Title: The replication factory targeting sequence/PCNA-binding site is required in G1 to control the phosphorylation status of DNA ligase I
Abstract: Article15 October 1999free access The replication factory targeting sequence/PCNA-binding site is required in G1 to control the phosphorylation status of DNA ligase I Rossella Rossi Rossella Rossi Istituto di Genetica Biochimica ed Evoluzionistica, CNR, Via Abbiategrasso 207, 27100 Pavia, Italy Search for more papers by this author Antonello Villa Antonello Villa M.I.A., Dipartimento di Farmacologia and DIBIT Scientific Institute S. Raffaele, Via Olgettina 60, Università di Milano, 20132 Milano, Italy Search for more papers by this author Claudia Negri Claudia Negri Istituto di Genetica Biochimica ed Evoluzionistica, CNR, Via Abbiategrasso 207, 27100 Pavia, Italy Search for more papers by this author Ivana Scovassi Ivana Scovassi Istituto di Genetica Biochimica ed Evoluzionistica, CNR, Via Abbiategrasso 207, 27100 Pavia, Italy Search for more papers by this author Giovanni Ciarrocchi Giovanni Ciarrocchi Istituto di Genetica Biochimica ed Evoluzionistica, CNR, Via Abbiategrasso 207, 27100 Pavia, Italy Search for more papers by this author Giuseppe Biamonti Giuseppe Biamonti Istituto di Genetica Biochimica ed Evoluzionistica, CNR, Via Abbiategrasso 207, 27100 Pavia, Italy Search for more papers by this author Alessandra Montecucco Corresponding Author Alessandra Montecucco Istituto di Genetica Biochimica ed Evoluzionistica, CNR, Via Abbiategrasso 207, 27100 Pavia, Italy Search for more papers by this author Rossella Rossi Rossella Rossi Istituto di Genetica Biochimica ed Evoluzionistica, CNR, Via Abbiategrasso 207, 27100 Pavia, Italy Search for more papers by this author Antonello Villa Antonello Villa M.I.A., Dipartimento di Farmacologia and DIBIT Scientific Institute S. Raffaele, Via Olgettina 60, Università di Milano, 20132 Milano, Italy Search for more papers by this author Claudia Negri Claudia Negri Istituto di Genetica Biochimica ed Evoluzionistica, CNR, Via Abbiategrasso 207, 27100 Pavia, Italy Search for more papers by this author Ivana Scovassi Ivana Scovassi Istituto di Genetica Biochimica ed Evoluzionistica, CNR, Via Abbiategrasso 207, 27100 Pavia, Italy Search for more papers by this author Giovanni Ciarrocchi Giovanni Ciarrocchi Istituto di Genetica Biochimica ed Evoluzionistica, CNR, Via Abbiategrasso 207, 27100 Pavia, Italy Search for more papers by this author Giuseppe Biamonti Giuseppe Biamonti Istituto di Genetica Biochimica ed Evoluzionistica, CNR, Via Abbiategrasso 207, 27100 Pavia, Italy Search for more papers by this author Alessandra Montecucco Corresponding Author Alessandra Montecucco Istituto di Genetica Biochimica ed Evoluzionistica, CNR, Via Abbiategrasso 207, 27100 Pavia, Italy Search for more papers by this author Author Information Rossella Rossi1, Antonello Villa2, Claudia Negri1, Ivana Scovassi1, Giovanni Ciarrocchi1, Giuseppe Biamonti1 and Alessandra Montecucco 1 1Istituto di Genetica Biochimica ed Evoluzionistica, CNR, Via Abbiategrasso 207, 27100 Pavia, Italy 2M.I.A., Dipartimento di Farmacologia and DIBIT Scientific Institute S. Raffaele, Via Olgettina 60, Università di Milano, 20132 Milano, Italy *Corresponding author. E-mail: [email protected] The EMBO Journal (1999)18:5745-5754https://doi.org/10.1093/emboj/18.20.5745 PDFDownload PDF of article text and main figures. ToolsAdd to favoritesDownload CitationsTrack CitationsPermissions ShareFacebookTwitterLinked InMendeleyWechatReddit Figures & Info The recruitment of DNA ligase I to replication foci in S phase depends on a replication factory targeting sequence that also mediates the interaction with proliferating cell nuclear antigen (PCNA) in vitro. By exploiting a monoclonal antibody directed at a phospho-epitope, we demonstrate that Ser66 of DNA ligase I, which is part of a strong CKII consensus site, is phosphorylated in a cell cycle-dependent manner. After dephosphorylation in early G1, the level of Ser66 phosphorylation is minimal in G1, increases progressively in S and peaks in G2/M phase. The analysis of epitope-tagged DNA ligase I mutants demonstrates that dephosphorylation of Ser66 requires both the nuclear localization and the PCNA-binding