Epithelial-to-mesenchymal transition and NF-kB pathways are promoted by a mutant form of DDB2, unable to bind PCNA, in UV-damaged human cells.

BACKGROUND: DNA-Damaged Binding protein 2 (DDB2) is a protein involved in the early step of Nucleotide Excision Repair. Recently, it has been reported that DDB2 is involved in epithelial-to-mesenchymal transition (EMT), key process in tumour invasiveness and metastasis formation. However, its role is not completely known. METHODS: Boyden chamber and cell adhesion assays, and ICELLigence analysis were performed to detect HEK293 adhesion and invasion. Western blotting and gelatine zymography techniques were employed to assess the EMT protein levels and MMP enzymatic activity. Immunofluorescence analysis and pull-down assays facilitated the detection of NF-kB sub-cellular localization and interaction. RESULTS: We have previously demonstrated that the loss of DDB2-PCNA binding favours genome instability, and increases cell proliferation and motility. Here, we have investigated the phenotypic and molecular EMT-like changes after UV DNA damage, in HEK293 clones stably expressing DDB2Wt protein or a mutant form unable to interact with PCNA (DDB2PCNA-), as well as in HeLa cells transiently expressing the same DDB2 constructs. Cells expressing DDB2PCNA- showed morphological modifications along with a reduced expression of E-cadherin, an increased activity of MMP-9 and an improved ability to migrate, in concomitance with a significant upregulation of EMT-associated Transcription Factors (TFs), whose expression has been reported to favour tumour invasion. We observed a higher expression of c-Myc oncogene, NF-kB, both regulating cell proliferation and metastatic process, as well as ZEB1, a TF significantly associated with tumorigenic potential and cell migratory ability. Interestingly, a novel interaction of DDB2 with NF-kB was detected and found to be increased in cells expressing the DDB2PCNA-, suggesting a direct modulation of NF-kB by DDB2. CONCLUSION: These results highlight the role of DDB2-PCNA interaction in counteracting EMT since DDB2PCNA- protein induces in HEK293 transformed cells a gain of function contributing to the acquisition of a more aggressive phenotype.

A functional in vitro cell-free system for studying DNA repair in isolated nuclei.

Assessment of DNA repair is an important endpoint measurement when studying the biochemical mechanisms of the DNA damage response and when investigating the efficacy of chemotherapy, which often uses DNA-damaging compounds. Numerous in vitro methods to biochemically characterize DNA repair mechanisms have been developed so far. However, such methods have some limitations, which are mainly due to the lack of chromatin organization in the DNA templates used. Here we describe a functional cell-free system to study DNA repair synthesis in vitro, using G1-phase nuclei isolated from human cells treated with different genotoxic agents. Upon incubation in the corresponding damage-activated cytosolic extracts, containing biotinylated dUTP, nuclei were able to initiate DNA repair synthesis. The use of specific DNA synthesis inhibitors markedly decreased biotinylated dUTP incorporation, indicating the specificity of the repair response. Exogenously added human recombinant PCNA protein, but not the sensors of UV-DNA damage DDB2 and DDB1, stimulated UVC-induced dUTP incorporation. In contrast, a DDB2PCNA- mutant protein, unable to associate with PCNA, interfered with DNA repair synthesis. Given its responsiveness to different types of DNA lesions, this system offers an additional tool to study DNA repair mechanisms.This article has an associated First Person interview with the first author of the paper.

Exploring new potential role of DDB2 by host cell reactivation assay in human tumorigenic cells.

BACKGROUND: The Host Cell Reactivation assay (HCR) allows studying the DNA repair capability in different types of human cells. This assay was carried out to assess the ability in removing UV-lesions from DNA, thus verifying NER efficiency. Previously we have shown that DDB2, a protein involved in the Global Genome Repair, interacts directly with PCNA and, in human cells, the loss of this interaction affects DNA repair machinery. In addition, a mutant form unable to interact with PCNA (DDB2PCNA-), has shown a reduced ability to interact with a UV-damaged DNA plasmid in vitro. METHODS: In this work, we have investigated whether DDB2 protein may influence the repair of a UV-damaged DNA plasmid into the cellular environment by applying the HCR method. To this end, human kidney 293 stable clones, expressing DDB2Wt or DDB2PCNA-, were co-transfected with pmRFP-N2 and UV-irradiated pEGFP-reported plasmids. Moreover, the co-localization between DDB2 proteins and different NER factors recruited at DNA damaged sites was analysed by immunofluorescence and confocal microscopy. RESULTS: The results have shown that DDB2Wt recognize and repair the UV-induced lesions in plasmidic DNA transfected in the cells, whereas a delay in these processes were observed in the presence of DDB2PCNA-, as also confirmed by the different extent of co-localization of DDB2Wt and some NER proteins (such as XPG), vs the DDB2 mutant form. CONCLUSION: The HCR confirms itself as a very helpful approach to assess in the cellular context the effect of expressing mutant vs Wt NER proteins on the DNA damage response. Loss of interaction of DDB2 and PCNA affects negatively DNA repair efficiency.

