ABSTRACT
Transcripts of both mitochondrial and nuclear DNA replication genes accumulate periodically during the cell cycle in Crithidia fasciculata. An octameric consensus sequence with a conserved hexameric core was found previously to be required for cycling of the TOP2 transcript, encoding the mitochondrial DNA topoisomerase. We show here that the rate of synthesis of the p51 protein, the large subunit of nuclear replication protein-A encoded by the RPA1 gene, varies during the cell cycle in parallel with RPA1 mRNA level. Plasmids expressing a truncated form of RPA1 ([Delta]RPA1) were used to identify cis elements required for cycling of the [Delta]RPA1 transcript. Sequences within the RPA1 5'-untranslated region (UTR) were found to be necessary for cycling of the [Delta]RPA1 transcript. These sequences also function when transposed 3' of the [Delta]RPA1 coding sequence. A 121 bp fragment of this sequence can confer cycling on a heterologous transcript, but is inactivated when two consensus octamers within the sequence are mutated. Mutation of these two octamers in the full-length 5'-UTR of [Delta]RPA1 is insufficient to abolish cycling of the mRNA unless three additional octamers having single base changes within the hexameric core are also mutated. Thus, common octameric sequence elements are involved in periodic accumulation of both the TOP2 and RPA1 transcripts.
DNA replication in trypanosomes is unusual in that both mitochondrial and nuclear DNA replication are cell cycle regulated and replication in the two organelles occurs in approximate synchrony (1 ,2 ). In other eukaryotic organisms, in contrast, replication of mitochondrial DNA takes place throughout the cell cycle (3 -5 ).
Trypanosomes contain a single mitochondrion in which the mitochondrial DNA is organized into a disc-shaped structure termed a kinetoplast. The mitochondrial DNA consists of two types of DNA molecules, minicircles and maxicircles, arranged into a network of topologically interlocked circles. Most of the studies of DNA replication in trypanosomes have dealt with mitochondrial (also called kinetoplast) DNA replication, focusing especially on minicircle replication. Replication of the minicircle DNA occurs free of the kinetoplast network (see 6 for a recent review). Minicircles are released from the interior of the network and after replication are re-attached to the network periphery (7 ,8 ). In contrast, maxicircles appear to be replicated while still attached to the network (9 ). A type II topoisomerase (10 ,11 ) and a [beta]-like DNA polymerase (12 ,13 ) have been localized to two antipodal sites on the kinetoplast periphery, along with minicircle replication intermediates (12 ), suggesting that proteins involved in DNA replication may be organized into a multi-protein complex at these sites.
Only a few enzymes with a possible role in nuclear DNA replication have been identified in trypanosomes. From the trypano- somatid Crithidia fasciculata, for example, a type I topoisomerase (14 ) and a type I ligase (15 ) have been purified, [alpha]-like and [beta]-like DNA polymerase activities have been partially purified (16 ) and, in addition, an RNase HI gene has been cloned (17 ) and a possibly distinct RNase HI activity has been detected (18 ). Of these enzymes, only the type I topoisomerase has been shown to be nuclear. The cellular location of the other activities remains to be determined.
In addition, a C.fasciculata homolog of the nuclear heterotrimeric, single-stranded DNA binding protein replication protein-A (RP-A) has been purified (19 ). RP-A was first identified as a component of a cellular fraction required for in vitro SV40 DNA replication (20 -22 ). RP-A is required in both the initiation and elongation stages of DNA replication (23 -25 ). Crithidia fasciculata RP-A consists of subunits of 51, 28 and 14 kDa and has been immunolocalized to the nucleus (19 ). Cfa RP-A is able to substitute for human RP-A in large T antigen-dependent unwinding of the SV40 origin and stimulates both priming and synthesis by human DNA polymerase [alpha]/primase (19 ). The genes encoding both the large (Cfa RPA1) and middle (Cfa RPA2) subunits of RP-A have been cloned and sequenced (26 ).