site of the enzyme. Finally, we show that DNA ligase I and PCNA interact in vivo in G1 and S phase but not in G2/M. We propose that dephosphorylation of Ser66 is part of a novel control mechanism to establish the pre-replicative form of DNA ligase I. Introduction DNA ligases play essential roles in DNA metabolism by catalysing the joining of both single- and double-stranded breaks in an ATP-dependent reaction (Kornberg and Baker, 1991). Of the four distinct DNA ligases that have been identified biochemically in mammalian cells (Lindahl and Barnes, 1992; Tomkinson and Levin, 1997) only DNA ligase I (LigI) appears to be involved in DNA replication (Waga et al., 1994). LigI is the main ligase of proliferating cells (Pedrali-Noy et al., 1974; Soderhall, 1976) and its expression is induced by both proliferative (Montecucco et al., 1992) and reparative (Montecucco et al., 1995a) stimuli. The stability of the protein and of its mRNA (Lasko et al., 1990; Montecucco et al., 1992) indicates that control of LigI expression at the transcriptional level takes place over a long time period (proliferating versus resting cells), and suggests that short-term regulation occurs at a post-translational level. The human DNA ligase I cDNA encodes a 102 kDa polypeptide (hLigI) (Barnes et al., 1990) organized in two well-defined functional domains: a C-terminal catalytic domain (residues 217–919), which shares a significant level of homology with all the ATP-dependent DNA ligases, and a hydrophilic N-terminal region (residues 1–216), which is not necessary for the catalytic activity and that has no counterpart in the other types of DNA ligases so far described in mammalian cells or in yeast. Although dispensable for the in vitro DNA-joining activity, the N-terminal domain is essential in vivo and its integrity is required for an ectopically expressed enzyme to rescue the lethal phenotype observed in mouse stem cells bearing a knocked out LigI gene (Petrini et al., 1995). A clue about the function of the N-terminal domain came from the observation that this region by itself is able to inhibit a cell-free DNA replication assay probably by preventing the interaction of hLigI with the replisome (Mackenney et al., 1997). More recently, it was shown that the N-terminal domain contains all the protein determinants necessary for the nuclear localization of the protein, its recruitment to sites of ongoing DNA replication, the so-called replication factories (Hozak et al., 1993), and interaction with the proliferating cell nuclear antigen (PCNA) (Montecucco et al., 1995b, 1998; Cardoso et al., 1997; Levin et al., 1997). The molecular mechanisms that govern these interactions are still largely unknown but post-translational modifications are most likely to be involved. With regard to this, it is worth underlining that several putative phosphorylation sites have been identified in the N-terminal domain by sequence analysis (Barnes et al., 1990; Savini et al., 1994) and LigI was shown to be the in vitro substrate for a number of protein kinases, including CKII and p34 kinase (Prigent et al., 1992). Finally, it was shown that hLigI is phosphorylated in vivo at multiple sites (Prigent et al., 1992), but the modified residues and the functional role of these modifications are still to be determined. In this paper we begin to address the in vivo phosphorylation of hLigI and we report the identification and characterization of a phospho-epitope recognized by monoclonal antibody (mAb) 1A4 and located in the N-terminal domain of the enzyme. The modified serine (Ser66) is part of a strong CKII consensus site and is phosphorylated in a cell cycle-dependent manner. We found that the cell cycle-dependent modification of Ser66 is controlled by dephosphorylation in G1 and probably in S phase. Dephosphorylation in G1 requires the nuclear localization of the enzyme and a functional PCNA-binding site (Montecucco et al., 1998). Notably hLigI interacts with PCNA in vivo both in G1 and S phase but not in G2/M. Results Isolation of a mAb against a hLigI phospho-epitope Monoclonal antibodies were raised against the baculovirus-expressed human DNA ligase I (rhLigI). Among them we selected two independent clones, 2B1 and 1A4, which in HeLa cell extracts recognized a single protein with the same electrophoretic mobility as hLigI, as revealed by the anti-LigI 1A9 mAb (kindly provided by Professor Tomas Lindahl; Figure 1A). The specificity of binding was also confirmed by the fact that recognition of the epitope in Western blotting was entirely competed by an excess of rhLigI (not shown). The two mAbs exhibited a different behaviour in indirect immunofluorescence (Figure 2A). 