Expression and Functional Characteristics of Calpain 3 Isoforms Generated through Tissue-Specific Transcriptional and Posttranscriptional Events

Abstract

Calpain 3 is a nonlysosomal cysteine protease whose biological functions remain unknown. We previously demonstrated that this protease is altered in limb girdle muscular dystrophy type 2A patients. Preliminary observations suggested that its gene is subjected to alternative splicing. In this paper, we characterize transcriptional and posttranscriptional events leading to alterations involving the NS, IS1, and IS2 regions and/or the calcium binding domains of the mouse calpain 3 gene (capn3). These events can be divided into three groups: (i) splicing of exons that preserve the translation frame, (ii) inclusion of two distinct intronic sequences between exons 16 and 17 that disrupt the frame and would lead, if translated, to a truncated protein lacking domain IV, and (iii) use of an alternative first exon specific to lens tissue. In addition, expression of these isoforms seems to be regulated. Investigation of the proteolytic activities and titin binding abilities of the translation products of some of these isoforms clearly indicated that removal of these different protein segments affects differentially the biochemical properties examined. In particular, removal of exon 6 impaired the autolytic but not fodrinolytic activity and loss of exon 16 led to an increased titin binding and a loss of fodrinolytic activity. These results are likely to impact our understanding of the pathophysiology of calpainopathies and the development of therapeutic strategies.

Study of calpain 3 received an important impetus after the demonstration of its involvement in limb girdle muscular dystrophy type 2A (Mendelian Inheritance in Man [MIM] 253600) (24). This neuromuscular disorder is characterized mainly by symmetrical atrophy and weakness of proximal limb muscles, by elevated creatine kinase in serum, and by a dystrophic pattern in muscle biopsies (4). Calpains are members of a family of intracellular nonlysosomal cysteine proteases (for reviews, see references 35 and 36). They are comprised of three ubiquitous calpains (μ, m, and μ/m); a skeletal muscle-specific calpain (calpain 3, CAPN3, nCL-1, or p94 [31]), a variant of which is also expressed in a lens-specific manner (17, 18); a digestive tract-specific calpain (nCL-4 [16]); and the stomach-specific calpains (nCL-2 and nCL-2′ [33]).

The human calpain 3 gene was reported to consist of 24 exons spanning approximately 45 kb (24). It encodes a 3.5-kb mRNA expressed predominantly in skeletal muscle tissues. The 821-amino-acid-long calpain 3 protein can be subdivided, like the other calpains, into four domains that include a proteolytic (domain II) and a calcium binding (domain IV) domain (26, 31). In addition, three short calpain 3-specific sequences (NS, IS1, and IS2 [36]) are present. These are located, respectively, at the N terminus, in the protease domain, and between domains III and IV. IS2 includes a titin (connectin) binding site (11, 34) as well as a putative nuclear localization signal (31). Calpain 3 differs from the ubiquitous calpains by its rapid autolysis, at least when it is expressed in COS-7 cells (32). Furthermore, under such conditions, calpain 3 can be detected in the nucleus (32). The IS2 sequence appears to be involved in these phenomena (32). In addition, it was shown that when wild-type calpain 3 is expressed in COS-7 cells, the 230-kDa intrinsic fodrin α subunit (7, 19) is proteolyzed, yielding a 150-kDa fragment (20).

Preliminary observations suggested the existence of alternatively spliced isoforms of the calpain 3 gene. To better understand the different modes of calpain 3 expression, we characterized alternatively spliced products of the mouse capn3 gene. In addition, we examined the impact of the murine capn3 isoforms on the known biochemical properties of calpain 3.

MATERIALS AND METHODS

RNA isolation.

Total cellular RNAs from skeletal muscles of Swiss mice and from NMRI mouse embryos at days 11.5 to 18.5 were isolated by the Fast RNA GREEN method (Bio 101). Total cellular RNA was prepared from primary cultures of human satellite cells, or from C2C12 mouse cell lines, by the guanidinium thiocyanate-CsCl gradient method. Total RNAs from embryonic (day 15.5) and 6-month-old adult mouse lenses and from primary culture of rat satellite cells were obtained by the RNA-Zol method (Bioprob System). Brain, smooth muscle (from small intestine), skeletal muscle, and heart poly(A)⁺ RNAs from 9- to 10-week-old BALB/c mice were purchased from Clontech.

RT-PCR amplification.

One microgram of total RNA was reverse transcribed to single-stranded cDNA with Superscript II reverse transcriptase (Gibco-BRL) and random hexamer primers in a 20-μl volume. The reverse transcription (RT) reaction was conducted at 42°C for 60 min. cDNA was amplified by PCR with calpain 3-specific primers (Table 1). PCR products were separated by agarose and polyacrylamide gel electrophoreses and revealed with ethidium bromide staining. The long-range PCR was performed with the Expand Long Template PCR System (Boehringer Mannheim) with two 25-mer primer pairs (Table 1). PCR amplification of the alternative first exon in mouse cDNA was performed with a primer, 5pRat.a, based on the alternative first exon of rat lens (EMBL accession no. U96367), and a primer, 5p.m, based on the second exon of mouse calpain 3 cDNA (Table 1).

TABLE 1.

