 transcripts by the nonsense posttranscriptional pathway.
The reason why pre-RasGRP1 mRNA is improperly processed in a subgroup of our SLE patients was not deduced in our study. Nevertheless, we were able to rule out a number of possibilities. The most prevalent abnormal variant identified in the PBMCs, T cells, and B cells of our SLE patients is splice variant A, which lacks exon 11. We failed to identify a point mutation in the gene’s exon 11 and its intron 10/exon 11 and exon 11/intron 11 splice junctions in the genomic DNA isolated from SLE patient 10, who had abundant exon 11-deleted isoforms. In addition, no mutation was found in the branch-point sequence of the human RasGRP1 gene, which is used to remove intron 10.
Fifteen of the introns (including intron 10) of the mouse, rat, and human RasGRP1 genes are U2-dependent. The exception in this gene is intron 3, which is U12-dependent (for review of U12-dependent introns, see Ref. 36). It has been estimated that only ∼0.1% of all introns in the human genome are U12 dependent. Our analysis of the nucleotide sequences of the human RasGRP2, RasGRP3, and RasGRP4 genes revealed the unexpected finding that all four RasGRP genes contained a single U12-dependent intron, even though the size of this intron varies considerably. We previously showed that the RasGRP4 gene is transcribed in HMC-1 cells but that this mast cell line cannot remove intron 3 in its processed RasGRP4 transcript (22). Intron 3 is correctly removed from RasGRP1 splice variants A to M. Nevertheless, it is possible that some SLE patients have a defect in the processing of U12-dependent introns in their lymphocytes that adversely impacts the removal of downstream introns in the precursor RasGRP1 transcript. The B cells in PBMCs express both RasGRP1 and RasGRP3. As noted in Fig. 1 D, we found no evidence of defective processing of the RasGRP3 transcript in these SLE patients. It therefore appears that SLE patients do not have a global defect in the U12-dependent processing of their transcripts and that the posttranscriptional defect we uncovered in RasGRP1 expression does not adversely affect the processing of the transcript that encodes another member of this family of signaling proteins in lymphocytes. We also discovered that the CD4 transcript is processed correctly in SLE patients, suggesting that the posttranslational defect is relatively specific to the RasGRP1 transcript in lymphocytes.
We considered the possibility that the aberrant splicing of the RasGRP1 precursor transcript in our SLE patients could be an indirect consequence of long-term PSL therapy. In this regard, SLE patient 8 had barely detectable levels of normal RasGRP1 transcripts (Fig. 1) and protein (Fig. 4). Because this patient had not been on PSL or any other therapeutic drug before his/her PBMCs were initially analyzed and because the patient continued to have aberrant RasGRP1 expression after he/she was treated with prednisolone and cyclophosphamide, the presence of defective RasGRP1 transcripts does not appear to be the primary consequence of the therapy used to treat our SLE patients. The data noted in Table II from the entire group of SLE patients support this conclusion. Nevertheless, SLE patient 3 did alter his/her expression of RasGRP1 after drug treatment to improve his/her clinical situation. Thus, there remains the possibility that RasGRP1 expression is affected by disease status or treatment in some SLE patients.
Unlike the RasGRP3 transcript and nearly all other transcripts in lymphocytes, the RasGRP1 transcript lacks a classical “AAUAAA” or “AUUAAA” polyadenylation signal sequence 10–30 nucleotides upstream of its poly(A) site. We did not determine the nucleotide sequences of the 3′ untranslated regions of the RasGRP1 transcripts in our SLE patients or healthy subjects. Nevertheless, it is unlikely that the unconventional polyadenylation site in the RasGRP1 gene is the problem, because reduced polyadenylation of eukaryotic pre-mRNA generally results in rapid catabolism of the transcript rather than defective splicing. A more likely explanation of our data is that a point mutation exists somewhere in the RasGRP1 gene in this subgroup of SLE patients that causes its precursor transcript to bind improperly to an undefined cis-acting factor in lymphocytes that ultimately controls the splicing of this transcript. In this regard, CD45/PTPRC is a tyrosine phosphatase found in T cells and other hemopoietic cells that has been linked to multiple sclerosis. Relevant to our study, two groups identified a polymorphism in exon 4 of the CD45 gene. This mutation does not result in a codon that encodes a new amino acid, but it somehow leads to aberrant splicing of the precursor transcript (37, 38). The latter studies provide a relevant example of how a translationally silent point mutation in a gene encoding a protein that participates in critical signaling events in T cells can lead to an autoimmune disorder.
