ABSTRACT
The mouse Sno gene, a Ski proto-oncogene homolog, expresses two isoforms, SnoN and SnoN2 (also called sno-dE3), which differ from each other in a location downstream from the site of alternative splicing previously described in the human SNO gene. SnoN2 is missing a 138 nt coding segment present in mouse SnoN and human SNON. We have cloned and sequenced the human ortholog of mouse SnoN2, the existence of which was predicted from conservation of the alternative splice donor site that produces the SnoN2 isoform. Mouse SnoN2 and SnoN are expressed throughout embryonic development, in neonatal muscle and in many adult tissues. SnoN2 is the major species in most tissues, but SnoN and SnoN2 are expressed at approximately equal levels in brain. In human tissues, SNON2 is the less abundantly expressed isoform. Expression of mouse SnoN and SnoN2 mRNAs is induced with immediate early kinetics upon serum stimulation of quiescent fibroblasts, even in the presence of the protein synthesis inhibitor cycloheximide, while Ski is not. Interestingly, although both isoforms of Sno are induced, SnoN2 induction is much higher than SnoN. These data are consistent with a role for Sno in the response to proliferation stimuli.
Ski and Sno each have the paradoxical property that they can promote both oncogenic transformation, in which cell proliferation increases, and terminal skeletal muscle differentiation, in which cells withdraw from the cell cycle (1 ,2 ). Thus, they may affect the responses of myoblasts to extracellular signals by proliferating or undergoing terminal differentiation. The v-Ski oncogene was isolated as the transforming component of Sloan-Kettering virus (SKV); when overexpressed, Ski can transform chicken embryo fibroblasts (3 -5 ). Paradoxically, overexpression of either v-Ski or chicken c-Ski can also convert non-myogenic avian cells to the myogenic lineage (1 ). A mutant Ski, tdM5i, fails to convert the cells to muscle, but retains its ability to transactivate the endogenous MyoD and myogenin genes (6 ). Since activation of MyoD and myogenin is insufficient to convert these cells to muscle in the presence of mutant SKI protein, SKI may play a role in myogenic conversion and may function in myogenesis in conjunction with MyoD. Transgenic mice overexpressing a truncated cloned chicken c-Ski cDNA in their skeletal muscles exhibit a marked enlargement of these muscles, specifically type IIb fast myofibers, and diminished body fat (7 ,8 ). The enlarged skeletal muscle phenotype suggests that SKI affects skeletal muscle development and/or maturation. Ski expression is low in developing muscle precursors in embryos and is thus probably not involved in determination (1 ,9 ), but is high in regions that are undergoing active proliferation and differentiation in the muscle (1 ,9 ), hematopoietic (10 -12 ) and neural lineages (13 ).
The human Sno gene (for Ski-related novel) was identified by cloning in low stringency hybridization screens of human cDNA libraries with the chicken v-Ski probe (14 ,15 ). Sno, like Ski, can convert non-muscle quail embryo cells to skeletal muscle and has weak transforming activity (2 ). SKI and SNO proteins are found in the nucleus, but are not members of any recognized transcription factor gene family. Human SNO and SKI share a 106 amino acid region in the N-terminal half of each protein that is 82% identical; the two proteins are 37% identical throughout their length, which is more similarity than members of many families. SKI protein can form a complex with SNON through a C-terminal region that includes a putative [alpha]-helical coiled coil structure (16 ,17 ). Three isoforms of human SNO, SNON, SNOA and SNOI, have been cloned as cDNAs (14 ,15 ). These isoforms differ by inclusion of mutually exclusive alternative exons A (
This paper extends the results of Pelzer et al., who described two alternatively spliced isoforms of Sno in mice, SnoN and sno-dE3, which we call SnoN2 (18 ). We describe the expression of these isoforms in additional tissues in mice and find that SnoN2 is 80% of the expressed Sno mRNA in most tissues, but ~50% in the brain. We have cloned and sequenced human SNON2. In humans SNON2 is a smaller fraction of the accumulated SNO mRNA population and previous workers were unable to detect it (18 ). We show that Sno expression is induced with immediate early kinetics within 30 min of serum stimulation of quiescent mouse fibroblasts, even in the presence of the protein synthesis inhibitor cycloheximide, but Ski is not. Interestingly, although both isoforms of Sno are induced, SnoN2 induction appears to be much stronger than SnoN. Our data are consistent with the possibility that Sno may have a regulatory role in cells such as myoblasts when they divide or differentiate in response to extracellular signals.
