Proliferation-associated SNF2-like Gene (PASG): A SNF2 Family Member Altered in Leukemia¹

Our understanding of the molecular mechanisms that balance hematopoietic cell proliferation, survival, differentiation, and apoptosis remains incomplete. However, a common feature regulating and supporting these processes is the dependence on soluble and/or membrane-bound cytokines and their cognate receptor systems (1, 2). After the binding of cytokines to their receptors, a series of signal transduction pathways are activated that result in changes in cellular physiology, growth, and/or differentiation. In addition, the withdrawal of particular cytokines or survival factors often leads to apoptosis (1, 2). These regulatory mechanisms also play important roles in leukemic cell transformation and growth (3).

The SNF2 subfamily of helicases is critical in the regulation of chromatin remodeling, DNA replication, repair, recombination,methylation, and transcription (4, 5, 6, 7, 8). SNF2 polypeptides are characteristically ATPases, share several functional domains, are differentially expressed, and associate in protein complexes to perform their effector functions (4, 5, 6, 7, 8). Yeast members of the SNF2 family are the basis for the derivation of the name SNF, which stands for “sucrose non-fermenter” (9, 10). Members of this family include genes involved in DNA repair (RAD5, RAD16, RAD54, and ERCC6), transcriptional regulation (Brahma, ATRX, and SNF2), and control of cell proliferation [MOT1, STH1, ISWI, BRG, and BRM(11, 12, 13, 14, 15, 16)]. Additionally, DDM1, a SNF2 homologue identified in flowering plants, is known to facilitate DNA methylation,possibly through altering chromatin accessibility to methylation-forming complexes during replication (17).

The multiple functions of SNF2 family members involving DNA homeostasis and transcription suggest that these genes may play important roles during embryogenesis and in the pathogenesis of cancer. For example,mutations in the SNF2-like helicase ATRX result in a syndrome characterized by X-linked mental retardation, congenital abnormalities,and transcriptional down-regulation of α-globin resulting inα-thalassemia (18). Mutations in RAD54 have been linked to tumor formation (19, 20). The human SNF5 gene product, a BRM-associated factor, has been reported to show germ-line and somatic mutations associated with the development of malignant rhabdoid tumors (21, 22). The HBRM protein interacts with Rb to repress the activation of E2F1, resulting in disrupted cell cycle regulation (23).

A role for SNF2 chromatin remodeling helicases in leukemia has not been established, although alterations in other chromatin remodeling proteins such as MLL and histone-modifying enzymes have been shown to be involved in translocations associated with various types of leukemia(24). To identify genes involved in cytokine-dependent leukemic cell proliferation, survival, or apoptosis, our laboratory used a differential RNA display screen after SCF³withdrawal from the cytokine-dependent human megakaryoblastic leukemia cell line MO7e (25). One of the mRNAs that was found to be profoundly down-regulated after cytokine withdrawal was identified as a member of the SNF2 gene family of chromatin remodeling helicases(26, 27, 28). This gene was originally identified as a partial cDNA called murine lymphocyte specific helicase, lsh, and its human counterpart is referred to as Hells (helicase lymphoid specific; Refs. 29 and 30). This report presents the full-length cDNA open reading frame for this human SNF2-like putative helicase, the locus for which is designated SMARCA6(SWI/SNF2-related, matrix-associated, actin-dependent regulator of chromatin, subfamily A, member 6).⁴The protein product of SMARCA6 is referred to as PASG(proliferation-associated SNF2-like gene) because of the association of expression with cell proliferation (31). Analysis of leukemia samples reveals a high incidence of an alternative PASG transcript containing a deletion of a region known to be critical for the regulation of transcription by yeast SNF2 gene products. These data suggest that PASG may play a role in leukemogenesis.