Primers for PCR covering the entire coding sequences of the murine capn3 transcripts^a

Primer pairb	Primer sequence (5′–3′)	Position within the ATG of the cDNA	PCR product size (bp)
p94sys1.a	CCTTTATTGCCTCTTCCTCA	−233 to −252	798
p94sys1.m	GCTGATTGTTGTATGTTGGC	546 to 565
p94sys2.a	CGGTTTATCATTGGTGGAGC	328 to 347	572
p94sys2.m	CTGAGCAGCGAGTTATCCAT	880 to 899
p94sys3.a	TTCACCAAATCCAACCACCG	571 to 590	540
p94sys3.mc	ACTCCAAGAACCGTTCCACT	1091 to 1110
p94sys4.a	GCACGAACATGACTTACGGA	806 to 825	555
p94sys4.m	AAGTGTCTGGGAAGTTTCGG	1341 to 1360
p94sys5.a	AGACAAAGATGAGAAGGCCC	1137 to 1156	640
p94sys5.m	GCCGATCCACAGAGATTGTA	1758 to 1777
p94sys5.a	AGACAAAGATGAGAAGGCCC	1137 to 1156	619
p94sys5b.mc	TTGCTGTTCCTCACTTTCCTG	1936 to 1956
S15int.a	CATGGATGGGGAGTTTAGCT	17 to 35 in intron 16	806
p94sys5b.mc	TTGCTGTTCCTCACTTTCCTG	1936 to 1956
p94sys5b.a	GACAGAGCACACAGCAACAA	1819 to 1838	327
S15int.m1	GCTGTTCTTGGCCTATCCTT	232 to 351 in intron 16
p94sys6.a	GCCAGAAGCCAAACCTACAT	1594 to 1613	679
p94sys6.m	GGTCTGTGTCATAGTGCTTG	2193 to 2212
p94sys7.a	GACAGAGCACACAGCAACAA	1819 to 1838	466
p94sys7.m	GTTGGCTGTTGAGATGGAAG	2265 to 2284
p94sys8.a	ACACAAGGGTTCACTCTGGA	2065 to 2084	639
p94sys8.m	GGACAACCCAGTACAAGCTA	2684 to 2703
5pRat.a	ACCGTCTCACAGCTTCTCA	135 to 153 in rat exon 1′	230
5p.m	ATGTCAGTCCTGTTGGCTC	344 to 362
LPCRmo1.a	GCTTTGTAATCGCTTCCTTTCCTTG	−79 to −55	2,640
LPCRmo1.m	TGGAGATGCACCTACTGGGTCTTTG	2537 to 2561
LPCRmo2.a	AATCCAACCACCGCAATGAGTTCTG	578 to 602	1,509
LPCRmo2.m	ACTCCAGCGTGAACCCTTGTGTCTT	2062 to 2086
TFIID.a	ACAGGAGCCAAGAGTGAAGAA	848 to 868	219
TFIID.m	CCAGAAACAAAAATAAGGAGA	1087 to 1107

^aThe position of each oligonucleotide and the predicted length of the corresponding PCR product are based on the published mouse cDNA sequence (EMBL accession no. X92523). 

^b5pRat.a is the primer derived for the published rat Lp82 cDNA sequence (EMBL accession no. U96367). Primers for TFIID amplification were obtained from the published mouse cDNA sequence (EMBL accession no. U63933). 

^cPrimer derived from the human calpain sequence (EMBL accession no. X85030). 

Cloning of the cDNA isoforms.

PCR products were subcloned into the pCRII or pCR2.1 plasmid (Invitrogen) in Escherichia coli XL1 Blue. When appropriate, clones were subjected to sequence analysis with internally specific primers and specific primers of the plasmid, by the dideoxy termination method on Applied Biosystems sequencers.

Quantitative RT-PCR.

Expression of the calpain 3 gene was investigated by a quantitative RT-PCR method with TaqMan probes (Perkin-Elmer). This technique allows real-time detection of PCR products by measuring the increase in fluorescence due to TaqMan probe degradation (8). Fluorescence emission was monitored with a sequence detector (Perkin-Elmer model 7700). The ubiquitous transcription factor TFIID was used to normalize the data across samples. The primer pairs used for amplification were M811CANP3.a (ACAACAATCAGCTGGTTTTCACC) and M954CANP3.m (CAAAAAACTCTGTCACCCCTCC) for calpain 3 and M616TFIID.a (ACGGACAACTGCGTTGATTTT) and M724TFIID.m (ACTTAGCTGGGAAGCCCAAC) for TFIID. The Taqman probes labeled with Tamra and with Fam were M884CAPN3.p (TGCCAAGCTCCATGGCTCCTATGAAG) and M654TFIID.p (TGTGCACAGGAGCCAAGAGTGAAGA) for calpain 3 and for TFIID, respectively.

Expression in COS-7 cells and Western blot analyses.

cDNA constructs corresponding to alternatively spliced isoforms, inserted in the correct orientation into a pSRD vector (37), were transfected into COS-7 cells by electroporation as described previously (2). After 48 h of incubation at 37°C, the transfected cells were harvested and sonicated in lysis buffer (100 mM Tris-HCl [pH 7.5], 10 mM EDTA, 1 mM dithiothreitol). Samples were fractionated by sodium dodecyl sulfate–10% polyacrylamide gel electrophoresis, blotted, and analyzed with antiserum against an α-fodrin peptide, kindly provided by T. C. Saido, and a peptide in the IS2-specific region of calpain 3 (32). The immunological reaction was revealed by peroxidase staining as described previously (20).

