| Nucleic Acids Research | Pages |
Cryptic intron activation within the large exon of the mouse polymeric immunoglobulin receptor gene: cryptic splice sites correspond to protein domain boundaries
Introduction
Materials And Methods
Plasmid construction
Transfection and RNA preparation
RT-PCR analysis
Results
Identification of cryptic splice sites
Mechanism of cryptic splice activation
An improved 3[prime] splice site at the deletional junction
Co-activation of cryptic 5[prime] and 3[prime] splice sites
Comparison of cryptic splice site locations with protein domains
Discussion
Acknowledgements
References
Cryptic intron activation within the large exon of the mouse polymeric immunoglobulin receptor gene: cryptic splice sites correspond to protein domain boundaries
Received May 25, 1999; Revised and Accepted July 21, 1999
ABSTRACT The fourth exon of the mouse polymeric immunoglobulin receptor (pIgR) is 654 nt long and, despite being surrounded by large introns, is constitutively spliced into the mRNA. Deletion of an 84 nt sequence from this exon strongly activated both cryptic 5[prime] and 3[prime] splice sites surrounding a 78 nt cryptic intron. The 84 nt deletion is just upstream of the cryptic 3[prime] splice site; the cryptic 3[prime] splice site was likely activated because the deletion created a better 3[prime] splice site. However, the cryptic 5[prime] splice site was also required to activate the cryptic splice reaction; point mutations in either of the cryptic splice sites that decreased their match to the consensus splice site sequence inactivated the cryptic splice reaction. The activation and inactivation of these cryptic splice sites as a pair suggests that they are being co-recognized by the splicing machinery. Interestingly, the large fourth exon of the pIgR gene encodes two immunoglobulin-like extracellular protein domains; the cryptic 3[prime] splice site coincides with the junction between these protein domains. The cryptic 5[prime] splice site is located between protein subdomains where an intron is found in another gene of the immunoglobulin superfamily.
INTRODUCTION
Vertebrate internal exons are generally between 50 and 300 nt in length and fewer than 1% are larger than 400 nt (1,2). The exon definition model proposes that this size limitation is functionally relevant to the splicing mechanism and that small nuclear ribonucleoproteins (snRNPs) bound to splice sites on both sides of an exon must interact as a first step towards assembling mature spliceosomes on the pre-mRNA (3). It is presumed that smaller exons pose a problem for the splice machinery because factors that bind to the 3[prime] and 5[prime] splice sites flanking the exon would be sterically hindered from binding simultaneously. Indeed, exons smaller than ~50 nt are not efficiently recognized by the splicing machinery and are often regulated alternative exons (e.g. 4-6). Internal exons may have an upper size limit because the interactions between the 5[prime] and 3[prime] splice site binding factors could be compromised by distance, thus not allowing stable spliceosome assembly. To identify the upper limit of internal exon size, the internal exon of three-exon minigene constructs were expanded with random DNA fragments. Exons up to 1400 nt were found to be efficiently incorporated into the mature mRNA when flanked by introns that were less than ~500 nt in length; large exons surrounded by large introns were excluded from the mRNA (7,8). These observations expand the exon definition model to suggest that snRNPs can interact and stabilize each other across an exon or an intron, depending upon which of the two is shorter, and indicate that distances between splice sites greater than ~500 nt can decrease splicing efficiency (8).
Only a few genes that contain large internal exons, all of which are alternatively processed, have been studied in any detail. These include the BRCA1 gene, which contains a 3.5 kb exon (9) and the neural cell adhesion molecule (NCAM) gene which has an 800 nt exon (10). Both of these exons can be included or excluded from the final mRNA product (9-11). It has been shown that alternative splicing of the large NCAM exon requires a suboptimal 5[prime] splice site; replacing this site with a strong 5[prime] splice site resulted in constitutive inclusion of this large exon (12). Also, sequences within the upstream common exon contribute to tissue-specific splicing of this large exon (13). The caldesmon gene contains a large internal exon that is alternatively spliced at competing 5[prime] splice sites to include either a 403 nt exon (internal 5[prime] splice site used) in most cells or a 1090 nt exon (terminal 5[prime] splice site used) in smooth muscle cells. Four long purine-rich sequences between the two splice sites were shown to enhance use of the internal 5[prime] splice site, but how the 1090 nt exon is spliced in smooth muscle cells has not been addressed (14).
The polymeric immunoglobulin receptor (pIgR) protein transports polymeric immunoglobulins, primarily IgA, from the basolateral to the apical surface of epithelial cells (reviewed in 15). pIgR is a member of the immunoglobulin (Ig) superfamily of structurally-related cell surface proteins. Typically, the extra-cellular domains of these proteins can be divided into specific Ig-like domains which are each encoded by a single exon. However, there are several exceptions to this `one domain/one exon' rule (reviewed in 16,17). The CD4 cell surface protein, the peripheral myelin protein Po and the NCAM genes each contain Ig-like domains encoded by two exons; the intron position varies among these genes (11,18,19). In contrast, there is only one known example of two extra-cellular Ig-like protein domains encoded by a single exon; this is the 654 nt fourth exon of the pIgR gene (20-24).