2B1 mAb stained all HeLa cell nuclei revealing the typical cell cycle-dependent sub-nuclear distribution of LigI, punctated in S phase and homogeneous during the remaining part of the cell cycle (Montecucco et al., 1995b). In contrast, 1A4 mAb decorated only a subset of nuclei, most of which showed the punctated pattern distinctive of S phase. To verify whether sites recognized by 1A4 mAb coincided with replication factories, HeLa cells were synchronized at different moments of S phase, labelled with bromodeoxy uridine (BrdU) for 45 min prior to fixation and immediately stained with anti-BrdU and 1A4 mAbs. Figure 2B shows the distribution of BrdU and 1A4 epitope in cells harvested 1, 5 and 8 h after releasing from the metabolic block. Co-localization of 1A4 epitope with sites of BrdU incorporation was detectable throughout S phase. On the other hand, in a proportion of cells harvested after 8 h, the enzyme was no longer exclusively confined to the replication factories but homogeneously distributed throughout the nucleoplasm, presumably as a consequence of the disassembly of the replication factories in late S–early G2 (Figure 2B). We interpreted the results of the immunofluorescence analysis as an indication that the 1A4 epitope could originate from a cell cycle-dependent post-translational modification of LigI. Since both the antigen and the cellular enzyme have been shown to be phosphorylated (Prigent et al., 1992; Gallina et al., 1995), we asked whether 1A4 mAb was directed toward a phospho-epitope. Thus, 2B1 and 1A4 mAbs were challenged in Western blot against the purified rhLigI, untreated or dephosphorylated with calf intestinal phosphatase (CIP). As shown in Figure 1B, while the ability of 2B1 to recognize rhLigI was not affected by CIP treatment, 1A4 mAb exclusively detected the non-dephosphorylated enzyme. Consistently, 1A4 mAb also failed to recognize the cellular enzyme immunopurified from a HeLa cell extract and then in vitro dephosphorylated with CIP (Figure 1C). We concluded that 1A4 mAb is directed toward a phospho-epitope of hLigI. Notably, this epitope is not conserved in mouse, as indicated by the inability of 1A4 to recognize the murine enzyme in Western blotting (Figure 1D). Figure 1.Characterization of 1A4 and 2B1 mAbs. (A) Western blot analysis of HeLa cell extracts with 1A4, 2B1 and 1A9 mAbs. Total cell extract (1 μg) was run on a standard 7.5% SDS–polyacrylamide gel, transferred to a nitrocellulose filter and probed with 1A4 and 2B1 mAbs at a 1:2000 dilution. The anti-DNA ligase I 1A9 mAb (kindly provided by Professor Tomas Lindahl) was used as a control. The antigen–antibody complexes were revealed with a peroxidase-conjugated goat anti-mouse IgG secondary antibody (Pierce) and a chemiluminescent substrate (Super Signal, Pierce). (B) 1A4 mAb is specifically directed to a phospho-epitope. Purified rhLigI was incubated with 1 U of CIP (+) or only in phosphatase buffer (−) and analysed by Western blotting with 1A4 and 2B1 mAbs (dilution 1:1000) and alkaline phosphatase-conjugated goat anti-mouse IgG secondary antibody (Promega). (C) hLigI was immunoprecipitated from HeLa cell extract as described in Materials and methods, incubated in the presence (+) or in the absence (−) of 1 U of CIP and run on SDS–PAGE (7.5% acrylamide:0.09% bis-acrylamide, to increase the difference in electrophoretic mobility between the phosphorylated and dephosphorylated protein). Proteins were transferred onto a nitrocellulose filter and probed with 1A4 and 2B1 mAbs as described in (A). (D) 1A4 mAb does not recognize the murine DNA ligase I. Western blot analysis of human HeLa (H) and murine NIH 3T3 (M) cell extracts with 1A4 and 2B1 mAbs performed as described in (A). Download figure Download PowerPoint Figure 2.