Titin binding assay.

cDNAs corresponding to alternatively spliced isoforms were inserted into a pAS2C-1 vector (modified pAS2-1 vector; Clontech) in frame with the GAL4 DNA sequence. Two titin cDNA clones (pCNT-N2 and pCNT-52 [32]) were used for a binding assay in a Saccharomyces cerevisiae two-hybrid system. Each calpain 3 isoform construct was cotransfected with either pCNT-N2 or pCNT-52 into the CG-1945 S. cerevisiae yeast strain by the Li acetate method. Yeast cells were grown for 48 h at 30°C in SD medium with leucine-tryptophan dropout supplement (Clontech). Interaction between calpain 3 isoforms and titin peptides was tested by monitoring (i) the growth capacity of transfected yeast in LW medium lacking histidine in the presence of 1.5 mM or 5 mM 3-amino-1,2,4-triazol and (ii) β-galactosidase activity with a chlorophenol red-β-d-galactopyranoside (CPRG) substrate as described previously (34). Yeast cultures at equal cell densities were used for the β-galactosidase assays, and substrate hydrolysis was measured in the linear response range.

Nucleotide sequence accession number.

The 2,513-bp sequence for murine intron 1 and 780-bp sequence for murine intron 16 have been deposited in the EMBL database under accession no. AJ224981 and AJ224980.

RESULTS

(i) Characterization of a putative splicing event in intron 16.

In the course of the cloning of the mouse calpain 3 cDNA (25), we isolated, from a mouse poly(A)⁺ muscle cDNA library (random lgt10 library from Stratagene), a clone (S15) containing a sequence between exons 16 and 17. This clone was sequenced, and the inserted sequence was found to be identical to the first 308 bp of the 780-bp-long murine intron 16 sequence (EMBL accession no. AJ224980).

The presence of these nucleotides in S15 suggested that an alternative 5′ splice donor site within this intron had been used instead of the regular one. In view of this observation, we undertook a systematic isolation of calpain splice variants.

(ii) Systematic PCR screening for additional isoforms.

A series of RT-PCR amplifications was performed with sets of primers covering the entire coding sequences of the murine capn3 transcripts (Table 1). These reactions were performed on total RNAs from established murine cell lines, at either the myoblast or the myotube stage. We observed differences in the gel profiles of the PCR products after amplification with the primer pairs p94sys3.a and p94sys3.m, p94sys5.a and p94sys5.m, and p94sys6.a and p94sys6.m (Fig. 1). Interestingly, these primer pairs flank either IS1 or IS2 (Fig. 2).

FIG. 1.

Electrophoresis of variant PCR products from the systematic screening of myoblast and myotube RNAs. Amplifications were performed with p94sys3 primers flanking the IS1 region (A) and with p94sys5 (B) and p94sys6 (C) primers flanking the IS2 region. MB, myoblast; MT, myotube. (A) The 396-bp band corresponds to the splicing out of exon 6. (B) The 601-, 505-, and 487-bp bands correspond, respectively, to the skipping of exon 15 and exon 16 and the splicing out of both exons 15 and 16. (C) The 603- and 774-bp bands correspond, respectively, to the retention of the int137 and int308 sequences.

FIG. 2.

Alternative splicing events of mouse calpain 3 mRNA. (A) Schematic diagrams of the calpain 3 protein with its four domains and its three specific sequences, NS, IS1, and IS2, and the exon structure of the mRNA. Positions of the primers used for long-range PCR are indicated. (B) Sites of alternative splicing with deletion of (i) exon 6, (ii) exon 15, (iii) exon 16, and (iv) exons 15 and 16. Arrows denote the positions of primers used in PCR for identifying alternative splicing events. (C) Variant forms of mouse calpain 3 mRNA produced by the association of alternative splicing of the three exons, 6, 15, and 16 (gaps in black lines), and retention of the sequences int137 and int308 (short and somewhat longer gray lines, respectively). (D) Retention of the sequences int137 and int308 in intron 16. AS and DS are 3′ acceptor and 5′ donor sites, respectively. Putative splice sites are in italics. The scores of comparison between consensus splice sites and normal or putative alternative splice sites are indicated.

PCR products whose fragment sizes differed from expected sizes were cloned and sequenced. The 396-bp band seen in Fig. 1A corresponds to the splicing out of exon 6; the 601-, 505-, and 487-bp products seen in Fig. 1B correspond, respectively, to the splicing out of exon 15, exon 16, and both exons 15 and 16. The 774- and 603-bp products seen in Fig. 1C correspond to the retention of two sequences internal to intron 16 (int308 and int137). The names of these internal sequences refer to the number of intronic (int) nucleotides retained. int308 and int137 correspond, respectively, to the insertion sequence present in clone S15 and to the last 137 nucleotides of int308. The different events are depicted in Fig. 2.

Products containing or lacking exon 6 were detected in both RNA populations, but their relative amounts were, however, in inverse proportions, the ex6⁻ isoform (lacking exon 6) being more abundant in myoblasts than in myotubes (Fig. 1A). While splicing out of exon 16 is seen in cDNAs from both myoblast and myotube cultures, splicing of both exons 15 and 16 seems to be present only in myoblast cDNA, where it is the major form (Fig. 1B). Amplification products containing int137 and int308 were detected in cDNA from myoblast cultures (Fig. 1C) and also in cDNA from myotube cultures, as was shown after subcloning of PCR products.