The exon definition model would suggest that pIgR exon 4 might be too large to be efficiently spliced. However, this exon is constitutively spliced into the mRNA in mice, humans, rats and cows (25-28). In rabbits, this exon is alternatively spliced by being either included or excluded from the mRNA (22). Large constitutively spliced exons have not been examined previously to identify features that contribute to their efficient splicing despite potential constraints due to exon size. However, as other suboptimal exons have been shown to require exonic splice enhancer (ESE) sequences that are bound by SR proteins (29-31), it is possible that the 654 nt pIgR exon 4 may also contain similar types of sequences. To begin a search for ESE-like elements that contribute to the efficient splicing of this large exon, we made several deletions within pIgR exon 4. We report here that, unexpectedly, one of these deletions strongly activated a cryptic splice reaction that divided the large exon into two smaller exons. We have identified the cryptic 5[prime] and 3[prime] splice sites that are used, have addressed possible explanations for why this deletion has activated this cryptic splice reaction, and explored the relationship between the cryptic 5[prime] and 3[prime] splice sites. These splice sites are always activated and inactivated together. Interestingly, the cryptic 3[prime] splice site maps to the junction of the two Ig-like protein domains encoded by exon 4. In addition, the activated cryptic 5[prime] splice site maps to a location that contains an authentic intron interrupting an Ig-like protein domain in the CD4 gene (18). The location of the cryptic splice sites may have implications for the evolutionary origins of the large pIgR exon.
MATERIALS AND METHODS
Plasmid construction
An 18 kb genomic DNA clone containing exons 2-5 of the mouse pIgR gene was obtained by screening a 129SV mouse genomic library (Stratagene, La Jolla, CA) with a cDNA probe encoding mouse pIgR mRNA (28). A 1.4 kb EcoRI-HindIII fragment containing pIgR exon 4 and portions of introns 3 and 4 was subcloned into the KpnI site of the DdBsm(H) plasmid (32) to make the chimeric Dd-pIgR gene (Fig. 1A). All mutations within the pIgR sequences were constructed in a pGEM7 subclone of the 1.4 kb EcoRI-HindIII fragment and subsequently cloned into the KpnI site of the DdBsm(H) plasmid. Dd-pIgR[Delta]Hc was constructed by removing an 84 bp HincII fragment from pIgR exon 4. Dd-pIgRHc(-<) was constructed by reinserting the 84 bp HincII fragment back into pIgR exon 4 in the opposite orientation. The mutation to recreate the sequence at the HincII deletional junction, Hc3, the mutation in the cryptic 3[prime] splice site, [Delta]Hc(3[prime]GA), and the mutation in the cryptic 5[prime] splice site, [Delta]Hc547D, were created using the megaprimer mutagenesis protocol (33). The mutagenic oligonucleotides used were: 5[prime]-CT-AACTGTATTCTATGTCGACCTCCAGGTACCTAACGCAC-AATGATG-3[prime], 5[prime]-AGGCGCTAGCAGTCGGAGGTGRACAT-3[prime] and 5[prime]-TGGGGTTCACATTCTCAGTAG-3[prime], respectively. The Hc3 oligo created a single nucleotide insertion in addition to four nucleotide changes. All mutations were confirmed by DNA sequencing.
Figure 1. Map of chimeric Dd-pIgR gene. (A) The 654 bp pIgR exon 4 and part of its surrounding introns were cloned into the third intron of the Dd gene at the KpnI (K) site. Open box, pIgR exon 4; filled boxes, Dd exons; hatched boxes, transcriptional control regions from the SV40 enhancer and Igk promoter. The sizes of the pIgR exon and surrounding introns are shown. (B) PCR strategy to identify spliced products from the chimeric Dd-pIgR gene. Arrows indicate the location of the PCR primers within Dd exons 3 and 4. The sizes of the PCR products derived from the possible RNAs are shown.
The µAVWT.BSC construct was provided by Dr Tom Cooper (Baylor College of Medicine) and contains a truncated and inefficiently spliced avian sarcoma virus env intron that is dependent on an ESE in the downstream exon in order to be spliced in vivo (34). The 84 bp pIgR HincII fragment was inserted in both orientations between BglII and SpeI restriction sites within the downstream exon of µAVWT.BSC to produce µAVHc(->) and µAVHc(-<). The GAR4 and MGAR4 plasmids are µAV derivatives that contain four copies of a synthetic wild-type ESE (GAR4) or mutant ESE (MGAR4) in the downstream exon and are positive and negative control plasmids provided by Dr Cooper.
Transfection and RNA preparation
HepG2 human hepatoma cells were grown in Dulbecco's modified Eagle's medium/F12 medium (1:1) supplemented with 10% fetal bovine serum (both from Life Technologies), 10 µg/ml insulin (Sigma) and 50 U/ml penicillin/streptomycin (Life Technologies). Transfections were performed using a calcium phosphate procedure (35). The DNA precipitate was removed after 6 h and the RNA harvested 48 h after transfection. Total cellular RNA was prepared from the transfected cells using a hot phenol RNA protocol (35). HT-29 human intestinal epithelial cells were treated with 100 U/ml interferon-[gamma] to increase levels of pIgR mRNA as described (36); RNA was harvested from HT-29 cells and mouse liver cells using the TRIzol reagent according to the manufacturer's protocol (Life Technologies).
