Characterization of putative human homologues of the yeast chromosome transmission fidelity gene, CHL1.

Helicases are components of numerous protein complexes, including those regulating transcription, translation, DNA replication and repair, splicing, and mitotic chromosome transmission. Helicases unwind double-stranded DNA and RNA homo- and hetero-duplexes. The yeast CHL1 helicase has been linked to maintenance of the high fidelity of chromosome transmission during mitosis. Mutations in this gene result in a 200-fold increase in the rate of aberrant chromosome segregation with a concomitant delay in the cell cycle at G2-M, suggesting that CHL1 is required for the maintenance of proper chromosome transmission. Two highly related human cDNA clones encoding proteins which are homologous to the yeast CHL1 gene product have been isolated. Here we show that these two distinct human CHL1-related mRNAs and proteins (hCHLR1 and hCHLR2) are expressed only in proliferating human cell lines. Quiescent normal human fibroblasts stimulated to re-enter the cell cycle by addition of serum begin to express the CHL1-related proteins as the cells enter S phase, concomitant with the expression of proliferating cell nuclear antigen. Furthermore, expression of the CHL1-related mRNAs is lost when human K562 cells cease to proliferate and terminally differentiate in response to phorbol ester treatments. Human hCHLR expression is not extinguished during hemin-induced differentiation of the same cell line, which produces erythrocyte-like cells that continue to proliferate. These experiments are consistent with the requirement of this putative helicase during either S or G2-M phase but not G1. In vitro transcribed and translated hCHLR1 protein binds to both single- and double-stranded DNA, supporting the possibility that these proteins are DNA helicases. Finally, affinity-purified hCHLR1 antisera was used to demonstrate the localization of the hCHLR proteins to the nucleolus by indirect immunofluorescence as well as by cell fractionation.

Studies in eukaryotes and prokaryotes have shown that helicases are involved in DNA replication, recombination and repair; in RNA transcription, translation, and splicing; and in differentiation, suggesting that their function is critical for normal cell growth and development and, in some cases, survival (1)(2)(3)(4)(5). These molecules are involved in the unwinding of double-stranded DNA and RNA coupled with the hydrolysis of nucleoside 5Ј-triphosphates. Some helicases have dual functions, such as the human ERCC2 gene, and its Saccharomyces cerevisiae homologue RAD3, and the human ERCC3 gene, and its S. cerevisiae homologue, RAD25, each participating in repair and transcription (4, 6 -8). Mutations in these genes affect DNA repair as well as RNA polymerase II transcription. Bona fide helicase mutations that result in defective cellular function(s) have been identified in xeroderma pigmentosum (XP) 1 and Cockayne's syndrome (9). XP is characterized by sensitivity to UV irradiation and by an increased frequency of skin cancers, whereas Cockayne's syndrome is characterized by neurological disorders and retarded growth. Recently, it has also been demonstrated that mutations in the human gene product homologous to the E. coli ReqQ helicases are responsible for Bloom's syndrome, an inheritable human genetic disease characterized by small physical size, photosensitivity, and immunodeficiency (10). In addition, mutations in another putative DNA helicase have been shown recently to be responsible for Werner's syndrome in humans, an inherited disease characterized by premature aging (11).
Helicase proteins are defined by several conserved protein motifs, all of which appear to be necessary for their function (12)(13)(14). The nucleotide triphosphate binding domains, encoded by the A and B boxes, are present in every helicase (15). The B box has been used to define subfamilies of helicases, such as the DEAD and DEAH families of helicases. Of the known helicases, the canonical RNA helicase eukaryotic translation initiation factor eIF-4A has been studied most extensively by mutational analysis (12,13). These molecular studies have allowed functions, such as ATP hydrolysis, RNA unwinding, and RNA binding, to be assigned to specific conserved domains. Even so, there are motifs for which function(s) remain unknown.
At least three members of the helicase family may take part in the process of chromosome segregation and transmission. The Drosophila melanogaster gene lodestar has been shown to be involved in the early divisions of the blastocyst (16). Deletion of this gene leads to the death of the D. melanogaster embryo by the 13th division and the presence of tangled and broken chromosomes. Several yeast genes involved in chromosome transmission, with strong similarities to helicases, have also been isolated. The yeast gene SGS1 was isolated as a suppressor of a type I topoisomerase mutant, top3 (17). SGS1 has since been shown to be necessary for maintaining the fidelity of chromosome segregation during both mitosis and meiosis (18). Deletion of this gene results in altered chromosome segregation, presumably due to nondysjunction of sister chromatids. The yeast CHL1 gene product is structurally related to DNA helicases as well (19). It was cloned by complementation of a Saccharomyces cerevisiae mutant, chl1, which is also defective in the fidelity of chromosome transmission. Yeast chl1 null mutants remain viable, but there is a substantial increase in the rate of missegregation (Ͼ200-fold) of specific chromosomes, with chromosome loss and nondysjunction contributing equally to this phenotype (19). The frequency of mitotic recombination in chl1 null mutants is near that of the wild-type, suggesting that it is not involved in replication or recombination. Furthermore, genetic studies in yeast indicate that the CHL1 protein does not function in the RAD9-dependent DNA repair pathway. The chl1 mutation also appears to activate a cell cycle checkpoint leading to a G 2 -M phase delay, suggesting that this protein acts prior to the completion of mitosis.
