1997 Oxford University Press
Nucleic Acids Research, 1997, Vol. 25, No. 22
4505–4512
Replication initiation and elongation fork rates within a differentially expressed human multicopy locus in early S phase Rona S. Scott1,2, K. Young Truong3 and Jean-Michel H. Vos1,2,* of Biochemistry and Biophysics, 2Lineberger Comprehensive Cancer Center and 3Department of Biostatistics, University of North Carolina at Chapel Hill, Chapel Hill, NC 27599-7295, USA 1Department
Received August 11, 1997; Revised and Accepted September 30, 1997
ABSTRACT Replication of the 400 copies of the 43 kb human ribosomal RNA (rDNA) locus spans most of the S phase. To examine the basis for the unusual pattern of rDNA replication, a sensitive strategy was developed to map origins of DNA replication and measure apparent rates of fork progression within a chromosomal locus. This technique, termed differential intragenomic replication timing, revealed that initiation within the actively transcribed rDNA occurred in early S within a 10.7 kb region spanning the promoter and 5′ external transcribed spacer. Forks emanating from this early bidirectional origin progressed at an apparent slow rate with the sense and anti-sense forks moving at 0.32 and 0.23 kb/min. Using a photochemical-based technique, the chromatin status of the rDNA repeats was assayed throughout the S phase. Approximately 85% of the rDNA repeats were in a transcriptionally active chromatin structure at the start of S phase. A progressive decrease in the transcription state of the rDNA loci was observed, reaching a minimum between 3 and 6 h in mid S phase. Altogether, the data suggest a link between RNA polymerase I mediated transcription and site-specific initiation of DNA replication within the rDNA multicopy locus. INTRODUCTION Although DNA replication is an essential aspect of cellular proliferation, the mechanisms of its regulation in mammalian cells are still poorly understood (1). One important reason for the lack of knowledge is the uncertainty about the DNA sequences involved in the initiation of replication along mammalian chromosomes (2–4). Early studies for mapping initiation sites for DNA replication have been based on the differential replication timing of specific genomic regions (reviewed in 1). To increase the resolution for mapping origins of replication, procedures have been developed based on the detection of non-linear replication intermediates by 2-dimensional (2-D) agarose gel electrophoresis (5,6) on the determination of replication fork direction by Okazaki and leading strand mapping (7,8), and on the examina-
tion of nascent strand length (9–11). Not only have these various techniques identified several potential eukaryotic origins of DNA replication (2–4), but they have also shown that the regulation of DNA replication is more complex than that of yeast, bacteria or viruses. The naturally amplified (400 copies) human ribosomal DNA (rDNA) locus, is an attractive system to study the regulation of initiation of DNA replication. The human rDNA locus is a 43 kb region that is tandemly repeated (∼400 copies) on the five acrocentric chromosomes 13, 14, 15, 21 and 22 (12). The transcribed region (13 kb) of the rDNA locus produces a 45S primary transcript that is processed into the 5S, 18S and 28S rRNA products. The promoter elements and the transcription termination site are located in the 30 kb non-transcribed spacer (NTS). In several eukaryotic systems, such as yeast (13), sea urchin (14), pea (15), frog (16) and mouse (16), origins of bidirectional DNA replication (OBR) have been located to the NTS. In the human rDNA locus, several studies have attempted to locate the initiation sites of DNA replication within the rDNA locus (17–20). Initiation of DNA replication in the human rDNA locus was shown to occur either in a specific location (20,21) or within a broad zone (17,18), a situation reminiscent to the Chinese hamster ovary (CHO) DHFR locus (3). Recent evidence suggests that the specification of initiation of DNA replication is under developmental and cell cycle controls (22,23). In the Xenopus rDNA locus, transition from early blastula to late blastula not only prompted transcription of the rRNA genes, but also restricted the initiation of DNA replication to the NTS (22). Similarly, the location of the DHFR origin of DNA replication from CHO cells was specified by a cell cycle regulated event. CHO nuclei isolated from early G1 phase showed random initiation throughout the DHFR locus when incubated in Xenopus extracts. However, nuclei isolated 3 h later initiated DNA replication at a specific site (23). To investigate the replication of the human rDNA locus throughout the S phase of synchronized human cells, a sensitive method, differential intragenomic replication timing (DIRT), was developed. With this strategy, we describe the localization of an early S phase initiation region within the rDNA locus spanning the promoter and 5′-end of the gene. The rates of the forks emanating from this origin were found to be slower than previous measurements by random autoradiography studies (24,25).