MO7e cells were cultured in RPMI 1640 containing 1%penicillin/streptomycin, 1% l-glutamine (all Life Technologies, Inc.), and 10% FBS (BioWhittaker, Walkersville, MD)supplemented with SCF (Genetics Institute, Cambridge, MA). Other cell lines included HL60, Spurlock, and K562 (myeloid leukemia-derived cell lines); Molt4 and CEM (T-cell-derived lymphoma/leukemia); Raji (B-cell lymphoma); PC-3, DU145, LNCAP, Lan2, Lan5, and Lan6 (prostate cancer-derived cell lines); EW-1 (Ewing’s sarcoma-derived cell line);and EBV-transformed lymphoblast cell lines. Apoptotic cell death after SCF withdrawal from MO7e cells was evaluated by acridine orange staining as described previously (32). Briefly, MO7e cells(1 × 10⁵ cells/well) were cultured in 96-well plates in the absence of SCF. Cells were collected at various times and stained with acridine orange (Sigma, St. Louis,MO) at a final concentration of 4 μg/ml for 10 min at room temperature. The stained cells were exposed to light with a wavelength of 570 nm and observed using a fluorescence microscope (Nikon Inc.,Melville, NY). Apoptotic cells were identified by their condensed and/or fragmented nuclei, whereas normal cell nuclei had a typically even and rounded fluorescence pattern (32).

MO7e cells were cultured with or without SCF for 6 h, at which time the cells were collected, and RNA was isolated using Trizol (Life Technologies, Inc.). Purified MO7e total RNA was analyzed for differential expression of mRNA as described by Liang and Pardee(33) with a Differential Display Kit (Display Systems, Los Angeles, CA). Each PCR product was separated by electrophoresis in 5%acrylamide gels followed by silver staining (34). Bands of interest were eluted from the gel and reamplified using the same set of primers to test for specificity and reproducibility, followed by cloning (TA Cloning vector pCR2.1; Invitrogen, Carlsbad, CA) and sequence analysis.

Quantitation of mRNA expression levels was determined by Northern blot hybridization in formaldehyde denaturing agarose gels. Membranes were hybridized overnight at 65°C with a [³²P]dCTP(New England Nuclear, Boston, MA)-labeled 1474-bp cDNA probe representing nucleotides 692-2165 of the PASG cDNA sequence. An 18S rRNA radiolabeled probe was hybridized to membranes that had been stripped to assess RNA loading for some blots. Visualization and quantitation of PASG mRNA or rRNA expression were done with either X-ray film or a PhosphorImager (STORM 860; Molecular Dynamics,Sunnyvale, CA). Total RNA from various normal adult tissues was from Stratagene (La Jolla, CA).

A cDNA library made from total MO7e RNA cloned into pUC19 was screened with a PASG cDNA fragment to isolate a full-length PASG cDNA. 5′ RACE was performed to extend the extreme 5′ end of the cDNA. Purified MO7e total RNA (5 μg) was reverse-transcribed using a downstream 20-mer gene-specific primer (GSP 1; 5′-CCACACAGAGATTAGTAGAG). This primer had 342 bp of overlap with the 5′ end of the most 5′ cDNA library clone. A polyadenylation tail was added to the 3′ end of the purified product, using terminal deoxynucleotidyl transferase (Life Technologies, Inc.). This resulting product was used as a template for PCR, which was done using primers A(5′-CCAGTGAGCAGAGTGAGGACTCGAGATTAATGCTACGTTTTTTTTTTT) and B(5′-TCCTCATAACTGGCTTCTC).

The PCR product was diluted 1:5,000 in H₂O and used as a template for nested PCR with the following upstream (primer C) and downstream (primer D) primers: (a) primer C, 5′-GACTCGAGATTAATGCTACGTT; and (b) primer D,5′-GACTCCTTTTTTCTCTCCAA. All 5′ RACE primers were made by Marshall University DNA Core Facility (Huntington, WV). Sequencing of cDNA clones was accomplished with either an automated DNA sequencer (PRISM,Model ABI377; Applied Biosystems, Inc. through the University of Cincinnati DNA Core) or a Sequenase kit (Ver. 2.0, Cleveland, OH) and ³⁵S-labeled dATP (specific activity of 1500 Ci/mmol; New England Nuclear). Regions of sequence overlap from multiple clones were used to confirm continuity of sequences.

BAC clones pBeloBAC 11 16208, 16209, and 17189 (Genome Systems,Inc., St. Louis, MO) were detected using specific primer pairs derived from human PASG cDNA: (a) 5′-CGAATGCTGCCAGAACTAAA and 5′-TGCTGAAGTTGAAATCTCTG (clones 16208 and 16209); and (b)5′-CGGGTGAGTGTCCAGGCATG and 5′-GCTGTTCTTCCTCTTCTAGC (clone 17189). BAC clones were expanded and purified according to the supplier’s protocol. Five overlapping cDNA probes covering the entire PASG open reading frame were used for exon mapping. Exon-specific primers were used to sequence across exon/intron boundaries.