Here we report the cloning of two human cDNAs with high homology to the product of the S. cerevisiae gene CHL1. Characterization of the corresponding cDNA clones revealed that they are encoded by a gene highly related to CHL1. Translation of the corresponding open reading frame (ORF) of this gene predicts a 102-kDa protein (actual mobility of the IVTT protein by SDS-PAGE is ϳ112 kDa) containing all of the conserved motifs of a helicase. Based on genomic studies, two distinct human CHL1-related cDNAs (hCHLR1 and hCHLR2) are generated by unique, but highly homologous (Ͼ98%), genes (20). 2 We also demonstrate that hCHLR1 mRNA and protein are expressed in dividing cells but not in cells signaled to terminally differentiate or cease their growth due to serum withdrawal. The timing of hCHLR protein expression indicates that these proteins are not required during G 1 , but that they could function during S, G 2 , or M phase. An IVTT hCHLR1 protein binds to single-and double-stranded DNA under salt conditions that are similar to other DNA-binding proteins, suggesting that this protein is also a DNA-binding protein. Finally, cellular fractionation studies and indirect immunofluorescence (IF) analysis of HeLa cells with an affinity-purified hCHLR antisera demonstrates that the protein(s) are present in the nucleolus.

EXPERIMENTAL PROCEDURES
Cloning of Human CHLR cDNAs-In a screen of human genomic DNA for cell cycle-related protein kinase genes, we fortuitously isolated a human cosmid containing a gene with motifs characteristic of a helicase. Single-copy DNA sequence from this cosmid clone, CDC2R-10, isolated by hybridization with a full-length human p34 cdc2 cDNA, was used to screen a lambda cDNA library corresponding to human K562 cells (Stratagene). It is presumed that the genomic cosmid clone containing the human CHL1-related gene was isolated by cross-hybridization with a low-copy repeat element contained within the 3Ј untranslated region of the human p34 cdc2 cDNA, because this region of the p34 cdc2 cDNA, but not the region containing the ORF, strongly hybridized with the CDC2R-10 cosmid. It may also be possible that a CDC2like family member could be closely linked to the human CHL1-related gene. Five of the hCHLR-positive clones identified from the K562 cDNA library were fully or partially sequenced as described previously (21). Oligonucleotides for DNA sequencing were produced on an Applied Biosystems 394 DNA/RNA synthesizer by the Molecular Resources Facility of St. Jude Children's Research Hospital. It was determined that a full-length hCHLR cDNA clone had not been obtained, and subsequent screens were performed using both HeLa and human fetal liver lambda cDNA libraries. Sequence analysis also revealed that a majority of the cDNA clones were derived from heterogeneous nuclear RNA (hnRNA) and contained intronic sequences. For this reason, HeLa cytoplasmic RNA was isolated (22) and used for both RACE and PCR cloning of the full-length cDNA. The RACE protocol was used to isolate the 5Ј end of the cDNA as determined by the presence of motif I and upstream stop codons in all three possible ORFs. Although overlapping cDNA clones and 5Ј RACE products did allow us to determine the sequence of the entire coding region and regions of the 5Ј and 3Ј UTR, the full-length clone was assembled as described below from cDNA clones and reverse transcription-PCR fragments (which were generated by using hCHLR1-specific primers). One hCHLR1 cDNA clone isolated from the K562 library, clone 19a, which contained the C-terminal third of the predicted hChlR1 peptide sequence, and a portion of the 3Ј UTR was subcloned into pBluescript II (Stratagene). Restriction analysis of the full-length sequence revealed a StuI site in clone 19a and a unique SacII site at nucleotide 1321 of the full-length cDNA. Primers were designed on either side of these two restriction sites, and reverse transcription-PCR, using HeLa cytoplasmic RNA, was used to generate a fragment corresponding to this region. The resulting PCR fragment was subcloned and sequenced to insure that there were no PCR artifacts. The fragment from this reverse transcription-PCR product was cut with SacII and StuI and subcloned into the SacII and StuI sites of pKS and clone 19a, respectively, creating pCHL. A second PCR fragment was generated spanning a unique SacI site in the 5Ј UTR 96 base pairs upstream of the translational start site and the unique SacII site. This PCR product was subcloned and sequenced, and a SacI and SacII fragment from this PCR reaction was subcloned into the corresponding sites of pCHL, creating pFLCHL, a full-length hCHLR1 cDNA.
Northern Blot Analysis and DNA Binding Assays-Total RNA was isolated from cells according to the method of Chomczynski and Sacchi and run on a 1% agarose-formaldehyde gel (23). The RNA was transferred to a Duralose (Stratagene) membrane and hybridized as described previously (21). Double-and single-stranded DNA binding assays were performed as described by others (24,25). The human E2A-HLF and CBF␤ cDNA clones for IVTT, used as controls for these experiments, were kindly provided by Drs. A. T. Look and S. Hiebert, respectively (26,27).