(A) 1A4 mAb detects a subset of HeLa cell nuclei. Exponentially growing HeLa cells were methanol-fixed and immunostained either with 1A4 or 2B1 mAbs and rhodamine-conjugated sheep anti-mouse IgG secondary antibody. Cell nuclei were stained with DAPI. Specimens were examined and photographed with a Leitz Orthoplan microscope. (B) HeLa cells were synchronized at the G1/S border by two successive thymidine blocks. In order to label replication factories, cells were incubated with BrdU for 45 min prior to fixation. Cells were co-stained with 1A4 mAb and FITC-conjugated anti-BrdU mAb. Antigen–antibody complexes were visualized by confocal laser microscopy. Images of the same cells stained with the two antibodies are shown. Early S, 1 h after release from the thymidine block; Mid S, after 5 h; Late S, after 8 h. Download figure Download PowerPoint The 1A4 phospho-epitope overlaps a CKII site in the N-terminal regulatory domain of hLigI In order to map the 1A4 phospho-epitope, we developed an in vivo assay that entailed Western blot analysis of epitope-tagged hLigI mutants expressed in transiently-transfected cells. We first checked whether the 1A4 epitope was located in the N-terminal regulatory domain of hLigI (residues 1–216), where several consensus sites for Ser/Thr kinases were identified by sequence analysis. We expressed in COS7 cells two epitope-tagged hLigI mutants, the first lacking the region between the replication factory targeting sequence (RFTS) and the nuclear localization sequence (NLS) (Δ31-118; Table I), and the second bearing a deletion of the residues of the N-terminal domain downstream of the NLS (Δ132–216; Table I). Western blot analysis of total cell extracts prepared 48 h after transfection (Figure 3A) showed that 1A4 mAb only recognized the Δ132–216 mutant thus locating the epitope between residues 31 and 118. Further mutants (see Table I) narrowed the interval to a stretch of 47 amino acids (residues 31–77; Figure 3A) comprising five serines and one threonine. We focused on Ser66, which is part of a strong CKII consensus site (S66EGEEEDE; Figure 3). The failure of 1A4 mAb to recognize the A66Δ132–216 mutant in which an alanine was substituted for Ser66 (Figure 3A; Table I) proved that the epitope is produced through phosphorylation of this serine. To test whether the epitope could result from the phosphorylation of hLigI by CKII, the recombinant enzyme was dephosphorylated in vitro with CIP, purified on phosphocellulose and re-phosphorylated in vitro with recombinant CKII. As shown in Figure 3B, 1A4 mAb recognized only the re-phosphorylated rhLigI. Altogether these results demonstrate that the 1A4 epitope is produced through phosphorylation, most likely by CKII, of Ser66. This finding explains why 1A4 mAb fails to recognize the murine LigI which, even though containing a putative CKII site at the same relative position (S65CEGEDEDE), differs in sequence from the human enzyme. Figure 3.Identification of the 1A4 phospho-epitope. (A) The indicated constructs were transfected into COS7 cells. Total cell extracts were prepared and analysed by Western blotting with anti-tag HUC1-1 (1:1000 dilution), 1A4 and 2B1 mAbs. The Western blot assay was as described in Figure 1A. The position of molecular mass markers is indicated on the left. (B) Recombinant rhLigI (10 μg) was dephosphorylated in vitro with 10 U of CIP. An aliquot was rephosphorylated with 0.4 mU of recombinant CKII. Both the dephosphorylated and re-phosphorylated proteins were analysed by Western blotting with 1A4 and 2B1 mAbs under conditions described in Figure 1A. At the bottom of the figure is the sequence of the hLigI region spanning residues 30–77. The CKII consensus site is underlined. Download figure Download PowerPoint Table 1. Schematic representation of the hLigI mutants used in this study All mutants consist of the entire C-terminal catalytic domain (residues 217–919) preceded by a mutated N-terminal regulatory region (residues 1–216) that contains the indicated deletions and/or substitutions. Mutants were produced with suitable primers and PCR-mediated mutagenesis. All the indicated constructs, including the wild-type protein (Lig-Tag-wt), bear at their C-terminus an epitope recognized by HUC1-1 mAb. The NLS (residues 