DISCUSSION
An increasing number of DNA repair proteins have been shown to have cell-cycle-dependent expression, including UNG2, RAD51, RAD52, APE, hNTH1, ANPG and hMYH (48,61–65). Northern blot hybridization of synchronized HaCat cells showed a cell-cycle-dependent mRNA expression of hOGG1 with increased transcription of the major nuclear and mitochondrial isoforms during S/G2 (Figure 1). Among the predicted DNA-binding sites for transcription factors in the HOGG1 promoter are two consensus sequences that potentially bind NF-Y (position 1884–1897, accession no. AJ131341) and E2F (position 1862–1869, accession no. AJ131341). Both NF-Y and E2F have been shown to be crucial regulators of the cell-cycle-dependent expression of several genes such as cdc2, CDC25, cyclin E, cyclin A, cyclin B, topoisomerase IIα and A-myb transcription factor (66–70). Moreover, although the murine and human OGG1 cDNAs are well conserved (85% over the ORF), there is little homology between the promoter regions of the genes (46% over 2 kb upstreams of the translation start). However, the mouse OGG1 gene also contains putative binding sites for NF-Y (position 2028–2040) and E2F (positions 967–974 and 231–240). In contrast, a recent study in which the promoter of HOGG1 was fused to the firefly luciferase gene carried out in HeLa cells showed that the expression of the reporter gene was not modulated during the cell cycle (71). As these experiments have been carried out in a different cell line, the participation of tissue-specific transcription factors in the transcriptional cell-cycle-dependent regulation of HOGG1 cannot be excluded.
We and others have tried to examine the intracellular distribution of endogenous hOGG1 in different cell lines. However, endogenous hOGG1 is barely detected using the antibodies that are presently available [(23) and L. Luna, M. Bjørås, V. Rolseth and E. Seeberg, unpublished data]. We, therefore, used the EGFP tag to study the subcellular localization of hOGG1. Using this approach we showed that EGFP-hOGG1 was relocalized during the cell cycle and it was associated with the nucleoli during S-phase and with condensed chromosomes during mitosis. The method by which cells were fixed (formalin, PFA or ethanol) had an effect on the localization of EGFP-HOGG1 within the nucleoli as can be seen when comparing Figures 2 and 5–8. In live cells, EGFP-hOGG1 appeared to be concentrated around the nucleoli while in fixed cells it appeared to have ‘leaked’ into the nucleolus. This has also been observed in other proteins like the high-mobility group box proteins 1 and 2 (HMGB1 and HMGB2) (72).
Some insights into the biological role of hOGG1 might be gained by examining putative role(s) of hOGG1 in the nucleolus. Ribosomal DNA is relatively GC-rich and it has been shown to be very sensitive to ionizing radiation (73). However, only a few studies have been carried out to investigate both the extent and the repair of DNA damage of rDNA in mammals. The severe hereditary progeroid disorder Cockayne syndrome is a consequence of a defective transcription-coupled repair (TCR) pathway. Some key players in TCR, such as the Cockayne syndrome A (CSA) and B (CSB) proteins have been identified (74). Recently, Egly and co-workers (75) showed that CSB is found in the nucleolus within a complex that includes RNA pol I, TFIIH and XPG. At the same time, Bohr and co-workers (76) reported a functional cross-talk between CSB and hOGG1. As neither a functional nor a physical direct interaction was detected between these two proteins, the authors hypothesize a protein complex containing among others CSB and hOGG1. Likewise, the protein coded for the XRCC1 gene participates in DNA single-strand breaks and in BER by its interaction with several of the enzymes involved in these processes. XRCC1 has been shown to be localized to the nucleolus and to physically interact with hOGG1 (77–80). Thus, we can picture a situation where multi-protein complexes composed of various combinations of polypeptides coexist in the nucleoli. The composition of these complexes depends on transient protein–protein interactions that most probably vary during cell-cycle progression and/or under specific conditions such as post-translational modifications. The actinomycin D inhibition results that show relocalization of EGFP-hOGG1 together with fibrillarin and polI to the caps in the segregated nucleoli strengthens the idea of a multimodular complex. This brings us to the question: is hOGG1 part of a complex of proteins that have a role in maintaining the integrity of rDNA, or are nucleoli simply a storage place for these proteins. Thus, the nucleolus might be a sequestering compartment of hOGG1 to avoid the accumulation of GC→TA transversions owing to the removal of 8-oxoG from 8-oxoG:A. Post-replicative repair of DNA mispairs is critical for effective maintenance of the genome. Thus, compartmentalization of hOGG1 to the nucleoli during S-phase might increase the repair efficiency of A:8-oxoG mispairs that arise during DNA replication, as dCMP or dAMP can be selectively incorporated opposite 8-oxoG (7). Accordingly, it has been shown that hMYH, an adenine-specific glycosylase responsible for removal of the mispaired adenine, localizes to the nucleus and during S-phase associates specifically with the replication foci for immediate post-replicative repair of adenine mispaired with 8-oxoG (64).
Nuclear localization signals (NLSs) are short stretches of amino acids that mediate the transport of nuclear proteins into the nucleus. NLSs are classified into three categories; the best studied are mono- and bi-partite motifs (81). Several sequences in a different number of proteins have been defined as nucleolus localization signals (NoLSs). In contrast to NLSs, there are no simple NoLSs and it has been proposed that localization of proteins to the nucleolus may occur through a variety of mechanisms (82). The nuclear localization of hOGG1 has shown to be driven by an NLS situated in the C-terminal segment of the protein (21). The serine to cysteine 326 mutation did not affect the nuclear localization of the protein; however, the association with the nuclear matrix and the chromatin, the nucleoli localization during S-phase and co-localization to the condensed chromosomes during mitosis were altered. As there appears to be no NoLS consensus similar to a mitochondrial or peroxisomal localization sequences, a scenario has emerged in which a nucleolar protein might have an interaction that causes it to be localized to the nucleolus. Thus, we suggest that the Ser-326 has to be phosphorylated to interact with a protein during S-phase that will translocate EGFP-hOGG1 to the nucleolus and to the condensed chromosomes during mitosis. Supporting this idea is the restoration of wild-type localization of the EGFP-hOGG1-Glu326 mutant protein that mimicked a phosphorylated serine. A link between aberrant protein localization and cancer has been established by several studies that include BRCA1, p53, APC and RCAS1 (83–87). A recent report has shown that the hOGG1-Cys326 protein has a lower ability to suppress mutations than the hOGG1-Ser326 protein in human cells in vivo (88). We suggest that the difference in the mutation suppressive ability between the two polymorphic hOGG1 proteins is due to the subnuclear localization brought about by a phosphorylation of the Ser-326.
Finally, we had identified a phosphorylated form of EGFP-hOGG1 in the chromatin fraction by western blotting using a phosphoserine-specific antibody (47). Interestingly, this form was still present in the chromatin fraction isolated from HeLa-EGFP-hOGG1-Glu326 but it was clear that this antibody did not recognize the mimicked phosphorylated Ser-326. We, therefore, suggest that an additional phosphorylation on a different serine is responsible for the association of the protein with the chromatin fraction. Further studies are needed to elucidate the complete phosphorylation status of hOGG1 during the cell cycle.