*To whom correspondence should be addressed at: Lineberger Comprehensive Cancer Center Room 326, CB#7295, University of North Carolina at Chapel Hill, Chapel Hill, NC 27599-7295, USA. Tel +1 919 966 6887; Fax: +1 919 966 3015; Email:
[email protected]
4506 Nucleic Acids Research, 1997, Vol. 25, No. 22 During S phase progression, the chromatin state of the rDNA repeats was observed to decrease from an initial high level of active transcription state. It is proposed that specification of initiation within the rDNA locus are not only under cell cycle and developmental controls, but may also respond to transcription mediated by RNA polymerase I. MATERIALS AND METHODS Cell culture and synchronization Monolayers of T98G (ATTC, CRL-1690), a glioblastoma multiforma, were maintained in Eagle’s minimal essential media (EMEM) supplemented with 5% fetal bovine serum (Hyclone), 2 mM glutamine, 100 µg/ml streptomycin and 100 U/ml penicillin. For synchronization and replication experiments, cells were plated at a density of 8.5 × 103 cells/cm2. Cells were first blocked in G0 by incubation in EMEM containing 0.5% FBS for 40 h. Then, cells were released with EMEM containing 10% FBS and blocked at the G1/S boundary with 10 µg/ml aphidicolin. After 24 h, cells were washed three times with PBS and released into media containing 10% FBS. Flow Cytometry The degree of synchronization was analyzed by 2-color fluorescence activated cell sorting (FACS) as previously described (26). Briefly, cells were pulsed with 10 µM bromodeoxyuridine (BrdU, Sigma) and 1 µM flourodeoxyurindine (FdU, Sigma) for 2 h at various times in S phase. Cells (1.5 × 106) were trypsinized, resuspended in 1.5 ml PBS and fixed with 3 ml of 95% ethanol. Cells were permeabilized by incubation in 0.04% pepsin for 20 min at 37C. The cells were centrifuged and resuspended in 1.5 ml 2N hydrochloric acid. After incubating for 20 min at 37C, 3 ml of 0.1 M sodium borate was added. The cell nuclei were pelleted and washed with 2 ml IFA/0.5% Tween 20 (10 mM HEPES, 150 mM NaCl, 4% FBS, 0.1% sodium azide). The cells were then labeled with a fluorescein–isothiocyanate conjugated antibody to BrdU (100 µl of a 1:5 dilution, Becton Dickinson) for 30 min at 4C, washed with IFA/0.5% Tween 20 and resuspended in 500 µl of IFA per 106 cells. The cell suspensions were incubated with RNaseA (10 µg/ml, Sigma) and propidium iodide (20 µg/ml, Sigma) at 37C for 15 min. The labeled cells were analyzed on a Becton Dickinson FACScan instrument. The S phase population was identified within a specific region of the FACS profiles. The number of cells within the region was counted by the FACScanner, and the S phase region was subdivided into four equal areas. The number of cells in each subregion was subsequently determined. The percent of cells in each subregion was calculated relative to the total cells in S phase (Fig. 2B). Replication timing and DNA isolation Aphidicolin blocked cells were released with media containing 10% FBS and pulsed with 10 µM BrdU and 1 µM FdU for 1 h intervals over an 8 h period. The cells were lysed with 2 ml of 0.5% SDS, 10 mM Tris pH 8.0, 1 mM EDTA and 100 µg/ml proteinase K (Sigma). The DNA was purified by phenol/chloroform extraction, precipitated and resuspended in 0.5 ml of 10 mM Tris pH 8.0 and 1 mM EDTA pH 8.0. The DNA was denatured with 0.2 N NaOH for 20 min at room temperature. To the denatured sample, 4.5 ml of alkaline CsCl solution (50 mM NaCl,
3 mM EDTA, 0.005% sarkosyl and cesium chloride to a refractive index of 1.4150; final refractive index 1.408) was added. Gradients were spun in a fixed angle type 50 rotor for 24 h at 82 000 g (37 000 r.p.m.) in a Beckman ultracentrifuge. The gradients were collected in 200 µl fractions. The resolution for the separation between the BrdU substituted DNA (H DNA) and that of the unsubstituted, parental DNA (L DNA) was examined by slot blotting 10 µl of each fraction and probing for Alu (Blur11) sequences. The position of H and L DNA in the gradient was also examined for each of the rDNA probes and no differences were observed as compared to the Alu probing (data not shown). These hybridizations showed that the first five fractions contained the H DNA and the last 10 fractions contained the L DNA. These fractions were pooled separately. H DNA (20 µl) and L DNA (2 µl) pooled samples were slot blotted onto nylon membranes (Micron Separation Inc.) and hybridized with various human rDNA probes and a human DHFR probe. A lower amount of L DNA was loaded onto the slot blots in order to maintain equal signal intensities between the H and L DNA after probing. These loading differences were accounted for in the analysis of the data. DNA probes The DNA probes Abe (1.6 kb BamHI–EcoRI restriction fragment), Dhx, (0.4 kb HindIII–XbaI restriction fragment), Dsh (1.0 kb SalI–HindIII restriction fragment), Chb (0.47 kb HindIII– BamHI restriction fragment) and the DHFR probe (1.6 kb EcoRI–ApaI restriction fragment) have been previously described (12,27). Corb is a 0.47 kb PCR fragment located at position 41088–41612, D3 is 0.53 kb PCR fragment located at position 17760–18290 and Bsn is a 0.25 kb SalI–NotI restriction fragment located at position 2922–3167 (28). Data analysis Pixel values for the intensity of each slot blot hybridization signal were determined by volume integration on a Molecular Dynamics phosphorimager. The amount of replication at each time interval (Rt) was calculated as shown in equation 1. R t HL