RT-PCR analysis
One microgram of total cellular RNA was reverse transcribed using oligo dT as a primer and BRL Superscript Reverse Transcriptase as per the manufacturer's instructions. The RT reaction was incubated at 42°C for 2 min then 1 µl of enzyme was added. The reaction was continued at 42°C for an additional 50 min. For the Dd constructs PCR analysis was performed using 5 µl of the RT reaction, 20 pmol of primers to Dd exon 3 (5[prime]-AGGCTGGTGCTGCAGAGA-3[prime]) and Dd exon 4 (5[prime]-CC-AGGTCAGGGTGATGTC-3[prime]) and BRL Taq polymerase as per the manufacturer's instructions. The cDNAs were amplified for 25 cycles of 94°C for 1 min, 58°C for 1 min and 72°C for 2 min. PCR analysis of the endogenous pIgR mRNA was performed with 20 pmol of primers to pIgR exon 3 (5[prime]-GATGTCAGCCTGGAGGTCA-3[prime]) and pIgR exon 5 (5[prime]-AYGGCCA-CAGAGCYTCCTG-3[prime]). cDNA amplification was carried out for 25 cycles of 94°C for 1 min, 55°C for 1 min and 72°C for 2 min. For the µAV constructs, the forward PCR primer was 5[prime]-CATTCACCACATTGGTGTGC-3[prime] and the reverse primer was 5[prime]-GGTTGGCAAAGTGATCCTAGACTAG-3[prime]. Amplification was for 25 cycles of 94°C for 1 min, 50°C for 1 min and 72°C for 2 min. Southern blot analysis was performed following standard protocols and a 32P random-prime labeled probe (Boehringer-Manheim kit) containing mouse pIgR exon 4-exon 5 genomic sequences.
PCR products were sequenced after being cloned into the SmaI site of pGEM 4Z. Sequencing was performed using the Sequenase Kit as per the manufacturer's instructions using primers for the SP6 or the T7 promoter.
RESULTS
Identification of cryptic splice sites
The 654 nt exon 4 from the mouse pIgR gene is constitutively spliced into the mRNA despite its large size. To determine whether sequences within the mouse pIgR exon 4 and part of its surrounding introns were sufficient to direct efficient exon 4 splicing, we placed these sequences into a chimeric context. A fragment containing pIgR exon 4, 262 bp of upstream intron and 420 bp of downstream intron was placed within the third intron of the mouse Dd MHC class I gene (Fig. 1A). The introns flanking pIgR exon 4 are large, both in the pIgR gene, as well as in the chimeric Dd-pIgR gene; in the pIgR gene this exon is flanked by introns of 2592 and 1237 nt (21) while in Dd-pIgR it is flanked by 814 and 1537 nt introns. It was important to maintain large surrounding introns since internal exons larger than ~500 nt were inefficiently spliced only when surrounded by introns that were larger than ~500 nt (8). RNA from the chimeric Dd-pIgR gene was analyzed by RT-PCR using primers for Dd exons 3 and 4; full inclusion of pIgR exon 4 would result in a PCR product of 878 bp and if the exon were skipped, a 224 bp product would be seen (Fig. 1B). Any PCR products of intermediate sizes would indicate partial exon 4 inclusion by use of cryptic splice sites. We have chosen to use the HepG2 human liver cell line in these studies because they transfect well, have been used for other RNA processing studies in our laboratory (32 and unpublished results), and the endogenous mouse pIgR gene is expressed in the liver (28). When the chimeric Dd-pIgR construct was transiently expressed in HepG2 cells, the large pIgR exon was found, by Southern blot of the RT-PCR products and by S1 analysis (data not shown), to be efficiently spliced into the mRNA by RT-PCR (Fig. 2A); there was no evidence of exon skipping or cryptic splice site use. Thus, sequences that direct efficient splicing of this large exon are contained within this pIgR gene fragment even in the context of a heterologous gene.
Figure 2. A deletion within pIgR exon 4 activates a cryptic splice reaction. (A) RT-PCR analysis of RNA from HepG2 cells transfected with Dd-pIgR and Dd-pIgR[Delta]Hc; M, marker lane with sizes shown on the left. The sizes of the PCR products are shown on the right. (B) Structure of the spliced RNAs from Dd-pIgR and Dd-pIgR[Delta]Hc, based on DNA sequence analysis of the RT-PCR products. [Delta]Hc denotes the 84 bp deletion made between the two HincII sites in pIgR exon 4. The nucleotide numbers shown below the RT-PCR diagrams at the splicing and deletional junctions correspond to the numbering in the mouse pIgR cDNA sequence, accession number U06431 (28). The 794 bp RT-PCR product is from the correctly spliced pIgR[Delta]Hc exon; the vertical line indicates the [Delta]Hc deletional junction. The 716 bp product results from the use of cryptic 5[prime] and 3[prime] splice sites, shown above the pIgR exon 4 as (5[prime]) and (3[prime]); the vertical line in the PCR product diagram indicates this cryptic splice junction. (C) Comparison of the sequence surrounding the cryptic intron within pIgR[Delta]Hc exon 4 with the wild-type sequence of pIgR; nucleotide numbers correspond to those in the cDNA sequence (28). Dashes indicate deleted sequence between the HincII restriction sites; bold nucleotides at 547 and 708 are the Gs at the intron/exon boundaries of the cryptic 5[prime] and 3[prime] splice sites.
