Báo cáo khoa học: Distinctive activities of DNA polymerases during human DNA replication ppt

Distinctive activities of DNA polymerases during human DNA replication Anna K. Rytko ¨ nen 1,2 , Markku Vaara 2 , Tamar Nethanel 3 , Gabriel Kaufmann 3 , Raija Sormunen 4 , Esa La ¨ a ¨ ra ¨ 5 , Heinz-Peter Nasheuer 6 , Amal Rahmeh 7 , Marietta Y. W. T. Lee 7 , Juhani E. Syva ¨ oja 2 and Helmut Pospiech 1 1 Biocenter Oulu and Department of Biochemistry, University of Oulu, Finland 2 Department of Biology, University of Joensuu, Finland 3 Department of Biochemistry, Tel Aviv University, Israel 4 Biocenter Oulu and Department of Pathology, University of Oulu, Finland 5 Department of Mathematical Sciences, University of Oulu, Finland 6 National University of Ireland, Department of Biochemistry, Cell Cycle Control Laboratory, Galway, Ireland 7 Department of Biochemistry and Molecular Biology, New York Medical College, Valhalla, NY, USA DNA polymerases (pols) have a central role in DNA replication and maintenance of chromosomal DNA [1]. At least 14 pols have been identified in the mam- malian cell, but only three – pols a, d and e – are needed to synthesize the bulk of DNA during nuclear DNA replication. These pols are structurally related, belonging to the family B DNA polymerases [2]. Nonetheless, all three perform additional roles in other DNA transactions as well as transduce signals of cell cycle control and DNA damage response [1]. Only pol a is capable of initiating DNA synthesis de novo owing to its associated primase activity [3]. The major function of pol a ⁄ primase is synthesizing a short RNA–DNA primer of  30–40 nucleotides that serves both to initiate leading strand DNA replication and to provide precursors of the 200 nucleotide-long Okazaki fragments on the lagging strand [4–6]. Pol a ⁄ primase is then replaced by the elongating pols d or e. This switch from pol a to pol d is controlled by rep- lication factor C, which loads the processivity factor, Keywords cell cycle; DNA polymerase; DNA replication; electron microscopy; UV cross- linking Correspondence H. Pospiech, Department of Biochemistry, PO Box 3000, FIN-90014 University of Oulu, Finland Fax: +358 8 553 1141 Tel: +358 8 553 1155 E-mail: helmut.pospiech@oulu.fi (Received 20 March 2006, revised 3 May 2006, accepted 5 May 2006) doi:10.1111/j.1742-4658.2006.05310.x The contributions of human DNA polymerases (pols) a, d and e during S-phase progression were studied in order to elaborate how these enzymes co-ordinate their functions during nuclear DNA replication. Pol d was three to four times more intensely UV cross-linked to nascent DNA in late compared with early S phase, whereas the cross-linking of pols a and e remained nearly constant throughout the S phase. Consistently, the chro- matin-bound fraction of pol d, unlike pols a and e, increased in the late S phase. Moreover, pol d neutralizing antibodies inhibited replicative DNA synthesis most efficiently in late S-phase nuclei, whereas antibodies against pol e were most potent in early S phase. Ultrastructural localization of the pols by immuno-electron microscopy revealed pol e to localize predomin- antly to ring-shaped clusters at electron-dense regions of the nucleus, whereas pol d was mainly dispersed on fibrous structures. Pol a and prolif- erating cell nuclear antigen displayed partial colocalization with pol d and e, despite the very limited colocalization of the latter two pols. These data are consistent with models where pols d and e pursue their functions at least partly independently during DNA replication. Abbreviations BrdU, bromodeoxyuridine; CLSM, confocal laser-scanning microscopy; EM, electron microscopy; immuno-EM, immuno electron microscopy; MCM2, minichromosome maintenance 2; NP-40, Nonidet P-40; PCNA, proliferating cell nuclear antigen; pol, DNA polymerase. 2984 FEBS Journal 273 (2006) 2984–3001 ª 2006 The Authors Journal compilation ª 2006 FEBS proliferating cell nuclear antigen (PCNA), onto the 3¢ end of the RNA–DNA primer [7]. Simian virus 40 (SV40) has provided the predomin- ant mammalian model system for DNA replication in the last three decades. In SV40 replication, the host cell provides all the replication factors, except for the viral large T antigen, which acts as the initiator protein and replicative helicase [8]. Interestingly, only pols a and d are required for the virus to replicate, whereas pol e seems dispensable [9–11]. In contrast, studies in yeasts and in animal systems indicate that both pols d and e are required for nuclear DNA replication [1,12]. Although pol e is essential for viability both in the budding yeast, Saccharomyces cerevisiae [13,14], and the fission yeast Schizosaccharomyces pombe [15,16], it is the C-terminal checkpoint domain [17], rather than the N-proximal catalytic pol domain, that executes the essential function [18–20]. Nevertheless, the catalytic activity of pol e seems to partake in DNA replication in a number of eukaryotic models [9,11,20–22]. Several hypotheses have been proposed to account for the requirement of both pols d and e in nuclear DNA replication. Most models placed the two pols on opposite arms of the replication fork [12]. This view is supported by genetic studies that demonstrate a strand bias in replication fidelity of proofreading-deficient pols d and e yeast mutants [23,24]. A bias for replica- tion errors on the leading and lagging strands also appears to be established by origins of replication [25]. However, pols d and e still await specific assignment to the leading or the lagging strand by this method. Moreover, the contributions of DNA checkpoint con- trol and DNA repair processes on strand-specific error bias also need to be established in more detail [26,27]. Other models have allocated a role for pol e during specific stages of DNA replication. Mostly, they impli- cate pol e in the initiation of replication [28–31]. On the other hand, a role in late DNA replication has been proposed for human pol e, based on confocal laser-scanning microscopy (CLSM) [32]. In the present study we addressed the specific contri- butions of pols a, d and e to nuclear replication by fol- lowing their behaviour during S-phase progression, using four different methods, namely (a) studying their cross-linking to newly synthesized DNA, (b) determin- ing their association with chromatin, (c) following the effect of cognate inhibitory antibodies on DNA repli- cation in isolated nuclei and (d) localizing the pols by immuno-electron microscopy (immuno-EM). The results suggest that pol a is continuously involved in replication throughout the S phase, pol e is more act- ive in early S phase and pol d is active during the later stages. Moreover, pol e colocalized with pol a in ring-shaped clusters within electron-dense regions of the nucleus, whereas pol d was mainly dispersed in fibrous structures. Taken together, these data are con- sistent with models where pols d and e pursue their functions independently during DNA replication. Results The cross-linking efficiencies of the three replicases to nascent DNA change during S-phase progression To evaluate the specific contributions of pols a, d and e to DNA replication, we studied their association with nascent DNA as a function of S-phase progression. HeLa cells were synchronized with mimosine, which blocks cells at the G1 ⁄ S border prior to initiation of replication [33,34]. Two hours after release from the block, cells have entered S phase, and after 14 h they were found to have entered the G2 ⁄ M phase of the cell cycle (Fig. 1A). We utilized the DNA polymerase trap technique to tag the pols with their DNA products [11,35]. Nascent nuclear DNA was briefly pulse- labelled with bromodeoxyuridine (BrdU) UTP and [ 32 P]dATP[aP] in a monolayer of nuclei isolated from synchronized cells. Subsequent digestion with DNase left the pols photolabelled with a residual radioactive DNA adduct. The photolabelled proteins were then separated from the bulk DNA and the pols were immunoprecipitated with an excess of specific anti- bodies. Analysis of precipitated protein and superna- tant indicated that the immunoprecipitation efficiency remained constant at different time points (data not shown). After resolution on SDS ⁄ PAGE and transfer to poly(vinylidene difluoride) membrane, the specific photolabelled products could be related to the corres- ponding immunoblotting signals (Fig. 1B). Although this method did not reveal the absolute level of the pols engaged in DNA synthesis, it allowed evaluation of the relative changes in their level as a function of S-phase progression. If pols a, d and e function co-ordinately in a com- mon replication fork, one would expect similar chan- ges in their photolabelling intensity during S-phase progression. However, as indicated by the results shown in Fig. 1B,C, pols a, d and e consistently dem- onstrated different behaviours during the S phase. The photolabelling intensity of pol a (Fig. 1C, upper panel) increased only slightly during the later stages of the S phase. Statistical modelling indicated that the photo- labelling of pol a could be fitted well into a linear model. The increase of relative photolabelling for pol a, expressed as a linear trend coefficient of the log A. K. Rytko ¨ nen et al. Human DNA polymerases a, d and e in replication FEBS Journal 273 (2006) 2984–3001 ª 2006 The Authors Journal compilation ª 2006 FEBS 2985 over the whole time range, was 0.025Æh )1 (SE ¼ 0.018), implying a 1.3-fold increase in relative photolabelling over the time range. Pol e behaved similarly to pol a, the estimated slope being 0.046Æh )1 (SE ¼ 0.018). However, the increase was not monotonic; a plateau was observed at mid S phase 8 h after release from mimosine block (Fig. 1C, bottom panel). In contrast, pol d showed a continuous rise in relative photolabelling throughout the S phase (Fig. 1B,C). The estimated slope was 0.113Æ h )1 (SE ¼ 0.018). This corresponds to a three- to fourfold increase from immediate early to late S phase and was clearly higher than the fluctuation of  1.5-fold observed for pols a and e (P ¼ 0.0036 for this compar- ison, see supplementary Doc. S1 for a more detailed description of the statistical analysis). Chromatin association of the replicases during S phase The increase in relative photolabelling of pol d could be attributed to an increase in cross-linking to nascent A C B Fig. 1. Photolabelling of DNA polymerases (pols) a, d and e during the S phase. The activity of the pols during the S phase was studied by using a UV cross-linking technique. HeLa cells were synchronized with mimosine, which blocks cells at the G1 ⁄ S border, then released from the block for 2 h (very early S phase), 5 h (early S phase), 8 h (middle S phase) or 12 h (late S phase) and photolabelled. Pols a, d and e, and their photolabelled derivatives, were monitored as described in the Experimental procedures. (A) Cell synchronization. Flow