1
H represents the intensity of the nascent DNA signal and L represents the intensity of the unreplicated DNA signal. To determine the replication timing for each probe, the Rt values for each probe were progressively summed as a function of time and normalized to the summation of the 0–8 h timepoint (equation 2):
R n
CRI
t
t0
R
2
8
t
t0
Cumulative replication indexes (CRI) corresponds to the cumulative replication index, Rt represents the amount of replication per time interval (equation 1), t is time post-aphidicolin release in hours, and n varies from 0 to 8 h. The CRI values, from four to six independent experiments (six experiments containing probes Abe, Corb, Dsh, Dhx and DHFR; five experiments with Bsn and Chb; four experiments with D3), were plotted as a function of time to determine the replication time for each of the rDNA probes. For each probe, the replication time was extrapolated manually from
4507 Nucleic Acids Acids Research, Research,1994, 1997,Vol. Vol.22, 25,No. No.122 Nucleic
4507
the CRI curves for each probe at various replicated levels (i.e. 10–80% replication). The mean and standard deviation were calculated and described as the average replication times. Finally, the average replication timing for each probe was plotted against its genomic position (Fig. 3B). Linear regression was used to identify the location of the origin of DNA replication (29). The intercept between the lines fitted to each half of the average replication timing curves identified the initiation region. The data were divided using the values from Dsh to Bsn as the leftward regression and Abe to Dsh for the rightward regression. Analysis of the six independent experiments allowed calculation of the mean and 95% confidence interval for the position of the initiation region (Fig. 3C). Replication fork rates Using Cricket Graph 3.2.1 software, linear regression analysis was performed on the replication timing curves for each replicated fraction (six independent experiments). The inverse slope of the line measured the rate of fork progression and was expressed as kb/min. In order to estimate the standard deviation of the mean, the average replication timing curves were not used for the analysis. Instead, the replication timing curves for each independent experiment were used. The rates were calculated for curves ranging from 10 to 80%, in increments of 5%. For both the sense and anti-sense forks, the measurements were made beginning at Bsn and ending at Dsh. Photochemical DNA modification The chromatin status of the rDNA locus was determined using the transcription-sensitive Psoralen and UVA (PUVA) assay previously described (30). Briefly, T98G cells were synchronized at various stages of the cell cycle. G0 cells were serum starved with 0.5% FBS for 40 h; G1 cells were released from the G0 block with 10% FBS for 12 h; cells at various intervals in S phase were obtained as described above. Cells were treated with 10 µg/ml 4′-hydroxymethyl-4,5′, 8-trimethyl psoralen (HMT) and UV-A for 1 h (31). The HMT was changed five times during the irradiation period. Cells were lysed and the DNA was purified by organic extraction. The DNA was digested with NotI and electrophoresed in 0.8% agarose in 1× TAE at 2 V/cm for 15 h. The Southern blot was probed with ABB which hybridizes to a 3.7 kb NotI fragment (12). The ratio of the slow migrating band to the summation of the slow and fast migrating bands measured the fraction of transcriptionally active rDNA loci (s/s + f). A Molecular Dynamics phosphorimager was used to measure the intensity of each band. The mean and standard deviation was calculated from four independent experiments. RESULTS Experimental strategy Figure 1 outlines the DIRT strategy. Briefly, tissue culture cells are synchronized at the G1/S border with low serum and aphidicolin, an inhibitor of DNA polymerase α and δ (32). The synchronized cells are released and pulsed at 1 h intervals throughout the S phase with BrdU to density label the newly replicated DNA (H DNA). The H DNA is resolved from the L DNA through an alkaline isopycnic centrifugation and hybridized with various probes (1, 2 and 3). Profiles for the percent of
Figure 1. Schematic illustration of the DIRT strategy which allows localization of an initiation region of DNA replication and measurement of fork rates during S phase progression within a locus of interest.