To begin to identify sequences that contributed to efficient exon 4 splicing, we constructed several small deletions within the pIgR exon 4. While most of the deletions did not have a striking effect on Dd-pIgR processing in HepG2 cells (manuscript in preparation), a deletion that removed an 84 bp HincII fragment dramatically affected processing of the pIgR exon (Fig. 2B). When RNA from HepG2 cells transiently transfected with Dd-pIgR[Delta]Hc was analyzed by RT-PCR, two bands were observed (Fig. 2A). The larger band of 794 bp is the size expected if the pIgR[Delta]Hc exon were fully spliced (878 minus the 84 bp deletion). The more abundant smaller band is likely to be from an mRNA that has used a cryptic splice site within exon 4. Both cDNA products were isolated and sequenced. This confirmed that the 794 bp cDNA product corresponded to full inclusion of the pIgR[Delta]Hc exon 4. Surprisingly, the smaller cDNA product (716 bp) resulted from the use of both a cryptic 5[prime] splice site and a cryptic 3[prime] splice site within the exon (Fig. 2B and C). The cryptic 5[prime] splice site is located 158 nt into pIgR exon 4, at nt 547, while the cryptic 3[prime] splice site is located 78 nt farther downstream in the pIgR[Delta]Hc exon 4, at nt 708 (this site is 320 nt into the wild-type pIgR exon 4). The use of both cryptic splice sites divides the large pIgR[Delta]Hc exon into two smaller exons of 158 and 334 nt, removing a 78 nt cryptic intron. The 84 bp HincII fragment was originally located between the cryptic 5[prime] and 3[prime] splice sites; deletion of this fragment has strongly activated the use of this cryptic splice reaction.
There was no evidence that pIgR exon 4 was either skipped or cryptically spliced from the chimeric Dd-pIgR gene containing full-length exon 4; cryptic splice sites were activated only by the [Delta]Hc deletion (Fig. 2A). To determine whether the cryptic splice sites we have identified are used to any measurable extent in the endogenous pIgR gene, we analyzed RNA samples from a human intestinal epithelial cell line and from mouse liver for pIgR gene expression by RT-PCR. Oligonucleotides corresponding to regions in pIgR exons 3 and 5 that are conserved between the mouse and human pIgR genes were used for the PCR analysis (Fig. 3C). By RT-PCR, only full inclusion of exon 4 was seen in both mouse and human RNA samples (Fig. 3A). Southern blot analysis of the RT-PCR products confirmed that cryptic splice sites are not used in the endogenous pIgR gene (Fig. 3B). We also saw no evidence that this exon was spliced out from the endogenous human or mouse pIgR mRNA as it is from the rabbit pIgR mRNA, confirming previous northern blot analysis results (26,28).
Figure 3. Endogenous mouse and human pIgR mRNA is not cryptically or alternatively spliced. (A) RT-PCR analysis of mouse and human RNA with primers in pIgR exons 3 and 5. M indicates marker lanes. (B) Southern blot of the RT-PCR reactions, probed with mouse pIgR exon 4-exon 5 genomic sequences. (C) Diagram of the pIgR gene, the primer locations, and the sizes of the possible PCR products.
Mechanism of cryptic splice activation
There are at least three possible explanations for why the 84 nt exon deletion resulted in predominant use of both a cryptic 5[prime] splice site and a cryptic 3[prime] splice site. One possibility is that the HincII fragment contains positive-acting sequences that promote full inclusion of exon 4, possibly by binding factors that would help bridge interactions between factors bound at the authentic 3[prime] and 5[prime] splice sites. In the absence of these sequences, the splicing machinery is not able to splice the large exon and therefore splits it into two smaller exons using the cryptic sites. A second possibility is that sequences within the HincII fragment act as a repressor of either or both of these cryptic sites; in the absence of these sequences the cryptic sites can be recognized as functional splice sites. A third possibility is that a better 3[prime] splice site was created by joining the two sequences on either side of the deletion. This improved 3[prime] splice site could then be recognized as a bona fide splice site; the cryptic 5[prime] splice site may have been activated to maintain pairing of splice sites. We performed several different experiments to distinguish among these possibilities. One way to determine whether sequences within the HincII fragment act to enhance splicing is to place the fragment into a gene that contains an intron whose splicing requires ESE activity. We used the µAV plasmid which contains two exons separated by the inefficiently spliced intron of the avian sarcoma virus env intron; an ESE in the downstream exon is required for efficient splicing. This gene has been used previously to measure heterologous ESE activity; GAR4 and MGAR4 contain four copies of either an ESE or a mutant ESE in the downstream exon and are positive and negative controls, respectively, for activation of µAV splicing (14,34). If the HincII fragment contains a similar ESE-like activity, we would expect to see enhanced splicing of the µAV intron in the presence of the HincII fragment, but not when the fragment is placed in the reverse orientation. The HincII fragment in this construct did not consistently activate splicing in either orientation (Fig. 4A). These data suggest that this region does not likely contain classical ESE sequences that enhance exon inclusion in this heterologous context.
Figure 4. Potential mechanisms of cryptic splice site activation. (A) The HincII fragment does not activate efficient splicing in a heterologous context. RT-PCR analysis of the HincII fragment in both orientations in the µAV construct, Hc(->) and Hc(-<); GAR4 and MGAR4 are positive and negative controls, respectively. GAR4 contains four copies of a purine-rich ESE that activates splicing of the µAV RNA; MGAR4 contains four copies of a mutant ESE that is unable to activate µAV splicing. M indicates marker lanes. (B) The HincII fragment inserted into pIgR exon 4 in the opposite orientation does not activate the cryptic splice reaction. RT-PCR analysis of RNA from HepG2 cells transfected with Dd, Dd-pIgR and Dd-pIgRHc(-<); the PCR products are shown schematically in Figure 1B. M indicates marker lanes.