Antibody Production and Characterization-Rabbit polyclonal antibodies to the hChlR1 protein were made to amino acids 2-130 (Hel1) and 427-542 (Hel3) of the predicted ORF of pFLCHL. PCR products, generated using primers containing BamHI and HindIII sites, were subcloned into KS-PCR for sequencing. Clones for each region, having the correct sequence, were subcloned into the pQE9 expression vector (Qiagen), which places a six-histidine tag at the 5Ј end of the sequence. Protein was induced by the addition of 2 mM isopropyl-␤-D-thio-galactopyranoside, followed by growth at 37°C for 3-4 h. Bacterial lysates for Hel1 were prepared according to the manufacturer's protocols, and the Hel1 fusion protein was purified by nickel-agarose (Qiagen) affinity chromatography. The Hel3 fusion protein was completely insoluble. Therefore, bacterial lysates containing the Hel3 protein were fractionated by SDS-PAGE, followed by excision of the gel slice containing the antigen which was then used, with Freund's adjuvant, as described previously (21)(Rockland, Inc.). The purified, and soluble, Hel1 protein was used to generate antibodies in an identical manner. The soluble, affinity-purified Hel1 fusion protein was linked to CNBr-activated Sepharose (Sigma), which was then used to affinity-purify the Hel1 antibody. The affinity-purified Hel1 antiserum was eluted from the matrix by using Actisep elution medium (Sterogen) and passed over a PD-10 desalting column (Pharmacia Biotech Inc.) to remove the Actisep medium. It was not possible to affinity-purify the Hel3 antibody because of its insoluble nature. However, we were able to use this antibody to effectively immunoprecipitate the IVTT hCHLR1 protein.
Immunoprecipitations and Western blots were used to determine the specificity of the Hel1 and Hel3 antisera as described previously (21). For the immunoprecipitations, the pFLCHL cDNA was in vitro transcribed and translated in the presence of [ 35 S]methionine, and a portion of the reaction was incubated at 4°C with each purified antibody in RIPA buffer (150 mM NaCl, 1% Nonidet P-40, 0.5% sodium deoxycholate, 0.1% SDS, and 50 mM Tris, pH 7.5) for 2-4 h. Immunoprecipitation and competition experiments utilizing the 6-His-Hel1 antigen were performed as described previously (21). The immunoprecipitated proteins were separated by SDS-PAGE gels. Western blots of the purified Hel1 protein, as well as in vitro transcribed and translated protein and tissue culture cell lysates (prepared as described previously)(21), were used to determine the effectiveness of the antisera in detection of the protein on membranes using the ECL system (Amersham Corp.) to visualize the protein.
Western blot analysis was also performed using the purified human RNA polymerase II protein complex corresponding to a MonoS column purified fraction from HeLa cells containing a subset of the basal transcription factors and human homologues of yeast proteins that are suppressors of RNA polymerase B (SRB), transcriptional coactivators which increase the activation of transcription, as well as essential components that complement XPF-, XPG-, and XPC-DNA repair-deficient extracts (28). This complex was kindly provided by Drs. Danny Reinberg and Helen Cho.
Cell Culture-K562 cells, a human myeloid cell line, were maintained in RPMI 1640 supplemented with 10% fetal bovine serum and 2 mM glutamine. HeLa cells obtained from American Type Culture Collection were maintained in DMEM supplemented with 10% fetal calf serum, 4 mM glutamine, 100 units/ml penicillin, and 100 g/ml streptomycin. WI-38 cells, normal human lung fibroblasts, were obtained from American Type Culture Collection and maintained in basal medium (Eagle) supplemented with 10% fetal bovine serum, 4 mM glutamine, 100 units/ml penicillin, and 100 g/ml streptomycin. For the differentiation experiment, K562 cells were seeded at 1 ϫ 10 5 cells/ml. Twenty-four h after seeding, the cells were induced to differentiate using 40 M hemin or 60 nM 12-0-tetradecanoylphorbol-13-Acetate (TPA) (29). A Me 2 SO control was also included. Three days after addition of the hemin and TPA, cells were harvested, and total RNA was isolated. For the growth cessation experiments, WI-38 cells (normal human diploid fibroblasts, passage 8) were seeded at 5 ϫ 10 4 cells/ml. Twenty-four h after seeding, a portion of the cells were harvested for RNA or protein analysis, and the medium on the remaining dishes was replaced with medium containing 0.1% fetal bovine serum. After 3 days in the low-serum medium, another sample was taken, and the medium was replaced with medium containing 10% fetal bovine serum. At stagedintervals following the addition of serum, cells were collected for isolation of protein and/or RNA as well as FACS analysis of their cell cycle.
