Human DNA polymerase delta is a pentameric complex with dimeric p12 subunit: Implications in enzyme architecture and PCNA interaction

Human DNA polymerase delta (Polδ), a holoenzyme consisting of p125, p50, p68 and p12 subunits, plays an essential role in all the three DNA transaction processes. Herein, using multiple physicochemical and cellular approaches we found that the p12 protein forms a dimer in the solution. In vitro reconstitution and pull-down of cellular Polδ by tagged p12 authenticates pentameric nature of this critical holoenzyme. Further, a consensus PIP motif at the extreme carboxyl terminal tail and a homodimerization domain at the amino-terminus of the p12 subunit were identified. Our mutational analyses of p12 subunit suggest that 3RKR5 motif is critical for dimerization that facilitates p12 binding to IDCL of PCNA via its PIP motif 98QCSLWHLY105. Additionally, we observed that oligomerization of the smallest subunit of Polδs is evolutionarily conserved as Cdm1 of S. pombe also dimerzes. Thus, we suggest that human Polδ is a pentameric complex with a dimeric p12 subunit; and discuss implications of p12 dimerization in regulating enzyme architecture and PCNA interaction during DNA replication.


Abstract
Human DNA polymerase delta (Pol), a holoenzyme consisting of p125, p50, p68 and p12 subunits, plays an essential role in all the three DNA transaction processes. Herein, using multiple physicochemical and cellular approaches we found that the p12 protein forms a dimer in the solution. In vitro reconstitution and pull-down of cellular Pol by tagged p12 authenticates pentameric nature of this critical holoenzyme.
Further, a consensus PIP motif at the extreme carboxyl terminal tail and a homodimerization domain at the amino-terminus of the p12 subunit were identified. Our mutational analyses of p12 subunit suggest that 3RKR5 motif is critical for dimerization that facilitates p12 binding to IDCL of PCNA via its PIP motif 98QCSLWHLY105. Additionally, we observed that oligomerization of the smallest subunit of Pols is evolutionarily conserved as Cdm1 of S. pombe also dimerzes. Thus, we suggest that human Pol is a pentameric complex with a dimeric p12 subunit; and discuss implications of p12 dimerization in regulating enzyme architecture and PCNA interaction during DNA replication.

Introduction:
Accurate and processive DNA synthesis by DNA polymerases (Pol) during chromosomal DNA replication is essential for the preservation of genomic integrity and for the suppression of mutagenesis and carcinogenesis (1). Eukaryotic chromosomal DNA replication is coordinated by three essential replicases: Polα, Polδ, and Polε (2,3). Based on biochemical and genetic studies mostly those carried out in the budding yeast, it has been proposed that Polα initiates DNA replication by synthesizing a short RNA-DNA primer, and is followed by loading of DNA clamp PCNA by its loader RFC. Polδ not only can synthesize "Okazaki fragments" in lagging strand, it can also take part in leading strand DNA synthesis (4,5). The precise role of Polε remains controversial but studies indicate that it plays a role in initiation and leading strand synthesis of DNA replication (3). The mechanism of DNA replication in higher eukaryotes is yet to be deciphered; however, the in vitro SV40 replication system suggests requirement of human Polα and Polδ to complete DNA synthesis (2,6). Irrespective of their different roles in DNA replication, these DNA polymerases possess certain commonalities like multisubunit composition, and the largest subunits only possessing signature sequences of a B-family DNA polymerase having catalytic activity (7). The accessory structural subunits of DNA polymerases do not show any sequence similarities between them, therefore understanding their functions is quite intriguing.