Recent studies on the expression of genes involved in nuclear and mitochondrial DNA replication in C.fasciculata suggest a possible mechanism for coordinating nuclear and kinetoplast S phases (2 ). The steady-state transcript levels of the genes encoding RPA1, RPA2, TOP2 and DHFR-TS all accumulate periodically during the cell cycle, with a peak in levels occurring at G1/S (2 ). Accumulation of mRNAs for both nuclear and mitochondrial DNA replication genes at the same point in the cell cycle may play an important role in coordinating nuclear and mitochondrial replication.
In other eukaryotic organisms expression of DNA synthesis genes is often cell cycle regulated. In the yeast Saccharomyces cerevisiae many of the genes involved in DNA replication are expressed periodically. Expression is regulated through an upstream activating sequence called the MluI cell cycle box (see 26 for a review), which is recognized by the DSC1 transcription factor (also called MBF). Periodic transcription of the DNA synthesis genes appears to be the result of periodic activation of the DSC1 transcription factor. The conserved hexameric core of this element corresponds to the recognition sequence for MluI restriction enzyme, hence the name. Other mechanisms for cell cycle-regulated expression of DNA synthesis genes have also been observed in eukaryotic cells. For example, the human topoisomerase IIa was shown to be regulated during the cell cycle by changes in mRNA stability (28 ).
It is unlikely that transcriptional regulation of DNA synthesis genes is utilized in C.fasciculata. Most gene expression in trypanosomes appears to be regulated post-transcriptionally through regulatory loops involving mRNA stability and regulation of trans-splicing and polyadenylation, rather than at the level of transcription (29 ,30 ). In trypanosomes, transcription of most nuclear genes is polycistronic (31 ). Maturation of messenger RNAs requires addition of a 39 nt sequence, called the spliced leader sequence or mini-exon, to the 5'-end of all transcripts (reviewed in 32 ,33 ). Although involving discontinuous transcripts, trans-splicing appears to be mechanistically very similar to cis-splicing (34 ). For trans-splicing, as for cis-splicing, the dinucleotide AG is the splice acceptor site. In addition, a polypyrimidine tract located upstream of the AG is essential for splicing (35 ). Addition of the spliced leader sequence and polyadenylation of the upstream transcript appear to be linked (36 ,37 ).
To determine whether periodic accumulation of mRNAs encoding nuclear and kinetoplast DNA replication proteins might be regulated by a common mechanism, experiments were undertaken to define the cis-acting sequences necessary for periodic accumulation of these transcripts. Recently, a consensus octameric sequence containing a conserved hexameric core, ATAGAA, was found to be necessary for periodic accumulation of the TOP2 transcript (37 ). We report here that such octameric sequences are also necessary for periodic accumulation of the transcript encoded by the nuclear DNA replication gene Cfa RPA1. In addition, we show that synthesis of the p51 protein, encoded by RPA1, is also periodic and that the peak in p51 synthesis occurs when RPA1 mRNA levels are maximum. These results are consistent with the coordinate regulation of nuclear and mitochondrial DNA replication genes by a post-transcriptional regulation of mRNA levels involving a common regulatory element.
A 4.7 kb MluI fragment of a [lambda]GEM11 clone containing nucleotides +1 to +932 of RPA1 and ~3.7 kb of 5'-flanking sequence was subcloned into pGEM7Zf(+) (Promega) to create plasmid p51GEM, which was used for sequencing. Construction of the [lambda]GEM11 C.fasciculata genomic library has been described (39 ). 5'-Flanking sequence from -770 to +1 was determined by dideoxy sequencing of both strands with Sequenase 2.0 (US Biochemicals). The sequence has been deposited in GenBank (accession no. Z23163).