A damaged DNA binding protein 2 mutation disrupting interaction with proliferating-cell nuclear antigen affects DNA repair and confers proliferation advantage.

In mammalian cells, Nucleotide Excision Repair (NER) plays a role in removing DNA damage induced by UV radiation. In Global Genome-NER subpathway, DDB2 protein forms a complex with DDB1 (UV-DDB), recognizing photolesions. During DNA repair, DDB2 interacts directly with PCNA through a conserved region in N-terminal tail and this interaction is important for DDB2 degradation. In this work, we sought to investigate the role of DDB2-PCNA association in DNA repair and cell proliferation after UV-induced DNA damage. To this end, stable clones expressing DDB2Wt and DDB2PCNA- were used. We have found that cells expressing a mutant DDB2 show inefficient photolesions removal, and a concomitant lack of binding to damaged DNA in vitro. Unexpected cellular behaviour after DNA damage, such as UV-resistance, increased cell growth and motility were found in DDB2PCNA- stable cell clones, in which the most significant defects in cell cycle checkpoint were observed, suggesting a role in the new cellular phenotype. Based on these findings, we propose that DDB2-PCNA interaction may contribute to a correct DNA damage response for maintaining genome integrity.

MCM5: a new actor in the link between DNA replication and Meier-Gorlin syndrome.

Meier-Gorlin syndrome (MGORS) is a rare disorder characterized by primordial dwarfism, microtia, and patellar aplasia/hypoplasia. Recessive mutations in ORC1, ORC4, ORC6, CDT1, CDC6, and CDC45, encoding members of the pre-replication (pre-RC) and pre-initiation (pre-IC) complexes, and heterozygous mutations in GMNN, a regulator of cell-cycle progression and DNA replication, have already been associated with this condition. We performed whole-exome sequencing (WES) in a patient with a clinical diagnosis of MGORS and identified biallelic variants in MCM5. This gene encodes a subunit of the replicative helicase complex, which represents a component of the pre-RC. Both variants, a missense substitution within a conserved domain critical for the helicase activity, and a single base deletion causing a frameshift and a premature stop codon, were predicted to be detrimental for the MCM5 function. Although variants of MCM5 have never been reported in specific human diseases, defect of this gene in zebrafish causes a phenotype of growth restriction overlapping the one associated with orc1 depletion. Complementation experiments in yeast showed that the plasmid carrying the missense variant was unable to rescue the lethal phenotype caused by mcm5 deletion. Moreover cell-cycle progression was delayed in patient's cells, as already shown for mutations in the ORC1 gene. Altogether our findings support the role of MCM5 as a novel gene involved in MGORS, further emphasizing that this condition is caused by impaired DNA replication.

A DDB2 mutant protein unable to interact with PCNA promotes cell cycle progression of human transformed embryonic kidney cells.

DNA damage binding protein 2 (DDB2) is a protein involved in the early step of DNA damage recognition of the nucleotide excision repair (NER) process. Recently, it has been suggested that DDB2 may play a role in DNA replication, based on its ability to promote cell proliferation. We have previously shown that DDB2 binds PCNA during NER, but also in the absence of DNA damage; however, whether and how this interaction influences cell proliferation is not known. In this study, we have addressed this question by using HEK293 cell clones stably expressing DDB2(Wt) protein, or a mutant form (DDB2(Mut)) unable to interact with PCNA. We report that overexpression of the DDB2(Mut) protein provides a proliferative advantage over the wild type form, by influencing cell cycle progression. In particular, an increase in the number of S-phase cells, together with a reduction in p21(CDKN1A) protein level, and a shorter cell cycle length, has been observed in the DDB2(Mut) cells. These results suggest that DDB2 influences cell cycle progression thanks to its interaction with PCNA.

Structure-activity relationship and role of oxygen in the potential antitumour activity of fluoroquinolones in human epithelial cancer cells.