The p50/Pol31/Cdc1 subunit makes a connecting bridge between the catalytic subunit p125/Pol3 and p68/Pol32/Cdc27, and is indispensable for cell viability. Even though Pol32 is not essential for cell survival in S. cerevisiae, in its absence cells exhibit sensitivity to both high and cold temperatures, and susceptibility to genotoxic stress (12). Contrarily, deletion of Cdc27 in S. pombe is not achievable (13). The non-essential p12 subunit is the Cdm1 homologue and is apparently absent in S. cerevisiae. Yeast two hybrid and coimmunoprecipitation analyses suggested a dual interaction of p12 with p125 and p50; however, the modes of binding are yet to be defined (14). In vitro reconstitution has facilitated purification of four different subassemblies of human Polδ such as p125 alone, p125-p50 (core complex), p125-p50-p68 and p125-p50-p68-p12 complexes for biochemical studies. Reports also suggest that the subunit composition of human Polδ may alter in vivo with cellular response to genotoxic burden (15,16). Upon treatment of human cells with genotoxins such as UV, methyl methanesulfonate, HU and aphidicolin, the p12 subunit undergoes rapid degradation to result in a trimeric human Polδ (p125/p50/p68) equivalent to yeast form with higher proof reading activity (17). Thus, p12 subunit seems to play a crucial role in regulating Polδ function.
The function of Polδ as a processive DNA polymerase mostly depends upon its association with PCNA that acts as a sliding clamp (18). The interaction of PCNA-binding proteins with PCNA is mediated by a conserved PCNA-interacting protein motif (PIP-box) with a consensus sequence Q-x-x-(M/L/I)-x-x-FF-(YY/LY). Previously, we have shown that all the three subunits of ScPolδ functionally interact with trimeric PCNA mediated by their PIP motifs (5). To achieve higher processivity in vitro, all the three PIP boxes are required; however for cellular function of ScPolδ, along with PIP of ScPol32, one more PIP box either from Pol3 or Pol31 subunit are essential. Similarly, reports from human Polδ studies suggest that all the four subunits of hPolδ are involved in a multivalent interaction with PCNA and each of them regulates processive DNA synthesis by Polδ (19). Except in p68 and p50, PIP motifs in p125 and p12 have not been mapped (20,21). Studies based on far-Western and immunoprecipitation analyses revealed that the first 19 amino acids of p12 are involved in PCNA interaction, although the region lacks a canonical PIP box sequences (14,22). Therefore, in this study we have re-investigated the interaction of p12 with other Polδ subunits and PCNA. Our results indicate that the smallest subunit p12 exists as a dimer in solution to establish a dual interaction with both p125 and p50 subunits of Polδ. The dimerization motif has been mapped to the aminoterminal end and a novel conserved PIP box has been identified in the c-terminal tail of p12. Importantly, dimerization of p12 regulates its interaction with PCNA. Based on our observations we propose that human Polδ exists in a pentameric form in the cell in addition to other subassemblies; and the effect of p12 dimerization on PCNA binding with various subassemblies have been discussed.

Oligomerization of p12 subunit of hPol
In several studies, human Pol holoenzyme has been purified either from insect cell line or bacterial systems by using standard chromatography techniques for biochemical characterizations (11, 23,24). Although most of the purified Pol used in various enzymatic assays was containing all the four subunits p125, p50, p68 and p12; in most cases the proposed stochiometry of each subunit in the holoenzyme 1:1:1:1 was less convincing. Especially the smallest subunit p12 was more than 1:1 ratio with respect to others. The band intensity of p12 was always either similar or more than that of other subunits (23,24). Even when we attempted to purify hPol holoenzyme by using GST-affinity beads, the band intensity of p12 protein was consistently higher than the other subunits ( Supplementary Fig. 1). Such discrepancy in Pol composition could arise due to either oligomeric status of p12 or multi-subunit p12 interaction with p125 and p50 or due to staining artifacts. To examine potential oligomerization status of p12 subunit, yeast two hybrid assay and native PAGE analysis were carried out (Fig. 1). p12 orf was fused in frame with both GAL4 activation (AD), and GAL4 binding domains (BD). Other hPol subunit orfs were fused with the GAL4 binding domain only. The HFY7C yeast reporter strain harboring the AD-p12 plasmid was co-transformed with any of the BD-Pol subunit plasmids, and selected on Leu -Trp -SDA plate. The interactions of p12 with any subunits of Pol in these transformants were analyzed by selecting them on plate lacking histidine. Growth on Hisplate demonstrates interaction between the two fusion proteins as only the binding of two proteins make it possible to form an intact GAL4 activator to confer HIS expression. As reported earlier, p12 interaction was observed with p125 and p50 but not with p68 ( Fig. 1 A, sectors 1, 2, and 3) (14).