To investigate the association between these different events in individual mRNA molecules, amplification of the entire cDNA was performed by long-range RT-PCR on mRNAs from myoblasts and myotubes. Screening of at least 40 independent clones from each mRNA population led to the identification of several variant forms of mouse calpain 3 mRNA. The majority of them, representing 12 independent isoforms, are produced by combinatorial associations of alternative splicing events involving exon 6, 15, or 16 and/or retention of int137 or int308 (Fig. 2C). Skipping of exon 4, 5, or 7 or retention of intron 18 was also occasionally encountered, but this occurred in too few clones to be taken into account.

(iii) Analysis of the sequences in intron 16.

Examination of the 3′ border of the murine int137 and int308 sequences (Fig. 2D) for the presence of potential consensus splice sites revealed the existence of a donor splice site (Shapiro’s rodent score of 74.6% [30]). It must be noted that this site corresponds to the nonconventional consensus sequence carrying a GC instead of the so-called invariant GT (10). Moreover, an acceptor splice site presenting a score of 90.6% with respect to the rodent consensus score is found in the murine sequence at the position corresponding to the 5′ border of the int137 sequence. Furthermore, a sequence corresponding to the consensus branch site is present 35 bp upstream of this putative acceptor site. These findings are thus compatible with the notion that int137 and int308 represent alternatively spliced products of the mouse capn3 gene. The corresponding murine calpain 3 polypeptides, if produced, would terminate before domain IV, due to the introduction of premature stop codons, and hence be devoid of the corresponding Ca²⁺ binding sites.

(iv) Additional events involving the NS region.

In our systematic screening of muscle cell lines, no alternative splicing products affecting NS, the N-terminus-specific region of calpain 3, were detected. However, several eye-specific calpain 3 isoforms carrying a variant and shorter first exon have recently been described: Lp82 (EMBL accession no. U96367 [17]), Lp85 (EMBL accession no. AF052540), both of which are present in lens tissue, and Rt88 (EMBL accession no. AF061726), which is present in retinoid and choroid tissues. The existence of these calpain 3 isoforms is therefore further proof of the existence of alternative splicing events.

To confirm whether the sequences encoding the different N termini reside in the same region of the genome and therefore to further characterize the genomic organization of the murine capn3 gene, lens-specific primers taken from the Lp82 sequence were used in PCR experiments on various genomic DNA fragments from the mouse capn3 region. The results obtained enabled us to locate lens-specific exon 1, named exon 1′, in the 3′ part of the first capn3 intron. We thus initiated the sequencing of murine intron 1. Analysis of the 2,513-bp-long sequence obtained showed the presence of a potential donor splice site with an excellent consensus score (82.4%) (30) at a position corresponding to the 3′ end of this exon. As for the rat sequence, a start codon included in a related Kozak consensus sequence (14) was present 311 bp upstream of this splice site. This exon is highly homologous (95.5%) to the corresponding published rat variant sequence (17).

As no significant acceptor splice site was found at the 5′ end of this exon, computer analyses were performed to detect the presence of a promoter. Use of the promoter prediction program Proscan (21) revealed a score of 88.27 (cutoff, 53) for the nucleotide sequence from positions −267 to −28, including a TATA box at position −96, before the putative initiation ATG. Comparison with sequences in the Transfac database (9) with the TFSEARCH and MatInspector programs (23) revealed the presence of GC box and CCAAT box promoter elements at nucleotides −112 and −87, respectively (1). In the same analysis, we observed two putative binding sites for the δ-crystallin enhancer binding protein, δEF1, at nucleotide positions −1042 to −1036 and −535 to −529 (29). These data suggest that the lens variant is transcribed from an alternative promoter.

(v) Tissue distribution of calpain 3 transcripts.

Mouse embryos of different developmental days (11.5 to 18.5) and several muscular and nonmuscular mouse tissues were investigated to verify the presence of calpain 3 mRNA isoforms and the extent to which other nonmuscular tissues may contribute to the production of calpain 3 isoforms. Primers covering alternatively spliced regions were used to generate RT-PCR products, which were visualized by electrophoresis (Table 1; Fig. 3). PCR amplification performed with primers specific to the ubiquitously expressed transcription factor TFIID was used as an internal control (Table 1).

FIG. 3.

Visualization of calpain 3 isoforms. Electrophoresis of RT-PCR products during mouse embryonic development from days 11.5 to 18.5 (A) and in different tissues (embryonic lens and lens, brain, heart, and smooth and skeletal muscle tissues from 9- to 11-week-old mice) (B). E, embryonic day. (a) PCR products obtained with p94sys3 primers. The 540- and 396-bp bands correspond, respectively, to the entire segment with and without exon 6. (b) PCR products obtained with primers p94sys5.a and p94sys5b.m. The 619-bp band is from the unspliced mRNA. The 601-, 505-, and 487-bp bands correspond, respectively, to the splicings of exon 15 and exon 16 and the association of splicings of exons 15 and 16. The embryonic lens profile was obtained with a Southern blot probed with the peroxidase-labeled primer sIS2 (5′-GGAAGCTGAAAATACAATCTCTG-3′, positions 1746 to 1768) (the figure is a photographic negative). (c) Amplification of a TFIID sequence.

