Three differentially expressed survivin cDNA variants encode proteins with distinct antiapoptotic functions

DNA restriction enzymes were obtained from Boehringer Mannheim Canada (Dorval, Quebec) or New England Biolabs (Mississauga, Ontario), and radioisotopes were from ICN Biomedicals (Mississauga, Ontario). Murine 14.5 days postcoitum (dpc) embryo λ phage cDNA and 129SV genomic DNA libraries were purchased from Clontech (Palo Alto, CA) and from Stratagene Cloning Systems (San Diego, CA), respectively. The murine PAC library was provided by Genome Systems (South Bend, IN). Recombinant human caspase-3 (catalogue no. 235 417) and the caspase-3 substrate colorimetric kit (catalogue no. 235 400) were purchased from EuroBiochem (Bierges, Belgium). The caspase-3 has a specific activity of approximately 1 U/mg protein, defined by the amount of enzyme that cleaves 1.0 μmol of the substrate Ac-DEVD-pNA per minute at 25°C. A panel of cDNAs derived from normal human tissues was obtained from Clontech—Multiple Tissue cDNA Panel 1, catalogue no. K1420-1.

Based on the published human EPR-1 cDNA sequence, a murine expressed sequence tag (EST mc61a11.r1) was identified by BLAST search. Identification of the cDNA encoding humansurvivin12,13 revealed that this EST actually corresponded to the murine homologue of survivin. Indeed, the predicted amino acid sequence corresponding to this EST exhibited 86% identity with the reported human survivin sequence, between amino acid residues 1 and 128. EST mc61a11.r1 was subsequently purchased from Genome Systems, the DNA sequence was confirmed, and the insert was radiolabeled with α-³²P-dCTP using the random primer synthesis method22 for use as a probe to screen both the murine embryo cDNA and genomic DNA libraries.

The λ phage cDNA library was prepared for plaque hybridization. Filters were hybridized in Quik-Hyb (Stratagene) for 1 hour at 65°C with the α-³²P-dCTP-labeled EST probe and finally washed at 55°C with 0.5 × standard sodium citrate (SSC), 0.1% sodium dodecyl sulfate (SDS). After 4 rounds of plaque purification, the cDNA inserts were isolated and subcloned into pBS KS(+) (Stratagene). The genomic DNA library was screened using the same α-³²P-dCTP-labeled probe, and a single clone, SRV10′, was isolated, containing the entire coding region of murine survivin. The ∼13 kb insert was purified from the λ phage and inserted into the Not1 site of pBS KS(+). A murine PAC library (Genome Systems) was also screened using the entire 13 kb SRV10′ insert as a probe. Three independent PAC clones were identified and were shown by Southern blotting with gene-specific radiolabeled oligonucleotide probes to contain the entire SRV10′ sequence.

Fluorescent in situ hybridization (FISH) analysis was performed following established procedures, by seeDNA Biotech Inc. (Toronto, Ontario) using the murine PAC survivin genomic clone as a probe.23,24 Lymphocytes were isolated from mouse spleen and cultured at 37°C in RPMI 1640 medium supplemented with 15% fetal calf serum, 3 μg/mL concanavalin A, 10 μg/mL lipopolysaccharide and 5 × 10⁻⁵ mol/L mercaptoethanol. After 44 hours, the cultured lymphocytes were treated with 0.18 mg/mL BrdU for an additional 14 hours. The synchronized cells were washed and recultured for 4 hours in α-minimum essential medium (MEM) with thymidine 2.5 μg/mL. Chromosome slides were made as reported.23 24 For FISH detection, the slides were heated at 55°C for 1 hour, treated with RNAse A, and denatured in 70% formamide in 2 × SSC for 2 minutes at 70°C, followed by dehydration with ethanol. The biotinylated probe was denatured at 75°C for 5 minutes in 50% formamide and 10% dextran sulfate, and hybridized overnight on the denatured slides. FISH signals and the DAPI banding patterns were recorded separately and photographed. The FISH signals were subsequently mapped to specific chromosomal bands by the superimposition of FISH and DAPI images.