The splice acceptor site at -556 was determined by RT-PCR (reverse transcription-polymerase chain reaction). cDNA was generated from poly(A)+ RNA as described with the RPA1-specific primer A75 (5'-GAGTCGATGGGCTGAATCTGGTGA-3', complementary to nt +18 to +41). The resulting cDNA was used in PCR as a template for oligonucleotides O29 (5'-AACGCTATATAAGTATCAGTTTCTG-3'), which is complementary to mini-exon sequences, and B13 (5'-CTTTCACGTTGTATGTGGTG- AC-3'). The PCR product obtained was cloned and sequenced. The splice acceptor site at -156 was determined previously (26 ), although slight differences in the 5'-flanking sequence obtained suggest that the resulting clone was from the other allele of RPA1. In this case, A75 was used in the RT-PCR reaction along with oligonucleotide O29, instead of B13 as described above.
Plasmid pDS3 was constructed by ligating a 1.7 kb fragment of RPA1 (containing 770 bp of 5'-flanking sequence and 932 bp of RPA1 coding sequence) into a pGEM5Zf(+) derivative containing the hygromycin drug resistance cassette of pX63HYG (40 ) downstream of a putative C.fasciculata polI promoter (details available on request).
To construct plasmid pDS5, sequences from -523 to -174 of the RPA1 5'-untranslated region (UTR) were deleted from pDS3 by the PCR overlap extension technique (41 ). The two overlapping, mutagenic oligonucleotides used were B55 (5'-CACGTATTCTGACAGGCTTGTAGGGCTGC-3') and B54 (5'-GCAGCCCTACAAGCCTGTCAGAATACGTG-3'), while the two outside oligonucleotides used were B56 (5'-CAGCGTGAAGGAGAAGAGCTTGCC-3') and the SP6 promoter primer (5'-TTAAGGTGACACTATAGAATACTCAAGCTATGC-3'). The initial reaction was performed on plasmid pDS3. The resulting ~0.9 kb PCR product was then digested with ApaI and NotI and used to replace the same ~0.9 kb NotI-ApaI fragment of pDS3.
To create plasmid pDS9N, PCR was performed on plasmid pDS3 with oligonucleotides B89 (5'-CCGCAAGCTTCTACAAGCCAAAGGAT-3') and C8 (5'-CGGGAAGCTTGATGTGTGAAATAG-3'). Both oligonucleotides contain HindIII sites near their 5'-end. RPA1 5'-flanking sequences from -532 to -221 were amplified and the resulting PCR product was digested with HindIII and then cloned into the HindIII site of pDS5. pDS9R is equivalent to pDS9N, except that the PCR product described above was cloned into HindIII-digested pDS5 in the opposite orientation to that of pDS9N. The ends of the HindIII-digested PCR product described for pDS9N were filled in with the Klenow fragment of DNA polymerase I (Promega) to create blunt ends and cloned into NotI-digested, Klenow-treated p[Delta]10Not to generate pDS11N. Construction of p[Delta]10Not has been described (39 ).
Plasmids pDS15 and pDS21 were constructed by the megaprimer PCR technique (42 ), using Vent DNA polymerase. The SP6 promoter primer was always one of the set of oligonucleotides used to construct the megaprimer. In all reactions, oligonucleotide B56 was utilized with the created megaprimer to generate the final PCR product, which was then cloned into EcoRV-digested pGEM5zf(+). For plasmid pDS15, H1 sequence (see Table 1 ) was first mutated. The mutagenizing megaprimer was constructed using oligonucleotide C44 (5'-GTTCTTCTTTTTAAGGATCCTCGCTGCGTTTC-3'). pDS3 was the template for both reactions. Cloning of the final PCR product created plasmid pRH1, which was used as the template for reactions mutating H2 sequence. To create megaprimer II, C48 (5'-GTGGAAGGGCAAAGATCTTGTGCGTCACCG-3') was the mutagenizing oligonucleotide. Cloning of the final PCR product created pRH1, which was then digested with Bsp120I and NotI. The ~0.9 kb Bsp102I-NotI fragment was used to replace the same fragment of pDS3, creating pDS15.