C2 (sometimes called C2C12; 19 ,20 ) and 10T1/2 cells were obtained from the American Type Tissue Collection and were cultured in Dulbecco's modified Eagle's medium (DMEM) (C2) or basal medium Earle's (BME) (10T1/2) medium with 10% fetal calf serum on gelatinized plates. FDC-P1 cells, growth factor- dependent mouse myeloid cells provided by Dr J.N.Ihle (21 ,22 ), were cultured in DMEM with 10% fetal calf serum and 25% WEHI-conditioned medium, which supplies IL-3 that the cells require for proliferation. 23A2 mouse myoblasts were derived from C3H 10T1/2 (clone 8) cells following 5-azacytidine treatment and subcloning of determined myoblasts (23 ). 23A2 cells were cultured on gelatinized plates in BME supplemented with 15% fetal calf serum.
10T1/2 cells were made quiescent by culture of confluent dishes in medium containing 0.5% fetal calf serum for 3 days. Serum stimulation consisted of replacing the low serum medium with medium containing 20% fetal calf serum, either with or without supplementation with 10 [mu]g/ml cycloheximide.
RNA was isolated from tissue culture cells or from mouse embryos or tissues by homogenization in guanidinium thiocyanate and ultracentrifugation through CsCl (24 ,25 ). Poly(A)+ RNA was selected by chromatography on oligo(dT)-cellulose using Pharmacia Type 7 (dT)-cellulose.
Poly(A)+ RNA isolated from subconfluent 23A2 mouse myoblasts (23 ) was used to make the cDNA library. Double-stranded cDNA was synthesized by the method of Gubler and Hoffman (26 ), using RT-XL AMV reverse transcriptase (Life Sciences Inc., St Petersburg, FL) for first strand cDNA synthesis. The cDNA was treated with EcoRI methylase (New England Biolabs) and EcoRI linkers were used for ligation to [lambda]gt10 phage arms (Stratagene); phage were packaged using Gigapack Gold extract (Stratagene). From 3.5 [mu]g of poly(A)+ RNA, 1.0 * 106 recombinants were isolated; the library was amplified once prior to screening. To isolate mouse Ski homologs, 400 000 plaque forming units (p.f.u.) of recombinant phage were plated at a density of 40 000 p.f.u./150 mm plate. Duplicate plaque lifts on nitrocellulose filters (Schleicher and Schuell) were prehybridized in 6* SSC, 10* Denhardt's (0.2% each of Ficoll 400, polyvinylpyrrolidone and bovine serum albumin) for 5 h and hybridized in 1 M NaCl, 50 mM Tris, pH 7.4, 2 mM EDTA, 10* Denhardt's, 50 [mu]g/ml sonicated denatured salmon sperm DNA, 2.5 [mu]g/ml each poly(A), poly(U), poly(I) and poly(C) and 0.5% SDS at 60oC for 60 h (27 ). The filters were hybridized with a random hexamer-primed fragment from the human SNOI cDNA that cross-hybridizes with any isoform of Sno (15 ). Filters were washed at 50oC as described (27 ), plus a final 0.3* SSC wash at 50oC. Several positives were rescreened and plaque purified; one was the 2018 bp SnoN2 described here.
cDNA clone fragments were subcloned in pGEM7zf(+) (Promega) and unidirectional nested deletions were made for sequencing (28 ) (Erase-A-Base; Promega). Double-stranded DNA primed by pUC/M13 universal or reverse sequencing primers or specific oligonucleotides was sequenced by the dideoxy chain termination method (29 ,30 ), sequencing both strands of the DNA.
Aliquots of 10 [mu]g total RNA from the indicated mouse tissue, cell line or bone marrow sample were reverse transcribed using random hexamer primers and AMV reverse transcriptase (Boehringer Mannheim). A portion of the first strand cDNA was then amplified using `hot start', Vent DNA polymerase (New England Biolabs), and the indicated specific primers at 0.4 [mu]M each plus 10 [mu]Ci [32P]dCTP at 3000 Ci/mmol. Following an initial denaturation at 95oC for 3 min, the 27 cycles consisted of 95, 62 and 72oC for 1 min each, plus a final elongation at 72oC for 7 min. The number of cycles and amount of reverse transcribed RNA that was used for PCR were chosen in pilot experiments so that the reactions were in the linear range with respect to amount of input RNA (31 ). A sample size of 1.4 [mu]g was used for Sno amplifications from tissue samples, 0.6 [mu]g for Sno induction in 10T1/2 cells and 0.1 [mu]g was used for GAPDH primer PCR. Aliquots of 10 [mu]l of each 50 [mu]l sample after PCR were subjected to 5% acrylamide gel electrophoresis (BioRad), then gels were dried, autoradiographed and analyzed using a phosphorimager (Molecular Dynamics). GAPDH primers were Clontech 5406-1 for human and 5409-1 for mouse. Additional oligonucleotide primer sequences are in Table 1 . Oligonucleotides oH075 and oH074 (Table 1 ) amplify human SNON as a 308 nt product and the human SNON2 variant that deletes the 138 nt as a shorter species (170 nt) (Figs 1 and 6 ). Oligonucleotides oM094 and oM065 amplify a 427 nt product (1302-1729) from mouse SnoN and a 289 nt product (1302-1591) from mouse SnoN2 (Figs 1 and 6 ).