Total RNA was isolated from confluent cell monolayers or from tissues by the method of Chomczynski and Sacchi.25 For adult tissue samples, adult mice (10-12 weeks) were killed and tissues were dissected and placed into liquid nitrogen. For embryonic samples, C57Bl/6 mice were mated, conception was assessed by the presence of a coital plug (the morning of coital plug being scored as 0.5 days postcoitum [dpc]), pregnant females were killed at various developmental time points, and embryos were carefully removed by dissection. Total RNA was extracted from adult and embryonic samples by homogenization of the tissues, and cDNA was synthesized by reverse transcription using murine leukemia virus (M-MLV) reverse transcriptase and a cDNA synthesis kit (Bethesda Research Laboratories, Burlington, Ontario). First-strand synthesis was primed using random hexanucleotides, and PCR was performed as reported,26 with the annealing temperature adjusted to 55°C. Pairs of oligonucleotide primers for PCR were synthesized such that they flanked the alternatively spliced mRNA forms. Primer srv86 (sense 5′-TCGCCACCTTCAAGAACTGGCCCTTCCTGGA), when paired with primer srvas311 (antisense 5′-GTTTCAAGAATTCACTGACGGTTAGTTCTT) was expected to generate PCR amplicons of 225 bp or 144 bp, depending on the mRNA splice form. Primer srv86, when paired with srvas6380 (antisense 5′-GGCTTCTGACAATGCTTG), was expected to generate a PCR amplicon of 332 bp. Resultant RT-PCR products were subsequently examined by agarose gel electrophoresis and by Southern blotting, using a radiolabeled oligonucleotide probe, srv295 (sense 5′-TGGCTGCGCCTTCCTCACTGT) corresponding to the sequence found within both amplicons. The amount of cDNA synthesized from each sample was calibrated according to the relative expression of hypoxanthine phosphoribosyl transferase (HPRT), as determined by agarose gel electrophoresis of RT-PCR products generated using the specific oligonucleotide primers, HPRT sense (5′–CACGGACTAGAACACCTGC) and HPRT antisense, (5′–GCTGGTGAAAAGGACCTCT), yielding a 221 bp product.27 

Nested RT-PCR of RNA derived from human cell lines or tissue was used to examine expression of human survivin transcripts.11 For the same purpose, cDNAs derived from the mRNA of several normal human tissues were purchased from Clontech . To determine whether a cDNA lacking exon 2 was present, primers in exons 1 and 3 were paired. Thus, primer hsrv2841 (sense 5′-CAGCCCTTTCTCAAGGAC) was first paired with hsrv5175 (antisense 5′-GGAAAGCGCAACCGGACGAAT), and 2% of the PCR product was subjected to a further 30 cycles of PCR using the primer pair hsrv2861 (sense 5′-CCGCATCTCTACATTCAAGAAC) and hsrv5159 (antisense 5′-CGAATGCTTTTTATGTTCCTC), yielding an expected product of either 193 bp or 83 bp, the latter expected if exon 2 is excised by alternative mRNA splicing. Nested RT-PCR was also used to determine whether the homologue of murine survivin₁₂₁ is present in human-derived tissue or cells. Primer hsrv2841 was paired first with primer hsrv5451 (antisense 5′-GAGGCAGGAGAATCACTTG), and 2% of the resultant amplicon was reamplified using primers hsrv2861 and hsrv5431 (antisense 5′-GAGAGGCAGAGGTTGCAGTG), yielding products of either 480 bp (retention of intron 3) or 370 bp (excision of exon 2). Resultant RT-PCR products were examined by agarose gel electrophoresis and by Southern blotting using a radiolabeled oligonucleotide probe, hsrv2888 (sense 5′-CTTCTTGGAGGGCTGCGCCTGCAC) corresponding to sequence in exon 1, which is found within both amplicons. For further verification, DNA sequencing of gel-purified amplicons was performed.