The photobehavior of ciprofloxacin, lomefloxacin and ofloxacin fluoroquinolones was investigated using several in vitro methods to assess their cytotoxic, antiproliferative, and genotoxic potential against two human cancer cell lines. We focused our attention on the possible relationship between their chemical structure, O2 partial pressure and photobiological activity on cancer cells. The three molecules share the main features of most fluoroquinolones, a fluorine in 6 and a piperazino group in 7, but differ at the key position 8, unsubstituted in ciprofloxacin, a fluorine in lomefloxacin and an alkoxy group in ofloxacin. Studies in solution show that ofloxacin has a low photoreactivity; lomefloxacin reacts via aryl cation, ciprofloxacin reacts but not via the cation. In our experiments, ciprofloxacin and lomefloxacin showed a high and comparable potential for photodamaging cells and DNA. Lomefloxacin appeared the most efficient molecule in hypoxia, acting mainly against tumour cell proliferation and generating DNA plasmid photocleavage. Although our results do not directly provide evidence that a carbocation is involved in photodamage induced by lomefloxacin, our data strongly support this hypothesis. This may lead to new and more efficient anti-tumour drugs involving a cation in their mechanism of action. This latter acting independently of oxygen, can target hypoxic tumour tissue.

DDB2 association with PCNA is required for its degradation after UV-induced DNA damage.

DDB2 is a protein playing an essential role in the lesion recognition step of the global genome sub-pathway of nucleotide excision repair (GG-NER) process. Among the proteins involved in the DNA damage response, p21(CDKN1A) (p21) has been reported to participate in NER, but also to be removed by proteolytic degradation, thanks to its association with PCNA. DDB2 is involved in the CUL4-DDB1 complex mediating p21 degradation; however, the direct interaction between DDB2, p21 and PCNA has been never investigated. Here, we show that DDB2 co-localizes with PCNA and p21 at local UV-induced DNA-damage sites, and these proteins co-immunoprecipitate in the same complex. In addition, we provide evidence that p21 is not able to bind directly DDB2, but, to this end, the presence of PCNA is required. Direct physical association of recombinant DDB2 protein with PCNA is mediated by a conserved PIP-box present in the N-terminal region of DDB2. Mutation of the PIP-box resulted in the loss of protein interaction. Interestingly, the same mutation, or depletion of PCNA by RNA interference, greatly impaired DDB2 degradation induced by UV irradiation. These results indicate that DDB2 is a PCNA-binding protein, and that this association is required for DDB2 proteolytic degradation.

Structure-activity relationship of resveratrol and its analogue, 4,4'-dihydroxy-trans-stilbene, toward the endothelin axis in human endothelial cells.

Resveratrol inhibits endothelin-1, a vascular tension regulator. We synthesized the resveratrol analogue 4,4'-dihydroxy-trans-stilbene with 2 hydroxyl groups in the 4 and 4' position to obtain a molecule more active than resveratrol (3,4',5-trihydroxy-trans-stilbene). The results demonstrate that 4,4'-dihydroxy-trans-stilbene led to a significant decrease in total endothelin-1 secretion and in endothelin-1 messenger RNA (mRNA) levels in human endothelial cells. In addition, resveratrol and its analogue decreased endothelin-converting enzyme-1 mRNA levels and further reduced the activity of the enzyme. 4,4'-dihydroxy-trans-stilbene was more active than resveratrol because the new molecule exerted greater activity at the level of endothelin synthesis and conversion, even at a lower concentration. Although 4,4'-dihydroxy-trans-stilbene and resveratrol inhibited formation of reactive oxygen species and lipid peroxidation, the treatment of cells with different oxidant agents did not modify the endothelin-1 release. This finding suggests that the inhibition of endothelin-1 secretion is independent of the antioxidant properties of the 2 compounds. On the basis of these results, the resveratrol analogue 4,4'-dihydroxy-trans-stilbene could be a promising chemopreventive agent against cardiovascular diseases.

p21CDKN1A participates in base excision repair by regulating the activity of poly(ADP-ribose) polymerase-1.