Surprisingly, p12 subunit also interacted with itself to give HIS expression (sector 4); while no growth was observed in the negative controls (sector 5 and 6). As no growth was observed to show p12 and p68 binding, our results suggest that p12 makes specific multivalent interaction with itself, p125 and p50 subunits; and such interactions are not mediated through any yeast proteins.
Further to ascertain oligomerization of p12, protein was purified to near homogeneity from bacterial cells by using GST-affinity column chromatography, GST-tag was cleaved off by PreScission protease, and analyzed by both native and SDS containing polyacrylamide gel electrophoresis (PAGE). The predicted MW of p12 is ~12 kDa with a pI of 6.3. p12 is known to possess abnormal migration in SDS-PAGE (25) and similarly we observed the protein to migrate at ~15 kDa molecular weight size position ( Fig. 1 B lane 2). By taking advantage of native PAGE analysis where proteins resolve based on their charge and hydrodynamic size, we found that p12 migrated at similar position with Carbonic anhydrase (CA) which is a protein of 30 kDa with pI 6.4 ( lane 4 and 5). Thus, the co-migration of two proteins indicate that both CA and p12 possess similar mass to charge ratio and the slower migrating p12 is potentially a dimeric complex.
By taking advantage of Isothermal calorimetry (ITC) technique, the oligomerization of p12 protein was examined (Fig. 1C i). p12 was placed both in the sample cell and in the syringe of the calorimeter and binding was analyzed by monitoring the change in entropy. The H was ~ -1.82 kcal/mol, the G was -9.48 kcal/mol, and the Kd for the complex was ~146 nM. The number of ligand binding site as derived from the ITC analysis was found to be ~0.6, which is closed to 1:1 binding of p12 monomers. Thus, our both in vivo and physiochemical study clearly indicate homodimerization of p12 subunit.

Cellular existence of oligomeric p12 in Pol complex and co-localization with PCNA
In order to decipher p12 oligomerization in its cellular state, Pol holoenzyme was immunoprecipitated from the cell lysate transfected with GFP-p12 by using either anti-GFP (i) or anti-p125 (ii) antibody ( Fig.   2 A). Irrespective of any antibody used for the pull-down assay, we detected presence of five subunits of Pol in the beads. Each of the four native subunits (p125, p68, p50 and p12) were detected by probing with subunit specific antibody where as GFP-p12 was detected by anti-GFP antibody. While GFP-p12 pulled down cellular heterotetrameric Pol complex, anti-p125 antibody precipitated Pol complex with both the forms of p12 (native and GFP-tagged). Thus, even in the cellular context, p12 exists as an oligomer in the Pol complex.
PCNA that functions as a cofactor for DNA polymerases orchestrates the replisome by recruiting multiple proteins involved in DNA transaction processes. Using microscopy, it was shown that PCNA forms distinct foci or replication factories indicative of active DNA replication entities within the nucleus (26). The p68 and p50 subunits of hPol were deciphered to form foci and co-localized with PCNA in several cell lines (27,28). To examine physiological relevance of p12 oligomerization and PCNA interaction, GFP-PCNA or GFP-p12 with RFP-p12 fusion constructs were transfected to CHO cell line.
As shown in Fig.2 B, irrespective of GFP (stained green) or RFP (stained red) fusion, p12 formed discrete compact foci and the patterns were very similar to reported p50 and p68 foci. Subsequent merging of foci in co-transfectant of GFP-p12 and RFP-p12 resulted in appearance of yellow foci (i). Thus, 100% coincidental accumulation of both the p12 foci suggests that both the proteins are part of replication unit and function together. Similarly, we have also observed sub-cellular co-localization of p12 with PCNA as yellow foci appeared by merging foci of GFP-PCNA and RFP-p12 (ii). However, in similar condition, we did not notice any co-localization of GFP-p12 and RFP-Pol foci (iii). We suggest from these observations that p12 could function in replication factories as an oligomeric protein with PCNA.