cytometric analy- sis indicates the DNA content of HeLa cells throughout the time course of a typical mimosine synchronization. (B) Autoradiogram and west- ern blot analysis of a representative experiment. (C) Photolabelling efficiency (autoradiography) and immunoreactive protein (western blot analysis) were densitometrically quantified and the ratios of these values were normalized against the average of the respective experiment. The results of five independent experiments on pols a, d and e, respectively, are presented with different marks. The average for each pol is shown as a bold line. Human DNA polymerases a, d and e in replication A. K. Rytko ¨ nen et al. 2986 FEBS Journal 273 (2006) 2984–3001 ª 2006 The Authors Journal compilation ª 2006 FEBS DNA in late S phase or to a decrease in the protein level associated with chromatin, because the calculated cross-linking intensities represent the ratio of these qualities. Inspection of our DNA polymerase trap experiments indicated an increase of the immunoreac- tive pol d protein level during S phase (Fig. 1B, lower panel). We therefore determined the association of pols a, d and e directly with chromatin. We utilized a sim- ple high-salt extraction scheme that permitted compar- ison with the results from the concurrent polymerase trap experiments. HeLa cells were synchronized with mimosine at the G1 ⁄ S boundary. After release from the block, cells at defined stages of the S phase were lysed in hypotonic buffer in the presence of Nonidet P-40 (NP-40) deter- gent to release detergent-soluble protein, including nucleosolic proteins (soluble fraction). The second fraction contained proteins released by high-salt extraction from the remaining monolayer of open nuc- lei and included the chromatin-associated proteins (‘bound’). The remaining material was solubilized in SDS (rest fraction). The quality of the fractionation was monitored by western blot analysis of marker pro- teins (Fig. 2A) from asynchronous cells. Markers for the soluble fraction included the Golgi marker GM130, the endoplasmic reticulum-specific marker protein disulfide isomerase (PDI) and b-tubulin. These proteins were found exclusively in the soluble fraction, indicating that the high-salt and rest fractions are largely free of soluble contaminants. The chromatin marker, minichromosome maintenance deficient-2 (MCM2), was distributed between the soluble and the bound fraction, as expected [36]. A similar distribution was found for PCNA (Fig. 2A). Lamins A ⁄ C were AB C Fig. 2. Association of DNA polymerases (pols) to chromatin during the S phase. Proteins were synchronized with mimosine and fractionated to result in a Nonidet P-40 soluble fraction, a high-salt (bound) fraction, and a remaining matrix fraction (rest), as outlined in the Experimental procedures. (A) Western blot analysis of marker proteins in a cell fractionation from asynchronous cells. Extracts representing an equal num- ber of cells were loaded from each fraction. The pan-histone antibody recognized multiple bands, corresponding to core and linker histones, as indicated by dots. (B) Levels of bound pols a, d and e during the S phase, as determined by western blot analysis. SYPRO orange staining was used to monitor and normalize loading of the gel. Lane A is an asynchronous control. (C) Densitometric quantification of the bound pro- tein levels. The results represent the average of two independent cell fractionations. Repetitions of the western blot analysis were averaged for each fractionation and pol. A. K. Rytko ¨ nen et al. Human DNA polymerases a, d and e in replication FEBS Journal 273 (2006) 2984–3001 ª 2006 The Authors Journal compilation ª 2006 FEBS 2987 detected in the high-salt and rest fractions, but not in the soluble fractions. Histones were identified mainly in the high-salt fraction. In asynchronous cells, pols a, d and e were distributed, to various extents, between the soluble and the high-salt fractions (Fig. 2A). These results indicate that the NP-40-resistant high-salt frac- tion represents a good approximation for the chroma- tin-bound pols. We then followed the NP-40-resistant high-salt frac- tion of the pols as a function of cell cycle progression. Results from representative western blot analyses of their high-salt fractions are presented in Fig. 2B. All three pols were detected in these fractions. Notably, pol e appeared as a double band after long runs in low-percentage gels. These bands could be accounted for by post-translational modification, proteolysis, alternative splicing or alternative promoter usage [37]. The relative abundance of the two forms did not vary during the S phase. As can be seen from the densitometric quantification of the experiments presented in Fig. 2C and their statistical modelling, NP-40-resistant levels of pol e appeared to be largely constant or slightly decreasing (slope )0.014Æh )1 ;SE¼ 0.016, corresponding to a 1.3- fold decrease during 17 h). In contrast, NP-40-resistant pol a seems to have an increasing trend (slope 0.024Æh )1 ,SE¼ 0.012, corresponding to a 1.5-fold increase). The pol d levels increased even more rapidly, approximately twofold during the S phase (slope 0.043Æh )1 ,SE¼ 0.013). The changes detected in the levels of the pols in the high-salt fraction are only moderate during the S phase; however, the difference between the replicative pols becomes apparent by pairwise comparison of the time trend of chromatin association. The difference