The cell cycle inhibitor p21(CDKN1A) has been shown to participate in nucleotide excision repair by interacting with PCNA. Here we have investigated whether p21 plays a role in base excision repair (BER), by analyzing p21 interactions with BER factors, and by assessing the response of p21(-/-) human fibroblasts to DNA damage induced by alkylating agents. Absence of p21 protein resulted in a higher sensitivity to alkylation-induced DNA damage, as indicated by reduced clonogenic efficiency, defective DNA repair (assessed by the comet test), and by persistence of histone H2AX phosphorylation. To elucidate the mechanisms at the basis of the function of p21 in BER, we focused on its interaction with poly(ADP-ribose) polymerase-1 (PARP-1), an important player in this repair process. p21 was found to bind the automodification/DNA binding domain of PARP-1, although some interaction occurred also with the catalytic domain after DNA damage. This association was necessary to regulate PARP-1 activity since poly(ADP-ribosylation) induced by DNA damage was higher in p21(-/-) human fibroblasts than in parental p21(+/+) cells, and in primary fibroblasts after p21 knock-down by RNA interference. Concomitantly, recruitment of PARP-1 and PCNA to damaged DNA was greater in p21(-/-) than in p21(+/+) fibroblasts. This accumulation resulted in persistent interaction of PARP-1 with BER factors, such as XRCC1 and DNA polymerase beta, suggesting that prolonged association reduced the DNA repair efficiency. These results indicate that p21 regulates the interaction between PARP-1 and BER factors, to promote efficient DNA repair.

Degradation of p21CDKN1A after DNA damage is independent of type of lesion, and is not required for DNA repair.

The inhibitor of cyclin-dependent kinases p21CDKN1A plays a fundamental role in several pathways involved in the DNA damage response, like checkpoint-mediated cell cycle arrest, transcription, apoptosis, and DNA repair. Although p21 protein level is regulated by proteasomal degradation, the relationship of this process with DNA repair pathways is not yet clear. In addition, the role of protein/protein interaction in regulating turnover of p21 protein, is controversial. Here, we show that in human fibroblasts treated with agents inducing lesions repaired through nucleotide excision repair (NER), or base excision repair (BER), p21 degradation was triggered more by the extent, than by the type of DNA damage, or consequent DNA repair pathway. In fact, lowering the amount of DNA damage resulted in an increased stability of p21 protein. Overexpression of p21 was found to obscure degradation, both for p21wt and a p21 mutant unable to bind PCNA (p21PCNA-). However, when expressed to lower levels, turnover of p21 protein after DNA damage was greatly influenced by interaction with PCNA, since p21PCNA- was more efficiently degraded than wild-type protein. Interestingly, a p21 mutant protein unable to localize in the nucleus because of mutations in the NLS region, was not degraded after DNA damage, thus indicating that nuclear localization is necessary for p21 turnover. Removal of p21 was not required for NER activity, since inhibition of p21 degradation by caffeine did not affect the UV-induced recruitment of repair proteins, such as PCNA and DNA polymerase delta, nor significantly influence DNA repair synthesis, as determined by autoradiography. These results indicate that degradation of p21 is not dependent on a particular DNA repair pathway, and is not required for efficient DNA repair.

Loss of p21 CDKN1A impairs entry to quiescence and activates a DNA damage response in normal fibroblasts induced to quiescence.

The cell cycle inhibitor p21(CDKN1A) induces cell cycle arrest under different conditions, including senescence and terminal differentiation. Still debated is its involvement in the reversible transition from proliferation to a non-dividing quiescent state (G(0)), in which a significant role has been attributed to cell cycle inhibitor p27(CDKN1B). Here we provide evidence showing that high p21 protein levels are necessary to enter and maintain the quiescence state following contact inhibition and growth factor withdrawal. In fact, entry into quiescence was impaired, both in human fibroblasts in which p21 gene has been deleted, or protein expression knocked-down by RNA interference. Importantly, in the absence of p21, human fibroblasts activate a DNA damage-like signalling pathway, as shown by phosphorylation of histone H2AX and Chk1 proteins. In addition, we show that in the absence of p21, checkpoint is activated by an unscheduled entry into S phase, with a reduced efficiency in DNA maturation, in the presence of high c-myc protein levels. These results highlight the role of p21 in counteracting inappropriate proliferation stimuli for genome stability maintenance.

Interaction of p21(CDKN1A) with PCNA regulates the histone acetyltransferase activity of p300 in nucleotide excision repair.

The cell-cycle inhibitor p21(CDKN1A) has been suggested to directly participate in DNA repair, thanks to the interaction with PCNA. Yet, its role has remained unclear. Among proteins interacting with both p21 and PCNA, the histone acetyltransferase (HAT) p300 has been shown to participate in DNA repair. Here we report evidence indicating that p21 protein localizes and interacts with both p300 and PCNA at UV-induced DNA damage sites. The interaction between p300 and PCNA is regulated in vivo by p21. Indeed, loss of p21, or its inability to bind PCNA, results in a prolonged binding to chromatin and an increased association of p300 with PCNA, in UV-irradiated cells. Concomitantly, HAT activity of p300 is reduced after DNA damage. In vitro experiments show that inhibition of p300 HAT activity induced by PCNA is relieved by p21, which disrupts the association between recombinant p300 and PCNA. These results indicate that p21 is required during DNA repair to regulate p300 HAT activity by disrupting its interaction with PCNA.