The RT-PCR products were separated by electrophoresis on a 2% agarose gel. Oligonucleotide probes were end-labeled with γ³²P-ATP. Following Southern transfer, Genescreen nylon filters (NEN Life Science Products, Brussels, Belgium) were hybridized with radiolabeled probe in Quik-Hyb for 1 hour, washed with 0.1% SDS, 0.5 × SSC at 55°C, and exposed to Kodak XAR-5 film at −70°C.

cDNAs corresponding to the 3 forms of murine survivin were subcloned in-frame into the expression vector pSecTagA (Invitrogen), to produce survivin fusion proteins bearing an N-terminal Ig κ-chain leader sequence and a C-terminal myc epitope/polyhistidine tag. The resultant plasmid vectors—surv₁₄₀/psec, surv₁₂₁/psec, and surv₄₀/psec—were then expressed stably into COS-7 cells using zeocin selection, conditioned media were collected, and the myc/His₆-tagged proteins were purified by nickel-column chromatography. Recombinant proteins were eluted with 250 mM imidazole, 0.3 mol/L NaCl in phosphate-buffered saline (PBS), and the presence of the expected forms of survivin was confirmed by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) and Western immunoblotting using murine monoclonal anti-myc antibodies, followed by detection with horseradish peroxidase (HRP)-conjugated goat–anti-mouse antibodies and visualization with the ECL Western blotting system (Pharmacia-Amersham, Rosendaal, Netherlands). Protein concentrations were determined using bovine serum albumin as a standard.

Lysates of transiently transfected cells were also evaluated for the expression of recombinant survivin variants. PBS-washed cell pellets were suspended in a buffer containing 25 mM Hepes pH 7.5, 5 mM MgCl₂, 5 mM EDTA, 5 mM dithiotreitol, exposed to 3 freeze-thaw cycles, and centrifuged at 16 000g for 30 minutes. The lysates were then divided into aliquots and frozen at −20°C for subsequent analysis.

The effect of recombinant murine survivins on caspase-3 activity was evaluated by using a specific colorimetric assay kit (EuroBiochem; catalogue no. 235 417) according to the manufacturer's instructions. Briefly, purified recombinant myc/His-tagged survivin (20-50 nmoles/L) was added to 30 ng (3 × 10⁻⁵ units) of human recombinant caspase-3 (EuroBiochem) in assay buffer (0.1 mol/L NaCl, 50 mM Hepes, 10 mM DTT, 1 mM EDTA, 10% glycerol, 0.1% CHAPS [3-([3-Cholamidopropyl] dimethylammoniol)-1-propane-sulfonate], pH 7.4) in a final volume of 120 μL. The reaction was started by addition of 10 μL of the colorimetric caspase-3 substrate I and the rate of change in absorbance at 405 nm was measured using a kinetic enzyme-linked immunosorbent assay plate reader. Controls without either caspase-3 or substrate resulted in no change in absorbance during 30 minutes of observation. Experiments were performed in triplicate, repeated 3 times, and the means and standard errors were calculated.

Polyclonal antibodies against murine survivin were generated by Eurogentec (Seraing, Belgium) using a standard immunization protocol whereby rabbits were immunized with KLH-conjugated peptides NH₂-ALPQIWQLYLKNYRI-COOH and NH₂-CFKELEGWEPDDNPI-CONH₂, representing amino acid residues 5 to 19 and 60 to 74, respectively, of the deduced murine amino acid sequence. Immunoglobulin was purified from immune sera by protein-A sepharose chromatography, and specific antisurvivin antibodies (Ig-srv .336) were subsequently obtained by affinity chromatography of the immunoglobulin fractions on EAH-sepharose 4B columns (Pharmacia-Amersham Biotech) to which the survivin peptides had been coupled.