between the slopes of pol d and pol a was 0.019Æh )1 (SE ¼ 0.018, P ¼ 0.3), between pol d and pol e was 0.057Æh )1 (SE ¼ 0.020, P ¼ 0.005), and for pol e vs. pol a it was )0.038Æh )1 (SE ¼ 0.020, P ¼ 0.05). The time trend deviated from a linear pattern for pol e, but allowing for curvature had no effect on the contrasts of its average slope vs. those of pol a and d, respectively. Therefore, the change in NP-40-resistant, high-salt-extractable pol e appears to be different from those of pols a and d. Inhibitory effects of antibodies against the replicases at different S-phase stages We further evaluated the temporal differences between the contributions of pols d and e to DNA replication by studying the effects of cognate neutralizing antibod- ies on the pol activities in nuclei isolated at different stages of the S phase [38–40]. We have previously shown that polyclonal antibody K18 against pol e inhibits replication in isolated nuclei from asynchro- nous HeLa cells to a level similar to that of the well-characterized, neutralizing antibody, SJK-132-20, against pol a [9,41] We extended this study by inclu- ding antibody 78F5, which neutralizes specifically the pol d activity [42] and by following the effect of the antibodies against the three pols as a function of S-phase progression. For this purpose, synchronized HeLa monolayer cells were released from the mimosine block, and the resultant G1 ⁄ S, early, middle and late S-phase cells were studied in the DNA replication assay. Antibody SJK-132-20 against pol a inhibited consis- tently  50% of the replicative DNA synthesis, irres- pective of the S-phase stage (Fig. 3A, estimated slope 0.3%Æh )1 ,SE¼ 0.51%). In contrast, the inhibition of replicative DNA synthesis by antibody 78F5 against pol d increased almost threefold, from 17 to 48%, as cells progressed from the G 1 ⁄ S boundary to the late S phase (Fig. 3A) (slope 2.6%Æh )1 ,SE¼ 0.70%). At the same time, inhibition of DNA replication by anti- body K18 against pol e dropped from 45 to 24%, reaching a minimum 8 h after release from mimosine block (slope )2.7%Æh )1 ,SE¼ 0.81% for the first 8 h). The difference between pols d and e was striking. The difference between the slopes of pol d and pol e was 3.7%Æh )1 (SE ¼ 1.0%, P ¼ 0.0003), using a model with separate linear and quadratic terms to allow for the nonlinear behaviour of pol e in late S phase. Mimosine, which has been utilized for cell synchron- ization in this study, has been found to induce DNA damage [43]. Therefore, we considered that the detec- ted differences between pols a, d and e could be influ- enced by checkpoint response, or may reflect DNA repair, rather than differences in the contribution to DNA replication. We therefore stimulated T98G cells to proliferate after prolonged serum deprivation and tested the effect of neutralizing antibodies on replica- tion in nuclei from these cells. Nuclei were from T98G cells, 12 h (early S phase) and 20 h (late S phase) after serum stimulation (see Fig. S1 for flow cytrometric analysis of a typical serum stimulation). Comparable to replication in mimosine-synchronized HeLa nuclei, DNA synthesis was found to be reduced by  60% by the anti-pol a Ig, both in early and late S phases (Fig. 3B). Inhibition by pol e antibodies dropped, in general, from 60% in early, to 18% in late, S phase. At the same time, inhibition by antibodies against pol d showed a general increase, from 8 to 69%. These results are comparable to the results obtained with mimosine-synchronized HeLa cells. The contrast Human DNA polymerases a, d and e in replication A. K. Rytko ¨ nen et al. 2988 FEBS Journal 273 (2006) 2984–3001 ª 2006 The Authors Journal compilation ª 2006 FEBS between pols d and e are even more pronounced in nuclei undergoing an unperturbed S phase than in nuc- lei synchronized by mimosine. The differences observed between pols d and e are therefore not provoked by possible DNA damage caused by mimosine synchron- ization. The results presented above indicate that the requirement of pol a activity remains constant throughout the S phase. As the other replicative pols depend on primer synthesis by pol a-primase, the effect of the anti-pol a Ig could well represent the cumulative inhibition of pol a and the subsequent elongating enzyme. Pol e activity contributes to DNA replication more at the G1 ⁄ S transition, and its relative importance diminishes as the S phase progresses. On the other hand, the requirement of pol d activity is lowest in the early S phase and increases as the S phase proceeds. Pols d and e localize differently during the S phase Next, we studied the nuclear localization of pols a, d and e as a function of S-phase progression. Human IMR-90 primary fibroblasts were synchronized with mimosine, after splitting from confluency, to achieve a sharp entry into S phase (Fig. S2). Cells were then col- lected at different time points to study the localization pattern of the three pols and PCNA during the indica- ted cell cycle stages by immuno-EM. We chose EM, because this technique permits studying localization at near molecular resolution, and localizations can be related to nuclear structures after standard contrasting. Moreover, there is no requirement for treatment with detergent or other manipulations that remove part of the protein from the nucleus. Ultrathin cryo-sectioning was performed directly from extensively fixed cells. AB Fig. 3. Effect of inhibitory antibodies on replicative