replication at each time point is plotted as CRI. From these curves, the replication timing for the various replicated populations (i.e. 50%) is determined and plotted against the genomic map. Probe 2, spanning the bidirectional origin, would be observed to replicate first, followed by the neighboring regions (probes 1 and 3, respectively). From this plot, the initiation region is located within the earliest replicating probe while the rate of fork progression is derived from the inverse slopes. Cell synchronization Human T98G cells were synchronized at the G1/S border by a double block of serum starvation and aphidicolin. Of the cells, >90% were released and shown to complete the S phase in ∼8 h as illustrated by 2-color FACS analysis of BrdU labeled cells (Fig. 2A and B). The degree of synchronization of the cells was examined by sub-dividing the S phase region from the FACS profiles into four sections, and calculating the percent of cells in each successive S phase section. The dissection of the S phase showed that the majority of cells progressed in unison into four
4508 Nucleic Acids Research, 1997, Vol. 25, No. 22
B
Figure 2. Replication timing of the human rDNA locus throughout S phase. (A) 2-color FACS profiles of T98G cells. The fluorescence of the anti-BrdU-FITC conjugate (Y-axis) visualizes the S phase population and the PI fluorescence separates the cells according to their DNA content into G1, S and G2/M populations (X-axis). (B) Synchronization of T98G cells. Aphidicolin synchronized T98G cells were released and pulsed with BrdU at 2 h intervals. The amount of cells in S phase at each timepoint through a 12 h period was calculated using the CYCLOPS software (Cytomation, Inc.) from the FACS profiles shown above. The percent of cells was plotted (j) in relation to the total number of cells counted. Using the CYCLOPS software (Cytomation, Inc.), the S phase from the FACS profiles was divided into four equal fractions (S1, S2, S3 and S4), each representing a 2 h interval. The percent of cells relative the total S phase population was plotted as a histogram.
successive S phase fractions (Fig. 2B), and indicated no loss of synchrony as cells progressed through the S phase. Initiation region The rDNA loci replicate throughout the S phase in human cells (33). To examine the basis for the unusual replication timing of the rDNA family, we compared replication within the rDNA loci in early and late S phase. As detailed in Figure 1, the DIRT procedure allows the quantitation of differences in replication timing within a defined genomic region. Slot blot and profiles of a representative DIRT experiment as applied to the 43 kb human rDNA locus are shown in Figure 3A. Differences in the replication timing within the rDNA locus were readily observed, with Corb and Bsn showing a significant amount of replication in very early S phase, 1 h post-release. In contrast, other rDNA probes showed little replication in early S phase, with most of their replication occurring in mid to late S phase (Fig. 3A). The
differences were also visible on the respective CRI curves spanning the entire S phase, and these differences can be noted by comparison to the housekeeping DHFR gene (Fig. 3A). To further quantify these replication profiles, the average replication timing of each rDNA probe was determined for various rDNA replicated fractions using the CRI curves and plotted against their respective genomic positions within the rDNA locus (Fig. 3B). The early replicated population, i.e. 10–40% rDNA replication, showed the greatest differences in replication timing between probes. Bsn and Corb were the first to replicate followed by a delay, in a bidirectional manner, of 1.5 h between the earliest and latest replicated region (Dsh). In contrast, there were minimal differences in replication timing between the various rDNA probes in late S phase, i.e. 50–80% replicated fractions. To accurately locate the initiation region, linear regression analysis was performed using the replication times of each of the probes relative to the probes genomic position. The earliest
4509 Nucleic Acids Acids Research, Research,1994, 1997,Vol. Vol.22, 25,No. No.122 Nucleic A
4509
B
C
Figure 3. Initiation of DNA replication within the human rDNA locus during S phase progression. (A) DIRT analysis of the human rDNA locus. The nascent DNA (H) and the unreplicated DNA (L) for an 8 h time period were hybridized with seven rDNA probes. The position of each rDNA probe is shown on the genomic map. Also, the organization of the human genomic rDNA locus is shown. The four EcoRI (E) restriction fragments are shown (A, B, C, D), with the direction of transcription denoted by an arrow. Shaded boxes represent the 5S, 18S and 28S rDNA transcripts. The profiles show the CRI for each rDNA probe (J) compared to the DHFR control (f). (B) Mapping of the initiation region of rDNA replication by the DIRT strategy. The average replication time for each probe was plotted according to its genomic position. The average replication timing for replicated populations in 10% increments (ranging from 10 to 80%) is shown with the standard deviations. (C) Position of the rDNA OBR. The intercept or OBR determined by linear regression analysis (F). Each point corresponds to a specific replication population as noted. The shaded region denotes the 95% confidence interval, and the horizontal line depicts the average OBR position for the 10–40% replication populations. The relative position of the rDNA transcription unit and probes are also shown.