Cell Fractionation-CEM-C7 cells were fractionated as described by Dickinson and Kohwi-Shigematsu (30). The cytoplasmic and nuclear fractions were dialyzed against phosphate-buffered saline and concentrated using Centriprep-3 and Centricon-10 concentrators (Amicon). Each was made in 1 ϫ RIPA prior to Western blot analysis. As localization controls, monoclonal antibodies to PCNA (Pharmingen) and/or NuMA (CalBiochem)(31) were used on these same Western blots.

Cloning and Structural Features of hCHLR cDNAs-The
DNA sequences of the two largest cDNA clones that hybridized to the single-copy DNA sequences from cosmid CDCR-10, 3a (2.2 kb) and 19a (1.0 kb), were determined. Analysis of the predicted ORFs of these cDNA clones using the SWISS-PROT and PIR protein data banks indicated that they encoded nonoverlapping sequences of a human protein most closely related to the CHL1 gene product of S. cerevisiae (Figs. 1 and 2A). This analysis also revealed that clone 3a was an in-frame fusion of two unrelated cDNAs. The region corresponding to the yeast CHL1 comprised the last 1 kb of this cDNA. CHL1 is a 99-kDa protein with the conserved domains of a DNA-dependent ATPase and helicase that is involved in the maintenance of high fidelity chromosome segregation (19). Translation of the CHL1-related region of clone 3a revealed an ORF with 22% identity to the N-terminal portion of CHL1, most notably in the two conserved helicase motifs, domains Ia and II (Fig. 2). This clone lacked conserved domain I (32)(33)(34). The amino acid composition of domain II places this protein in the DEAD/DEAH subfamily of RNA and DNA helicases (33). Clone 19a contains an ORF that is 40% identical to the C-terminal portion of the CHL1 protein. This ORF contains conserved helicase domains IV, V (a helix-turn-helix domain), and VI, where the sequence identity is greatest.
The lack of several conserved helicase domains and upstream stop codons indicated that we had not isolated a fulllength clone. In addition, Northern blot analysis with the hCHLR1 cDNAs demonstrated a 4.7-kb mRNA species in proliferating cell lines (data not shown) and in human tissues that contained dividing cells (Fig. 3). Therefore, we decided to screen additional cDNA libraries in an attempt to isolate fulllength clones. Two additional cDNA libraries were screened in an attempt to isolate overlapping cDNAs. Analysis of the two largest clones from this group of overlapping clones revealed homologous sequences encoding an ORF containing conserved helicase domain III, as well as additional 5Ј sequences. However, domain I was not contained in any of these clones. One of the cDNA isolates also contained a conserved proline-glutamic acid-serine-threonine sequence located downstream of helicase domain I, identical to the location of a similar proline-glutamic acid-serine-threonine sequence in the yeast Chl1 protein (19). Sequence analysis of additional cDNAs revealed that these Rad3 is included because this protein had the highest homology to CHL1 prior to the cloning of the hCHLR1 cDNA, and it is indicative of how the location of the conserved motifs varies between these proteins. clones could be divided into two distinct groups based on 31 nucleotide differences dispersed throughout the coding sequences. Seventeen of these nucleotide differences resulted in amino acid changes (Fig. 1). The significance of these amino acid differences is unknown at this time; however, four of these changes are conservative substitutions, and none of the altered amino acids are located in the conserved helicase domains or putative nuclear localization signal sequences. These two related cDNA clones are most likely encoded by distinct, but homologous, genes located on human chromosome 12p11 and 12p13 (20).
Using a PCR-based strategy, RACE, we isolated sequences encoding the N-terminal region of one of the two human CHL1related proteins (35). The major 1200-base pair product produced by this method was subcloned, sequenced, and found to contain helicase conserved motif I, as well as additional 5Ј sequences (Fig. 1). In-frame termination codons were located near the 5Ј end of the PCR product, suggesting that this fragment encoded the entire 5Ј portion of the hCHLR1 mRNA, including the 5Ј UTR. Several other RACE products were also isolated as well as DNA sequence from this reaction and a second independent reaction using oligonucleotide primers located 3Ј of motif I. A number of these RACE products appear to result from alternative splicing in the 5Ј UTR and coding region, whereas others may represent cDNA products from the other locus. These points are still under investigation. The two distinct cDNA classes isolated in these studies have been named hCHLR1 and hCHLR2, respectively, for human CHL-1 related (Figs. 1 and 2). A schematic diagram comparing the positions of conserved helicase domains of the S. cerevisiae CHL1 and Rad3 (another homologous protein related to this family) proteins, as well as the hCHLR1 protein, is shown in Fig. 2B. The spacing between these conserved helicase domains is usually well conserved in functional homologues (36,37). The hCHLR1 and S. cerevisiae CHL1 protein are nearly identical with regard to this spacing, whereas the S. cerevisiae Rad3 protein is distinct in this regard (Fig. 2B). The similarity between the spacing of the helicase domains of the hCHLR proteins and the yeast CHL1 suggests that the hCHLR genes may be human homologues of the yeast CHL1 gene.