Reverse transcribed 11.5 day mouse embryo total RNA, which expresses both SnoN2 and SnoN isoforms, was primed with oM051 and oM054 using Expand High Fidelity enzyme mix (Boehringer Mannheim Inc.) to obtain SnoN and additional SnoN2 clones. The PCR was programed for 3 min at 95oC, 10 cycles of 94oC for 1 min, 55o for 1 min, 72oC for 2 min, 10 cycles of the same with 20 s annealing cycle extension, then 7 min at 72oC. PCR product was digested with BamHI and SalI, then cloned into pGEM37f(+) cut with BamHI and SalI. The same protocol was used for cloning human SNON2 and variants, with human kidney RNA and oligonucleotide primers oH070 and oH072.
Table 1
The riboprobe for SnoN was prepared from a subcloned 5'-end cDNA fragment terminating at the PstI site at nt 1678 (Fig. 3 A) in the pGEM 3zf(+) vector (Promega). The plasmid was cleaved at the BstEII site at nt 1207, then used with the SP6 primer and [32P]UTP (800 Ci/mmol; NEN) in vitro to transcribe a uniformly labeled single-stranded RNA probe. A cyclophilin template (Ambion) was used to make a cyclophilin probe in parallel. Both probes were gel purified on a denaturing 6% acrylamide gel prior to use. In pilot studies we determined that the abundance of cyclophilin over Sno RNA required 20-fold dilution of the cyclophilin probe to keep the band intensities within range. Thus, we mixed 105 c.p.m. of SnoN and 5000 c.p.m. of cyclophilin gel purified, in vitro transcribed [32P]UTP-labeled riboprobes with each 50 [mu]g total RNA sample. These were denatured at 90oC and annealed overnight at 42oC. Samples were treated with RNase A/RNase T1 as described (Ambion #1410), ethanol precipitated and electrophoresed on a denaturing 6% acrylamide gel. Gels were dried and signal intensities visualized and analyzed with a phosphorimager (Molecular Dynamics).
Alignment of the mouse, human and chicken Sno sequences at the site of the 138 nt deletion shows that mouse and human share a consensus splice donor sequence, whereas chicken does not (Fig. 2 A), as noted previously (18 ). To determine if human cells express the SNON2 mRNA species that is predicted from conservation of the consensus splice donor sequence and its use in mouse SnoN2, we performed RT-PCR using human total RNA and primers that flank the SNON2 alternative splice site. These primers do not span the SNOI/SNOA alternative splice location and thus should detect SNON (311 nt) and SNON2 (173 nt), but not SNOA or SNOI (Fig. 6 ). Humans express a SNON2 species at low levels in addition to the major SNON species (Fig. 2 C). We confirmed that these PCR bands both contain SNON sequence by PCR amplification without radionuclides and hybridization with a human SNO probe (data not shown). We cloned and sequenced three SNON2 clones from human tissue. These nucleotide sequences confirm that human SNON2 is spliced as in the mouse, remaining in-frame and predicting the translation of a similar SNON2 protein missing 46 amino acid residues (Figs 1 and 6 ). This is the first report of the human SNON2 isoform. Phosphorimager quantitation in both blot and 32P PCR experiments showed that human SNON2 is expressed at 1-2% of the level of SNON RNA. It is possible that this is an overestimate of human SNON2, since smaller species can PCR amplify preferentially (although this did not happen with mouse SnoN2; Fig. 5 A and B). This contrasts with the mouse, where SnoN2 is often the most abundant Sno isoform expressed (Figs 3 B and 5 ).