replication time for the various replication populations ranging from 10 to 80% was extrapolated using intercept of each fitted line. As shown in Figure 3C, the initiation region for the early S phase population (i.e. 10–40%) was positioned with 95% confidence to a 10.7 kb segment spanning the promoter and 5′-end of the transcription unit. The size of the initiation region was determined as the average of the 95% confidence intervals for each replication population. In contrast, the 50–80% replication populations showed with 95% confidence the initiation region to span a 22.2 kb domain encompassing the entire rDNA gene and its 5′ and 3′ flanking area.
Chromatin status The placement of the initiation region at the promoter and transcription unit of the rDNA loci may suggest a link to transcription. Hence, the transcriptional status of the rDNA chromatin was investigated. A study in mouse rDNA showed that two types of ribosomal chromatin structures co-existed approximately in equal amounts in asynchronously dividing cells, which represented transcriptionally active and inactive rRNA genes (30). Using this in situ photochemical technique on aphidicolinsynchronized T98G cells, the rDNA transcription state was
4510 Nucleic Acids Research, 1997, Vol. 25, No. 22 investigated in relation to the human S phase. As shown in Figure 4A, the percent of transcriptionally active chromatin for the asynchronous, G0, G1 cells and G2/M cells was measured at 77.6 ± 6.1, 78.2 ± 2.8, 78.6 ± 3.3 and 90.4 ± 0.85%, respectively. It should be noted that these values are not absolute as the controls of purified DNA treated with and without PUVA resulted in values of 91.9 ± 6.6 and 12.0 ± 3.67%, respectively. The values for the S phase populations were plotted in Figure 4B. In early S phase, transcriptionally active chromatin for the rDNA was observed to be highest for the aphidicolin-blocked G1/S population (88.7 ± 4.2%) while a progressive decrease in the rDNA transcription state occurred between 0 and 3 h (Fig. 4B). A maximal reduction in the transcription status of the rDNA was reached between 3 and 6 h post-aphidicolin release, resulting in ∼40% of the repeats being silent during this period. When most rDNA units had been replicated, 6 h post-aphidicolin release, the transcription state of the rDNA locus progressively increased back to a level similar to that in early S phase. Fork rates The apparent rates of the diverging replication forks which are equivalent to the slopes (kb/min) of the replication gradients emanating from the rDNA origin (Fig. 3B) were determined by linear regression analysis from six independent DIRT experiments. As summarized in Figure 4B, the apparent rates of the sense fork, through the 45S transcription unit, and anti-sense fork through the NTS, were 0.32 ± 0.14 and 0.23 ± 0.01 kb/min, respectively. Consequently, the early replicating rDNA population was duplicated by symmetrically diverging forks progressing at similar apparent rates. As observed with rDNA genes from other species (13,14,34–36), the sense fork replicating the human rDNA was collinear with the 45S transcription unit. By late S phase, both the sense and anti-sense apparent fork rates had reached values similar to previous measurements by random fiber autoradiography (37). In addition, the time of increase of the apparent rate of fork progression coincided with a decrease in transcriptionally active rDNA chromatin (Fig. 4B). DISCUSSION To study the regulation of human rDNA replication throughout the S phase, a sensitive assay was developed based on the fine measurement of DIRT. Differences in the temporal order of replication were observed within the 43 kb human rDNA genomic locus. In early S phase, a site-specific initiation region was located with 95% confidence to a 10.7 kb region spanning the promoter and 5′ external transcribed spacer revealing a bidirectional replication gradient centered at the 5′-end of the rDNA genes. Furthermore, the apparent rates of fork