Identification of motifs involved in p12 dimerization and interaction with PCNA
To identify dimerization and PIP motifs in p12, primary sequences from human, mouse, bovine and S. pombe were aligned (Fig. 3 A). The CLUSTAL W alignment analysis showed a high degree of amino acid conservation of p12 sequences in mammals, and they showed only ~17% identity with S. pombe orthologue Cdm1. The carboxyl termini of these proteins displayed better conservation than the amino termini. Cdm1 is composed of 160 aa and the divergence is apparently due to possession of an insert of about 38 amino acids exactly in the middle of the p12 homology sequences in Cdm1. We thought to examine the role of two highly conserved sequences located at the extreme ends of p12: a basic tripeptide sequence 3RKR5 (henceforth referred as RKR motif) motif and a putative PIP box sequence 98QCSLWHLY105 (henceforth referred as PIP motif). The putative PIP motif is located in the carboxyl terminal tail of p12, which is the usual position of PIP in most of the DNA polymerases and appears to be very similar to known PIP sequences (Fig. 3 B). Since, an earlier study reported 4KRLITDSY11 (a part of RKR motif) as a PCNA binding region of p12 (14), we wanted to compare the model structure of this peptide with 98QCSLWHLY105. Structurally PIP box sequences are highly conserved and formation of a 310 helix is a characteristic of such sequences. The amino acid stretch encompassing RKR (1-  and PIP (92-GDPRFQCSLWHLYPL-106) domains were used for peptide structure prediction by using PEP-FOLD3 server (http://bioserv.rpbs.univ-paris-diderot.fr/services/PEP-FOLD3/) rather than using a known template based prediction to avoid any biasness. Further the models were validated by SAVES and Ramachandran plot ( Supplementary Fig. 2 A and B), which showed most of the residues in allowed regions.
Our structural prediction suggested first 10 amino acids of RKR motif to form a -helix whereas the PIP motif of p12 forms a typical 310 helix, the structure that snugly fits into the IDCL domain of PCNA. The p12 PIP structure was further aligned with available X-ray crystal structures of PIP peptide from p21 (1AXC) and p68 (1U76). The superimposition shows a high degree of similarity between the PIP motifs ( Fig. 4 C). Similarly, p12 peptide structure was aligned with that of p68 PIP-hPCNA co-crystal structure, and a remarkable overlapping between the structures was observed (Fig. 3 C). Thus, our in silico analysis indicated that the C-terminal PIP sequence 98QCSLWHLY105 is the most probable motif to interact with PCNA than the N-terminal RKR motif 4KRLITDSY11.
To provide experimental evidence to our in silico prediction, two p12 mutants were generated by mutating R3, K4, R5 and L104, Y105 to alanines; and their binding to wild type p12 and PCNA were analyzed by yeast two hybrid approach (Fig. 3 D). As depicted, while transformants of BD-p12 with AD-p12 or AD-PCNA grew on SDA plate lacking leucine, tryptophan and histidine amino acids (rows 5 and 6); both R3A, K4A, R5A and L104A, Y105A p12 mutants failed interact with PCNA and thus no growth was observed (rows 2 and 4) similar to the vector control (row 7). Interestingly, R3A, K4A, R5A mutant is also defective in p12 interaction in yeast cells but not the L104A, Y105A mutant (compare row 1 with 3).
These in vivo results suggest that while RKR-motif plays a dual role in dimerization and PCNA interaction, 98QCSLWHLY105 is only involved in PCNA binding.
Further, we wanted to examine whether dimerization of p12 is also required for other Pol subunit interactions. Yeast two hybrid analyses of co-transformants harboring AD-R3A, K4A, R5A with BD-p125 or BD-p50 demonstrated that mutation in this motif has no effect on Pol's subunit interaction as transformants grew efficiently on Leu-Trp-His-plate ( Fig. 3 D, sectors 8-12). Thus, dimerization of p12 is not required for p125 or p50 binding. However, it is not clear whether p125 and p50 bind to the same or different regions in p12. Analyses of various deletion constructs in p12 did not reveal any significant results to suggest any interaction domain (data not shown). Nonetheless, dimerization will facilitate binding of p125 and p50 subunits of Pol to separate monomer of p12 dimer.