Analysis of the hCHLR mRNA and Protein-The expression of the hCHLR genes has been examined by Northern blot analysis of mRNA from K562, HeLa, and WI-38 cell lines (data not shown), as well as a panel of human adult tissues (Fig. 3). A 4.7-kb mRNA transcript, which presumably encodes both the hCHLR1 and hCHLR2 proteins, was detected in all of the cell lines and specific adult tissues. In fact, the 4.7-kb hCHLR mRNA was most abundant in tissues containing a large population of dividing cells (e.g. T cells, the spleen, and B cells) or cells undergoing division and/or recombination (e.g. the thymus), suggesting that expression of this gene is restricted to actively dividing and/or nondifferentiated cells (Fig. 3). Conversely, the larger hCHLR mRNAs are prominent in thymus, testis, ovary, small intestine, and pancreas, all tissues that contain a significant population of dividing cells and/or cells undergoing high levels of recombination. Preliminary attempts to complement yeast chl1 mutants with the human cDNAs has been unsucessful thus far. 3 It should be noted, however, that another group has isolated one of the putative human CHLR1 proteins reported here, 2 and they suggest that it is, indeed, the human CHL1 homologue.
Two distinct polyclonal antibodies, Hel1 and Hel3, were generated to fusion proteins corresponding to different regions of the hCHLR1 protein (see "Experimental Procedures" for details). Both the Hel1 and Hel3 antisera are capable of immunoprecipitating IVTT hCHLR1 protein (data not shown). The soluble Hel1 antigen was an effective competitor of the Hel1 antibody. Furthermore, the purified Hel1 antisera also immunoprecipitates an identically sized protein from [ 35 S]methionine-labeled HeLa cell lysates. Because the Hel1 antisera could be easily purified in large quantities by affinity chromatography, it was used exclusively for the remaining experiments. This affinity-purified Hel1 antisera was used to examine the expression of the hCHLR protein(s) in a number of different proliferating human cell lines (Fig. 4). Once again, an identically sized p112 hCHLR protein(s) was identified in all of these cell lines; smaller polypeptides detected in some of these lysates are most likely due to proteolysis, but this has not been confirmed. In addition, although identical amounts of each cell lysate were used for this blot, differing amounts of the hCHLR protein(s) were detected. In general, the relative level of hCHLR protein(s) correlated with the time required for cell division, with higher steady-state hCHLR protein levels in cell lines such as CEM, Jurkat, and 293 (all with relatively short doubling times and, therefore, more rapidly dividing), and lower steady-state hCHLR protein levels in cell lines such as HepG2, pNET1, and Saos-2 (all with relatively long doubling times, reflecting their slower progression through the cell cycle). This apparent correlation between the proliferative capacity of a particular cell line and the steady-state level of hCHLR protein(s) was examined further using normal human diploid fibroblasts as a model (see below). Interestingly, several of these cell lines also express a much less abundant ϳ150-kDa species (CEM-C7, Jurkat, and 293) or a ϳ140-kDa species (PNET1 and PNET2). These larger bands may, in fact, be derived from the larger 6.0 -9.5-kb hCHLR RNAs detected on the human tissue Northern blot (Fig. 3). These data are consistent with the possible expression of one or more alternatively spliced isoforms from one or both of the hCHLR genes.
Analysis of hCHLR Expression during the Transition from Quiescence to Growth, as Well as during Chemically Induced Cell Differentiation-The hypothesis that the hCHLR proteins are only expressed in actively dividing cells was tested by examining hCHLR expression in proliferating and quiescent WI-38 fibroblasts. Serum-stimulated WI-38 fibroblasts were collected at 4-h intervals for FACS analysis and the isolation of cellular proteins for Western blot analysis with purified Hel1 and PCNA antisera (Fig. 5). The 112-kDa hCHLR protein was not detected until 16 -20 h after serum addition, which coincides with the entry of these cells into S phase (Fig. 5). Steadystate levels of the 112-kDa hCHLR protein continued to increase for the duration of the experiment (28 h after serum addition; ϳ50% of the cells in S phase). The steady-state levels of a well studied component of the cell cycle machinery, PCNA, was used as a control for this experiment (Fig. 5). The PCNA protein is a 36-kDa subunit of DNA polymerase ␦ and is responsible for stimulating DNA replication (38). Steady-state levels of PCNA are known to decrease markedly in quiescent fibroblasts and increase dramatically as the cells enter S phase (Fig. 5) (38). From these experiments, we can conclude that hCHLR mRNA and protein expression is not required during G 1 in these normal fibroblasts but is required during S, G 2 , and/or M phase.
Further evidence supporting the requirement of the hCHLR gene product(s) in proliferating, but not quiescent or terminally differentiated, cells was obtained by examining hCHLR gene expression during differentiation. A human erythroleukemia cell line, K562, can be induced to differentiate along either the erythroid pathway, by using hemin, or along the monocyte/ macrophage pathway by using the phorbol ester, TPA (39 -41). Differentiation of K562 cells with TPA also induces growth arrest, whereas hemin-induced cells continue to proliferate, although a small percentage of cells (Ͻ20%) cease to divide (41). We found that steady-state hCHLR mRNA expression was slightly reduced (15%) in hemin-induced cells, as compared to both an untreated and Me 2 SO-treated controls, but that its expression was dramatically diminished (85%) in TPA-induced cells (data not shown). These data are consistent with the previous observations indicating that hCHLR function is restricted to proliferating cells.