progression in early S phase were calculated to be slower than previous measurements by random fiber autoradiography studies (37) and appeared to increase during S phase progression. Finally, the fraction of rDNA in a transcriptionally active state was observed to be highest at the onset of S phase and progressively decreased during the first half of S phase reaching a minimum in mid S phase. A bidirectional origin In early S phase, the mapping results in this paper suggest that initiation of DNA replication occurred within a 10.7 kb fragment spanning the 3′-end of the NTS and the 5′-end of the rDNA
Figure 4. Replication rates and transcription status in the human rDNA locus. (A) Coexistence of transcriptionally active and inactive rDNA chromatin structures in the human S phase. The transcription state of the rDNA locus was visualized by using a chromatin-sensitive photochemical assay (30). Synchronized cells were treated with psoralen and UVA (PUVA) at various times through the cell cycle (28). PUVA (+) and (–) are control samples of purified T98G DNA that were treated with or without PUVA, respectively. Asy corresponds to asynchronous cells; G0 cells were serum starved for 40 h, G1 cells were serum starved and released for 12 h and G2/M cells were released for 12 h post-aphidicolin block. The S phase timepoints are noted as 0–8 h postaphidicolin release. The lower, fast-migrating band (f) represents the inactive rDNA repeat while the upper, slow-migrating band (s) represents the transcriptionally active rDNA repeats (30). (B) Comparison of sense (f) and anti-sense (F) replication forks rates to transcription state (z) within the rDNA locus. Apparent rates were determined from the replication gradients observed in Figure 3B and plotted according to the average replication time of Bsn for each replicated fraction. Standard deviations of the means are noted as half-bars. The apparent average rate in early S (0–3.25 h) was calculated to be 0.32 ± 0.14 kb/min for the sense fork and 0.23 ± 0.01 kb/min for the anti-sense fork. The apparent rate of fork progression is compared to the percent of transcriptionally active rDNA repeats (z). The mean with the standard deviation calculated from four independent experiments is plotted for the S phase timepoints. The arrow denotes the approximate value for other timepoints not in the S phase.
transcription unit. The size of the initiation region may be limited by the sensitivity of the DIRT strategy and may be further refined by increasing the number of probes and independent experiments. Alternatively, the initiation region within the early replicated rDNA loci may be delocalized within this 10 kb fragment. A delocalized initiation region has been previously reported for the human rDNA locus (17,18). In late S phase, the putative initiation region was extended to a 22 kb fragment (95% confidence) spanning the entire rDNA gene including 3′ and 5′ flanking sequences. This apparent delocalization was not a result of loss of synchronization, as the degree of sychronization did not appear to change during S phase progression. However, the greater delocalization of the rDNA initiation region in late S phase may be due to a combination of elongation forks and late initiation events occurring within the rDNA repeats. A shorter BrdU pulse
4511 Nucleic Acids Acids Research, Research,1994, 1997,Vol. Vol.22, 25,No. No.122 Nucleic
4511
Figure 5. Comparison of rDNA initiation of DNA replication. The repeated unit of the rDNA locus is shown. The boxes denote the transcription unit with an arrow showing the direction of transcription. The four EcoRI (E) fragments (A, B, C, D) are shown, and the location of the probes used in this study is represented. Below the rDNA locus, the replication initiation zones observed from four previous reports are summarized and compared to the results obtained from this report. The stripped boxes represent regions where a higher frequency of initiation of DNA replication was observed.