No capn3 transcripts could be seen in embryos before day 12.5, in agreement with in situ hybridization results (see below). Isoforms present in myoblast and myotube cultures (ex6⁻, ex15⁻, ex16⁻, int137⁺, and int308⁺) were also found in mouse embryos (Fig. 3A). Transcripts lacking exon 6, though present in embryos from days 12.5 to 18.5, represented the predominant forms from embryonic days 12.5 to 15.5 (Fig. 3A). Figure 3B shows that isoforms lacking both exons 15 and 16 are the major isoforms in the early stage and that their expression decreases during development in favor of the complete isoform. In addition, the int137⁺ and int308⁺ transcripts were found at some stages and in low copy numbers (data not shown).

Different mouse tissues (adult skeletal, cardiac and smooth muscle, brain, and lens tissues and lens tissue of a day 15.5 embryo) were also subjected to similar analyses. While both spliced and unspliced isoforms were seen in all these tissues (Fig. 3B), our results clearly showed that the relative amounts of the calpain 3 RNA isoforms vary from tissue to tissue. Whereas in the striated tissues the complete forms are the most common ones, isoforms lacking exon 6 or both exons 15 and 16 are expressed at high levels in adult brain, adult smooth muscle, and embryonic lens. The embryonic lens pattern is also seen in adult lens but at lower intensities.

These experiments were completed by measuring, by a semiquantitative RT-PCR method, the calpain 3 mRNA levels among mRNA preparations extracted from the same embryonic and adult mouse tissues. The experimental design was such that the total amount of calpain 3 was taken into account, independently of alternative splicing events. The results presented are relative to the amount of calpain 3 mRNA present in the skeletal muscle (Fig. 4). Although calpain 3 mRNA is present in trace amounts in most tissues examined (at levels from 100- to 1,000-fold lower than the level present in skeletal muscle), it is present at an exceptionally higher level in one tissue, the embryonic lens. In addition, this experiment enabled us to demonstrate a dramatic decrease of calpain 3 expression in lens between the mouse embryonic and adult stages.

FIG. 4.

Measure of calpain 3 mRNA level by quantitative RT-PCR. Calpain 3 transcription was monitored in six different tissues by a quantitative PCR method with TaqMan probes. The level of TFIID mRNA was used to normalize the results across different tissues. An arbitrary value of 1 was assigned to the skeletal muscle value, and the values measured in the other tissues are expressed as ratios to the skeletal muscle content, these ratios being drawn in a logarithmic scale.

(vi) In situ hybridization.

A previous in situ hybridization study of murine embryo slices demonstrated that an oligonucleotide probe specific to exon 1 (NS region) failed to yield any lens signal but that probes for exon 6 and exon 16 (in the IS1 and IS2 regions, respectively) unambiguously demonstrated the presence of calpain 3 transcripts in this tissue (5). To complete these data, isoform-specific oligonucleotides (var5p.AS, int308.AS, int137.AS, ep6.AS, ep15.AS, ep16.AS, and ep1516.AS) were used in similar experiments (Table 2). Hybridization of the oligonucleotide probe derived from the 5′ sequence of the mouse lens-specific capn3 gene (var5p.AS) yielded intense labeling, which was seen only in the lens (Fig. 5). The probes specific for splicing events involving the IS1 and IS2 regions also yielded a restricted and intense lens-specific signal (Table 3). It should be noted that the signal obtained with the isoform-specific oligonucleotides was far more intense than the signals obtained with the exon 6- and exon 16-specific probes, suggesting that alternatively spliced transcripts predominate in embryonic lens tissue. This result is in agreement with RT-PCR results. In addition, the lens capn3 signal is present in embryos at day 12.5, well before the visualization of calpain 3 expression in skeletal muscles, and is still present 1 week after birth (6). The signal corresponding to the probes ep16.AS and ep1516.AS disappeared after embryonic day 13.5, although the corresponding RNA isoforms were still detected by RT-PCR. The apparent discrepancy between these results may be explained by the differential sensitivities of the two methods used.

TABLE 2.

Specific antisense probes for splicing events used in in situ hybridization^a

Primer	Sequence (5′–3′)	Position
var5p.AS	TGGGAGCCAAGGGCCTGCACACCTGCTGGG	Exon 1′
	AAGACACGGTCTTCAAAGAGGCACTTCTGA
int308.AS	CAGACAATTCCAAACCCCACAGTGTTGAAGA	Intron 16
	GACCCTCCCCTGAGAGCCACCATAACCCC
int137.AS	CCACCCTGGGACTGGTTTGGTGTGGCAGCG	Intron 16
	ACCTTAAAAGACACTTACCCAAGAAATACC
ep6.AS	CACCGGAACAATTGTGTCAATGGAGCAGCC	Exon 5–exon 7
ep15.AS	AACGAAGATGATGGGCACTGGCCGATCCAC	Exon 14–exon 16
ep16.AS	GTGGCCAGGCCTTGGCTTGTTTTTTTTCTT	Exon 15–exon 17
ep1516.AS	GTGGCCAGGCCTTGGCACTGGCCGATCCAC	Exon 14–exon 17

^aAntisense probes specific for exon skipping events (last four primers) are designed in such a way that the first and last (in italics) 15 nucleotides correspond to the end and the beginning, respectively, of the exons which flank the exons which have been spliced out. 