Another way to address whether specific sequences that promote full exon inclusion or that repress the use of the cryptic sites are present within the HincII fragment is to place it in its natural location in pIgR exon 4, but in the reverse orientation. This will replace the original sequence with a totally different sequence while maintaining a constant exon size. If the HincII fragment contains specific active sequences, they will not be present when the fragment is in the opposite orientation, so the cryptic sites should still be used. On the other hand, if the deletion has brought together sequences that create a better splice site, these will be interrupted by the sequence in the opposite orientation and the cryptic splice sites should no longer be used. By RT-PCR analysis we observed total exon inclusion and no cryptic splice site use from Dd-pIgRHc(-<) transiently expressed in HepG2 cells (Fig. 4B). This suggests that loss of sequences within the 84 bp HincII fragment has not activated the use of the cryptic sites, but rather, the deletion has brought together sequences that activate splicing.
An improved 3[prime] splice site at the deletional junction
Taken together, the experiments described above suggest that the cryptic sites are activated because the deletion has created a better 3[prime] splice site. A consensus 3[prime] splice site consists of a pyrimidine tract of 11 or more nucleotides, followed by NYAG at the intron/exon boundary; the pyrimidine tract is found downstream of the branch point which has the consensus YNYYRAY where Y represents pyrimidine; R, purine and N, any nucleotide (37-39). The sequence just upstream from the cryptic 3[prime] splice site in the wild-type pIgR exon 4 contains only nine pyrimidines in the 24 nt (38% pyrimidine) between the NYAG and a potential branch point sequence with only a 4 out of 7 nt match to consensus (Fig. 5A). The [Delta]Hc mutation however, brings a better polypyrimidine tract and two possible 6 out of 7 nt matches to the branch point consensus sequence adjacent to the cryptic 3[prime] splice site (Fig. 5A). In [Delta]Hc, 12 out of 18 nt between the NCAG and the potential branch points are pyrimidines (67% pyrimidine). Since both the length and pyrimidine content of the polypyrimidine tract are important determinants for 3[prime] splice site recognition (40,41), this improvement of the cryptic 3[prime] splice site sequence, in combination with a good branch point, may allow for more efficient recognition by splicing factors. Also, in conjunction with the authentic downstream 5[prime] splice site, use of the cryptic 3[prime] splice site defines a shorter 334 nt exon that may be more easily recognized by the splicing machinery than the full-length 654 nt exon.
Figure 5. The deletion in [Delta]Hc has created an improved 3[prime] splice site sequence within the pIgR exon 4. (A) Sequences surrounding the cryptic 3[prime] splice site in [Delta]Hc and [Delta]Hc(3[prime]GA) are shown compared to the sequence of the undeleted pIgR exon (top) and the authentic 3[prime] splice site upstream from exon 4 (bottom). The branch point and 3[prime] splice site consensus sequences are shown: Y, pyrimidine; R, purine; N, any nucleotide (37-39). Underlined nucleotides, pyrimidines between potential branch points and the 3[prime] splice sites; overlined bases and *, potential branch point sequences based on match to the consensus; outlined AG, dinucleotide at intron/exon boundaries; bold nucleotides in [Delta]Hc(3[prime]GA), nucleotides that were changed to create this mutation. The numbers shown above the nucleotides correspond to those in the mouse pIgR cDNA sequence as in Figure 2. (B) Mutation made to reconstruct the sequence at the HincII deletional junction within the pIgR exon. The open box represents the sequence just downstream from nt 615, which is shown below; the hatched box represents the sequence just downstream from nt 700, which is shown below. The bold AG is the AG recognized as the cryptic splice in [Delta]Hc. In pIgRHc3, the sequence downstream from nt 615 has been changed to the sequence downstream from nt 700; mutated nucleotides are underlined. (C) The mutated sequence in pIgRHc3 is efficiently recognized as a cryptic 3[prime] splice site. RT-PCR analysis of RNA from HepG2 cells transfected with Dd-pIgR and Dd-pIgRHc3; the RT-PCR product of pIgR is shown schematically in Figure 2B; the smaller product made by pIgRHc3 is a result of a cryptic splice reaction between the cryptic sites at nt 547 and 623 (the new site created in the Hc3 mutation). M indicates the marker lane and sizes of the markers and RT-PCR products are shown.
To test whether the deletional junction created an improved cryptic 3[prime] splice site sequence, we reconstructed this sequence in the context of the wild-type pIgR exon, in pIgRHc3, and we mutated the cryptic 3[prime] splice site in [Delta]Hc away from consensus in [Delta]Hc(3[prime]GA) (see below). If the change in sequence due to the HincII deletion activated splicing, then recreating the sequence by site-directed mutagenesis should also activate cryptic splicing. The 9 nt just downstream from the 615 Hc site were altered to match the sequence just downstream from the 700 Hc site to make pIgRHc3 (Fig. 5B). Thus, this sequence is duplicated, but only the sequence upstream of the 615 Hc site contains the better pyrimidine tract and branch point sequences. By RT-PCR analysis, a single band, smaller than the full length pIgR product is seen from pIgRHc3 transiently expressed in HepG2 cells (Fig. 5C). By sequence analysis, this band was identified to be the result of RNA spliced between the 547 cryptic 5[prime] splice site and the new cryptic 3[prime] splice site created around the 615 Hc site.