PCR on human/rodent somatic cell DNA hybrid panels from Coriell Institute for Medical Research (Camden, NJ) allowed determination of which chromosome contained the PASG gene. Radiation hybrid mapping allowed for more specific chromosomal localization (Stanford G3 Radiation Hybrid Panel, RH01.02; Research Genetics, Inc., Huntsville,AL). For FISH, a 30.9-kb BAC (BAC plasmid clone 16209; Genome Systems,Inc.) was shown to contain a genomic insert of 23.5 kb, which spanned part of the PASG gene. This plasmid was used as a probe for FISH(35). The purified genomic clone was labeled with digoxigenin-11-dUTP by nick translation (Boehringer Mannheim,Indianapolis, IN). The probe mixture, including the labeled PASG genomic clone and repetitive human DNA selected at C₀t 1 (initial concentration of DNA multiplied by time in seconds) dissolved in Hybrisol VI (Oncor, Gaithersburg,MD), was hybridized to normal cytogenetic metaphase slide preparations. The slides were treated with fluorescein-labeled antidigoxigenin antibody and counterstained with propidium iodide,4′,6-diamidino-2-phenylindole/antifade blue, and antifade, allowing visualization under light microscopy of the areas where labeled DNA hybridized to chromosomes. Giemsa staining selectively marked specific chromosomal regions to create unique banding patterns for each chromosome. These distinct bands served as landmarks along the chromosome length and enabled identification of the specific site or band to which the cloned genomic DNA probe had hybridized.

A complete open reading frame for PASG cDNA was amplified using a 5′28-nucleotide oligomer (5′-TCAGGAGCTCAGGATGCCAGCGGAACGG-3′) that contained a SacI site and a Kozak translation initiation sequence (AXXATGC). A 3′ 30-nucleotide oligomer(5′-TGGATCCCGGGCAAACAAACATTCAGGACT-3′) included a XbaI site. Green fluorescent protein vector (pEGFP-N1; Clontech) was used for cloning and subsequent expression analysis. To investigate the role of a 5′ putative NLS signal, the motif was mutated using PCR-mediated site-directed mutagenesis. An inactivating mutation in the NH₂ terminus putative NLS was constructed using the mutagenic primer 5′-ATGAGGAAAAATAGAGGAAG-3′, where the underlined nucleotide creates a K to N substitution in the NLS (see Fig. 7 for a schematic; Ref. 36). Resulting PCR fragments were cloned into pGEMTeasy vector (Promega, Madison, WI) and then subcloned into the PASG-EGFP construct used for nuclear localization studies.

These constructs were transfected into COS-7 or NIH 3T3 cells using Effectene (Qiagen) or DOSPER (Boehringer Mannheim) liposomal reagents according to the manufacturers’ instructions. Forty-eight h later,cells were split at low density onto multiwell LabTek (Nalge Nunc International, Naperville, IL) slides. Cells were allowed to adhere for 24 h and pulsed with 10⁻⁵ m BrdUrd(Sigma) for 16 h. Cells were rinsed once with PBS, fixed for 15 min at 25°C with 3.7% paraformaldehyde (Sigma), permeabilized with 0.3% Triton X-100 in PBS for 15 min at 25°C, and incubated with anti-BrdUrd monoclonal antibody (Accurate Scientific, San Diego, CA) at a dilution of 1:500 for 1 h at 37°C in a solution of 5 mg/ml BSA, 0.5% NP40, 20 mm MgCl₂, 0.02 unit/μl DNase I (Roche) in PBS. Cells were washed with PBS and incubated with 1:200 dilution of rhodamine-labeled donkey antirat antibody (Jackson Immunoresearch, West Grove, PA) for 1 h at 37°C. Cells were washed with PBS, mounted with Fluoromount-G(Southern Biotechnology Associates, Birmingham, AL), visualized with a Nikon Optiphot microscope, and photographed using a DC120 digital camera (Eastman Kodak, Rochester, NY). Photos were imported into Photoshop (Adobe) and processed for publication.

To determine the degree of disruption of nuclear targeting, 100 EGFP-positive cells were counted and visually scored for nuclear-only localization of the transfected protein. The individual performing the scoring was blinded as to the constructs being scored. The transfections were repeated three times, and the average percentage of EGFP-positive cells with nuclear-only localization was compared. Significant differences between the means were determined using Student’s t test in Microsoft Excel.