We examined several questions about the mammalian Sno exon structure. We confirmed that mouse Sno shares the exon 1/2 junction position with chicken Sno and with Ski by nucleotide sequencing of a genomic clone that includes exon 1 (data not shown). Chicken Sno has an intron in the 5'-untranslated region (33 ). We used PCR to compare reverse transcribed RNA with genomic DNA templates and found that mouse Sno does not interrupt the 500 nt 5'-untranslated region with an intron; PCR across this region produces the same size product whether the template is reverse transcribed RNA or genomic DNA (data not shown). Thus, the largest coding exon, which is exon 2 in chicken Sno, is exon 1 in mammalian Sno (Figs 1 and 6 ).
A small exon 2 in mouse Sno has been proposed and its junction locations were confirmed by sequencing the genomic DNA around exon 3 (18 ). This exon 2 is not a separate exon in the chicken Sno gene (Fig. 1 ; 18 ,33 ). [This should not be confused with the unrelated `exon 2' that is present in v-ski and alternatively spliced in chicken c-Ski (34 ,35 ), which is not conserved in Sno or in mammalian Ski.] It has been shown that the alternatively spliced exon 3 is continuous in genomic mouse DNA and that alternative splicing of the SnoN2 region is not due to exclusion of an exon that consists only of the region that is deleted in SnoN2 (18 ). We determined the nucleotide sequences of RT-PCR cloned human variants that lack exon 3 and those that lack both exons 2 and 3 and found that the 2/3 and 3/4 exon boundaries are conserved between the mouse and human Sno genes. The human SNO variant clones that correspond to the omission of exon 3 (SNOD3 for
Many nuclear proto-oncogenes and/or transcription factors are induced by serum stimulation of quiescent cells, responding within the first 2 h after stimulation (36 -38 ). We have previously shown that Sno and Ski are cell cycle regulated in synchronized cycling myeloid cells, peaking in the early to mid G1 phase of the cell cycle (12 ). To examine whether Sno and Ski exhibit cell cycle regulation in another cell type and as part of the characterization of Sno, we examined Sno and Ski expression in response to serum stimulation of serum-deprived fibroblasts. Quiescent 10T1/2 cells were stimulated with serum and RNA was isolated at various times thereafter. These RNA samples were then electrophoresed, blotted and hybridized with the SnoN probe.
The mouse Sno gene encodes two variants of SnoN, SnoN and SnoN2 (which has also been called sno-dE3; 18 ). Mouse orthologs of human SNOA or SNOI have not been detected. Human and mouse Sno genes are single copy (15 ). Mouse Sno variants SnoN2 and SnoN differ from each other only by inclusion (SnoN) or exclusion (SnoN2) of a 138 nt segment encoding 46 amino acids in-frame. Pelzer et al. have proposed that these 46 amino acids encode several potential kinase recognition domains (18 ). SnoN2 has also been called sno-dE3 for
We cloned and sequenced SNON2 mRNAs from human tissue and confirmed that human SNON2 is spliced as in the mouse, remaining in-frame and predicting translation of a similar SNON2 protein missing 46 amino acid residues (Fig. 1 ). SNON2 constitutes only 1-2% of the SNO mRNA (Fig. 2 C). SnoN2 is the most abundant Sno isoform in many mouse tissues (Fig. 3 ; 18 ). Three out of 43 clones obtained during RT-PCR cloning were SNON2, agreeing with our RT-PCR data that SNON2 is not extremely rare. These were full-length clones and thus did not have a significant size advantage in the PCR reaction or cloning.
A schematic summary of the Ski/Sno family is presented in Figure 6 . We note that there is alternative splicing in Ski but not Sno in birds and in Sno but not Ski in mammals. The mouse SNON and SNON2 isoforms contain 10 fewer amino acid residues than the human because of small internal deletions (Fig. 1 ). Human SNOA, SNOI and SNON have been previously described (14 ,15 ), but the exon structure and additional SNON2 isoform are just emerging (this paper; 18 ,33 ).