Embryonic samples, 12.5 dpc, were pooled and homogenized in PBS with 5 mM EDTA, 5 mM MgCl₂. The clear cell lysate, obtained following centrifugation at 14 000g at 4°C for 30 minutes, was incubated with biotin (EZ-Link Sulfo-NHS-LC-LC-Biotin, Pierce, Polylab N.V., Antwerp) for 60 minutes at 4°C, after which the reaction was stopped by the addition of 0.1 mol/L Tris-HCl. Biotinlyated proteins were separated from free biotin by size fractionation on a PD-10 column (Pharmacia) with PBS as the running buffer. The resultant biotinylated protein solution was incubated in PBS with 0.5 mol/L NaCl with either 20 μg of Ig-srv336 or preimmune immunoglobulin for 2 hours, after which protein A-sepharose was added for a further 1 hour. The immunoprecipitated pellets were washed 3 times with PBS/0.5 mol/L NaCl and twice with PBS. Samples were boiled in nonreducing Laemmli buffer and separated on an 18% SDS-polyacrylamide gel. After transfer, the nitrocellulose filters were blocked for 2 hours in Blotto, incubated with HRP-conjugated streptavidin for 1 hour, and then washed. The Amersham ECL Western-blotting system, followed by exposure of the blots to Hyperfilm-ECL, was used to detect specific bands. Prestained standard markers were used to determine the approximate molecular weight.

Alignments of DNA and amino acid sequences were performed with the aid of MacVector software (Oxford Molecular Ltd, Oxford, UK).

A murine survivin cDNA probe was identified by searching the dbest database for a murine EST—clone mc61a11.r1—that matched closely the cDNA encoding human EPR-1.11 By FASTA analysis,28 EST mc61a11.r1 exhibited 85% and 79% sequence similarity to the corresponding portions of human survivin andEPR-1, respectively. Using this EST as a probe, 1 genomic and 9 cDNA clones (ranging in size from 800 bp to 2.1 kb) were identified by λ phage library screening. Each of the cDNA clones was sequenced in its entirety on both strands, while approximately 11 kb of the ∼13 kb genomic clone was also sequenced. A murine genomic PAC DNA library was subsequently screened with the purified survivin genomic DNA probe, resulting in the identification of 3 independent, overlapping PAC clones. A biotin-labeled PAC DNA clone containing the entiresurvivin coding region (see below) was then used for FISH to localize the survivin gene to chromosome 11E2 (Figure1). Restriction mapping and partial sequencing of DNA fragments from the PAC clones confirmed that the λ phage genomic clone and the PAC clones were overlapping and represented a single gene (data not shown).

Four murine survivin cDNAs (clones 5.25, 92.16, 41.1, and 42.5) corresponded very closely to their full-length human counterparts. The largest of these, clone 5.25, comprised 48 bp of 5′ untranslated sequence (UTR) upstream of an ATG translation initiation codon, a 420 bp open reading frame (ORF), and 872 bp of 3′ UTR following an in-frame TAA stop codon, and predicted a 140 amino acid protein—survivin₁₄₀— of molecular weight 16 286 Da, with 84% sequence similarity to human survivin (Figure2). Comparison of the DNA sequence of this full-length cDNA clone with that of the murine survivin λ phage and PAC genomic clones revealed that the structural organization of the murine survivin locus was very similar to that of its human counterpart (Figures 3 and4). The murine survivin gene comprises 4 exons. Exon 1, containing the ATG translational start site (position 2865), is preceded by a putative TATA-less promoter and a GC-rich region extending approximately 250 nucleotides in the 5′ direction. Exons 1 to 4 (111, 110, 118, and 81 bp, respectively) are separated by 3 introns 292, 2797, and 2448 nucleotides in length. Exon 4 contains an in-frame TAA stop codon at nucleotide position 8822.