Co-activation of cryptic 5[prime] and 3[prime] splice sites
The cryptic 3[prime] splice site in [Delta]Hc does not appear to be as strong as the authentic 3[prime] splice site at the 5[prime] end of the large pIgR exon, which contains a very good polypyrimidine tract (17 out of 20 nt are pyrimidines; 85% pyrimidine) as well as a 6 out of 7 nt match to the branch point consensus sequence (Fig. 5A). Thus, the cryptic site is probably not able to compete directly with the authentic site for splicing to a common upstream 5[prime] splice site. Rather, both splice sites are recognized in the pIgR [Delta]Hc exon; for this to occur, a suitable 5[prime] splice site must be located downstream from the authentic 3[prime] splice site and upstream from the cryptic 3[prime] splice site to maintain proper pairing of splice sites. The cryptic 5[prime] splice site chosen in the cryptic splice reaction is located 158 nt into the exon, 78 nt upstream from the cryptic 3[prime] splice site, and is the best match to the 5[prime] splice site consensus sequence in this region.
If the [Delta]Hc mutation has created a better 3[prime] splice site by improving the branch point site and the pyrimidine tract, and this in turn is what activates the cryptic 5[prime] splice site, then it should be possible to decrease use of the cryptic splice reaction by weakening the cryptic 3[prime] splice site. To test this, we mutated the cryptic 3[prime] splice junction of [Delta]Hc from AG to GA [Figs 5A and 6A, [Delta]Hc(3[prime]GA)]. While the AG is an essential sequence for the splice reaction, it has been shown that the first AG downstream from the branch point sequence is usually chosen as the 3[prime] splice site (42-44). Since there is an AG 6 nt downstream from the AG used as the cryptic 3[prime] splice site in the pIgR[Delta]Hc exon, this mutation is likely to weaken, but not necessarily eliminate use of this site. In [Delta]Hc(3[prime]GA), the sequence between the potential new NYAG and the branch point sequence is 68% pyrimidine, similar to that in [Delta]Hc (Fig. 5A). However, the 3[prime]GA mutation completely eliminated use of the cryptic splice reaction (Fig. 5B); the pIgR exon in Dd-pIgR[Delta]Hc(3[prime]GA) is fully spliced into the mRNA and there is no evidence of cryptic site use. Thus, the cryptic 5[prime] splice site at nt 547 is also activated and inactivated by changing the sequence at the cryptic 3[prime] splice site.
Figure 6. The cryptic 5[prime] and 3[prime] splice sites are co-inactivated. (A) Mutations made to the cryptic 5[prime] and 3[prime] splice sites. The cryptic 3[prime] splice site AG at nt 708 and 709 were mutated to GA in [Delta]Hc(3[prime]GA). Another AG dinucleotide is located 6 nt downstream at nt 715. [Delta]Hc547D contains a G to T change at the -1 position of the cryptic 5[prime] splice site at nt 547 which decreases the match of this site to the consensus 5[prime] splice site sequence; the sequence of the cryptic 5[prime] splice site and mutated splice site is shown. (B) Mutating the cryptic 3[prime] splice site from AG to GA eliminates the cryptic splice reaction. RT-PCR analysis of RNA from HepG2 cells transfected with Dd-pIgR[Delta]Hc and Dd-pIgR[Delta]Hc(3[prime]GA); the RT-PCR products are shown schematically in Figure 2B. M indicates marker lanes and sizes of the markers and RT-PCR products are shown. (C) Mutating the cryptic 5[prime] splice site away from the consensus sequence substantially inactivates the cryptic splice reaction. RT-PCR analysis of RNA from HepG2 cells transfected with Dd-pIgR[Delta]Hc and Dd-pIgR[Delta]Hc547D; the RT-PCR products are shown schematically in Figure 2B. Lane 3 was loaded with five times more RT-PCR reaction than was loaded in lane 2. The 665 nt band in lane 3 is due to the weak activation of an upstream cryptic 5[prime] splice site at nt 497. M indicates marker lanes and sizes of the markers and RT-PCR products are shown.
To further examine the connection between the two cryptic splice sites, we decreased the match of the 547 cryptic 5[prime] splice site to the consensus 5[prime] splice site sequence at the -1 position in [Delta]Hc547D (Fig. 6A). Several outcomes from this mutation are possible: another sequence that is a suitable 5[prime] splice site may be activated; if there is no other suitable splice site, the cryptic 3[prime] splice site may be put in direct competition with the authentic 3[prime] splice site for splicing to the upstream Dd exon, or the cryptic splice reaction may be inactivated as it was for [Delta]Hc(3[prime]GA). In fact, it is this last possibility that was seen predominantly when Dd-pIgR[Delta]Hc547D was transiently expressed in HepG2 cells; most of the RNA was no longer spliced at the cryptic sites (Fig. 6C, lane 2). A small amount of a shorter RT-PCR product was seen when 5-fold more reaction volume was loaded onto the gel (Fig. 6C, lane 3); this product was identified by sequencing to be the result of splicing between a new cryptic 5[prime] splice site at nt 497 (TGT|GTAAGA) and the cryptic 3[prime] splice site at nt 708. That the cryptic splice reaction is inactivated by mutations in both the cryptic 3[prime] and 5[prime] splice sites suggests that the cryptic splice sites are co-recognized by the splicing machinery.