RKR-motif of p12 is critical for dimerization
Multiple basic amino acid motifs such as RKR/KKR/KRK in other proteins are known to play crucial roles in a variety of cellular processes such as their retention and exit from ER, nuclear localization as well as the gating of K+ channels (29)(30)(31)(32). Such motifs are also found to be involved in protein-protein interaction (33)(34)(35). To ensure the involvement of RKR-motif in p12-p12 dimerization, mutant proteins R3A, K4A, R5A and L104A, Y105A were purified to near homogeneity and analyzed in SDS-PAGE ( Supplementary   Fig. 3 lanes 3 and 4). The mutant proteins were co-migrated with the wild type p12 close to the dye front and thus mutations in these residues had no obvious effect on protein mobility and stability. When the proteins were further resolved in non-denaturating PAGE, both wild type and L104A, Y105A proteins were co-migrated; and their migration was similar to CA as shown earlier ( Fig. 4 A, compare lanes 1 with 2 and 4). However, R3A, K4A, R5A mutant p12 was migrating much faster than the other two p12 proteins in native gel as a monomer (Fig. 4 B compare lane 3 with lanes 2 and 4). Therefore, we concluded that RKR motif is absolutely required for dimerization, and mutations in 98QCSLWHLY105 motif have no effect on such a function of p12.
To rule out the possibility that the faster migration of RKR mutant in non-denaturating gel is not due to because of change in residual charge in the protein rather due to abolition of dimerization, we compared the gel filtration elution profiles of wild type and R3A, K4A, R5A mutant p12 by separating equal amount of proteins through S200 molecular exclusion chromatography at physiological salt concentration. While the wild type p12 protein eluted in two peaks of volume at ~1.4 ml and ~2.2 ml corresponding to an oligomeric and monomeric status of the protein (red line), R3A, K4A, R5A mutant only eluted (grey line) later at a single peak volume of ~2.2 ml (Fig. 4 B). This also rules out any change in stokes radius of p12 proteins due to mutation in the dimerization domain as a portion of wild type co-elutes with mutant p12.
Co-immunoprecipitation experiment was also carried out in the cell lysates harboring FLAG-p12 and GFP-p12 or GFP-p12 R3A, K4A, R5A by using anti-FLAG antibody conjugated beads; and further probed with anti-GFP antibody ( Fig. 4 C). While anti-FLAG antibody could pull down wild type p12 as detected by anti-GFP antibody, it did not precipitate the RKR mutant (compare lane 3 and 5). Corroborating with our yeast two hybrid results, native PAGE, size exclusion chromatography and pull-down assays clearly demonstrate that indeed RKR-motif is a protein-protein interaction domain and is essential for p12 dimerization.

Formaldehyde cross-linking reveals dimerization of p12
Our native PAGE and ITC analyses suggested potential dimeric nature of p12 as it was migrating at similar position with CA (30 kDa). To estimate the exact number of p12 molecules in the oligomeric complex, we employed formaldehyde cross-linking assay. Reagents like formaldehyde or glutaraldehyde cross-links neighboring lysine or arginine residues of proteins to form a stable complex that can even withstand SDS (36,37); moreover, our mutational analysis deciphered the involvement of RKR motif in dimerization.
Therefore, purified recombinant proteins were cross-linked with formaldehyde, analyzed on 12 % SDS-PAGE and detected by Coomassie brilliant blue staining (Fig. 5 A). Upon treatment with the cross-linker, both wild type and L104A, Y105A mutant p12 proteins showed a concentration dependent cross-linked dimers (lanes 2, 3, 4, 10, 11 and 12) those were migrating below 32 kDa position and did not form any higher order oligomers, whereas R3A, K4A, R5A mutant protein remained as monomer (lanes 6, 7 and 8).
Without any cross linker, all the three proteins migrated at the bottom of the gel (lanes 5, 9 and 13). In addition to 3RKR5 motif, p12 also possess yet another similar multi-basic motif 15KKR17 in only mammalian p12 ( Fig. 3 A colored in blue). As R3A, K4A, R5A p12 mutant failed to form any oligomer in the presence of formaldehyde, we suggest that 15KKR17 sequences has no role in such a function and dimerization is specifically mediated by R3K4R5 motif.