119–131) and the signal that directs the polypeptides to the replication factories (RFTS, residues 1–20) are indicated. The ability of hLigI mutants to associate with replication factories was scored as '+' (targeting proficient) or '−' (targeting deficient). The presence (+) or the absence (−) of the 1A4 epitope (1A4) in the mutated polypetides is indicated. The ability of the different mutants to undergo (+) or not (−) cell cycle-dependent dephosphorylation of Ser66 is also indicated. Phosphorylation of Ser66 is cell cycle-dependent The indirect immunofluorescence experiment in Figure 2A shows that phosphorylation of Ser66 is cell cycle regulated. To investigate this phenomenon in more detail, HeLa cells were synchronized in mitosis by nocodazole and shake-off treatment, and then released from the block by replating in fresh medium. Total cell extracts were prepared from mitotic cells and from cells harvested 1, 2, 3, 6, 9, 12, 15 and 21 h after removing the drug. The efficacy of synchronization was assessed both by FACS (not shown) and by Western blot analysis of total cell extracts with anti-cyclin A, anti-cyclin B and anti-cyclin E antibodies (Figure 4A). It is known that cyclin A is degraded in mitosis (normally before the nocodazole block), is absent in G1 and starts to accumulate as cells reach the G1/S border. Cyclin B is degraded as soon as nocodazole is washed out and cells are allowed to complete mitosis. Finally cyclin E is detectable only in a narrow window centred around the G1/S border (Pines and Hunter, 1994; Sherr, 1994; Clute and Pines, 1999). As shown in Figure 4A, cyclin A was barely detectable in extracts prepared either from mitotic cells or from cells grown for up to 12 h in nocodazole-free medium, in agreement with the results obtained by Pagano et al. (1992). In contrast, the cyclin B level was high in mitosis and during the first hour of incubation in the absence of the drug, and fell drastically during the successive hour. Cyclin E was detectable only in extracts prepared 9, 12, 15 and 21 h after reseeding cells in fresh medium and peaked between 12 and 15 h. The same cell extracts analysed with anti-cyclin antibodies were then probed with 1A4 and 2B1 mAbs. As expected, comparable levels of 2B1 epitope, and hence of hLigI, were detectable throughout the experiment. In contrast, the level of 1A4 epitope changed during the cell cycle (Figure 4A): it was maximal in mitosis, decreased during the first 2 h of growth in fresh medium and was undetectable for most of G1 (3, 6 and 9 h). The reappearance of the epitope 12 h after release from the block occurred concomitantly with the increase in cyclin A levels. Notably, mitotic levels of 1A4 epitope were not reached even in late S/G2 phase (21 h after replating in nocodazole-free medium). In conclusion, phosphorylation of Ser66 starts as soon as cells enter S phase, although the level of 1A4 epitope increases mostly after completion of S phase. Western blot analysis of extracts prepared from cells in G1 (6 h), S phase (12 and 15 h) or mitosis (M) and fractionated in SDS–PAGE, under conditions that amplify differences in electrophoretic mobility, showed that the apparent molecular mass of hLigI was significantly higher in M than during the rest of the cell cycle (Figure 4B). This difference in electrophoretic mobility is likely to be due to a change in the phosphorylation level since, as we showed in Figure 1C, the apparent molecular mass of hLigI is significantly affected by this post-translational modification. It is worth underlining that the fraction of the enzyme stained by 1A4 in S phase still migrated faster than mitotic hLigI, suggesting that the change in electrophoretic mobility is due to phosphorylation of as yet unidentified residues. This interpretation is in agreement with the results of Prigent et al. (1992), according to which the enzyme is a substrate for p34cdc2 in vitro. Figure 4.Cell cycle-dependent phosphorylation of Ser66. HeLa cells were synchronized in mitosis by a 16 h nocodazole (40 ng/ml) and shake-off treatment. (A) Mitotic cells were divided into nine aliquots. One aliquot (M) was immediately lysed in sample buffer to be successively analysed by Western blot. The remaining aliquots