Table 1
pDS21 was constructed in the same manner, again using oligonucleotides SP6 and B56. To mutate the conserved hexameric core of H5 sequence (Table 1 ), C76 (5'-GGCTTCTGTTTCGAGCTCATTATTTTG-3') was used to create the mutagenic megaprimer. The template for both reactions was pDS15. Cloning of the final PCR product created plasmid pRH5, which was then used as template in mutating H4 hexameric sequence (Table 1 ). Oligonucleotide C75 (5'-GTGCTGCGCTGTCTAGAGTTGTGGTAGTTG-3') was utilized to create megaprimer IV. Cloning of the final PCR product created pRH4. Megaprimer V was generated with oligonucleotide C74 (5'-GTCTTTCACGGGTACCGTGGTGACACAG-3') to mutate the hexameric core of sequence H3 (Table 1 ), using pRH4 as template. The final plasmid, pRH3, was digested with Bsp120I and NotI and the ~0.9 kb fragment was used to replace the same Bsp120I-NotI fragment of pDS3, creating pDS21.
Plasmids pDS17 and pDS19 were constructed by performing PCR on either plasmid pDS3 (for construction of pDS17) or pDS15 (to create pDS19). The oligonucleotides used for PCR were C68 (5'-ATAAGAATGCGGCCGCATCACCAACGCTCACAGAAAC-3') and C69 (5'-ATAAGAATGCGGCCGCTGAAATAGGTTACGTGGGAAGG-3'), which amplify RPA1 5'-flanking sequences from -343 to -231. Both oligonucleotides contain NotI sites at their 5'-ends for cloning. The resulting PCR products were then digested with NotI and cloned into NotI- digested p[Delta]10Not.
All plasmids were electroporated into C.fasciculata cells as described previously (37 ).
Synchronization of C.fasciculata cultures by hydroxyurea and Northern blot analyses were performed as described previously (2 ,38 ). In all experiments the starting cell density was 2-3 * 107 cells/ml. The time between the first and second peaks in the number of cells having two nuclei was taken as the doubling time of each culture. Total RNA was isolated using RNeasy columns according to the manufacturer's specification (Qiagen). Hybond-N nylon membranes were used under hybridization conditions suggested by the manufacturer (Amersham). Northern blots were probed with either an ~0.8 kb Bsp120I-MluI fragment of RPA1 (nt + 82 to +932, probe 1, Fig. 3 A) or with RPA1 5'-flanking sequences from -773 to -523 and from -174 to -158 (probe 3, Fig. 3 A). Probe 3 was generated by PCR on template pDS5 with oligonucleotides D2 (5'-CTGCACGTATTCTGACAG-3') and the SP6 promoter primer, which anneals to pGEM-5Zf(+) (Promega) sequence. The blot of total RNA isolated from hydroxyurea- synchronized C.fasciculata (pDS3) cells was, in addition, stripped and reprobed with a PCR product generated from template pDS3 by the SP6 promoter primer and oligonucleotide B13, which is complementary to RPA1 5'-flanking sequences from -377 to -356. The CaBP probe is a 260 bp fragment derived from a partial cDNA of the C.fasciculata calcium binding protein (2 ). Quantitation of Northern blot analyses was performed with a Molecular Dynamics PhosphorImager and ImageQuant software. Poly(A)+ RNA was isolated with Dynatech Oligo(dT) Magna Beads under conditions suggested by the manufacturer. Poly(A)+ RNA was prepared from total RNA isolated at 180 min after release from hydroxyurea arrest.