FIG. 5.

In situ hybridization on an embryonic lens region. Shown are transverse sections of NMRI mice embryos at day 12.5 (A to C) and sagittal sections at day 17.5 (D to F). In the phase-contrast images (A and D), the bars represent 600 mm. The other panels show results of dark-field in situ hybridization with a probe specific for the exon 1′ sequence (var5pAS) (B and E), a sense probe (var5p.S) (F), and a titin-specific probe (C) (6). EOM, extrinsic ocular muscle; FM, facial muscle; L, lens, NL, neural layer of retina; O, orbital plate of frontal bone; PNC, primitive nasal cavity; PV, primitive vitreous humor. Note that pigments of the neural layer of the retina and primitive ossification within the medial part of the roof of the orbit are due to nonspecific labeling.

TABLE 3.

Lens-specific in situ hybridization results from mouse embryo slices

Probea	Nature of the products	Signal on day of embryonic developmentb:
		11.5	12.5	13.5	14.5	15.5	16.5	17.5	18.5	P0
ep6.AS	ex6−	ND	+		+	+	+	+	+	+
ep15.AS	ex15−	ND	+	+	+	+	+	−	−	−
ep16.AS	ex16−	ND	+	(+)	−	−	−	−	−	−
ep1516.AS	ex15−16−	ND	(+)	(+)	−	−	−	−	−	−
int137.AS	int137+	ND	+	+	ND	+	+	ND	+	+
int308.AS	int308+	ND	+	+	ND	+	+	ND	+	+
var5p.AS	ex1′+,1−	−	+	+	+	ND	+	+	+	+

^aIt should be noted that the probes yielded a restricted lens-specific signal. 

^b+, presence of the lens signal; −, absence of the lens signal; (+), very faint labeling; P0, day of partum; ND, not done. 

(vii) Investigation of functional characteristics of the translated products.

While the exact physiological functions of calpain 3 remain unknown, some characteristics of this protein are already accessible to testing (20). These include calpain 3’s titin binding ability and its autolytic and fodrinolytic capacities. The extent to which the spliced regions are involved in mediating these properties was investigated with COS-7 cells and in the yeast two-hybrid system. The advantage of these assay systems is that they allow dissection of the various splicing events, including the testing of isoforms that are not seen in vivo, and thereby assessment of the impacts of individual events.

Cell extracts from COS-7 cells in which the different isoforms are transiently expressed were subjected to Western blot analyses. First, calpain 3 polypeptides were detected with a polyclonal antibody directed against a peptide from the NH₂-terminal region of the IS2 region, spanning the end of exon 14, exon 15, and the beginning of exon 16 (32). Under the conditions used, the full-length products were present only if the autolytic property was impaired. The results showed that the calpain 3 products derived from complete p94-mRNA were not detectable (Fig. 6B). A similar result was seen for the variant in which exon 1′ was substituted for exon 1 (ex1′⁺,1⁻). In contrast, unproteolyzed products derived from all other isoforms lacking exon 6 or exon 15 and/or both exons 15 and 16 were detected (Fig. 6B). These results indicated that the preservation of both IS1 and IS2 is necessary for the rapid autolysis of calpain 3.

FIG. 6.

Investigation of the autolytic and fodrinolytic capacities and titin binding abilities of the different isoforms. (A) The isoforms investigated for biochemical characteristics are drawn under the diagram of the calpain 3 protein (Lp82 corresponds to the ex1′⁺,1⁻6⁻15⁻16⁻ isoform). Their names and molecular masses are given at the left of the line. The position of the C129S mutation, which inactivates the catalytic site, is indicated. Western blot analyses were performed on lysates of transfected COS-7 cells to visualize proteolytic calpain 3 fragments with an antibody against the IS2-specific region of p94 (B) and to assess fodrinolysis with an antibody specific to the 150-kDa α-fodrin fragment (C). Open arrowheads indicate the full-length products, and filled arrowheads indicate proteolyzed fragments. The titin binding capacity of each isoform was monitored in a yeast two-hybrid system by measuring β-galactosidase activity with CPRG as the substrate (1 U of β-galactosidase is defined as the amount which hydrolyzes 1 mmol of CPRG in 1 min). (D and E) Histograms of β-galactosidase expression for each isoform following cotransfection into S. cerevisiae cells of a calpain 3 isoform and pCNT-N2 (D) and pCNT-52 (E) titin peptide-encoding clones. Bars indicate the mean β-galactosidase activities from three independent experiments; ranges are indicated by the vertical lines. (Lp82 corresponds to the ex1′⁺,1⁻6⁻15⁻16⁻ isoform). The first column corresponds to transfection with the vector pAS2C-1 alone.

Second, with the same COS-7 protein extracts, appearance of the 150-kDa proteolyzed fodrin α subunit (7, 19) was monitored with an antibody against the N-terminal sequence of this protein (27). While fodrinolytic activity is preserved in half of the extracts, it is barely demonstrable for extracts of ex1′⁺,1⁻ isoforms and undetectable for isoforms lacking both exons 15 and 16 (Fig. 6C). As isoforms lacking exon 15 possess fodrinolytic activity, these results suggest that exon 16 is essential for fodrin cleavage.