Western blotting of PASG-EGFP products was accomplished as follows. Cells were washed twice with PBS, exposed to RIPA2 buffer [20 mm Tris (pH 7.4), 10% glycerol, 137 mm NaCl,0.1% SDS, 1% Triton X-100, 2 mm EDTA, 0.5% sodium deoxycholate with 0.6 mm phenylmethylsulfonyl fluoride, 4.5μg/ml aprotinin, and 1 μg/ml pepstatin], and collected with a rubber policeman. The cell lysate was shaken vigorously for 20 min at 4°C and microfuged at high speed for 10 min at 4°C. Aliquots of the supernatant were snap-frozen in liquid nitrogen and stored at −80°C. Protein concentrations were determined using Protein Assay (Bio-Rad,Hercules, CA) according to the manufacturer’s instructions. Approximately 50 μg of total cell lysate were denatured in 6×SDS/2-mercaptoethanol loading buffer (37) at 100°C for 10 min, electrophoresed on a 10% polyacrylamide gel, and transferred to nylon membranes (Amersham). These membranes were reversibly stained with Ponceau S to determine the evenness of transfer and loading(38). Membranes were blocked with 10% nonfat dry milk in TBST for 1 h at 37°C and exposed to anti-GFP (Clontech) mouse monoclonal antibody at a dilution of 1:1000 in 10% milk/TBST for 1 h at 25°C. Membranes were washed in TBST and incubated for 1 h at 25°C in alkaline phosphatase-conjugated goat antimouse secondary antibody (Promega) at a 1:5000 dilution in 10% milk/TBST. After washing in TBST and TBS, blots were developed with nitroblue tetrazolium/5-bromo-4-chloro-3-indoyl phosphate (Sigma) according to the manufacturer’s instructions. Blots were scanned and processed for publication using Adobe Photoshop.

Total RNA was extracted from bone marrow, peripheral blood mononuclear cells, fresh or frozen tissue, and cultured cells using Trizol reagent. Full-length cDNA was synthesized using Superscript RNase H-Reverse Transcriptase (Life Technologies, Inc.). To detect a 75-bp deletion(from nucleotides 2014–2088) first identified in the MO7e cDNA library, primers were designed that would amplify either a 255-bp wild-type or a 180-bp deleted region: (a) 5′ primer,5′-CAGAGATTTCAACTTCAGCAG-3′; and (b) 3′ primer,5′-AGGCGATAAACAACAACTGG-3′. Reverse-transcribed templates were amplified for 30 cycles (95°C for 30 s, 60°C for 45 s,and 72°C for 30 s) using a Gene Amp PCR system 9700(Perkin-Elmer/Applied Biosystems, Foster City, CA). The amplified products were analyzed by agarose gel electrophoresis. Samples containing a shorter product on the agarose gel were cloned into pGEM-T-easy TA cloning vector (Promega). Cloned inserts were sequenced to confirm the presence of the transcript with the 75-bp deletion. Negative controls for deletion analysis included standard water controls and patient total RNA samples that had not been subjected to reverse transcription.

To analyze the gene products associated with growth and apoptosis in cytokine-dependent leukemia, a differential display screen was used to identify mRNAs whose expression was altered after cytokine withdrawal(SCF) from the human megakaryoblastic leukemia cell line MO7e. The percentage and timing of apoptosis after SCF withdrawal are shown in Fig. 1,A. One of the cDNAs identified from this screen had significant homology to a conserved domain shared by a variety of DNA and RNA helicases (data not shown). Northern blot analysis demonstrated that this cDNA recognized an approximately 3.5-kb mRNA and confirmed the decrease in expression after cytokine withdrawal (Fig. 1,B). A higher molecular weight species of approximately 7.5 kb (data not shown) is hypothesized to represent a precursor of the 3.5-kb mRNA, but this has not been formally proven. To determine whether the decrease in PASG expression was related to the induction of apoptosis or growth arrest, MO7e cells were cultured in the absence of FBS in the presence of 20 ng/ml SCF (Fig. 1,C). Cells cultured under these conditions undergo growth arrest but show minimal apoptosis (Fig. 1,D). Northern blot analysis demonstrates that under such conditions of growth arrest without the induction of apoptosis, the level of PASG mRNA declines dramatically (Fig. 1,C). However, 24 h after returning cells to FBS, the level of PASG mRNA again increases, demonstrating that the cells were still viable and that the induction of proliferation by the addition of FBS results in an increase of PASG mRNA (Fig. 1 C). At approximately 8–12 h after the addition of FBS, a rise in[³H]thymidine incorporation is detectable,which continues to increase along with cell numbers over 24 h(data not shown). These results link PASG transcript expression more closely with proliferation than with the inhibition of apoptosis.