Crithidia fasciculata cultures were synchronized with hydroxyurea as described, except that cultures were grown in KD medium (43 ). At 30 min time intervals, 12.5 ml cells were harvested by centrifugation for 2 min in a clinical centrifuge at 5000 g and washed twice at 28oC in KD medium lacking methionine (KD-MET). The cells were then resuspended in 2.5 ml KD-MET and 35S-Trans label (ICN) was added to a concentration of ~150 [mu]Ci/ml. After incubation for 15 min at 28oC, NaN3 was added to 0.2% and the cells were then washed twice in ice-cold 1* TBS (137 mM NaCl, 2.7 mM KCl, 25 mM Tris, pH 8.0), 0.04% NaN3. The cells were then resuspended and lysed in 0.25 ml RIPA buffer (150 mM NaCl, 1% NP-40, 0.5% deoxycholate, 0.1% SDS, 50 mM Tris, pH 8.0). Pancreatic DNase I (Sigma) was then added to lysates to 200 [mu]g/ml and RNase A (Sigma) was added to 50 [mu]g/ml. After a 20 min room temperature incubation, the insoluble material was pelleted in a microcentrifuge and 1 [mu]l anti-RP-A rabbit polyclonal antibody (19 ) was added to the supernatant. Immune complexes were collected using Protein A-Sepharose Fast Flow (Pharmacia) and suspended in 20 [mu]l SDS gel loading buffer. Half of each sample was electrophoresed in individual lanes on a 10% polyacrylamide gel. PhosphorImager quantitation of the dried 10% SDS-polyacrylamide gel was as for Northern blots.
For Western blots of cycloheximide-treated cells, 100 [mu]g/ml cycloheximide was added to exponentially growing cells. A control aliquot of cells was removed prior to addition of cycloheximide. At 30 min intervals, 4.5 ml aliquots were removed, harvested as described earlier and resuspended in 90 [mu]l 0.1 M Tris, pH 6.8, 100 [mu]g/ml cycloheximide. SDS-PAGE with a 10% resolving gel was used to analyze 10 [mu]l extract and Western blots were performed as previously described (19 ). The secondary antibody used was goat anti-rabbit IgG-alkaline phosphatase (Sigma). Cell lysates were prepared from 1.5 ml cells at 30 min intervals for Western blots of whole cell extract from hydroxyurea-synchronized cells. Nitrocellulose blots were blocked for 1 h in 5% powdered milk, 1* TNT (100 mM Tris, pH 7.5, 150 mM NaCl, 0.05% Tween-20) and then incubated in TNT, 3% BSA with rabbit polyclonal anti-RP-A antibody at a 1:5000 dilution. The primary antibody was detected with [125I]protein A added to a concentration of 0.5 [mu]Ci/ml in TNT, 3% BSA. PhosphorImager quantitation of Western blots was as for Northern blots.
Experiments were undertaken to define the cis-acting sequence elements necessary for periodic accumulation of the RPA1 transcript. A plasmid encoding a truncated RPA1 transcript ([Delta]RPA1) was constructed, allowing expression of the [Delta]RPA1 transcript to be monitored during the cell cycle. Plasmid pDS3 (Fig. 3 A) contains the N-terminal two thirds of RPA1 coding sequence and 773 bp of RPA1 5'-flanking sequence (Fig. 1 ) in a vector expressing the hygromycin phosphotransferase gene. To determine if cis-acting sequences necessary for periodic accumulation of the RPA1 transcript are present on plasmid pDS3, C.fasciculata (pDS3) cultures were synchronized with hydroxyurea (2 ,44 ) and Northern blot analysis was performed on whole cell RNA isolated at 30 min intervals after release from hydroxyurea block.