were reseeded in fresh medium and harvested 1, 2, 3, 6, 9, 12, 15 and 21 h later. Equal numbers of cells were analysed by Western blotting with 1A4 and 2B1 mAbs, anti-cyclin A rabbit antiserum, anti-cyclin B and anti-cyclin E mAbs after fractionation on a 10% SDS–PAGE. (B) Samples were run on 7.5% acrylamide:0.09% bis-acrylamide SDS–PAGE to increase the difference in electrophoretic mobility among proteins with different levels of phosphorylation. Download figure Download PowerPoint In summary, the results in this section seem to indicate that the cell cycle-dependent phosphorylation of Ser66 can distinguish the post-replicative form of hLigI in S and G2 from the pre-replicative enzyme in G1. This issue will be addressed further in the Discussion. Dephosphorylation of Ser66 requires the RFTS The proximity of Ser66 to both the RFTS (residues 1–20; Montecucco et al., 1998) and the NLS signals (residues 119–131; Montecucco et al., 1995b) prompted us to investigate whether phosphorylation of this residue could affect the nuclear and/or sub-nuclear targeting of hLigI. Thus, we produced two tagged mutants of hLigI in which Ser66 was replaced either with an alanine or with an aspartic acid residue to mimic the permanently dephosphorylated or permanently phosphorylated enzyme, respectively. Mutants were expressed in transiently-transfected cells and their localization was assessed by indirect immunofluorescence with anti-tag HUC1-1 mAb. As shown in Figure 5A, both mutants were nuclear and efficiently recruited to replication factories in S phase, ruling out the possibility that phosphorylation of Ser66 alone could affect the sub-cellular distribution of LigI during the cell cycle. Figure 5.(A) Recruitment of Ser66 substitution mutants to the replication factories. A66-LigI and D66-LigI were transfected into 46BR.1G1 cells and their sub-cellular distribution was determined by indirect immunofluorescence 48 h later. For 45 min prior to fixation, cells were grown in BrdU-containing medium in order to label replication factories. Cells were co-stained with HUC1.1 mAb and a FITC-conjugated anti-BrdU mAb as described in Montecucco et al. (1995). Confocal laser images of the transfected cells are shown. (B) Cytoplasmic phosphorylation of Ser66. The Δ2-20/119–216 mutant, which is confined to the cell cytoplasm, was expressed in NIH 3T3 cells and the presence of the 1A4 phospho-epitope was verified by indirect immunofluorescence. Nuclei were stained with DAPI. Specimens were photographed with a Leitz Orthoplan microscope. Download figure Download PowerPoint Next we explored the alternative possibility that the sub-cellular distribution of the enzyme could instead influence the phosphorylation status of Ser66. In order to do this, we tested the ability of 1A4 mAb to recognize the Δ2-20/119–216 mutant (Table I), which lacks both the NLS and the RFTS and is therefore confined to the cytoplasm of the transfected cells. Mouse NIH 3T3 cells were selected for this experiment since the murine enzyme is not recognized by 1A4 mAb (see Figure 1D). As shown in Figure 5B, 1A4 stained the cytoplasm of the transfected cells, unequivocally proving that phosphorylation of Ser66 did not require the proper sub-cellular localization of the enzyme. Finally, we investigated whether the sub-cellular distribution could control Ser66 phosphorylation during the cell cycle. The cytoplasmic Δ2-20/119–216 mutant described above was expressed in COS7 cells, and total cell extracts were prepared either from cells synchronized in mitosis with nocodazole or harvested in G1, 6 h after releasing from the block. Western blot analysis with both 1A4 and anti-tag HUC1-1 mAbs showed that, contrary to the cellular enzyme, this mutant was still phosphorylated in G1 (Figure 6A). To rule out the possibility that this difference could be simply due to the overexpression of the recombinant protein (Figure 6A, compare the intensity of the 1A4 signals due to the endogenous and to the transfected protein, respectively), a wild-type epitope-tagged hLigI was challenged in the same assay. The fact that the transfected wild-type enzyme was correctly dephosphorylated in G1 indicated that the sub-cellular distribution did indeed have a role in the