Third, titin binding ability was tested in a yeast two-hybrid system using as bait two different regions of titin: an internal and C-terminal fragment corresponding, respectively, to the sarcomeric N₂ and M lines. These regions were previously shown (11, 34) to act as binding sites for calpain 3, with binding to the N2 line region being much stronger than binding to the M line region (34). Our data demonstrate that binding to the N2 line region is affected only in isoforms with exon 1′ (Fig. 6D). In contrast, binding to the C-terminal region is enhanced (between three- and fivefold) for all non-lens isoforms lacking exon 16 (Fig. 6E). No significant differences with respect to N2 line binding is seen between the calpain 3 isoforms lacking exon 15 or exons 15 and 16. This result is to be contrasted with the effect on C-terminal titin binding, suggesting that the calpain 3 sites responsible for the association with these two titin segments are not identical. Thus, in addition to the two titin binding sites, there may be at least two calpain 3 interaction mechanisms which may correspond to distinct functional and/or regulatory biological properties. Considering the fact that splicing of exons 6, 15, and 16 does not completely abolish titin binding, we infer that these exons are not essential for this association, though exon 16 might regulate this binding, as its absence led to a stronger interaction with the C-terminal titin region.

DISCUSSION

In this report, we describe the isolation and analyses of calpain 3 gene isoforms. We identified different transcriptional and/or posttranscriptional events in mice which lead to alterations involving the NS, IS1, and IS2 regions and/or the calcium binding domains. These events can be divided into three groups: (i) splicing of exons that preserve the translation frame; (ii) inclusion of two distinct intronic sequences between exons 16 and 17 that disrupt the frame and would lead, if translated, to a truncated protein lacking domain IV; and (iii) use of an alternative first exon specific to lens tissue. The splicing out of exons 6, 15, and 16 cannot represent transient intermediate steps towards the production of mature calpain 3 mRNA, and the status of the molecules retaining parts of intron 16 is still unclear. The fact that the splicing of exons 6, 15, and 16 were seen with both total and poly(A)⁺ RNAs lends credence to the fact that the corresponding RNAs represent authentic mRNA isoforms.

The mouse myoblast cell line C2C12, which can be induced to differentiate into myotubes, was chosen as a model for the study of calpain 3 transcription during in vitro skeletal muscle differentiation. The results presented here show that differentiation in this cell culture system is accompanied by a change in the expression pattern of calpain 3 RNA isoforms, suggesting that the transcription of the different isoforms is regulated during muscle differentiation. We also noted an increase in the relative overall abundance of calpain 3 transcripts, and particularly of mature 3.5-kb mRNA, in myotubes compared to that in myoblasts.

As the initial observations were carried out ex vivo, there remained a possibility that the reported RNA splicing events do not occur in vivo. There was thus a need to substantiate these data by the examination of different murine tissue samples. Splice variants were observed at various developmental stages (from days 12 to 18) of fetal gestation as well as in adult mouse tissues. These data imply that multiple capn3 RNA isoforms are also generated in vivo and, furthermore, that they are developmentally regulated. Fougerousse et al. (6) reached a similar conclusion, upon observation of the spatiotemporal transcription patterns of CAPN3 during human prenatal development. They demonstrated the presence of alternatively spliced products in smooth muscles in humans by in situ hybridization using oligonucleotides specific for the NS, IS1, and IS2 regions as probes. Our results further corroborate and extend the observations of Ma et al. (17), i.e., that a lens-specific exon stemming most likely from the use of a lens-specific promoter exists and that it is subject to alternative splicing involving the IS1 and IS2 regions.

The examination of the impact of the loss or substitution of defined exons on a number of biochemical parameters provides us with a refined and sensitive tool to dissect calpain 3 and establish structure-function relationships. Furthermore, since the segments affected by these events span the sequences which are unique to calpain 3, such analyses also enable us to address the specificity of this protease compared to those of the ubiquitous calpains. Translation of the calpain 3 isoform transcripts identified in this study would obviously result in substantial alterations of calpain 3 structure and properties. Transfection experiments with COS-7 cells demonstrate that the corresponding proteins are synthesized, at least in these cells. Assuming that these proteins are also synthesized in vivo (and the report of Ma et al. demonstrated that in the rat lens this is definitively the case), and on the basis of the results of the functional analyses performed in this study, the consequences of the observed splicing events from the NH₂ to COOH termini can be considered to either prevent or alter the presumed functions of the domains involved.

The consequences of replacing the NH₂ terminus containing the NS sequence with the lens-specific peptide are not known. Clearly, this substitution needs also to be viewed in the context of a simultaneous loss of parts of IS1 and IS2. The main lens isoform (Lp82) seems to have lost fodrinolytic as well as autolytic activity, although the rat lens protein still possesses caseinolytic activity in vitro in the presence of 20 mM Ca²⁺ (18). It should be noted that the experiments presented herein were performed without addition of Ca²⁺. It is therefore possible that, unlike p94 (12), this isoform requires exogenous Ca²⁺ for exercising proteolytic activity. In support of this hypothesis, autolysis was enhanced in our experiments upon addition of calcium to 5 mM for all tested calpain 3 isoforms lacking exon 16 (data not shown).