A cDNA including a putative complete open reading frame was constructed using overlapping cDNA clones obtained from a MO7e cDNA library and 5′RACE (Fig. 2). This open reading frame sequence was confirmed by sequencing cDNAs from normal human thymus. The position of two in-frame stop codons in the extreme 5′ region followed by a Kozak consensus ATG start site(39, 40) as well as the presence of a 3′ TAA termination codon suggests that the 2514-nucleotide sequence (838 amino acids) is the full-length open reading frame (Fig. 2). The cDNA sequence includes a 105-nucleotide 5′ UTR and a 477-nucleotide 3′ UTR containing a cognate polyadenylation sequence (Fig. 2). We believe that the 5′ UTR should be approximately 500 bp long to account for the 3.5-kb transcript size estimated from Northern blotting, but reverse transcription through a highly GC-rich area in the 5′ region upstream from the start site was unsuccessful. This human sequence includes an open reading frame that is 230 amino acids longer at the NH₂ terminus than the previously identified murine cDNA and adds the 5′ and 3′ UTRs to the previously published sequence (29, 30). These sequences have also been confirmed by sequencing genomic BAC clones. Domains originally shown to be characteristic of other DNA and RNA helicases are shown in Fig. 2,as highlighted in black (I, Ia, II, III, IV, V and VI),whereas those domains more recently shown to also share significant homology among the SNF2 subfamily are depicted by gray highlighting (domains IVa and IVb; Refs. 26, 27, 28).

Based on protein sequence alignments, this cDNA is most homologous to DDM1 and SNF2L (the human homologue of Drosophila ISWI)while sharing various degrees of homology to the rest of the SNF2 family (Fig. 3, A and B) (27, 28). Significant homology with DDM1 and SNF2L also exists outside of the classical helicase domains, suggesting a relatively recent evolutionary divergence (Fig. 3, A and B; Refs.27 and 28). This cDNA is one of the smallest known members of the SNF2 gene family identified thus far, being approximately the same size as RAD54 (Fig. 3,B). However,PASG is significantly more homologous to SNF2L than to human RAD54,suggesting a shared function with SNF2L (Fig. 3,A; Refs.27 and 28). Of interest is that the spacing of conserved motifs in PASG differs from other known members of the SNF2 family in that PASG has additional amino acids inserted between domains IV and V (Figs. 2,3). The functional significance of these additional amino acids is unclear at this time.

Initial chromosomal assignment of the PASG gene was accomplished by using a human/rodent somatic cell library and performing genomic PCR for regions in the 5′ and 3′ regions of the cDNA. These experiments demonstrated that the PASG gene was located on human chromosome 10(data not shown). A radiation hybrid human genomic panel was then used to further map the PASG gene to chromosomal 10q23–24 (data not shown). This chromosomal assignment was confirmed using in situhybridization with a BAC clone containing the PASG gene (Fig. 4).

Two separate BAC clones containing overlapping regions of the PASG gene were used to determine the intron/exon boundaries (Table 1). The human PASG gene has 22 exons that encode the open reading frame(Table 1; Fig. 5). The location of conserved helicase domains is also noted. The definition of conserved helicase domains is based on the analyses by Gorbalenya et al. (26) as well as Richmond and Peterson (9). Using these criteria, the following observations can be made for PASG genomic organization: most of the conserved helicase domains lie within a single exon (Fig. 4; Table 1);however, domain II, which contains the DEGH motif, could be considered to be split between exons 10 and 11. Whereas the DEGH and immediately surrounding sequences are contained within exon 11, there is an intron situated between the glutamine (Q) and histidine (H) located 10 and 9 amino acids 5′ to the DEGH motif, respectively. Thus, two amino acids(L and Q) are contained within exon 10, whereas the remainder of domain II lies within exon 11. Domain IVb is also split between exons 16 and 17. Domain IV lies completely within exon 13, based on the schema described previously (Fig. 5; Table 1; Ref. 9). Intron size was determined by direct sequencing or estimated by the size of PCR products spanning two consecutive exons.