DNA synthesis in isolated, permeabilized nuclei during the S phase. Replicative DNA synthesis using isolated nuclei in the presence of excess cytoplas- mic extract was measured as incorporation of radioactive dCMP into newly synthesized DNA. Levels of inhibition by specific anti- bodies from independent replication reac- tions are plotted for each DNA polymerase (pol). The line represents the average for each pol. (A) Inhibition of replication in isolated HeLa cell nuclei synchronized with mimosine. (B) Inhibition of replication in serum-stimulated T98G cells. Results of ind- ividual experiments for pols a, d and e are plotted as triangles, filled circles and dia- monds, respectively. Lines indicate the aver- age inhibition by antibodies against the cognate pols. A. K. Rytko ¨ nen et al. Human DNA polymerases a, d and e in replication FEBS Journal 273 (2006) 2984–3001 ª 2006 The Authors Journal compilation ª 2006 FEBS 2989 After selection of suitable antibodies and optimization of the conditions, double staining was carried out with two successive antibodies and protein A conjugated with 5 and 10 nm gold particles, respectively. As can be seen from the double stainings of pols a and e, mouse mAb CL22-2-42B, against the catalytic subunit of pol a, is well suited for immuno-gold label- ling (Fig. 4, left panel) [44]. The same antibody has been previously employed successfully for immuno-EM of resin-embedded cells [45,46]. Similarly to the previ- ous studies, pol a could be detected mainly at electron- dense regions of the nucleus. Pol a labelling appears, in part, as ring-shaped, focal structures alone, or it colocalizes with pol e (Fig. 4, left panel, asterisks), or as more disperse staining of discrete nuclear regions, which is particularly visible at later stages of the S phase (Fig. 4D,E). In earlier studies, the ring-shaped foci of pol a were shown to coincide with sites of DNA synthesis [45,46]. They were also shown to repre- sent replication factories that appear as ovoid bodies attached to the nucleoskeleton in thick sections [46– 48]. Although the foci are largest and most abundant during G 1 ⁄ S transition and early S phase, pol a appears to be rather evenly distributed between ring- shaped and dispersed structures. When enumerating the gold particles from 17 pol a ⁄ e double staining ser- ies from two independent synchronizations, we found that 41% of pol a localized in foci (Table 1). The rel- ative level of pol a in foci was rather constant until late S phase⁄ G 2 transition, where the levels appeared to decrease (data not shown). This is consistent with the work of Lattanzi and coworkers [46], who reported the ring-shaped pol a foci to disappear in the G 2 ⁄ M phase. As evident from Fig. 4 (left panel), pol a colocalizes at a near-molecular level with pol e stained with mAb H3B [49]. mAb G1A [49], against pol e, gave a similar localization pattern as mAb H3B, and both mAbs colocalized in double stainings (data not shown), indi- cating the specificity of the staining. The colocalization of pol a and e is largely confined to the ring-shaped foci. In fact, more than half of all detectable foci con- tained both pol a and e (data not shown). This is not surprising, because pol e staining is concentrated in foci, 75% of pol e being focal in pol a ⁄ e double stai- nings (Table 1). Similarly to pol a, pol e in foci appears to be most pronounced from G 1 up to early S phase (Fig. 4A–C, asterisks), but pol e levels in the staining decreased relative to pol a as the S phase pro- gressed (Fig. 5). For detection of pol d, we utilized rat mAb PDK- 7B4 against p50, the B subunit of human pol d [50] (Fig. 4, right panel, 5 nm gold particles). p50 has previously been shown, by immunofluorescence micro- scopy, to colocalize with the catalytic subunit [51]. From double stainings of pol d and e, it became apparent that pol d mainly localizes outside the ring- shaped foci, which are typical of pols a and e (Fig. 4, right panel). In 17 pol d ⁄ e double staining series from different S-phase stages, only 30% of pol d-directed gold particles were found in the foci of three or more particles, whereas  74% of pol e was focal in the same series (Table 1). This difference persisted throughout the cell cycle period studied from G 1 until late S phase. Although some pol d-directed gold parti- cles could be detected in foci, the abundance of pol d in the foci was small compared with Pol e. Pol d stain- ing was instead dispersed, but restricted to distinct ter- ritories of the nucleus. It is notable that whereas some areas of the nucleus showed strong staining, neigh- bouring regions remained largely free of pol d (Fig. 4I,K). Pol d directed gold particles located in the vicinity of fibrous structures and often adopted a ‘beads-on-a-string’ structure (Fig. 4I,K, arrowheads). The overall staining intensity of pol d relative to pol e increased as the S phase progressed, and peaked in mid ⁄ late S phase 8–12 h after release from mimosine block. This was accompanied by a sharp drop, of 32%, in the fraction of the pol e-directed gold particles from 0 to 8 h (Fig. 5), consistent with an augmented role of pol d in later S phase. We next repeated pol d ⁄ e double labelling in T98G cells at different time points after cells were stimulated to proliferate by serum addition. Major features of the pol d and e staining appeared to be conserved between mimosine-synchronized fibroblasts and serum-stimula- ted T98G cells. Analysis