cell cycle-dependent modification of Ser66 (Figure 6B). Surprisingly, however, dephosphorylation rather than phosphorylation appeared to be affected by the sub-cellular localization of the transfected protein. To investigate this aspect further, we tested additional hLigI mutants in the same assay (see Table I). The first mutant (NLS-M2) was confined to the cytoplasm because of amino acid substitutions that inactivate the NLS (Montecucco et al., 1995b). The second mutant, which bore an 84-amino-acid deletion in the N-terminal domain downstream of the NLS (Δ132–216), was nuclear and was recruited to the factories during S phase (Montecucco et al., 1998). The last mutant (Δ2-20) lacked a RFTS and, consequently, although nuclear, was not targeted to the BrdU foci in S phase (Montecucco et al., 1998). As shown in Figure 6C–E, only Δ132–216 was correctly dephosphorylated in G1, while NLS-M2 remained phosphorylated in G1 in agreement with the result obtained with the Δ2-20/119–216 mutant (Figure 6A). Unexpectedly, the Δ2-20 mutant which lacked the RFTS did not undergo dephosphorylation in G1. Since Ser66 is close to the region deleted, the possibility existed that a structural rearrangement occurring in this mutant could affect the recognition of the phosphoresidue. To rule out this possibility, and to understand whether dephosphorylation did indeed require a functional RFTS, we exploited the L(F/G)8,9 mutant in which two consecutive phenylalanines in the RFTS were replaced with glycines. We showed previously that this mutant is unable to interact with PCNA and, even though correctly targeted to the cell nucleus, is never recruited to the replication factories (Montecucco et al., 1998). The L(F/G)8,9 mutant was co-transfected with the Δ132–216 construct into COS7 cells, and phosphorylation of Ser66 was assessed by Western blotting. As shown in Figure 6F, contrary to the result for Δ132–216, L(F/G)8,9 remained phosphorylated in G1. Thus, Ser66 dephosphorylation in G1 depends both on the nuclear localization of the enzyme and on the functionality of the RFTS motif, previously thought to be required only during S phase. Figure 6.Identification of the protein motifs required for dephosphorylation of Ser66 in G1. COS7 cells were transfected with the indicated constructs. The day after transfection, cells were synchronized in mitosis by nocodazole and shake-off treatment. Half the population was immediately lysed in sample buffer (M). The remaining cells were replated in fresh medium and harvested 6 h later (G1). Total cell extracts were analysed with 1A4 and HUC1-1 mAbs as described in Figure 1A. (A) Analysis of the Δ2-20/119–216 mutant. Left and right panels are different exposures of the same filter. The arrow points to the endogenous DNA ligase I. (B), (C), (D) and (E) show the behaviour of the wild-type protein (Lig-Tag) and of the indicated mutants (Δ132–216, NLS-M2, Δ2-20). The arrow in (C) points to the endogenous DNA ligase I. Lane C in (E) shows, for comparison, the levels of 1A4 epitope due to the endogenous enzyme in an equivalent amount of non-transfected mitotic cells. (F) The behaviour of the L(F/G)8,9 and Δ132–216 mutants co-transfected into COS7 cells. Download figure Download PowerPoint In vivo interaction of hLigI with PCNA is cell cycle-dependent The results in the previous section demonstrate that dephosphorylation of Ser66 in G1 is prevented by the same mutations that inhibit the interaction of hLigI with PCNA in vitro as well as inhibiting the recruitment of the enzyme to the replication factories. It is conceivable, therefore, that both dephosphorylation of Ser66 in G1 and recruitment to the factories in S phase are correlated to the interaction of hLigI with PCNA. On the basis of this consideration, we decided to test whether such an interaction did occur in vivo. We used a rabbit polyclonal antiserum to hLigI to immunoprecipitate the enzyme from a total cell extract prepared from exponentially growing HeLa cells. We then checked by Western blotting whether PCNA could be co-immunoprecipitated under conditions that preserve protein–protein interactions. As a control, the same experiment was performed with the pre-immune serum. As shown in Figure 7A, the polyclonal anti