Western blot analysis of a purified RNA polymerase II complex from HeLa cells, containing all of the required proteins to direct RNA polymerase II transcription as well as proteins involved in selected DNA repair pathways (28), with the Hel1 antibody indicated that the hCHLR proteins are not contained in this complex (data not shown). This suggests that the hCHLR proteins are unlikely to function in RNA polymerase II transcription and/or the DNA nucleotide excision repair activities associated with this complex (4,28,42,43). Additional experiments performed with UV-irradiated WI-38 fibroblasts, which contain normal p53, did not demonstrate any increase in the level of hCHLR protein(s), whereas p53 levels were increased (data not shown). These results are consistent with the absence of the hCHLR protein(s) in the RNA polymerase II complex described above and suggest that the hCHLR protein(s) are most likely not involved in DNA repair pathway(s).
The hCHLR1 Protein Binds to ssDNA and dsDNA-Because the predicted ORF of the hCHLR1 cDNA exhibits homology to a yeast DNA helicase, the ability of IVTT hCHLR1 protein to bind to single-and/or double-stranded total human DNA was assayed. If hCHLR1 is, indeed, a DNA helicase, we would expect it to bind to nonspecific DNA sequences. IVTT hCHLR1 protein that had been labeled with [ 35 S]methionine was incubated with either single-or double-stranded DNA affinity columns. The columns were washed with buffer containing increasing amounts of salt, which allows one to determine the relative binding capacity of the labeled protein for the DNA template. As a positive control, the known DNA-binding protein, E2A-HLF, a transcription factor that binds a specific DNA sequence with relatively high specificity, was used (26). As a negative control, the CBF␤ transcription factor protein, which cannot bind to DNA without its heterologous partner, was used (27). The results are shown in Fig. 6. The hCHLR1 protein binds to both single-and double-stranded DNA, eluting from the column under moderately high salt conditions (200 -400 mM). The hCHLR1 protein seems to bind preferentially to the single-stranded DNA template, consistent with its possible function as a DNA helicase (Fig. 6). As expected, the HLF protein was bound to both single-and double-stranded DNA, whereas the CBF␤ protein did not (Fig. 6). However, the bind- FIG. 5. Western blot analysis of hCHLR protein expression in normal WI-38 fibroblasts that have been stimulated to re-enter the cell cycle. WI-38 fibroblasts were rendered quiescent by serum deprivation for 48 h. They were then stimulated to re-enter the cell cycle by the addition of serum. Cells were collected at 4-h interval and either analyzed by FACS or total cell lysates were prepared for further analysis. Asynchronously growing WI-38 fibroblasts were used as a control (Asynch). Approximately 50 g of total cell lysate were transferred to a membrane, which was then probed with either the affinity-purified Hel1 antisera or a purified antisera made to PCNA. The positions of the hCHLR proteins and the PCNA are indicated on the right. Molecular weight markers are shown on the left (in thousands). The percentage of cells in either G 0 -G 1 , S, or G 2 -M (as determined by FACS analysis) is indicated below the blots.
ing capacity of the E2A-HLF protein for the nonspecific DNA template contained in the column is slightly less than the hCHLR1 protein, as might be expected for a sequence-specific DNA-binding transcription factor.