To determine whether sequences within the 349 bp sequence implicated in periodic accumulation of the [Delta]RPA1 transcript can function out of the context of the 5'-UTR of RPA1, RPA1 5'-flanking sequences from -532 to -221 were cloned into a HindIII site of pDS5 engineered 3' of the [Delta]RPA1 coding sequence to create plasmid pDS9N (Fig. 4 A). Crithidia fasciculata (pDS9N) cells were synchronized with hydroxyurea and [Delta]RPA1 transcript levels were determined throughout the cell cycle. As shown by PhosphorImager quantitation of the Northern blot (Fig. 4 B), the pDS9N-encoded transcript exhibited cyclic accumulation. A peak in accumulation of the pDS9N-encoded transcript occurred at 180 min after release from hydroxyurea arrest and was 3.5-fold higher than the minimum level at 90 min. Therefore, this sequence is able to confer periodic accumulation on the truncated [Delta]RPA1 transcript even when located 3' of the [Delta]RPA1 coding sequence. In contrast, insertion of the same 312 bp sequence in the reverse orientation (plasmid pDS9R, Fig. 4 A) did not restore normal cycling to the truncated transcript (Fig. 4 B). Rather, levels of the pDS9R-encoded transcript vary only slightly during the cell cycle, similar to that of the 1.7 kb plasmid pDS3-encoded transcript. Thus, this 312 bp sequence functions in an orientation-dependent manner and can function outside the context of the 5'-UTR.
To determine whether the 312 bp sequence cloned into pDS9N can also confer periodic accumulation on a heterologous transcript, this sequence was cloned into p[Delta]10Not. Plasmid p[Delta]10Not (Fig. 5 A) contains as a reporter gene a partial cDNA sequence from a C.fasciculata homolog of the Trypanosoma cruzi flagellar calcium binding protein gene (CaBP; 2 ). The CaBP gene was chosen because steady-state mRNA levels of the endogenous CaBP transcript appear to be constant during the cell cycle in synchronized C.fasciculata cells (2 ). p[Delta]10Not also contains TOP2 5'-UTR sequences from -883 to -609 and from -40 to +1. This sequence includes the major TOP2 splice acceptor site at -668 and upstream polypyrimidine tract, but lacks the cis-acting sequences shown to be necessary for cyclic accumulation of a plasmid-encoded, truncated TOP2 transcript (39 ). A NotI site was engineered at the boundary between sequences at -609 and -40, allowing insertion of RPA1 5'-flanking sequence into this site.
A comparison of the 5'-UTR of RPA1, TOP2 and DHFR-TS identified a common octameric motif with a completely conserved hexameric core, ATAGAA (30 ). Two of these octameric sequences (H1 and H2, Table 1 ) are contained within the 5'-UTR of RPA1. Both H1 and H2 are within the 312 bp sequence that was shown to confer periodic accumulation on a heterologous transcript. Since it has been shown that this conserved sequence is necessary for periodic accumulation of a plasmid-encoded TOP2 transcript (39 ), we have investigated the role of sequences H1 and H2 in periodic accumulation of the [Delta]RPA1 transcript.
To determine whether sequences H1 and H2 are necessary to confer periodic expression on the chimeric transcript, RPA1 sequences from -343 to -231, including wild-type H1 and H2 sequences, were amplified by PCR and cloned into p[Delta]10Not to create pDS17 (Fig. 6 A). Plasmid pDS19 (Fig. 6 A) contains the same sequences found in pDS17, except that both H1 and H2 sequences have been mutated (Table 1 ).
To determine whether mutation of H1 and H2 sequences would also abolish periodic accumulation of the 2.1 kb [Delta]RPA1 transcript, plasmid pDS15 (Fig. 7 A), in which both H1 and H2 sequences are mutated (Table 1 ), was constructed. PhosphorImager quantitation of a Northern blot of total RNA from synchronized C.fasciculata (pDS15) cells is shown in Figure 7 B. Mutation of these sequences did not abolish normal cycling of the plasmid- encoded 2.1 kb [Delta]RPA1 transcript. Rather, the plasmid-encoded transcript accumulated periodically, with a peak in transcript levels at 240 min that is 3.4-fold higher than the trough at 120 min. Although the maximum and minimum transcript levels were shifted by 60 min from those of the endogenous RPA1 transcript in Figure 3 B, in this experiment the minimum and maximum levels of the endogenous transcript were also shifted by the same amount (data not shown). Thus, additional sequences within the 5'-UTR of RPA1 must be able to confer periodic accumulation on the pDS15-encoded transcript.