or isolation of small nascent strands may allow discrimination of these two possibilities, i.e. overlap of replication initiation and fork readthrough versus origin delocalization. Ambiguities between the various mapping strategies as applied to the human rDNA locus have been previously observed (Fig. 5). These ambiguities, reminiscent of the CHO DHFR locus (7,38), revealed that initiation of DNA replication within the human rDNA locus could occur either site-specifically (19,20) or over a wide domain (17,18). Specifically, a study by Yoon et al., reported that initiation of DNA replication as measured by nascent strand size occurred throughout the rDNA locus including the transcription unit (18). However, the study by Little et al., using 2-D gel electrophoresis showed that initiation of DNA replication occurred throughout the rDNA NTS with no initiation events detectable within the transcription unit (17). Interestingly, both these studies noted a higher frequency of initiation events at the 3′-end of the NTS. In contrast, two other studies reported site-specific initiation to be limited to the 3′- and 5′-ends of the rDNA NTS. Nevertheless, the overlaps between the various mapping studies appear in agreement with our report describing initiation of DNA replication within the 3′-end of the NTS human rDNA locus (17–20) (Fig. 5). Fork rates at a human origin The placement of the early S phase origin at the promoter and 5′ external transcribed spacer resulted in replication forks that were collinear with the transcription unit which should minimize collisions between the transcription and replication apparatus. Interestingly, the apparent rates of sense and anti-sense replication forks diverging from the early origin were determined to be five times slower than the average fork rate (1.5 kb/min in HeLa cells) previously measured for human cells by random fiber autoradiography (37). As initiation of DNA replication should be the predominant replication event in early S phase, the apparent slow rates in early S phase may suggest that initiation of DNA replication is a rate limiting step. Experimental effects are an unlikely explanation for this slow fork rate since replication timing analysis within the silent mouse immunoglobulin heavy chain locus also resulted in a higher average elongation fork rate,
i.e. 1.7–1.9 kb/min (39). Since replication forks can dislodge transcription complexes (40), slow fork progression may reduce potentially disruptive effects on gene expression. Hence, early replicated regions carrying active genes may generally be replicated at slower rates than late ones with silent genes. This apparent rate of fork progression was observed to increase in middle to late S phase with a concomitant decrease in the fraction of transcriptionally active rDNA loci. It should be noted that the rate of fork progression measured in middle to late S phase is termed an apparent rate as the individual rates for initiation, elongation and termination of DNA replication cannot be measured. Intriguingly, a similar increase in the apparent fork rates through S phase progression has previously also been observed in HeLa and CHO cells by random fiber autoradiography studies (24,25). DNA replication and chromatin structure A study by Conconi et al. showed that two types of chromatin structures which co-exist in the mouse rDNA locus corresponded to actively and inactively transcribed rDNA repeats. The same psoralen crosslinking procedure was used to investigate the transcription status of the human rDNA repeats during S phase progression. In early S phase, most, if not all, rDNA loci were in an actively transcribed chromatin structure. Hence, initiation of DNA replication most likely occurred within the transcribed repeats. Interestingly, the rDNA origin coincided a region bound by a number of transcription factors. For example, binding factors driving RNA polymerase I transcription such as the UBF complex and associated proteins (41) are associated with the promoter. Binding sites for ubiquitous nuclear factors such as the cell cycle regulator p53 and the proliferation-dependent transcription/replication factor OTF-1/NFIII can be found a few kb upstream from the rRNA promoter (42, Briley,L.P., Scott,R.S., Russo,J.R. and Vos,J.-M.H., unpublished observations). As it has previously been shown that initiation of DNA replication in viral systems is influenced by transcription factor binding sites (for review see 43), it will be interesting to investigate whether
4512 Nucleic Acids Research, 1997, Vol. 25, No. 22 transcription factors are crucial for the regulation of initiation of DNA replication in the human rDNA loci. ACKNOWLEDGEMENTS We would like to thank J.Sylvester for the generous gift of rDNA probes, J.Wortman for establishing the synchronization protocols, D.Svoboda, M.Grosz, L.Briley and M.Persmark for their discussions and readings of the manuscript, J.Kelly for secretarial assistance, and the National Cancer Institute for its gift of aphidicolin (ref.# NSC-234714). This work was supported by the National Cancer Institute and the American Cancer Society. REFERENCES 1 Goldman, M.A. (1988) BioEssays, 9, 50–55. 2 DePamphilis, M.L. (1993) J. Biol. Chem., 268, 1–4. 3 Hamlin, J.A. and Dijkwel, P.A. (1995) Curr. Opin. Genet. Dev., 5, 153–161. 4 Huberman, J.A. (1995) Cell, 82, 535–542. 5 Brewer, B.J. and Fangman, W.L. (1988) Cancer Cells, 6, 229–234. 6 Nawotka, K.A. and Huberman, J.A. (1988) Mol. Cell. Biol., 8, 1408–1413. 7 Burhans, W.C., Vassilev, L.T., Caddle, M.S., Heintz, N.H. and DePamphilis, M.L. (1990) Cell, 62, 955–965. 8 Handeli, S., Klar, A., Meuth, M. and Cedar, H. (1989) Cell, 57, 909–920. 9 Vassilev, L. and Johnson, E.M. (1989) Nucleic Acids Res., 17, 7693–7705. 10 Vassilev, L. and Johnson, E.M. (1990) Mol. Cell. Biol., 10, 4899–4904. 11 Vassilev, L.T., Burhans, W.C. and DePamphilis, M.L. (1990) Mol. Cell. Biol., 10, 4685–4689. 12 Sylvester, J.E., Whiteman, D.A., Podolsky, R., Pozsgay, J.M., Respess, J. and Schmickel, R.D. (1986) Hum. Genet., 73, 193–198. 13 Brewer, B.J. and Fangman, W.L. (1988) Cell, 55, 637–643. 14 Botchan, P.M. and Dayton, A.I. (1982) Nature, 299, 453–456. 15 Hernández, P., Martín-Parras, L., Martínez-Robles, M.L. and Schvartzman, J.B. (1993) EMBO J., 12, 1475–1485. 16 Gogel, E., Langst, G., Grummt, I., Kunkel, E. and Grummt, F. (1996) Chromosoma, 104, 511–518.