Although 9 independent survivin cDNA clones were isolated, only 4 corresponded to a full-length form comprising exons 1 to 4. The remaining 5 clones corresponded to 2 variant cDNAs that appeared to have arisen by alternative pre-mRNA splicing (see Figures 2 and 4). Clones 5.20, 91.1, and 5.24 corresponded to a shorter transcript that lacked exon 2-derived sequences, and therefore contained an ORF of only 120 nucleotides followed by a 3′ UTR, predicting a truncated 40 amino acid residue protein—survivin₄₀—with a molecular weight of 4667 Da (see Figure 4). In contrast, clones 7HB and 7L .15 contained exons 1 to 3 and retained intron 3 as well. Acquisition of a new in-frame stop codon within intron 3 results in an ORF of 363 nt, predicting a 121 amino acid protein (survivin₁₂₁) with a molecular weight of 14 172 Da (see Figure 4).

The amino acid sequences of the 3 variant murine survivin proteins were aligned (see Figure 2) and compared with those of other BIR-containing IAP protein members (not shown). The amino-terminal 113 amino acid residues of survivin₁₂₁ are identical to those of survivin₁₄₀. The remaining carboxy-terminal portions of these 2 variants are distinct. Amino acids 15 to 87 comprise the cysteine/histidine-containing zinc-binding BIR domain of survivin₁₄₀ (and survivin₁₂₁). This region exhibits 30% to 40% sequence identity with each of the BIR-domains of murine XIAP, NAIP, and BRUCE. In common with human survivin, NAIP, and BRUCE, murine survivin does not contain a carboxy-terminal RING finger. Survivin₄₀ shares sequence with only the amino-terminal 30% of the BIR domains of survivin₁₄₀ and survivin₁₂₁ and does not contain a region rich in cysteine or histidine residues, suggesting that it likely does not have direct antiapoptotic function.

Using mRNA variant-specific RT-PCR, we determined the relative distribution of the 3 survivin transcripts in a variety of murine tissues, including brain, heart, kidney, liver, lung, ovary, pancreas, spleen, testis, and thymus. As illustrated in Figure5, survivin₁₄₀-specific mRNA was detectable only in thymus and testis. In contrast, survivin₁₂₁-specific transcripts were detectable in all tissues, although the signal was strongest in brain and ovary.Survivin₄₀ mRNA, on the other hand, was not detectable in any of these tissues.

Previous studies have demonstrated that survivin is expressed in mouse embryos as early as 11.5 dpc.19 We examined the developmental expression of the 3 survivin mRNA variants by RT-PCR (Figure 6). In contrast to what we had observed in adult tissue, all 3 survivin mRNA species, including those corresponding to survivin₄₀, were readily detectable throughout development from 7.5 dpc, the earliest time point examined, until 14.5 dpc. In all cases, the identity of the amplicons generated by RT-PCR was confirmed by Southern blotting using the radiolabeled internal oligonucleotide srv295. In addition, representative PCR products of each size were purified and subcloned, and their identity was verified by DNA sequencing.

To confirm directly whether the different survivin transcripts detected in mouse embryos resulted in translation of distinct protein products, we isolated 12.5 dpc murine embryos and biotinylated the lysates for immunoprecipitation with an antisurvivin immunoglobulin that specifically identifies all 3 recombinant forms. We could detect bands representing survivin₁₄₀ and survivin₁₂₁, confirming that in embryos that are 12.5 dpc, at least 2 of the transcripts are translated into survivin proteins of the predicted sizes (Figure 7). The immunoprecipitation did not yield a band with a predicted molecular weight consistent with survivin₄₀.