of a series of pol d ⁄ e double- stained cells from the G 1 ⁄ S boundary until the late S phase (22 h) revealed a similar pattern of mainly focal staining for pol e (65%) and predominantly dis- persed staining for pol d (30%) (Fig. 6 and Table 1). Pol e staining was strongest in the early S phase, where large foci prevailed. In several foci, residual pol d staining could also be detected. As S phase proceeds, relative pol e staining and abundance of foci decreased, as well as the size. Foci contained, on aver- age, nine gold particles in early S phase, but only about five gold particles per focus in the mid and late S phase (Fig. 6). Late S-phase samples showed more heterogeneity, probably as a result of cells that failed to proliferate (Fig. S1). For pol d, the dispersed stain- ing detected in fibroblasts prevailed also in the T98G cells throughout S phase, with a minor part of pol d colocalizing to large pol e foci, or forming, less fre- quently, small own foci (typically three gold particles), that may well have arisen from a single pol d molecule Human DNA polymerases a, d and e in replication A. K. Rytko ¨ nen et al. 2990 FEBS Journal 273 (2006) 2984–3001 ª 2006 The Authors Journal compilation ª 2006 FEBS owing to the amplification process using secondary antibodies and protein A. PCNA colocalizes partly with pol d and partly with pol e Attempts to detect sites of ongoing DNA replication by means of BrdU incorporation failed because various methods of DNA denaturation, required for immunode- tection of BrdU, destroyed the fine structure of the cryo- sections. In order to obtain further insight into the function of pols d and e, we determined the locations of these proteins, relative to PCNA, by immuno-EM dou- ble staining. PCNA is a processivity cofactor of both pols d and e. Therefore, it is considered an important marker for active replication [52–55]. Still, not necessar- ily all PCNA participate in DNA replication, as PCNA is more abundant inside the cell than DNA replication forks at a given time, and also partakes in other DNA transactions [54]. As can be seen from a comparison AF GB CH ID EK Fig. 4. Replicative DNA polymerases (pols) show distinctive localization patterns in human IMR-90 fibroblasts synchronized with mimosine. The cells were synchronized to cell cycle stages, as indicated on the left, after release from mimosine block. Ultrathin cryosections were then subjected to immu- nostaining of pol a followed by staining of pol e (images A–E), or immunostaining of pol d followed by staining of pol e (images F–K). Immunolabelling was visualized under the electron microscope by linking the pri- mary antibody to protein A coupled to 5 nm (small: pol a and d) or 10 nm (large: pol e) gold particles. Ring-like focal staining of at least four particles is marked by asterisks, and examples of beads-on-a-string like stain- ing of pol d is shown by arrowheads. The scale bar is 100 nm. A. K. Rytko ¨ nen et al. Human DNA polymerases a, d and e in replication FEBS Journal 273 (2006) 2984–3001 ª 2006 The Authors Journal compilation ª 2006 FEBS 2991 between Fig. 7 with Fig. 4, PCNA behaved similarly to pol a, yielding a staining pattern that is partly focal (asterisks) and partly disperse or ‘beads-on-a-string’-like (arrowheads). Similar patterns of PCNA staining, coin- ciding at least partly with sites of DNA synthesis, have been reported in previous EM studies of mammalian and plant cells [56–60]. Focal staining is most apparent from G 1 to early S phase, and foci contain often also pol e (Fig. 7F,G, asterisks), and less frequently pol d (Fig. 7A–C, asterisks). It is noteworthy that although foci containing only PCNA were rarely detected, pol e foci free of PCNA were common, in particular in the G 1 and early S phases (Fig. 7F,G,I, open circles). This indi- cates that pol e is present in preformed structures. As S phase progresses, pol e staining decreases relative to PCNA, although the decrease is weaker compared with the pol d ⁄ e double staining (Fig. 5). In contrast, the levels of pol d-directed gold particles remain constant, or show a slight increase, relative to PCNA during the S phase (Fig. 5). In double staining of pol d and PCNA, pairs of small and large gold particles were visualized. They indicate intimate colocalization of the two proteins (Fig. 7D,E, arrows). As immunolabelling is obviously incomplete, such double-labelling probably detects only part of potential pol d–PCNA complexes. Taken together, the immuno-EM studies indicate that pol e adopts mainly a ring-shaped focal staining that dominates during the early S phase, whereas pol d is detected mostly as disperse or beads-on-a-string-like staining that prevails in late S phase. As for pol a and PCNA, they show staining patterns that combine focal and dispersed features. Discussion An outstanding question in eukaryotic DNA replica- tion is how the elongating replicases pols d and e co-operate to achieve efficient and faithful duplication of the nuclear DNA. We addressed this question in the present study by combining biochemical and cell biolo- gical approaches aiming to determine the spatial and temporal co-ordination of the two pols and additional replication proteins throughout the S phase. The main conclusion emerging from this study is that pols d and e pursue their functions during DNA replication with- out being physically connected, although they may per- form complementary functions at the same replication