Since, we did not detect dimerization of RKR-motif mutant in any of our assay, we wanted to rule out the possibility that this effect resulted from a significant change in p12 conformation. For this reason, we compared the CD spectra of the wild-type and R3A, K4A, R5A mutant p12 proteins ( Fig. 5 B). The CD spectra determined in the "far-UV" region (200 to 260 nM) showed p12 to be enriched in -structure as evident from the characteristic negative peaks at 208 nm and 222 nm. It also indicates that the mutant protein retains similar level of secondary structures as that of wild-type protein, which would suggest that the mutations do not cause a major perturbation of p12 structure. Even the ITC assays failed to detect any binding between wild type p12 and R3A, K4A, R5A p12 mutant ( Fig. 1 C ii and D ii). Considering all these evidences, we conclude that p12 forms a dimer solely mediated by R3K4R5 motif.

In vitro reconstitution of pentameric human Pol holoenzyme
Various subassemblies of human Polδ such as p125 alone, p125-p50, p125-p50-p68, and p125-p50-p68-p12 have been purified in vitro by mixing various combination of purified proteins (11,38). In this study, Pol holoenzyme was expressed by co-transforming two bacterial expression constructs GST-p125 and pCOLA234 (p50-p68-His-Flag-p12); and complex was purified to near homogeneity as described in the methods section. Taking advantage of strategically located PreScision protease site, cleaved Pol complex was obtained in which only p12 subunit was amino-terminally FLAG tagged. To conclusively show two different forms of p12 in the holoenzyme, untagged p12 protein purified from bacterial GST-p12 system was added and the mixture was further incubated at 4 C for 4 hrs. If p12 forms a dimer and exist in the Pol complex, untagged p12 will compete out some of the Flag p12 and a Pol with five subunits will be appeared. Thus, the mixture was loaded into S200 minicolumn for separation and fractions were collected in a 96 well plate. Since we could not detect enough protein by coomassie staining despite our repeated trails, various fractions were analyzed in SDS-PAGE followed by detection of various subunits by probing with specific antibody (Fig. 6). To detect the presence of enriched holoenzyme fractions, the membrane was first probed with anti-p68 antibody (A) and further selected fractions were again separated in SDS-PAGE (B).
As analyses, we conclude that intrinsically human Pol is a pentameric complex possessing two p12 subunits.

Dimerization of p12 is essential for PCNA interaction
Since, both R3K4R5 and 98QCSLWHLY105 motifs are involved in PCNA interaction and the former motif additionally involved in protein dimerization; to ascertain any regulatory role of this motif in p12 function, a PCNA overlay experiment was carried out. Proteins were resolved on native PAGE to keep the natural folding and dimer structure intact as in Fig. 4A, transferred to PVDF membrane and blocking was performed in the presence of PCNA. After several washings, the blot was developed with anti-PCNA antibody ( Fig. 7 A). As depicted in the figure, although p12 and its L104A, Y105A mutant formed dimers (upper panel, compare lane 1 and 3), signal for only wild type was detected (lower panel) suggesting that 98QCSLWHLY105 motif is a true PIP motif required for PCNA interaction. Despite retaining PIP motif in the extreme c-terminal tail, due to its inability to form dimer, R3A, K4A, R5A mutant failed to bind to PCNA. Similarly, a pull down experiment was also carried out by taking a mixture of stoichiometric equivalents of purified GST-p12 or various p12 mutants and PCNA in Tris-buffer containing 150 mM NaCl salt concentration (Fig. 7 B). The mixture was incubated with GST-beads at 4 C for 3 hrs in rocking condition, further beads were washed thrice and bound PCNA was eluted by SDS containing sample buffer.
The eluted PCNA was detected by anti-PCNA antibody only when it was mixed with wild type p12 but not with L104A, Y105A or R3A, K4A, R5A mutant (compare lane 3 with 6 and 9).
To determine the binding affinity of p12 with PCNA, ITC assay was carried out by placing p12 (10 M) in the sample cell and PCNA (120 M) was injected. The H was ~ -80 kcal/mol, G was ~ -6.80 kcal/mol, and the KD for the complex was 10 M (Fig. 7 C). In similar assay condition, no significant heat exchange was observed while RKR-or PIP-motif p12 mutants was kept in the cell and PCNA in the syringe, suggesting no interaction between the proteins. Thus, our results suggest that dimerization at R3K4R5 motif promotes p12 interaction with PCNA via 98QCSLWHLY105 motif.