To test directly the function of the variant murine survivin proteins, recombinant myc/His₆-tagged survivin fusion proteins were created. The cDNAs encoding the survivin forms were subcloned into the expression vector pSecTagA, and the resultant vectors surv₁₄₀/psec, surv₁₂₁/psec, and surv₄₀/psec cDNAs were each transfected stably into COS-7 cells. Western immunoblotting of cell lysates with anti-myc or antisurvivin antibodies (Ig-srv336) revealed single bands of appropriate molecular size for each of the survivin forms. Following purification of recombinant myc/His-tagged survivin variant forms from conditioned media, varying concentrations were tested in caspase-3 activity assays. In a representative set of experiments summarized in Table 1, both survivin₁₄₀ and survivin₁₂₁ suppressed exogenous caspase-3 activity with similar dose-response curves. At submaximal concentrations, survivin₁₄₀ and survivin₁₂₁ had an additive effect with respect to inhibiting caspase-3 activity. As predicted, survivin₄₀ had no effect on caspase-3 activity, and furthermore, we did not observe any effect of survivin₄₀ on the caspase-3 inhibiting function of survivin₁₄₀ or survivin₁₂₁.

Using the published genomic DNA sequence of humansurvivin,11 we designed a series of oligonucleotides that were expected to identify similar humansurvivin cDNA variants by RT-PCR. Oligonucleotide primers hsrv2841, hsrv2861, and hsrv2888 lie within exon 1, hsrv5159 and hsrv5175 are within exon 3, whereas hsrv5431 and hsrv5451 lie within intron 3. As noted in Figure 8A, mRNA transcripts corresponding to full-length survivin were detected in all cell lines surveyed, including those derived from patients with leukemia, colon cancer, neuroblastoma, melanoma, hepatoma, neurofibrosarcoma, and carcinoma. Transcripts were also detected in mRNA from 4 patients with mucosa-associated lymphoid tissue-type (MALT) lymphomas, as well as in normal human tonsil, lymphoid tissue, kidney, pancreas, liver, brain, and lung (not shown). The identity of the transcripts was confirmed by Southern blotting with an internal oligonucleotide probe specific to survivin or EPR-1. As demonstrated in Figure 8B, we readily detected survivin transcripts corresponding to murine survivin₁₂₁ in RNA from cell lines derived from 2 human leukemias, a germ cell tumor, a teratocarcinoma, melanoma, hepatoma, and carcinoma, but not in a colon cancer, neurofibrosarcoma, or in normal lymphoid tissue, tonsil, kidney, pancreas, brain, or lung. Weak signals corresponding to this transcript were also found in all 4 MALT lymphoma samples (not appreciated in Figure 8). Transcripts corresponding to the human counterpart of survivin₄₀were detected in RNA derived from a neuroblastoma cell line and from 1 of the MALT lymphomas.

The human survivin gene has been reported to be highly similar to the reverse complement of the EPR-1gene.11 Indeed, the human survivin gene was originally identified by screening a genomic library with a humanEPR-1 cDNA. Several lines of evidence indicate that the murine gene and cDNAs that we have isolated correspond to the murine homologue of survivin, rather than of EPR-1. First, the humanEPR-1 gene comprises 5 exons, with the mature protein being encoded by an ORF of 1014 nucleotides. In contrast, the human and murine survivin genes contain 4 exons, and the mature proteins are encoded by ORFs of 426 and 420 nucleotides, respectively. Second, both the sizes and nucleotide sequences of the murine survivinexons are similar to those of their human counterpart. Third, although the reverse complement of the most 3′ end of human EPR-1exhibits sequence similarity to survivin, and the corresponding exons—although transcribed in opposite directions—demonstrate considerable overlap, despite extensive DNA sequencing of our murinesurvivin genomic clones, regions homologous to exon 1 or the 5′ part of exon 2 of EPR-1 (those regions ofEPR-1 which are not similar to the reverse complement of human survivin) could not be found. Finally, attempts to transcribe the opposite strand of the murine survivin gene by computer analysis did not reveal additional potential intron-exon structures that could encode a protein similar to EPR-1. We are confident, therefore, that we have isolated the murine homologue ofsurvivin, rather than of EPR-1. Furthermore, in view of recent claims that survivin and EPR-1 may be members of a closely linked family of genes,14 we tried to determine whether the murine genome also contains a family ofsurvivin/EPR-1-related genes (data not shown). Southern blot analyses of murine ES cell genomic DNA and a murine survivinPAC genomic DNA clone did not suggest the presence of more than a single gene, nor did FISH analysis detect more than a single locus. Nonetheless, these are negative results and do not exclude definitively the possibility that a family of genes does exist.