If translated, all but splice variant M should encode RasGRP1 isoforms with an intact GEF domain. The major problem resides in the regulatory C-terminal half of the protein. The most prominent abnormal RasGRP1 isoform identified in our SLE patients was splice variant A, which lacks exon 11. Loss of this exon does not cause a frame-shift abnormality or a premature translation termination codon in the processed transcript. Nevertheless, if translated, the resulting abnormal RasGRP1 protein loses the 35-mer sequence of PLTPSKPPVVVDWASGVSPKPDPKTISKHVQRMVD that links the functional GEF domain to the regulatory DAG/phorbol ester- and calcium-binding domains. This amino acid sequence is present in human, chimpanzee, rhesus monkey, bovine, rat, and mouse RasGRP1. Because of its conservation throughout evolution, it is likely that loss of this 35-mer domain leads to a defective signaling protein in lymphocytes.
Nine of the other RasGRP1 splice variants identified in our study have a premature translation termination codon. If translated, these splice variants would encode truncated nonfunctional RasGRP1 isoforms that have lost as many as 624 of their C-terminal amino acids. Numerous studies have documented the biologic importance of the C-terminal half of RasGRP1 and its family members. The most thoroughly studied domain in this portion of RasGRP1 is its DAG/phorbol ester-binding sequence. Activation of T cells via their TCRs result in the rapid phospholipase γ1-dependent generation of DAG. DAG binds to the C1 domain in RasGRP1, thereby promoting translocation of the signaling protein to the inner leaflet of the cell’s plasma membrane and then to the Golgi complex and endoplasmic reticulum. A naturally occurring RasGRP4 splice variant in the mast cells of the C3H/HeJ mouse strain resembles the RasGRP1 splice variant L noted in Fig. 2. We previously showed that this RasGRP4 isoform is unresponsive to phorbol esters (24). Thus, even if translated, RasGRP1 splice variants C, D, G, H, I, K, L, and M should result in RasGRP1 isoforms that are unresponsive to DAG and phorbol esters. Okamura et al. (39) recently reported that the 127-mer amino acid sequence downstream of the DAG/phorbol ester-binding C1 domain in RasGRP3 also helps regulate the intracellular movement of this signaling protein inside cells. If the corresponding domain in RasGRP1 has a similar intracellular targeting function, splice variants B, E, and F also should not be fully functional in SLE patients even if translated.
We generated RasGRP1-specific Abs to evaluate the consequences of defective RasGRP1 transcripts on the levels of normal RasGRP1 in T cells. After documenting the specificity of these anti-peptide Abs (Fig. 3), we used an immunoblot approach to monitor RasGRP1 protein levels in the T cells purified from SLE patients and healthy individuals. In confirmation of the RNA data noted in Fig. 1, SLE patients 8 and 9 contained very little, if any, ∼95-kDa RasGRP1 protein in lysates of their T cells (Fig. 4). The only surprise was our failure to detect smaller sized proteins in the other SLE patients that would be expected to be derived from the relatively abundant splice variants A, B, or D. Based on this observation, we next transfected HEK-293 cells with expression constructs that encode normal RasGRP1 and its splice variants A, B, and D (Fig. 5). The appropriate RasGRP1 transcripts were found in all four transfectants. If anything, the level of the splice variant B transcript in its transfectant was ∼3-fold higher than the level of the normal RasGRP1 transcript in its transfectant. Despite the high levels of splice variant B and D mRNA in their respective transfectants, no recombinant proteins were found in the lysates of these cells. Although it is possible that RasGRP1 spice variants B and D are metabolized in T cells and HEK-293 cells differently due to the fact that epithelial cells normally never express RasGRP1, our transfection/expression data and cell-free transcription:translation data suggest that many defective RasGRP1 transcripts are not efficiently translated in T cells by a control mechanism that remains to be identified. Whatever the mechanism, the net result is greatly diminished levels of functional RasGRP1 protein in the patient’s T cells as noted in SLE patients 8 and 9 (Fig. 4,A). Although the functional importance of having splice variants that encode abnormal isoforms of RasGRP1 remains to be determined, the T cells isolated from RasGRP1-defective SLE patient 8 were unable to produce significant amounts of IL-2 when activated (Fig. 4 E). These preliminary data suggest that some alternatively spliced RasGRP1 isoforms and/or lower expression levels of RasGRP1 can profoundly alter T cell function. Considering the importance of RasGRP1 in lymphocyte development and function in the mouse, our data suggest that dysregulation of RasGRP1 expression is a contributing factor in the development of autoimmunity in a subset of SLE patients.
The authors have no financial conflict of interest.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
This work was supported by the Japanese Ministry of Health, Labor, and Welfare, the Japanese Ministry of Education, Culture, Sports, Science, and Technology, the Japanese Society for the Promotion of Science, and the National Institutes of Health (grant AI-54950).
Abbreviations used in this paper: SLE, systemic lupus erythematosus; DAG, diacylglycerol; GEF, guanine nucleotide exchange factor; KLH, keyhole limpet hemocyanin; PSL, prednisolone; qPCR, quantitative PCR; RasGRP, Ras guanyl nucleotide-releasing.