forks. We infer it from the following observations. First, the relative contribution of pol d to replicative DNA synthesis increases steadily with progression of the S phase at the expense of pol e. This is judged from the different behaviour which the two pols exhib- ited in cross-linking nascent DNA (Fig. 1) and binding chromatin (Fig. 2), as well as the degree of inhibition of replicative DNA synthesis attained with cognate inactivating antibodies (Fig. 3). Second, immuno-EM visualization revealed that pols d and e localize to mainly different nuclear sites and structures through- out the S phase (Figs 4 and 6). The more pronounced contribution of pol e in early S phase agrees with the proposed role in replication initiation. Namely, in the budding yeast, chromatin immunoprecipitation has demonstrated that pols a and e load concurrently onto origins of replication [28–30]. Subsequently, these pols transferred from origin to nonorigin DNA concomitantly with Cdc45 and MCM2-7, possibly reflecting their retention at the rep- lication fork junction as the replicated ori DNA moves away [28]. Similarly, pol e loads onto chromatin prior to initiation in Xenopus egg extracts [31]. The inde- pendent behaviour of pols d and e observed in this study could further reflect distinct roles of the two enzymes during elongation, possibly participation in the lagging and leading strand DNA synthesis, respect- ively [61,62]. In a recent chromosome-wide scan in the budding yeast, Hiraga et al. [61] were able to demonstrate that Table 1. Distribution of DNA polymerases (pols) between ring-shaped, focal structures and dispersed staining in immuno-electron microsco- py. The number of 5- and 10-nm gold particles were quantified from pol a ⁄ e and pol d ⁄ e double stainings. Clusters of three or more gold par- ticles were considered as foci. Four to 29 separate images, representing typically eight to nine nuclei, were counted for each of 17 series derived from two independent synchronizations (IMR-90 cells) or seven series derived from one synchronization (T98G cells). Staining Particles counted Particles in foci % particles in foci Nuclei counted Images counted IMR-90 Pol a (5 nm) 7029 2880 41.0 129 293 Pol e (10 nm) 4242 3180 75.0 Pol d (5 nm) 6967 2097 30.1 139 304 Pol e (10 nm) 3181 2342 73.6 T98G Pol d (10 nm) 1939 581 30.0 54 131 Pol e (5 nm) 1934 1258 65.0 Human DNA polymerases a, d and e in replication A. K. Rytko ¨ nen et al. 2992 FEBS Journal 273 (2006) 2984–3001 ª 2006 The Authors Journal compilation ª 2006 FEBS all three replicative pols a, d and e are associated with early firing origins in cells arrested early in S phase. These data suggest that all three replicases participate in the synthesis at each active origin. What is more, the authors recognized a delayed association of pol d with origin ARS305 compared with pols a and e. This is consistent with our data presented here. The functional differences between pols d and e were underscored by their ultrastructural visualization using immuno-EM. This revealed that the level of immuno- reactive pol e decreases more than threefold relative to pol d during S phase and that each enzyme exhibits a strikingly different localization pattern. Whereas pol e stained mainly as ring-shaped foci, pol d adopted a more dispersed staining of discrete nuclear territories with little focal clustering. Both pol a and PCNA show staining patterns more similar to pol e in early S phase and to pol d in late S phase. Where studied, pol a and PCNA partially colo- calize with pol d as well as pol e. However, colocaliza- tion of pols d and e was very limited in the double staining. Hence, pol a and PCNA are present in struc- tures that contain either pol d or e, but rarely, if at all, both. Although pol a and PCNA are both well-estab- lished markers for the sites of DNA synthesis using immuno-EM and resin-embedded samples [45–47,52– 60,63], it is still uncertain if all the sites of their colo- calization with pols d and e are actually DNA replica- tion sites. In other words, we cannot absolutely exclude the possibility that only a minority of the detected protein is actively engaged in DNA replica- tion, while most observed structures have other func- tions (e.g. storage sites for the replication factors). Fuss & Linn [32] studied the localization of pol e in proliferating primary fibroblasts by CLSM. The authors found that pol e formed foci throughout the cell cycle. These foci colocalized with PCNA and sites of DNA synthesis only in late S phase, but were adja- cent to PCNA foci in early S phase, suggesting a role of pol e in DNA replication late in S phase. It is diffi- cult to relate these results to the data presented here. The small foci detected by CLSM in early S phase were 300–400 nm across with an optical plane of  600 nm [32]. This is considerably larger than the ring-shaped foci observed in immuno-EM (Figs 4 and 6). The latter are, in most cases, between 50 and 100 nm across, using ultrathin sections of 70–80 nm thickness. Therefore, the ring-shape foci described here are probably below the detection limit of fluorescence microscopical techniques. In contrast, the larger foci described by Fuss & Linn [32] in late S phase corres- pond well in size to the nuclear regions of dispersed staining of PCNA and pol d that are predominant in late S phase. These regions span several hundreds of nm, and contain both dispersed PCNA and focal pol e (Fig. 7I,K). Nonetheless, direct colocalization is not Fig. 5. Quantitative analysis of DNA polymerases (pols) a , d and e, and proliferating cell nuclear antigen (PCNA) in immuno-electron microscopy. The number of nuclear gold particles representing the indicated proteins were quantified from images taken from the respective double stainings from synchronized IMR-90 cells. The graphs represent the abundance of a given gold particle relative to the total number of all gold particles in a given staining series. Each curve represents an independent experiment. Typically, 17–20 ima- ges from seven to nine nuclei were analysed per time point in each experiment. A. K. Rytko ¨ nen et al. Human DNA polymerases a, d and e in replication FEBS Journal 273 (2006) 2984–3001 ª 2006 The Authors Journal compilation ª 2006 FEBS 2993 [...]