It is of interest that exon 1′ variants were seen exclusively in the lens. Furthermore, they all have impaired binding to both known titin sites, which is congruent with the absence of titin in this tissue. Taken together, our observations suggest that exon 1′ isoforms fulfill lens-specific functions. Calpain 3 would not be the first enzyme found to also be expressed in the lens. The related m calpain was reported to be implicated in proteolysis of crystallins during normal maturation of rat lens (3). In addition, it is noteworthy that various metabolic enzymes acquire a second function as taxon-specific lens structural proteins (22, 39). It is thus conceivable that calpain 3 belongs to this class of proteins. Whereas our study confirms that calpain 3 mRNA is less abundant in the adult lens, it also clearly shows that in the embryonic lens, the situation is quite the opposite, there being more calpain 3 mRNA in this tissue than in adult skeletal muscle. The presence of such calpain 3 variants in lens may have to do more with the formation and maturation of the lens than with exercising visual activities.

The fact that two isoforms for exon 6 coexist in a variety of tissues suggests the presence of different calpain 3 activities in the same tissues or perhaps even fibers. The function of the IS1 sequence, which is encoded partially by exon 6, is unknown. It is located in proteolytic domain II. Polypeptides lacking exon 6, while capable of proteolyzing α-fodrin, have impaired autolytic activity. This observation is in agreement with the location of autolytic cleavage sites in the IS1 region (12). Cleavage of calpain 3 follows a three-step process yielding, sequentially, 60-, 58-, and 55-kDa products. The first two steps involve sites encoded by exon 6. Therefore, the absence of two of three cleavage sites may explain the impairment of autolysis.

IS2 is presumed to have an important role in the rapid autolysis of the protein (32). Furthermore, it comprises a binding site for titin (34) and a putative nuclear translocation signal (31). Splicing modulation of exon 15, which carries the nuclear translocation signal, may affect, among other things, the subcellular localizations of the resulting proteins by directing them either to the cytoplasm or to the nucleus, as well as the selection of the titin binding site (Fig. 6). Our data indicate that the loss of exon 16 has two effects: increased titin binding at its C-terminal end and a loss of fodrinolytic activity. As titin is not present in COS-7 cells, we can infer that it is not the sequestration of calpain 3 through its association with titin that prevents fodrinolysis. Furthermore, the amino acids encoded by exon 16 do not include residues that participate in the catalytic site. They may, however, be necessary for substrate recognition. These data (Fig. 6) also suggest that the two titin binding sites map outside of the sequence contained in exons 15 and 16 (though sequences within these exons may impact differentially the titin binding sites) and furthermore that the N2 line binding site resides between amino acids 570 and 595 (34).

In addition to splicing events resulting in loss of particular exons, we also noted the presence of a splicing site and branch sites in murine intron 16, leading either to the formation of a new exon of 137 bp (int137) or to the addition of a 308-bp sequence 3′ to exon 16 (int308). Unlike with Lp85, which retains intron 18 without consequences to the reading frame (data not shown), inclusions of intron 16 sequences lead to premature in-frame stop codons (18). As these sequences are situated between exons 16 and 17, the presumed consequence is the loss of protein domain IV, which is thought to participate in the calcium regulation of calpain activity. The resulting isoforms may therefore be calcium independent or less calcium dependent by using the remaining calcium-binding EF hand motif, present in the third domain. Similar observations were reported for the stomach-specific calpains nCL-2 and nCL-2′ (33) and for a Drosophila atypical calpain (38). Elucidation of these mechanisms and of their significance awaits further investigations.

Since differences in levels of gene expression between mouse and human are not exceptional, we examined whether human calpain 3 was also subject to alternative splicing. RNAs extracted from human cultured muscle cells were analyzed by RT-PCR as described above. Preliminary results performed on these RNAs demonstrated the existence of similar splicing events involving the IS1 region. Skipping of exon 6 can be evidenced in the presence of the ex6⁻ cDNA, both at the myoblast and myotube stages (data not shown). We also demonstrated that in lymphoblastoid cell lines, exon 15 is systematically spliced out but that only half of the molecules have exon 6 spliced out. Evidence of alternative splicing in vivo in humans is also corroborated by the presence of splicing events during development, as was revealed by in situ hybridization (6).

Muscle cells are notorious for their utilization of alternative splicing as a means to generate multiple isoforms for a given gene (28). Calpain 3 thus belongs to this category of genes. It is worth remembering that titin also exists in alternatively spliced forms (15). In particular, the two titin binding sites recognized by calpain 3 are affected by tissue-specific alternative splicings. These events generate at the N2 line cardiac or skeletal muscle-specific isoforms and at the M line a mixture of isoforms whose ratios vary from muscle to muscle (13). The generation of these isoforms suggests that calpain 3, which interacts with at least one other protein that is also subject to alternative splicing (titin), may be involved in a complex tissue-specific spectrum of combinatorial possibilities. The specific topography and characteristics of muscle involvement in patients with limb girdle muscular dystrophy type 2A may be related to this potentially complex set of interactions.

The differential splicing of the calpain 3 gene and the modulation of the expression of the different isoforms need to be taken into account in investigations of the biological role(s) of calpain 3. It is of course of crucial importance to demonstrate the existence in vivo of the translated products generated from the variant calpain 3 mRNA molecules as reported by Ma et al. for the rat lens (18). It would also be of interest to determine the exact (sub)cellular locations of all these isoforms. Finally, the demonstration of the existence of different calpain 3 protein isoforms is also likely to have important consequences on our understanding of human pathophysiology and of the phenotypes of capn3^−/− animals as well as on the establishment of therapeutic strategies.