Comparison of cryptic splice site locations with protein domains
The pIgR protein is composed of a cytoplasmic domain, a membrane spanning region, and five extracellular Ig superfamily protein domains, D1-D5 (Fig. 7A). In most members of the Ig superfamily, each separate extracellular domain is encoded by a single exon of ~300 nt. Exon 4 of the pIgR gene is an exception because it encodes two separate extracellular protein domains, D2 and D3. Interestingly, the cryptic 3[prime] splice site we have identified is located exactly at the junction between these two Ig-like domains (Fig. 7B). Thus, the cryptic 3[prime] splice site divides the large 654 nt pIgR exon 4 into two exon segments of 320 and 334 nt, each encoding one Ig-like domain. The cryptic 5[prime] splice site at nt 547 splits the D2 protein domain in half, resulting in two half domains of 158 and 162 nt (Fig. 7B).
Figure 7. (A) Domain structure of the pIgR protein compared with the exonic structure of the pIgR mRNA. The pIgR protein contains five extracellular Ig superfamily domains D1-D5, membrane-spanning (M) and cytoplasmic domains. Three of the five Ig-like domains are encoded by individual exons; D2 and D3 are encoded by the large exon 4 (21,23,28). (B) Location of the cryptic splice sites within pIgR exon 4 in relation to the coding potential of the exon. The cryptic 3[prime] splice site falls exactly between the D2 and D3 domains and the cryptic 5[prime] splice site divides the D2 domain almost in half. The GT and AG at the cryptic exon/intron boundaries are shown in bold.
DISCUSSION
An 84 nt deletion within the large 654 nt pIgR exon 4, created in a search for cis-acting sequences that contribute to efficient exon splicing, was found to activate cryptic 5[prime] and 3[prime] splice sites within this exon. Use of these cryptic sites generated two smaller exons of 158 and 334 nt, separated by a 78 nt intron. It was possible that this deletion activated the cryptic splice reaction because the deleted fragment contained sequences that either promoted full inclusion of the large exon or that repressed the use of the cryptic sites; in the absence of these sequences, the cryptic sites became active. However, based on our results, it is much more likely that the deletion has created an improved 3[prime] splice site. Indeed, the sequence upstream from the cryptic 3[prime] splice site in the pIgR[Delta]Hc exon is more pyrimidine-rich and contains sequences with a better match to the branch point consensus than the sequence upstream from this site in the wild-type pIgR exon.
While the cryptic 3[prime] splice site was activated because of the improved sequence created by the deletion, a suitable cryptic 5[prime] splice site existed nearby and this was also critical to activation of the cryptic splice reaction. In fact, these two splice sites are activated and inactivated together; both the [Delta]Hc(3[prime]GA) and [Delta]Hc547D mutations, which decrease the match to consensus sequence of the cryptic 3[prime] and 5[prime] splice sites, respectively, individually inactivated the cryptic splice reaction. That the [Delta]Hc547D mutation at the cryptic 5[prime] splice site inactivated the cryptic splice reaction, strongly implies that this site is required for activation of the cryptic reaction; simply creating an improved 3[prime] splice site sequence was not sufficient. The initial unit recognized by the splice machinery has been shown in different situations to be either the exon or the intron. It has been suggested that the relative sizes of the introns and exons, whichever is shorter, determines how the splice sites will be co-recognized (3,8,45). Since the cryptic intron is inactivated by a mutation at either the 5[prime] or 3[prime] cryptic splice sites, it is likely that splicing of this small 78 nt cryptic intron within the large pIgR exon is an example of intron definition. The two exons on either side of the cryptic intron are 158 and 334 nt, within the size range of most internal exons, and thus could be recognized by exon definition. However, if this were the case, the mutations at the cryptic splice sites would have likely caused one of the cryptic exons to be spliced out rather than causing the cryptic intron to be retained.
The locations of the cryptic splice sites activated in pIgR[Delta]Hc, with respect to the coding capacity of the exon, are very intriguing. pIgR exon 4 is unique in that it encodes two of the five extracellular Ig-like protein domains (D2 and D3) of the pIgR protein; most Ig-like protein domains are encoded by individual exons (16,22,46). The cryptic 3[prime] splice site divides the large exon at the boundary of the two Ig-like protein domains, which is the location of authentic introns in most Ig superfamily-encoding genes. In fact, the pIgR exons encoding domains D1, D4 and D5 are 345, 333 and 333 nt, respectively (21,28), very similar in size to the 334 nt cryptic exon. The cryptic 5[prime] splice site divides the D2 Ig domain about in half, between sequences that encode two [beta] sheets; this is the same location of an intron dividing an Ig-like protein domain in the CD4 gene (18). The fact that both the activated cryptic 3[prime] and 5[prime] splice sites in pIgR exon 4 map to locations where introns exist in other genes, suggests that these sites may represent ancient exon/intron boundaries in the pIgR exon. The pIgR genes from five mammalian species have been cloned and all have the same exon-intron structure, including a large fourth exon. A more primitive relative of the pIgR gene, one that might have all Ig-like domains encoded by single or multiple exons, has not been found.
There is no obvious reason why the sequence of the cryptic splice sites should have been preserved in the mouse pIgR exon 4. It is likely fortuitous that the GT and AG required for splicing were retained at the cryptic 5[prime] and 3[prime] splice junctions, respectively, since these sequences have not been preserved in all species. The sequences at the cryptic sites do conform to the `proto-splice site' sequence of C/AAG|G which were proposed to be potential sites of intron integration (47), but is also the sequence that remains after two exons have been joined by splicing (38,48). Thus, these sequences could have been left after splicing of ancient introns from these locations.