Localization of the hCHLR-related Proteins to the Nucleolus-An attempt was made to determine the location of the hCHLR proteins in human cells by using the affinity-purified Hel1 antisera for Western blot analysis of various CEM-C7 cell fractions as well as IF of HeLa cells. CEM-C7 cells were chosen for the fractionation experiments because of the abundance of hCHLR protein(s) in this highly proliferating human T-cell line (Fig. 4). However, because these cells grow in suspension, they are not as easily analyzed by IF, requiring that the cells be prepared by cytospin centrifugation before analysis. Conversely, HeLa cells, a human cervical carcinoma, grows as an adherent monolayer in culture, eliminating any need for mechanical manipulations of the cells (such as centrifugation) for IF analysis. CEM-C7 cells were fractionated into cytoplasmic, nuclear, DNase-treated nuclear, low salt-extracted nuclear, high salt-extracted nuclear, nuclear matrix, and nucleolar fractions. The DNase treatment of nuclei facilitates the migration of nuclear matrix-associated proteins into SDS-PAGE, which would otherwise be lost due to their association with nucleic acid. Likewise, low salt and high salt extraction of nuclear proteins helps to distinguish proteins, such as NuMA (31), that are more tightly bound to the nuclear matrix. Finally, the remaining insoluble material consists of the most insoluble components of the nuclear matrix as well as proteins associated with the nucleolus. This type of cellular fractionation has been used to confirm the localization of a nuclear matrix-associated protein, NuMA (44). Western blot analysis of the various cellular fractions shows that the majority of the 112-kDa hCHLR protein is localized to the insoluble fraction associated with portions of the nuclear matrix and nucleolus (Fig. 7). The larger M r ϳ150,000 molecular weight protein observed in previous Western blot analysis (Fig. 5) was not observed in this experiment, presumably due to its relatively low abundance (ϳ50 g of protein used in Fig. 5, whereas only ϳ10 g of protein was used in Fig. 7). To ensure that the fractionation procedure worked as desired, the location of a normally nuclear, but not nucleolar, PCNA protein as well as the NuMA protein was examined using the same Western blot. The PCNA protein was present in the nuclear, low salt nuclear extraction, high salt nuclear extraction, and the insoluble nuclear matrix fraction but was completely absent from the cytoplasmic, DNasetreated nuclear, and nuclear matrix fractions (Fig. 7). Furthermore, analysis of this Western blot with a NuMA-specific antibody (Calbiochem) indicated that the high molecular weight, nonproteolyzed NuMA was confined to the nuclear matrix and insoluble nuclear fractions (Fig. 7), as expected. Because NuMA was readily detected in both the nuclear matrix and insoluble nuclear fractions and the hCHLR protein(s) were only detected in the latter, it suggests that the hCHLR protein(s) are more abundant in a component of the insoluble nuclear/ nucleolar fraction. Cellular fractionation studies of the human HeLa cells demonstrated that the hCHLR protein(s) were, once again, primarily associated with the insoluble nuclear and nucleolar fractions (data not shown).
Final confirmation of the exact location of the hCHLR protein(s) was obtained by IF using the affinity-purified Hel1 antisera and HeLa cells (Fig. 8). This antibody specifically detects a single protein species of 112 kDa in HeLa cells (Fig.  4B). IF analysis with the Hel1 antisera demonstrated prominent nucleolar staining (Fig. 8A). A higher magnification of the Hel1-specific staining in HeLa nuclei is also shown (Fig. 8B). If the Hel1 antiserum was pretreated with competing Hel1 anti- The location of the PCNA protein is indicated to the left. In C, the same Western blot was stripped once more and then probed with a NuMAspecific antibody. Proteolytic products of NuMA can be seen in the DNase-treated, low salt nuclear extract, and insoluble nuclear fraction. A small amount is also seen in association with the nuclear matrix fraction.
gen, the nucleolar staining pattern was lost (Fig. 8C). Similarly, controls using only the goat anti-rabbit secondary antibody (fluorescein isothiocyanate-conjugated; Fig. 8D), as well as phosphate-buffered saline buffer controls (data not shown), did not produce this nucleolar staining pattern. This data, combined with the results of the cellular fractionation studies described above (Fig. 7), confirm that the hCHLR protein(s) are localized to the nucleolus. DISCUSSION The predicted hCHLR ORFs place these proteins in the DEAD/DEAH or DEXH family of DNA and RNA helicases. Several members of this family are known DNA helicases, such as the Rad3 protein of S. cerevisiae, or putative DNA helicases based on function, such as the SNF2 family (45,46). These proteins lack the SAT motif found in RNA helicases (45,46). Other members of this subfamily are proven RNA helicases or are putative RNA helicases, as suggested by their functions during embryogenesis (47), mRNA splicing (48 -52), or ribosomal RNA processing/assembly (53)(54)(55). Each of the RNA helicase, or putative RNA helicase, members of this family share the SAT motif in domain III (45,46). DNA sequence analysis implies that the hCHLR proteins are DNA helicases because they lack the SAT motif. This conclusion is also supported by the DNA binding capability of IVTT hCHLR1 protein, as well as its sequence similarity to the yeast CHL1 protein, a putative DNA helicase required for the fidelity of chromosome segregation.
Although the overall sequence identity to the S. cerevisiae CHL1 protein is very high, especially in the conserved helicase domains, there is no direct evidence that hCHLR proteins are the functional human homologues of this protein. It is significant that the spacing of the conserved domains is almost identical between the yeast CHL1 and human hCHLR proteins. Also, many of the members of the helicase family, best characterized by the SNF2 subfamily, have large N-terminal and/or C-terminal extensions that the hCHLR and CHL1 proteins lack (46). Thus, it may be that the hCHLR and CHL1 proteins are not homologues, but rather they comprise a subfamily of re-lated helicases, which may also include the S. cerevisiae Rad3 protein and its homologues.
Expression studies indicate that the function of the hCHLR proteins is linked to cellular proliferation. The dramatic increase in the steady-state levels of hCHLR protein(s) at the G 1 -S transition, concomitant with PCNA, suggests that the function of these protein(s) is required during either S, G 2 , or M phase of the cell cycle. Interestingly, some of the proteins necessary for DNA replication are similarly induced during entry from G 0 into the cell cycle (56 -59). DNA polymerase ␣ and the associated primase complex are absent in nonproliferating cells, and both are up-regulated just prior to entry into S phase. As with the DNA polymerase ␣-primase complex, steady-state levels of the hCHLR protein(s) remain constant throughout the cell cycle in proliferating cells (data not shown). The levels of hCHLR protein(s) in cell lines also roughly correlate with how rapidly these cells transit the cell cycle. The greater the time required to transit the cell cycle (e.g. HepG2 and Soas-2), the lower the level of protein; and conversely, for the more rapidly dividing suspension cell lines (e.g. Jurkat and CEM-C7), the higher the level of protein.