Replication-dependent DNA damage response triggered by roscovitine induces an uncoupling of DNA replication proteins.

The cyclin-dependent kinase (CDK) inhibitor roscovitine is under evaluation in clinical trials for its antiproliferative properties. Roscovitine arrests cell cycle progression in G(1) and in G(2) phase by inhibiting CDK2 and CDK1, and possibly CDK7 and CDK9. However, the effects of CDK2 inhibition in S-phase cells have been not fully investigated. Here, we show that a short-term treatment with roscovitine is sufficient to inhibit DNA synthesis, and to activate a DNA damage checkpoint response, as indicated by phosphorylation of p53-Ser15, replication protein A, and histone H2AX. Analysis of DNA replication proteins loaded onto DNA during S phase showed that the amount of proliferating cell nuclear antigen (PCNA), a cofactor of DNA replication enzymes, was significantly reduced by roscovitine. In contrast, chromatin-bound levels of DNA polymerase delta, DNA ligase I and CDK2, were stabilized. Checkpoint inhibition with caffeine could rescue PCNA disassembly only partially, pointing to additional effects due to CDK2 inhibition and the presence of replication stress. These results suggest that in S-phase cells, roscovitine induces checkpoint-dependent and -independent effects, leading to stabilization of replication forks and an uncoupling between PCNA and PCNA-interacting proteins.

Spatiotemporal dynamics of p21CDKN1A protein recruitment to DNA-damage sites and interaction with proliferating cell nuclear antigen.

The cyclin-dependent kinase inhibitor p21CDKN1A plays a fundamental role in the DNA-damage response by inducing cell-cycle arrest, and by inhibiting DNA replication through association with the proliferating cell nuclear antigen (PCNA). However, the role of such an interaction in DNA repair is poorly understood and controversial. Here, we provide evidence that a pool of p21 protein is rapidly recruited to UV-induced DNA-damage sites, where it colocalises with PCNA and PCNA-interacting proteins involved in nucleotide excision repair (NER), such as DNA polymerase delta, XPG and CAF-1. In vivo imaging and confocal fluorescence microscopy analysis of cells coexpressing p21 and PCNA fused to green or red fluorescent protein (p21-GFP, RFP-PCNA), showed a rapid relocation of both proteins at microirradiated nuclear spots, although dynamic measurements suggested that p21-GFP was recruited with slower kinetics. An exogenously expressed p21 mutant protein unable to bind PCNA neither colocalised, nor coimmunoprecipitated with PCNA after UV irradiation. In NER-deficient XP-A fibroblasts, p21 relocation was greatly delayed, concomitantly with that of PCNA. These results indicate that early recruitment of p21 protein to DNA-damage sites is a NER-related process dependent on interaction with PCNA, thus suggesting a direct involvement of p21 in DNA repair.

Anthocyanidins decrease endothelin-1 production and increase endothelial nitric oxide synthase in human endothelial cells.

Epidemiological and intervention studies correlate anthocyanin-rich beverages and a low incidence of coronary heart diseases. Since endothelin-1 (ET-1) and nitric oxide (NO) produced by endothelial NO synthase (eNOS) are vascular tension regulators secreted by endothelial cells, we studied the influence of two anthocyanidins, namely cyanidin (CY) and delphinidin (DP), on the regulation of ET-1 and eNOS in cultured human umbilical vein endothelial cells (HUVECs). Aglycon anthocyanidin forms, such as CY and DP, may be present in vivo after the first deglycosylation step occurring in the jejunum and in the liver. DP showed a major action compared to CY inducing a significant dose-dependent inhibitory effect on both protein and mRNA levels of ET-1. CY and DP both increased the protein level of eNOS, but DP showed the major effect raising eNOS protein in a dose-dependent manner. To correlate the vasoprotective effect of CY and DP with their antioxidant activity, we analysed also the antioxidant effect of anthocyanidins both in vitro and in HUVECs. In particular, we examined the effect of anthocyanidins on endothelial heme oxygenase-1 (HO-1), an inducible stress protein. In all tests, DP showed a higher antioxidant activity than CY. Finally, the antiproliferative effect induced by DP was detected in HUVECs. DP and CY differ in the number and position of hydroxyl groups in their structure; therefore, the greater biological activity by DP, compared with CY, seems to be due to the presence of the three hydroxyl groups on the B ring in the molecular structure of DP.