The first in-frame ATG that follows two stop codons has a surrounding sequence consistent with a Kozak consensus motif translational start site (GGC-ATG-CCA; Fig. 2; Ref. 39). In addition, this first ATG appears to be represented in the dominant RNA transcript as assessed by RT-PCR (Fig. 6,A). Of interest, however, is the presence of an alternative splicing variant that results in the deletion of the first ATG (Fig. 6,B). The donor and acceptor sites that form the sequence basis for the alternative splicing are shown in Table 2. This second ATG (at position +49) also contains surrounding nucleotides for an acceptable Kozak box(GCA-ATG-GTT). The alternative splicing variant results in a predicted protein that is 16 amino acids shorter at the NH₂ terminus than that encoded by the dominant transcript. This shorter transcript has been observed in all tissues and cell lines sampled thus far as a minor mRNA species (data not shown). These data demonstrate that the size of the human PASG product is encoded by a locus of approximately 40,000 nucleotides situated at chromosomal 10q23-q24; the gene locus contains 22 exons contributing to the open reading frame. In addition, there is evidence for alternative splicing at the 5′ end of the gene that results in a protein with a predicted alternative NH₂ terminus.

Sequence analysis identified a putative NLS in the NH₂ terminus of the protein and is indicated by the underlined residues in Fig. 7,A. To investigate the role of this putative NLS, we used PCR-mediated site-directed mutagenesis (Fig. 7,A). The mutation was cloned into a full-length PASG construct that was fused to the EGFP, and the construct was transfected into COS-7 cells followed by fluorescence microscopy. Mutation of the residue significantly(P < 0.001) decreases the nuclear localization of PASG when compared with EGFP-PASG. (Fig. 7,C). Fig. 7 B shows the dramatic degree of disruption of PASG addressing observed when the NLS signal is mutated. Whereas the vast majority of cells transfected with the EGFP-PASGΔNLS show disrupted subcellular localization, a small percentage of cells maintain proper nuclear location, suggesting that PASG may contain an additional NLS motif. A comparable degree of disruption is seen when constructs are transfected into NIH 3T3 or COS-7 cells,demonstrating that this effect is not cell-type specific.

To ensure that the constructs produced full-length protein products,Western blotting of PASG-GFP products was performed. At 48 h after transfection, whole cell lysates were harvested. Anti-GFP Western blotting shows that GFP-reactive bands of the appropriate size are present in cells transfected with EGFP-PASG and EGFP-PASGΔNLS. Based on the sequence data presented here, PASG is predicted to have a molecular weight of approximately 97,000. The EGFP portion of the fusion protein is predicted to add approximately M_r 27,000 in size, suggesting that a full-length EGFP-PASG protein should have a molecular weight of approximately 124,000. As observed in Fig. 7 D, one of the two GFP-positive bands migrates at approximately M_r 124,000. A band appearing to run M_r 10,000 less could be explained by an alternative translational start at a downstream methionine (M86),which is also surrounded by a conserved Kozak translational start sequence. A PASG polypeptide beginning at M86 would have a predicted molecular weight of 9,400 less than that of full-length PASG.

The expression pattern of PASG mRNA in normal adult tissues was determined using Northern blot hybridization. The highest levels of expression of PASG mRNA are observed in the adult thymus and testis(Fig. 8,A). However, expression is also observed at dramatically lower levels in other tissues, such as the uterus, small intestines,colon, and peripheral blood mononuclear cells (Fig. 8, A and B). PASG mRNA is also present at relatively high levels in mononuclear cells isolated from normal bone marrow (Fig. 8,B, Lanes 2–5). Peripheral blood mononuclear cells,representing mostly PBLs freshly isolated from normal individuals,demonstrate detectable levels of PASG mRNA, although significantly less than that observed in the thymus or bone marrow [Fig. 8,B,compare Lanes 2–5 (bone marrow) with Lanes 10–11 (thymus) and Lanes 6–8 (PBLs)]. However,activation of PBLs by a 3-day exposure to ConA results in a profound increase in the level of PASG mRNA to the level of that seen in thymus(Fig. 8 B, Lane 9).