Although it was shown previously that RPA1 mRNA accumulates periodically during the cell cycle (2 ), it was not known whether synthesis of the p51 protein is also periodic. To address this question, we have examined the rate of p51 synthesis in synchronized C.fasciculata cultures. At 30 min intervals after release from hydroxyurea arrest, Trans 35S-Label (ICN) was added to an aliquot of the culture and after a 15 min incubation whole cell extract was prepared and analyzed by immunoprecipitation with rabbit polyclonal anti-RP-A antibodies (19 ).
PhosphorImager quantitation of the SDS-PAGE analysis of the immunoprecipitates (Fig. 8 B) shows that the p51 protein is synthesized periodically, in parallel with RPA1 transcript levels. The maximum level of incorporation of label into p51 occurred in the cells harvested at 150 min. When adjusted for the 15 min required for harvesting and washing the cells in pre-warmed growth medium (~15 min) and for the 15 min labeling period, this time corresponds to 180 min after release from drug treatment, the time at which maximum levels of RPA1 transcript are achieved (Fig. 3 ).
mRNA levels of both kinetoplast and nuclear DNA replication genes of the trypanosomatid C.fasciculata have been shown to cycle in parallel as cells progress through the cell cycle (2 ). The presence of kinetoplast DNA replication genes within the cell nucleus raises the possibility that nuclear and kinetoplast DNA replication genes share a common regulatory mechanism. The present study investigates the mechanism of cell cycle regulation of the RPA1 gene, encoding the large subunit of C.fasciculata nuclear single-stranded DNA binding protein RP-A.
Periodic accumulation of a plasmid-encoded transcript of a truncated RPA1 gene ([Delta]RPA1) allowed identification of the cis-acting sequences necessary for periodic accumulation of the [Delta]RPA1 transcript. Two plasmid-encoded transcripts were detected by Northern blot. Transcripts that utilized the major splice acceptor site at -556 accumulated periodically, whereas transcripts that used the minor splice acceptor site at -156 did not cycle in the manner of endogenous RPA1, suggesting that the necessary sequences must be present on the message to confer periodic accumulation. Their presence in 5'-flanking DNA sequences (which is the case for the transcript spliced at the minor splice acceptor site) was not sufficient. Similar results were also found for a plasmid-encoded TOP2 transcript (39 ), consistent with a post-transcriptional regulation of expression of both genes.
Deletion or mutation of sequences necessary for normal cycling of the truncated transcript resulted in both a large decrease in cycling of the transcript level and in a shift in minimum expression levels to the period in the cell cycle when the maximum level of expression normally occurred. A similar decrease in transcript levels and shift in the timing of the minimum level has also been observed for plasmid-encoded TOP2 transcripts having deletions within the 5'-UTR of the major transcript (39 ). The significance of this shift in timing of the minimum in transcript level is unknown.
An octameric sequence with a completely conserved hexameric core was shown previously to be necessary for periodic accumulation of the TOP2 message (39 ). Five similar sequences are present in the 5'-UTR of RPA1 (Table 1 ). Mutation of all five octameric sequences in the 5'-UTR of RPA1 completely abolished periodic accumulation of the truncated transcript. Thus, as for TOP2, one or more of these octameric sequences are necessary for periodic accumulation of the RPA1 message. The similarity of the octameric sequences found to be necessary for periodic accumulation of RPA1 and TOP2 suggests that periodic accumulation of these transcripts involves a common regulatory mechanism. Similar sequences may also be involved in periodic accumulation of the transcripts encoding the middle subunit of RP-A (Cfa RPA2) and DHFR-TS, both of which are also expressed periodically (2 ). The importance of each nucleotide in the octameric sequence and the possible contribution by other sequences within the transcript remain to be determined, however. Identification of the trans-acting factor(s) involved will enable us to further investigate this unusual cell cycle regulatory mechanism.