Intracellular localization of the cyclin-dependent kinase inhibitor p21CDKN1A-GFP fusion protein during cell cycle arrest.

The cyclin-dependent kinase (CDK) inhibitor p21CDKN1A is known to induce cell cycle arrest by inhibiting CDK activity and by interfering with DNA replication through binding to proliferating cell nuclear antigen. Although the molecular mechanisms have been elucidated, the temporal dynamics, as well as the intracellular sites of the activity of p21 bound to cyclin/CDK complexes during cell cycle arrest, have not been fully investigated. In this study we have induced the expression of p21CDKN1A fused to green fluorescent protein (GFP) in HeLa cells, in order to visualize the intracellular localization of the inhibitor during the cell cycle arrest. We show that p21-GFP is preferentially expressed in association with cyclin E in cells arrested in G1 phase, and with cyclin A more than with cyclin B1 in cells arrested in the G2/M compartment. In addition, we show for the first time that p21-GFP colocalizes with cyclin E in the nucleolus of HeLa cells during the G1 phase arrest.

Anthocyanins induce cell cycle perturbations and apoptosis in different human cell lines.

To investigate the mechanistic basis for the biological properties of anthocyanins, two aglycone anthocyanins [delphinidin (DY) and cyanidin (CY)] were used to examine their effects on cell cycle progression and on induction of apoptosis in human cancer cells (uterine carcinoma and colon adenocarcinoma cells) and in normal human fibroblasts. These compounds differ in the number and position of hydroxyl groups on the beta ring in the molecular structure. Cellular uptake of anthocyanins was confirmed by HPLC analysis and no metabolites were detected. The clonogenic assay showed that CY induces a dose-dependent growth inhibitory effect only in fibroblasts. This effect was confirmed by flow cytometric analysis, showing a significant reduction of cells in S phase. In contrast, DP inhibited cell growth in normal and tumour cell lines. This event is accompanied in fibroblasts by an accumulation of cells in the S phase suggesting a block in the transition from S to G2 phase. On the other hand, in tumour cell lines we observed a reduction of cells in G1 phase, paralleled by the appearance of a fraction of cells with a hypodiploid DNA content, thus demonstrating an apoptotic effect by DP. The occurrence of apoptosis induced by DP was confirmed by morphological and biochemical features, including nuclear condensation and fragmentation, annexin V staining, DNA laddering and poly(ADP-ribose) polymerase-1-proteolysis. Furthermore, the mitochondrial membrane potential of apoptotic cells after treatment with DP was significantly lost. The different effects exerted by DP as compared with CY suggest that the presence of the three hydroxyl groups on the beta ring in the molecular structure of DP may be important for its greater biological activity.

p21CDKN1A does not interfere with loading of PCNA at DNA replication sites, but inhibits subsequent binding of DNA polymerase delta at the G1/S phase transition.

The ability of the cyclin-dependent kinase (CDK) inhibitor p21CDKN1A to interact with PCNA recruited to DNA replication sites was investigated to elucidate the relevance of this interaction in cell cycle arrest. To this end, expression of p21 protein fused to green fluorescent protein (GFP) was induced in HeLa cells. G1 phase cell cycle arrest induced by p21GFP occurred also at the G1/S transition, as shown by cyclin A immunostaining of GFP-positive cells. Confocal microscopy analysis and co-immunoprecipitation studies showed that p21GFP co-localized and interacted with chromatin-bound PCNA and CDK2. GFP-p21 mutant forms unable to bind to PCNA (p21PCNA-) or CDK (p21CDK-) induced cell cycle arrest, although immunoprecipitation experiments showed these mutants to be unstable. Expression of HA-tagged p21wt or mutant proteins confirmed the ability of both mutants to arrest cell cycle. p21(wt)HA and p21CDK-HA, but not p21PCNA-, co-localized and co-immunoprecipitated with chromatin-bound PCNA. Association of p21 to chromatin-bound PCNA resulted in the loss of interaction with the p125 catalytic subunit of DNA polymerase delta (pol delta). These results suggest that in vivo p21 does not interfere with loading of PCNA at DNA replication sites, but prevents, or displaces subsequent binding of pol delta to PCNA at the G1/S phase transition.