In light of the expression pattern of PASG mRNA in normal tissues, we examined a variety of malignant cell lines, including those derived from myeloid and lymphoid leukemias (Fig. 8,C). Whereas the level of PASG mRNA varied considerably among the different cell lines,some expression was detectable in all cell lines tested by Northern blot hybridization. To examine the expression of PASG mRNA in hematopoietic malignancies, samples of AML and ALL were analyzed. The results showed that all samples of AML expressed relatively high levels of PASG mRNA, regardless of whether they were obtained at diagnosis or relapse (Fig. 8,D, Lanes 3–6). Whereas an extensive analysis of all FAB subtypes of AML has not been performed,samples from all FAB subtypes (M6 subtype not tested) have shown PASG mRNA expression (Fig. 8; data not shown). In addition, samples of both T-cell and pre-B-cell ALL also showed expression of PASG mRNA, although there was more variability, with some samples having relatively low levels compared with that observed in AML samples [Fig. 8 D,compare Lanes 3–6 (AML) with Lanes 7–11(ALL)].

These results show that PASG mRNA is differentially expressed in normal and neoplastic human tissue and that it does not appear to be restricted to lymphocytes. In addition, the level of human PASG mRNA shows a significant increase when PBLs are activated and stimulated to proliferate after exposure to ConA, similar to that observed for the mouse homologue (29).

During the screening and sequencing of cDNAs from the original cDNA library derived from the myeloid leukemia cell line MO7e, a cDNA clone was identified that had an in-frame 75-bp deletion in the COOH-terminal half of the molecule. To determine whether this deletion was unique to MO7e or might be present in leukemic blasts from patients, we screened both AML and ALL samples by RT-PCR as described in “Materials and Methods.” The RT-PCR product for wild-type mRNA produces a 255-bp amplicon, whereas the mRNA with the deletion gives a 180-bp fragment.

Fig. 9,A shows the presence of both transcripts in approximately equal amounts obtained from leukemic blasts from a patient with AML(Lane 1) and one from a patient with ALL (Lane 2). Samples, which show only the 255-bp wild-type product, are included from a myeloid leukemia cell line, Spurlock (Lane 3), an ovarian cancer (Lane 4), and an EBV-transformed lymphoblastoid cell line (Lane 5). Sequence analysis shows that the longer amplicon represents the wild-type PASG mRNA, whereas the shorter amplicon contains the in-frame 75-bp deletion (Fig. 9,B). The analysis of additional samples of AML and ALL demonstrates that the presence of the PASG mRNA with the 75-bp deletion(PASG^-75bp) occurs in approximately 57% of AML samples and 37% of ALL samples (Table 3). The expression of the PASG^-75bp mRNA was observed in all AML FAB subtypes, except the M6 subtype, which was not tested, as well as in both T-cell and pre-B-cell ALL (Table 3; data not shown). Interestingly, none of 20 samples of ovarian cancer or a variety of transformed cell lines showed this deletion (Table 3). In addition, the PASG^-75bp mRNA form was not observed in normal tissues, including bone marrow, thymus, and quiescent or activated lymphocytes (Table 3). However, these data cannot completely rule out the possibility that some normal tissues could express this alternative mRNA in a small subset of cells or at a very low level.

This 75-nucleotide deletion causes the loss of 25 amino acids from the 3′ half of exon 18 (Figs. 9,10). However, because the deletion results in an in-frame loss of nucleotides, a complete mRNA is formed with the exception of the deleted sequence. When DNA from several samples of leukemia or MO7e was analyzed for genomic mutations at the donor/acceptor sites for intron/exon boundaries in this region or in the downstream intron, no mutations or abnormal sequences were found (data not shown). However,there is a satisfactory donor splice sequence found within exon 18 that serves as a donor site for the generation of the PASG^-75bp mRNA found in AML and ALL (Fig. 10). It is unclear at this time why this site is used in addition to the normal donor splice site in these acute leukemias. Of further interest is that the region that is deleted contains an amino acid motif(STRAGGLG) known to be critical for transcriptional activation by the yeast SWI2/SNF2 polypeptide (9). The PASG^-75bp mutant mRNA in human leukemia thus appears to arise as an abnormal, alternative RNA processing event. These results suggest that the mutant PASG^-75bpmRNA may contribute to the altered physiology of leukemic blasts from AML and ALL.

The initial strategy that we used to identify gene products regulated by cytokines in factor-dependent hematopoietic malignancies led to the isolation of a gene encoding a protein belonging to a heterogeneous family of helicases and chromatin remodeling molecules. The analysis of the cDNA reveals that this putative helicase is most closely related to the SNF2L family of genes, with the highest sequence and structural homology to the DDM1 and SNF2L genes. The genetic locus has been designated SMARCA6. The mRNA level encoding this protein is down-regulated in response to cytokine withdrawal followed by growth arrest and apoptosis, down-regulated after growth arrest without the induction of significantly increased apoptosis when cells are cultured without serum but with cytokine (SCF), up-regulated in lymphocytes stimulated to proliferate by exposure to ConA, and expressed in a variety of tissues and cell types. The sequence homology, expression pattern, and association with proliferation have led us to refer to the gene product as PASG (31).