Because mutation of these octameric sequences did not eliminate expression of the RPA1 transcript, but rather abolished periodic accumulation of the transcript, these sequences appear to be negative regulatory elements. Two models proposed for the periodic accumulation of TOP2 (39 ) are consistent with the results for RPA1. One possibility that has been suggested is that these sequences function in regulation at the level of trans-splicing. Periodic activity of a factor which repressed trans-splicing during the G2 and M phases of the cell cycle would result in enhanced degradation of the transcript during these periods of the cell cycle. This factor could interact directly with the octameric sequence or function through additional interactions with other factors. As was noted (39 ), S phase regulation of the mouse thymidylate synthase gene occurs at the level of cis-splicing (45 ). Although cis-splicing serves a different purpose than trans-splicing, regulation of a DNA synthesis gene at the level of splicing has been observed in other systems.
Alternatively, the octameric sequence might function as an instability element, targeting RPA1 mRNA for increased degradation during the G2 and M phases of the cell cycle. Periodic degradation could be affected by either a cell cycle-regulated nuclease or a periodic activity that affected accessibility of a nuclease to that region of the mRNA. Instability elements have been found in the 3'-UTR of various genes in higher eukaryotes (reviewed in 46 ). The ability of sequences from the 5'-UTR of RPA1 to function when transposed to the 3'-UTR is consistent with this latter possibility. However, since the 3-dimensional structure of the RPA1 transcript is unknown, a possible role for the octameric sequences in trans-splicing is not excluded. It is worth noting, however, that octameric sequences can be found in the 3'-flanking sequence of TOP2 (+5687 to +5692; 38 ) and DHFR-TS (+1708 to +1713; 47 ), although it is not known if they play a role in conferring cycling on these transcripts.
Expression of the p51 protein is highest during the period in the cell cycle when RPA1 steady-state mRNA levels are also at a maximum. The p51 protein appears to be stable, with no decrease in p51 protein levels detectable by Western blot over approximately one cell doubling in the absence of protein synthesis. As a result, synthesis approximately doubles the amount of p51 present just prior to S phase. It is unknown, however, whether the activity of the protein varies during the cell cycle. For example, Welch and Wang (48 ) have shown that although the level of the cdc2 protein remains relatively constant over the cell cycle, the majority of newly synthesized human cdc2 exists in a modified form that suggests it is complexed with cyclin and will likely function in mitosis.
Synthesis of the mitochondrial type II topoisomerase has also been shown to be periodic (J.Hines and D.Ray, in press), as was synthesis of p28, the middle subunit of RP-A (data not shown). Thus, synthesis of many DNA replication proteins together at the end of G1 may contribute to the coordination of the two S phases. Identification of the trans-acting factors involved in cell cycle modulation of mRNA levels of nuclear and mitochondrial DNA replication genes will be essential for determining the mechanism of regulation of these genes.
We thank Jane Hines for valuable technical assistance and for providing cell extracts and oligonucleotides and Sally Pasion for assistance with initial cell cycle experiments. This research was supported by NIH research grant GM53254 and by a pre-doctoral traineeship award in Cellular and Molecular Biology to L.B. from NIH grant GM07185.
*To whom correspondence should be addressed. Tel: +1 310 825 4178; Fax: +1 310 206 7286; Email: danray@mbi.ucla.edu
Octamer
Sequence
Locationa
Mutated sequence
H1
CATAGAAA
-315 to -308
AGATCTTA
H2
CATAGAAA
-262 to -255
GGATCCTA
H3
CATACAAC
-370 to -363
CGGTACCC
H4
CATAGACC
-455 to -448
CTCTAGAC
H5
TATAAAAG
-503 to -496
TGAGCTCG
REFERENCES