In these studies, we report the isolation and characterization of 3 independent survivin mRNA species from a murine embryonic cDNA library. Examination of the corresponding cDNA and genomic DNA sequences indicated that each of these transcripts originates from the same gene. The cDNA encoding murine survivin₁₄₀ is derived from all 4 exons. The corresponding deduced amino acid sequence predicts a protein with striking similarity to full-length survivin of human origin, containing a single amino-terminal BIR repeat of approximately 70 amino acids, and lacking a RING finger domain, yet containing a 40 residue coiled-coil carboxy-terminal domain. A second cDNA, derived from the same survivin gene, but lacking exon 2-derived sequence, predicts the existence of a truncated 40 amino acid protein—survivin₄₀— that does not contain the BIR element believed to be critical for antiapoptotic function. Finally, a third cDNA, which retains intron 3, and thereby acquires a new in-frame stop codon, predicts the existence of an additional 121 amino acid protein—survivin₁₂₁. Both survivin₁₄₀ and survivin₁₂₁ contain an intact BIR domain, and as expected, interfere with caspase-3 activity. In contrast to survivin₁₄₀, survivin₁₂₁ does not contain the tubulin-interacting carboxy-terminal coiled-coil region. Although we have not yet demonstrated that survivin₁₄₀ and survivin₁₂₁ differ in their ability to interact with tubulin-containing structures, we anticipate that in the absence of the coiled-coil domain, survivin₁₂₁ activity will not be restricted spatially in a cell cycle-dependent manner.

Consistent with previous reports for humansurvivin,11 expression of murinesurvivin₁₄₀, as detected by RT-PCR, is most prominent in proliferating adult tissues. In contrast, murinesurvivin₁₂₁ mRNA transcripts appear to be more widely distributed, being found in most adult tissues surveyed, a finding in keeping with the observations of Kobayashi et al21 who also showed that murine survivin is present in several adult tissues and all T-cell and B-cell lines. We were unable to detect survivin₄₀ mRNA transcripts in any of the adult murine tissues sampled. The failure of Ambrosini et al11 to detect humansurvivin₁₂₁-specific RNA by Northern blot analysis likely indicates that the 3′ UTR survivin probe that they used lies beyond the region encompassed by this variant transcript. This notion is borne out by our studies in which each of the survivin mRNA forms is detectable by RT-PCR in a variety of human cells and tissues. Examination of the human survivingenomic DNA sequence11 predicts that the human counterparts for murine survivin₁₂₁ and survivin₄₀ would be 128 (human survivin₁₂₈) and 40 (human survivin₄₀) amino acid residues in length, respectively. Human survivin₁₂₈, similar to its murine counterpart, would retain the BIR domain and lose its C-terminal coiled-coil structure. Human survivin₄₀ would lack both a BIR domain and the coiled-coil structure. In several normal adult tissues, only transcripts corresponding to full-length human survivin were detected. In contrast, in addition to transcripts for full-length human survivin, most human tumors and tumor cell lines examined also expressed transcripts corresponding to survivin₁₂₈, whereas a neuroblastoma cell line and 1 of the MALT lymphomas contained survivin₄₀-specific transcripts as well. Based on our observation of an apparent change in the pattern of transcription of the variant survivin forms, it seems reasonable to assume that the differential expression of 1 or more of the functionally distinct survivin variants might affect malignant behavior of a tumor, and thereby alter prognosis. A recent report attempted to correlate the tumor-specific expression of survivin with prognosis in patients with neuroblastoma.20 Notably, this study detected survivin expression immunohistochemically using antibodies that would not be expected to distinguish among the 3 survivin forms. Our studies did not allow the roles of survivin variants in defining tumorigenicity to be assessed. We believe, however, that a more extensive evaluation, coupled with specific in situ localization of the survivin variants, will provide further insights into the role of the survivin variants in normal and abnormal cell growth and proliferation.