Different families of helicases play important roles in normal growth and development as well as in malignant cell transformation and cancer cell physiology (21, 22, 41, 42). The demonstration that the PASG gene is located on chromosome 10q23-q24 raises the possibility that it may also contribute to the development or physiology of several types of malignancies. This region of chromosome 10 has been associated with loss of heterozygosity in a variety of malignancies including prostate and ovarian cancer, glioblastoma, and high-grade astrocytomas as well as in subtypes of leukemia and lymphoma (43, 44, 45, 46, 47, 48, 49, 50, 51). Our laboratory is currently screening such malignancies by single-strand conformational polymorphism analysis to determine whether and to what extent PASG might be altered.

In our initial assessment of PASG expression in leukemia, we observed the presence of a highly expressed transcript containing a 75-nucleotide deletion (PASG^-75bp). This alternatively processed PASG transcript was observed in approximately 35% and 60% of ALL and AML samples, respectively. The predicted protein resulting from this in-frame deletion lacks 25 amino acids from exon 18 in the COOH-terminal half of the PASG protein. The 25-amino acid region contains the fourth conserved helicase motif including the sequence STRAGGLG. This conserved sequence has been shown by mutational analysis of the yeast SWI2/SNF2 polypeptide to be critical for the transactivation of genes involved in sucrose fermentation (9). Specifically, the expression of SWI2/SNF2 mRNA containing the deletion of the sequence STRAGGLG results in a dominant negative phenotype and the inability of yeast to transactivate specific enzymatic pathways involved in sugar metabolism(9). By analogy, the expression of PASG mRNA with a deletion involving this conserved sequence could also function in a dominant negative fashion in human leukemia. This might in turn lead to altered regional specific chromatin remodeling, resulting in altered gene expression (24). The high homology of PASG to the DDM1 helicase also raises the possibility that PASG could be involved in chromatin remodeling essential for accessibility of DNA methylation enzymatic machinery (17, 52). The effect of a dominant negative form of PASG would then potentially result in the loss of gene specific methylation after DNA replication. A consequence of hypomethylation may be inappropriate expression of genes that play roles in altering cell proliferation and possibly drug resistance(52). We are currently extending our analysis of PASG in leukemia as well as exploring the functional consequences of the 75-bp deletion.

Mutations and translocations involving histone acetylases and associated proteins are involved in the development of both myeloid and lymphoid leukemias (24). For example, the MLL gene may contribute to altered chromatin remodeling and transcription by recruiting histone acetylases and/or SWI/SNF-like helicases(24). If there is abnormal expression and/or localization of MLL secondary to a translocation event, the abnormal derepression of genes regulated by MLL might also occur due to aberrant recruitment and function of chromatin remodeling proteins (53). Whether PASG is a component of such chromatin remodeling complexes remains to be established.

We have described several alternative forms of a SNF2-like helicase,PASG, which could potentially alter its function. The alternatively processed transcript, which results in the deletion of the first ATG,would produce a predicted protein that has lost 16 amino acids from the NH₂ terminus. It will be of interest to know whether this NH₂ terminus truncation affects PASG function. In addition, the critical role of a NLS in transporting PASG to the nucleus predicts the possibility that mutations changing or deleting this sequence would result in a PASG protein incapable of performing its nuclear function(s). The observation that a subset of human acute leukemias expresses an altered PASG transcript with a 75-nucleotide (25-amino acid) deletion involving a conserved domain known to result in a dominant negative polypeptide of the yeast homologue SWI2/SNF2 suggests that this altered form of PASG may also disrupt normal hematopoietic cell functions. Whether and by what mechanism the altered form of PASG contributes to leukemogenesis will require further work linking structural changes to functional consequences.

We thank the staff of the Hematology/Oncology Division who helped with the collection of tissue samples and cell lines. We are grateful to the Children’s Cancer Group and the group’s leukemia cell banks for some of the samples used in these studies. Dr. Steven Cannistra (Beth Israel Hospital, Boston, MA) generously supplied us with samples of ovarian cancer. We thank Dr. Jin Li for performing sequence analysis of the PASG genomic clones.