Inter-domain connecting loop region of hPCNA mediates its interaction with p12
In view of the fact that our predicted model structure showed p12 PIP peptide binding to IDCL domain of hPCNA; for confirmation, yeast two hybrid assay was carried out with p12 fused to Gal4 binding domain and two PCNA mutants namely, pcna-79 and pcna-90 fused to Gal4 activation domain. In pcna-79, two key hydrophobic residues L126 and I 128 of inter-domain connecting loop were mutated to alanines; whereas pcna-90 possesses two mutation P253,K254-AA in the extreme C-terminal tail of PCNA. Most of the interacting proteins bind to any of this two region of a trimeric PCNA ring (37). While wild type and pcna-90 were able to interact with p12 as evident from growth on SDA plate lacking leucine, uracil and histidine ( Fig. 8 A, sectors 1 and 3); pcna-79 did not support the survival as it failed to form intact Gal4 by interacting with p12 (sector 2). p12 PIP mutant was used as a -ve control (sector 4). To strengthen our finding, GST pull down assay was carried out. Equal stochiometry of wild type and IDCL mutant of PCNA proteins were incubated with GST-p12, and pull-down assays were performed on glutathione-Sepharose affinity beads as described previously (39). In accordance with the data shown in Fig. 8

Cdm1, a p12 homologue of Schizosaccharomyces pombe also forms dimer
The other fourth subunit of DNA polymerase delta that has been well characterized is Cdm1, a p12 orthologue from S. pombe (10). Cdm1 consists of 160 amino acids, MW of 18.6 kDa and pI of 7.73. Like p12, it shows abnormal mobility in the SDS PAGE and migrates as ~22 kDa protein ( Fig. 9 A). As it also has conserved RKR-motif (K2K3R4in Cdm1) at its N-terminal end, we wanted to examine whether homodimerization property of the smallest subunits of Pol is evolutionarily conserved. Wild type and K2A, K3A, R4A mutant Cdm1 proteins were purified to near homogeneity from bacterial over-expression systems and analyzed on native PAGE. Just like p12, the wild type Cdm1 migrated slower than its mutant.
The slower migration of Cdm1 complex comparison to p12 dimer could be attributed to the differences in their pI and MW (pI 7.3 vs 6.3; MW 12 vs 18.6). The dimeric p12 incidentally migrates at about the same position with monomeric Cdm1 mutant protein (Fig. 9 B). Thus, we concluded that homodimerization is the intrinsic property of the 4 th subunits of Pol and is mediated by a stretches of conserved basic amino acids located at the extreme amino terminal end (RKR/KKR-motif).

Discussion:
DNA polymerase  is a high fidelity essential DNA polymerase not only plays a central role in DNA replication, it also participates in DNA recombination and several DNA repair pathways from yeasts to human (40,41). Several mutations in the mouse and human Pol subunits have been mapped to cause various cancers and genome instability in yeasts (42)(43)(44). Thus, it is important to understand function of each subunits and their precise role in processivity and fidelity of the holoenzyme. In this report, we have re-investigated role of the smallest subunit of hPol, p12 in holoenzyme architecture and PCNA interaction.
As reported earlier and also in this study p12 subunits interact with p125 and p50, whereas p50 makes a connecting bridge between catalytic p125 and accessory p68 subunits (45). Thus, hPol is proposed to be a heterotetrameric holoenzyme. Our yeast two hybrid assay, cellular co-localization in replication foci, mutational analyses and several physio-biochemical assays including formaldehyde crosslinking clearly demonstrate that p12 exists as a homodimer both in vivo and in vitro; and dimerization is dependent on amino terminal tripeptide basic amino acids sequence, 3RKR5-motif. It also argues against the tetrameric nature hPol. As it has been previously shown, various sub-assemblies of hPol such as tri-and tetra-meric holoenzymes could exist in the cell based on Pol's function in either replication or repair (46), we propose that in addition to these sub-complexes, pentameric Pol is also formed, which could be the native form of Pol. Pull-down of cellular Pol by a tagged p12 and in vitro reconstitution of Pol5 definitely authenticates our prediction (Fig. 2 A and Fig. 7). Purification or in vitro reconstitution of Pol holoenzymes by several groups also indicated higher stochiometry of p12 in comparison to other subunits in Pol (19, 25,38,[45][46][47].