... (1994) Anatomy of a DNA replication fork revealed by reconstitution of SV40 DNA replication in vitro Nature 369, 207–212 7 Tsurimoto T & Stillman B (1991) Replication factors required for SV40 DNA replication in vitro II Switching of DNA polymerase alpha and delta during initiation of leading and lagging strand synthesis J Biol Chem 266, 1961–1968 8 Waga S & Stillman B (1998) The DNA replication fork... specificity studies that yeast DNA polymerases d and e Human DNA polymerases a, d and e in replication 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 replicate different DNA strands at an intracellular replication fork J Mol Biol 299, 405–419 Shcherbakova PV & Pavlov YI (1996) 3¢ fi 5¢ exonucleases of DNA polymerases e and d correct base analog induced DNA replication errors on opposite DNA strands in Saccharomyces... Components and dynamics of DNA replication complexes in S cerevisiae, redistribution of MCM proteins and Cdc45p during S phase Cell 91, 59–69 Aparacio OM, Stout AM & Bell SP (1999) Differential assembly of Cdc45p and DNA polymerases at early and late origins of DNA replication Proc Natl Acad Sci USA 96, 9130–9135 Masumoto H, Sugino A & Araki H (2000) Dpb11 controls the association between DNA polymerases a and... a human member of the MCM2-3–5 family, is displaced from chromatin during DNA replication J Cell Biol 129, 1433–1445 Huang D, Pospiech H, Kesti T & Syvaoja JE (1999) ¨ Structural organization and splice variants of the POLE1 gene encoding the catalytic subunit of human DNA polymerase e Biochem J 339, 657–665 Krude T, Jackman M, Pines J & Laskey RA (1997) Cyclin ⁄ Cdk-dependent initiation of DNA replication. .. Generation of monoclonal antibodies against DNA polymerase a-primase and their use for investigating protein–protein interactions during viral DNA replication PhD Thesis Munich University, Munich, Germany Stadlbauer F, Brueckner A, Rehfuess C, Eckerskorn C, Lottspeich F, Forster V, Tseng BY & Nasheuer HP Human DNA polymerases a, d and e in replication (1994) DNA replication in vitro by recombinant DNApolymerase-a-primase... step of human chromosomal DNA replication causes DNA damage J Cell Sci 17, 4897–4908 44 Nakamura H, Morita T, Masaki S & Yoshida S (1984) Intracellular localization and metabolism of DNA polymerase a in human cells visualized with monoclonal antibody Exp Cell Res 151, 123–133 45 Jaunin F, Visser AE, Cmarko D, Aten JA & Fakan S (2000) Fine structural in situ analysis of nascent DNA movement following DNA. .. H, Sugino A & Waga S (2004) Distinct roles of DNA polymerases d and e at the replication fork in Xenopus egg extract Genes Cells 9, 179–191 63 Philimonenko AA, Hodny Z, Jackson DA & Hozak P (2006) The microarchitecture of DNA replication domains Histochem Cell Biol 125, 103–117 64 Garg P & Burgers PMJ (2005) DNA polymerases that propagate the eukaryotic DNA replication fork Crit Rev Biochem Mol Biol... S-phase progression and DNA replication, rather than a DNA repair or damage response function Furthermore, the polymerase trap and inhibition of replication in isolated nuclei are both functions of DNA synthesis of the cognate pols It is difficult to conceive that a DNA synthesis function, other than DNA replication, prevails for G C Fig 7 Replicative DNA polymerases (pols) d and e partly colocalize with... for the study of DNA replication in cultured cells J Electron Microsc Tech 18, 91–105 60 Sobczak-Thepot J, Harper F, Florentin I, Zindy F, Brechot C & Puvion E (1993) Localization of cyclin A at sites of cellular DNA replication Exp Cell Res 206, 43–48 61 Hiraga S-I, Hagihara-Hayashi A, Ohya T & Sugino A (2005) DNA polymerases alpha, delta, and epsilon localize and function together at replication forks... Steitz TA (1998) A mechanism for all polymerases Nature 391, 231–232 3 Lehman IR & Kaguni LS (1989) DNA polymerase a J Biol Chem 264, 4265–4268 4 Bullock PA, Seo YS & Hurwitz J (1991) Initiation of simian virus 40 DNA synthesis in vitro Mol Cell Biol 11, 2350–2361 5 Nethanel T & Kaufmann G (1990) Two DNA polymerases may be required for synthesis of the lagging strand DNA of simian virus 40 J Virol 64, 5912–5918 . Distinctive activities of DNA polymerases during human DNA replication Anna K. Rytko ¨ nen 1,2 , Markku Vaara 2 , Tamar Nethanel 3 ,. bulk of DNA during nuclear DNA replication. These pols are structurally related, belonging to the family B DNA polymerases [2]. Nonetheless, all three perform additional roles in other DNA transactions. contributions of human DNA polymerases (pols) a, d and e during S-phase progression were studied in order to elaborate how these enzymes co-ordinate their functions during nuclear DNA replication.