The Northern blot data of adult human tissues show very low levels of hCHLR mRNA expression in the brain. This is not surprising because this organ lacks mitotic cells that can be found, to varying extent, in most tissues. On the other hand, tissues with a larger number of proliferating cells, such as the ovary and testis, have significant levels of hCHLR mRNA. The thymus has an overwhelming amount of expression; but this is a tissue in which cell division is not prominent, but targeted DNA rearrangements (e.g. T-cell receptor) are. Therefore, the function of the hCHLR protein(s) in the thymus may reflect a function unrelated to growth, such as genetic recombination that is required for generation of the T-cell receptor, or apoptosis. It is not unprecedented for a helicase protein to have more than one function. The best studied example of a helicase with multiple functions is the S. cerevisiae Rad3 protein and its human homologue, ERCC2/XPD (36). Both the yeast and human Rad3/ERCC2 proteins are known to take part in the repair

FIG. 8. IF of hCHLR protein(s) in HeLa cells using the Hel1 antisera.
HeLa cells were grown, fixed on coverslips, and processed as described under "Experimental Procedures." A, 60ϫ magnification of HeLa cells using the affinitypurified Hel1 antisera. The regions detected by the Hel1 antisera correspond to nucleolar regions of each cell. B, 100ϫ magnification of HeLa cells using the same Hel1 antisera. C, the Hel1 antisera used in A and B was preincubated with the 6-His-Hel1 fusion protein (as a competitor) and then used for IF. D, HeLa cells were treated with phosphate-buffered saline, followed by the fluorescein isothiocyanate-conjugated secondary antibody used in A-C.
of DNA damage, as evidenced by the UV sensitivity of yeast rad3 mutants and human XP patients. A second function for these proteins is their involvement in RNA polymerase II transcription. Both have been shown to be members of the TFIIH transcription complex (6,60). Recently, a third function has been proposed for the human ERCC2 protein (61). Namely, it may be involved in p53-mediated apoptosis. Overexpression of wild-type p53 in cell lines normally results in programmed cell death. Identical experiments in ERCC2/XPD-deficient cell lines leads to abrogation of the apoptotic phenotype. Furthermore, overexpression of wild-type p53 and wild-type ERCC2 in the ERCC2/XPD-deficient cells results in an apoptotic phenotype that is indistinguishable from normal cells.
Finally, the nucleolar location of the hCHLR protein(s) may also provide clues to its function(s). The nucleolus is the location of rRNA synthesis and maturation and ribosome assembly. Because rDNA transcription is necessary for cell proliferation, this organelle is an attractive target for growth control. Indeed, the retinoblastoma (pRb) protein down-regulates rDNA transcription in U937 cells induced to terminally differentiate by treatment with the phorbol ester TPA (62). This occurs by translocation of pRb to the nucleolus and its subsequent interaction with UBF, a component of the RNA polymerase I transcription complex. This interaction between pRb and UBF has been shown to negatively affect transcription initiated from a polymerase I promoter by using an in vitro RNA polymerase I transcription assay. Many nucleolar antigens have been described that are present only in cells that are undergoing active cell division (Refs. 63-65 and this study). The best characterized of these is the nucleolar protein p120 (66 -68). The level of this protein, a proliferation marker in tumor cells, correlates with cell growth rates, and it is absent in quiescent and differentiated, nondividing cells. High levels of the p120 protein are indicative of rapidly dividing cells, often associated with a transformed phenotype. Antisense oligonucleotides directed to a pre-mRNA splice junction site of p120 caused cells to arrest at the G 1 -S transition (69). The hCHLR proteins could also be considered a member of this group of proteins considered to be nucleolar proliferative markers because its expression is upregulated during cell growth and correlates with the apparent proliferative capacity of these different cell types.
Several members of the helicase superfamily containing the conserved motifs have now been shown to associate with the nucleolus by IF or by their function, usually a mutant phenotype leading to faulty ribogenesis (53)(54)(55). All of these proteins are potential RNA helicases based on function and/or sequence homology to known RNA helicases (i.e. a combination of DEXD/ DEXH and SAT motifs). The hCHLR helicase would be the first member of the helicase superfamily with potential DNA unwinding capabilities to be localized to this organelle. Given the timing of the reappearance of this protein at the G 1 -S boundary and its potential as a nuclear matrix-bound DNA helicase, it is possible that the function of these protein(s) is related to nucleolar chromatin restructuring during the cell cycle. Future studies will attempt to determine whether the hCHLR protein(s) are DNA helicases, as their nucleotide and protein sequences and DNA binding abilities would suggest, as well as determine exactly how these proteins function in the nucleolus.