It is possible that pIgR exon 4 was generated by fusion of Ig half-domains and then by fusion of adjacent complete Ig domains, each fusion event resulting in the loss of an intron. In order not to drastically affect the function of the pIgR protein, this last fusion event would have been tolerated only if splicing of the now abnormally large exon were not impaired. The fortuitous presence of cis-acting sequences that enhance exon recognition could explain efficient exon 4 splicing. In most mammals, these sequences in pIgR exon 4 presumably have been retained since only the full-length mRNA is made. However, the rabbit makes two mRNAs, one containing and one lacking the large exon. This would argue that sequences in the rabbit pre-mRNA must have diverged so that splicing of the large exon is now less efficient. A protein encoded by pIgR mRNA that lacks the large exon would only have domains 1, 4, and 5 of the extracellular region. In vitro binding studies have demonstrated that all five extracellular domains of human and bovine pIgR are required for high affinity binding of dimeric IgA (49,50). However, rabbit pIgR lacking domains 2 and 3 retains high affinity binding of dimeric IgA, likely because changes in other parts of the protein have compensated for the loss of two of the extra-cellular Ig domains (51,52). Thus, there seem to be two solutions to the problem presented by the formation of an abnormally large exon: (i) to preserve RNA sequences that ensure efficient splicing (as seen in most mammalian pIgR genes), or (ii) to modify other sequences within the gene to compensate for changes in protein activity (as seen in the rabbit pIgR gene).
While the cryptic splice sites chosen are likely to be the `next best' sites, after the authentic splice sites, based upon their match to the consensus sequences, it is surprising that both of these sites fall at locations that contain introns in other Ig superfamily domains. It is possible that some information resides within the exon sequences that not only aids in recognition of the exon, but also contributes to specifying the location of these cryptic splice junctions. Results from the [Delta]Hc(3[prime]GA) and [Delta]Hc547D mutations can be interpreted to support this idea. Since 3[prime] splice sites are usually the first AG downstream from the branch point, mutating the AG at the cryptic 3[prime] splice site could potentially activate splicing to the AG located 6 nt downstream. The pyrimidine content of the sequence in this region is preserved between [Delta]Hc and [Delta]Hc(3[prime]GA) (Fig. 5A) and the branch point sequences and the cryptic 5[prime] splice site remain intact, yet the [Delta]Hc(3[prime]GA) mutation inactivated the cryptic reaction. This suggests either that this mutation has weakened this cryptic 3[prime] splice site below a certain threshold required to be recognized or that, for some reason, the AG at the protein domain boundary is a more favorable splice site than the AG located 6 nt away. Also, when the cryptic 5[prime] splice site was mutated away from the consensus sequence, only a very minor portion of the RNA was spliced to another cryptic 5[prime] splice site 50 nt upstream which, based on the match to the consensus 5[prime] splice site, is only slightly less favorable than the 547 cryptic site. In the presence of the 547D mutation, this upstream cryptic site, which is within a [beta] sheet protein-coding domain, is not able to replace the 547 5[prime] cryptic splice site in the cryptic splice reaction. Thus, it appears that cryptic splice sites can be activated because of their match to consensus, but there also may be other, ill-defined sequences that contribute to use of these sites. If these other sequences can be bound by general splice factors such as SR proteins (29), they may also contribute to efficient splicing of the intact pIgR large exon 4. SR proteins may act as protein-protein bridges to connect the authentic 3[prime] and 5[prime] splice sites that border this abnormally large exon.
One implication of our finding that cryptic splice sites are activated at junctions of protein coding domains is that other alternative splice reactions, either authentic or cryptic, may also frequently occur between protein coding domains or between former exon/exon boundaries. One example is the alternative splice reaction in the caldesmon large internal exon which separates the myosin and calmodulin binding domains from the central helical domain (14). Also, in the experiments to determine the upper limit of internal exon size (8), the internal exon was expanded with hypoxanthine phosphoribosyl-transferase cDNA fragments. As the exon size increased, activation of cryptic splice sites within the cDNA sequence was observed; at least one of these cryptic splice sites appears to be located at an exon/exon border. There are also two examples of transcripts that have been spliced and then respliced at a splice junction formed by the first splice reaction to generate the mature mRNA. Both in the rat AMP deaminase gene (53) and the Drosophila ultrabithorax gene (54) there are alternative RNAs generated by resplicing a partially spliced precursor RNA at a 5[prime] splice site that was formed by an exon/exon junction.
ACKNOWLEDGEMENTS
We wish to thank Kim Phillips and Tim Boze for expert technical assistance with early parts of this work and Drs Brett Spear and Jeffrey Davidson for helpful comments on the manuscript. This work was supported by grants from the National Institutes of Health (CA51998) to C.S.K. and the National Science Foundation (MCB-9507513 and MCB-9808637) to M.L.P.
REFERENCES
*To whom correspondence should be addressed. Tel: +1 606 257 5478; Fax: +1 606 323 2094; Email: mlpete01{at}pop.uky.edu
This page is run by Oxford University Press, Great Clarendon Street, Oxford OX2 6DP, as part of the OUP Journals
Comments and feedback: jnl.info{at}oup.co.uk
Last modification:
Copyright© Oxford University Press, 1999.
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  S. R. BRUCE, R.W. C. DINGLE, and M. L. PETERSON B-cell and plasma-cell splicing differences: A potential role in regulated immunoglobulin RNA processing RNA, October 1, 2003; 9(10): 1264 - 1273. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 S. R. Bruce and M. L. Peterson Multiple features contribute to efficient constitutive splicing of an unusually large exon Nucleic Acids Res., June 1, 2001; 29(11): 2292 - 2302. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||