Definition of the exon structure in mouse Sno (18 ) and its confirmation in human (this paper, Figs 1 and 6 ) suggests that the human SNOA structure previously described (14 ) results from usage of an alternative version of exon 2. The SNOI sequence divergence was previously thought to result from an alternative version of exon 3 in which the divergent insertion region is flanked by identical regions (15 ). That was called exon 3 in anticipation of finding mammalian homologs for the `exon 2' that is alternatively spliced in chicken c-Ski, but which we now know is absent from mammalian Ski and Sno. However, since there is a small exon 2 in mammalian Sno that is not a separate exon in the chicken Sno gene, it is after all exon 3 (Fig. 1 ; 18 ,33 ). Now that the exon structure in mouse and human Sno is defined, it is clear that the SNOI divergence begins at the start of what we now call exon 3 (Figs 1 and 6 ) and may arise from splicing of an alternative version of exon 3. SNOI could not be produced by failure to excise an intron, since there is not a consensus 5' splice site in the SnoI cDNA sequence at the point of divergence nor a 3' consensus site at the point where SNOI unique sequence rejoins SNON, as noted previously (15 ). We have not experimentally confirmed that the SNOI sequence is an alternative exon flanked by consensus splice sites in the genome; sequencing of the human SNO gene will resolve this. SnoN2 described here and previously (18 ) arises from alternative use of a splice donor within exon 3; this occurs rarely in humans, but SnoN2 is often the most abundant Sno species expressed in the mouse. Future experiments will define whether these isoforms function differently.
SnoN/N2 expression is diminished in adult mouse skeletal muscle compared with neonatal skeletal muscle (Fig. 3 ). Ski expression is elevated in the regenerating axolotl limb blastema (40 ). It will be of interest to examine whether SnoN/N2 expression is increased in regenerating muscle when satellite cell proliferation is stimulated.
Sno expression is induced upon serum stimulation of quiescent fibroblasts, even when stimulated in the presence of the protein synthesis inhibitor cycloheximide. In contrast, Ski is not induced by serum stimulation in this cell type. Thus, Sno is an immediate early gene but Ski is not. Immediate early genes are those that are induced without requiring protein synthesis; delayed early genes are those that require protein synthesis for induction. Some authors refer to immediate early genes as those induced minutes after stimulation and delayed early as those induced hours after stimulation, both in the absence of protein synthesis (41 ). Sno induction is detectable by 30 min, so it is also immediate early by this criterion. Another term is proliferation stimulus-responsive gene (41 ). Such early response genes are thought to play important roles in regulation of the re-entry of cells into the cell cycle. We found that the SnoN2 isoform is much more strongly induced than SnoN. This raises the possibility that there may be regulatory influences on the alternative splicing that produces the SnoN2 isoform.
Many immediate early genes are cell cycle regulated but some are not; this may depend on the cell type assayed (41 ). We showed earlier that Ski and Sno expression is modulated in a different cell type, peaking in early to mid G1 in synchronized myeloid cells in response to progress in the cell cycle (12 ). Thus, Sno is both proliferation stimulus responsive in fibroblasts and cell cycle regulated in myeloid cells. In contrast, Ski is not proliferation stimulus responsive in fibroblasts but is cell cycle regulated in myeloid cells. In addition Ski responds to IL-3 or GM-CSF growth factor depletion and addition in myeloid cells (12 ), whereas Sno does not (Pearson-White, unpublished results). Ski is induced 6-fold by phorbol ester administration in megakaryocytic hematopoietic cells (11 ). Thus, there are differences in growth factor responses and cell type-specific cell cycle regulation between the Ski and Sno genes. These data are consistent with potential differing roles for Ski and Sno in regulating the response of different cell types to extracellular signals in which they either enter (proliferate) or exit from (terminally differentiate) the cell cycle.
We thank Mark Lovell for the rhabdomyosarcoma tumor sample and primary human myoblasts and Bill Pearson for comments on the manuscript. We thank the University of Virginia Child Health Research Center Molecular Genetics Core for plasmid purification. This work was supported by NIH grant RO1 HD27202 and Muscular Dystrophy Association research grants to S.P.-W.
*To whom correspondence should be addressed. Tel: +1 804 982 0756; Fax: +1 804 982 3993; Email: sp3i@virginia.edu
Name
Location
Sequence (5' -> 3')
Size (nt)
oH075
nt 1950-1970
CTTCTGCTGTATCCCAGTCTA
oH074
Anneals nt 2237-2258
TCACAAGAAGCGGAGATGAAGC
308, 170
oM094
nt 1301-1318
CTGCTGCGTCCCAGTCTA
oM065
Anneals nt 1702-1728
TGAACTGCTCAGCATCTCCACCTCCAT, anneals SnoN2 nt 1565-1591
427, 289
oH070
nt 688-721
AACAGAAGAGT
oH072
Anneals nt 2758-2791
CAGTAATACACA
2103, 1965
oM051
nt 43-81
GTTTAAAGCACCA
oM054
Anneals nt 2107-2133
ACACAAAGCCCT
2092, 1954
REFERENCES