17 Little, R.D., Platt, T.H.K. and Schildkraut, C.L. (1993) Mol. Cell. Biol., 13, 6600–6613. 18 Yoon, Y., Sanchez, J.A., Brun, C. and Huberman, J.A. (1995) Mol. Cell. Biol., 15, 2482–2489. 19 Coffman, F.D., Georgoff, I., Fresa, K.L., Sylvester, J., Gonzalez, I. and Cohen, S. (1993) Exp. Cell Res., 209, 123–132. 20 Gencheva, M., Anachkova, B. and Russev, G. (1996) J. Biol. Chem., 271, 2608–2614. 21 Coffman, F.D., Fresa, K.L., Hameed, M. and Cohen, S. (1993) J. Cell. Biochem., 51, 157–164. 22 Hyrien, O., Maric, C. and Mechali, M. (1995) Science, 270, 994–997. 23 Wu, J.-R. and Gilbert, D.M. (1996) Science, 271, 1270–1272. 24 Housman, D. and Huberman, J.A. (1975) J. Mol. Biol., 94, 173–181. 25 Painter, R.B. and Schaefer, A.W. (1971) J. Mol. Biol., 58, 289–295. 26 White, A.E., Livanos, E.M. and Tlsty, T.D. (1994) Genes Dev., 8, 666–677. 27 Vos, J.-M.H. and Hanawalt, P. (1987) Cell, 50, 789–799. 28 Gonzalez, I.L. and Sylvester, J.E. (1995) Genomics, 27, 320–328. 29 Truong, Y.K., Scott, R.S. and Vos, J.-M.H. (Submitted). 30 Conconi, A., Widmer, R.M., Koller, T. and Sogo, J.M. (1989) Cell, 57, 753–761. 31 Misra, R.R. and Vos, J.-M.H. (1993) Mol. Cell. Biol., 13, 1002–1012. 32 Liu, P.K., Chang, C.-C., Trosko, J.E., Dube, D.K., Martin, G.M. and Loeb, L.A. (1983) Proc. Natl. Acad. Sci. USA, 80, 797–801. 33 Balazs, I. and Schildkraut, C.L. (1971) J. Mol. Biol., 57, 153–158. 34 Bozzoni, I., Baldari, C.T., Amaldi, F. and Buongiorno-Nardelli, M. (1981) Eur. J. Biochem., 118, 585–590. 35 Funderud, S., Andreassen, R. and Haugli, F. (1978) Nucleic Acids Res., 5, 713–721. 36 Bernard, M., Lagnel, C. and Pierron, G. (1995) Nucleic Acids Res., 23, 1447–1453. 37 Edenberg, H.J. and Huberman, J.A. (1975) Annu. Rev. Genet., 9, 245–284. 38 Dijkwel, P.A. and Hamlin, J.L. (1995) Mol. Cell. Biol., 15, 3023–3031. 39 Brown, E.H., Iqbal, M.A., Hatton, K.S., Valinsky, J. and Schildkraut, C.L. (1987) Mol. Cell. Biol., 7, 450–457. 40 French, S. (1992) Science, 258, 1362–1365. 41 Jantzen, H.-M., Chow, A.M., King, D.S. and Tjian, R. (1992) Genes Dev., 6, 1950–1963. 42 el-Deiry, W.S., Kern, S.E., Pietenpol, J.A., Kinzler, K.W. and Vogelstein, B. (1992) Nature Genet., 1, 45–49. 43 DePamphilis, M.L. (1993) Trends Cell Biol., 3, 161–167.