Based on the putative structure of human survivin, investigators hypothesized that its function would be similar to that of other members of the IAP family.11 The antiapoptotic function of IAP-class proteins is believed to be partially mediated via their BIR-domains. In Drosophila, several proteins, including the so-called death proteins, reaper and doom, are known to interact with IAP member proteins, thereby altering the binding of the BIR-containing protein to caspases, and diminishing their antiapoptotic function.10,29,30 Both human and murine full-length survivins have been shown to bind specifically to caspases 3 or 7 (or both) via the BIR-domain and to inhibit apoptosis in in vitro systems.21,31 Regulation of this interaction has been further elucidated by recent studies in which survivin expression in HeLa cells was shown to be cell-cycle dependent and largely restricted to the G₂/M phase, a period during which the C-terminal coiled-coil structure of survivin binds to microtubules of the mitotic spindle, thereby enhancing interaction with target caspases.32 It has been hypothesized that the overexpression of survivin observed in most human cancer cell lines31 reflects, at least in part, escape from this apoptotic checkpoint during mitosis.32 

These new insights into the mechanisms by which survivin may regulate the apoptotic pathway are particularly intriguing in light of our finding that there exist 3 variant forms of survivin, each containing distinct structural domains. Survivin₁₄₀, similar to human survivin, contains both a BIR-domain and a coiled-coil carboxy-terminal region, and consequently would be expected to regulate apoptosis in a cell cycle-dependent fashion. In contrast, murine survivin₁₂₁, and its human counterpart, human survivin₁₂₈, lacks the coiled-coil structure, but contains a BIR-domain. We have further confirmed that both murine survivin₁₄₀ and murine survivin₁₂₁ can directly inhibit caspase-3 activity. Because survivin₁₂₁ is likely unable to interact with microtubules, one would not expect it to regulate terminal caspases 3 and 7 in a localized manner that is similarly coupled to the cell cycle. Thus, murine survivin₁₄₀ and survivin₁₂₁, despite their similarity, may have quite different spectra of biologic activity. The complexity of this system is further underscored by the existence of survivin₄₀, which lacks both a complete BIR domain and a coiled-coil region. Therefore, it is reasonable to speculate that the differential expression of these forms of survivin during the cell cycle might affect the balance between cell proliferation and programmed cell death.

We propose that the various forms of survivin differ in their ability to interact with other proteins involved in the regulation of apoptosis (or possibly to self-associate), and therefore, that the role of survivin in apoptosis regulation may vary in a manner that is defined by the spectrum of survivin proteins present, and consequently, by the combinatorial protein-protein interactions that are permitted. In a purified system, we were not able to detect any effect of survivin₄₀ on the inhibition of caspase-3 activity by either survivin₁₂₁ or survivin₁₄₀. Despite these preliminary findings, it is still conceivable that through protein-protein interactions, survivin₄₀, for example, might inhibit and thus regulate the antiapoptotic function attributed to survivin. It follows that alterations in expression of 1 or more of the survivin variants may be predicted to result in a variety of disease processes, either degenerative or proliferative. In that respect, mutations in the NAIP gene have been linked to motor neuron degeneration in one of the spinal muscular atrophies,33 while genetic deregulation of human IAP1 has been associated with the development of MALT lymphomas.34The wide tissue distribution and early developmental expression of the variant forms of survivin make it a likely candidate to be involved in several disease processes. Elucidation of the site-specific function(s) of these survivin variants may therefore provide further insights into the regulation of the apoptotic pathway in a variety of physiologic and pathologic conditions.