We also found that the dimerization of the 4 th subunit of Pol is not restricted to human as Cdm1 of SpPol also forms a dimer that is again dependent on KKR motif. As the RKR/KKR motif has been retained in other p12 homologues as well, it appears that such a property of the smallest subunit of Pol is evolutionarily conserved. Interestingly, the small accessory subunit of yet another B-family polymerase Pol (a complex of Rev3 and Rev7) Rev7 is found to function as a dimer (48 (p125+p50+p68+p12), Pol3 (p125+p50+p68), Pol2 (p125+p50) and p125 alone (Fig. 8 A) (45). Accordingly, the addition of p68/Pol32 to core (p125+p50 or Pol3+Pol31) results in high processivity, thus its binding to PCNA appears to be critical. Considering all these, we propose a model for the network of protein-protein interactions of the Pol-PCNA complex (Fig. 10). In a pentameric state of hPol, along with p68, any other two subunits among p125/p50/p12 will bind to PCNA in any combinations as shown in the figure 10 B (i to iv). Upon p68 degradation or its phosphorylation, p125, p50 and one monomer of p12 dimer can bind to PCNA (v). Similarly, upon p12 proteosomal degradation as a response to DNA damage; other three subunits will make contacts with PCNA (vi); whereas due to cleavage of both p68 and p12 in certain situation, p125 and p50 bind to PCNA (vii), although the Pol core functions with low processivity. Thus, this study warrants extensive mutational analyses as had been carried out in yeast Pol holoenzymes rather than analyzing sub-complexes to decipher precise role of these PIPs in cellular function and processive DNA synthesis by hPol.
In conclusion, here we show that RKR-mediated dimerization plays a vital role in p12 binding to PCNA and Pol5 architecture, and the phenomenon appears to be conserved throughout evolution.
Dimerization of p12 could play an instrumental role in re-orchestrating pentameric Pol-PCNA complex at the replication fork and thereby it regulates Pol's function.

Experimental Procedures:
Plasmids, oligonucleotides, antibodies and enzymes: Human DNA polymerase delta constructs pCOLA-hPold234 and pET32-p125 (kind gift from Prof. Y. Matsumoto) were used as precursor plasmids for subsequent manipulation (57). The oligonucleotides (Supp. Other hPol subunits such as p125, p50 and p68 were also PCR amplified by the primer sets NAP252-NAP248, NAP254-NAP255, and NAP258-NAP257, respectively; digested with EcoRI-BamHI and cloned into same sites of pGBT9. Wild type and K2A, K3A, R4Amutant of Cdm1 were PCR amplified from S. pombe genomic DNA using primers NAP361-NAP451 and NAP361-NAP452, respectively; and the BamHI digested products were cloned into BglII site in pNA716 for bacterial expression. Similarly, GST-hp125 expression plasmid was generated by cloning a BamHI digested PCR product amplified from pET32p125 as a template using NAP247 and NAP248 primers into BglII site of pNA716. All of these constructs were authenticated by DNA sequencing. Protein Purifications: All GST tagged proteins were expressed in Escherichia coli BL21 DE3 and purified by affinity chromatography using glutathione sepharose beads (GE-healthcare). The proteins were expressed as amino terminal GST-fusion proteins under T7 promoter. Briefly, 5 ml pre-culture of the transformant was added to 500 ml LB + 50 μg/ml ampicillin and grown at 37 °C till the OD600 reaches to 0.6. Next the culture was induced with 1 mM IPTG and allowed to grow for another 8 hrs. Cells were harvested and about 3 gm of frozen cells were resuspended in 1X cell breaking buffer (50 mM Tris-HCl pH 7.5, 10 % sucrose, 1 mM EDTA, 500 mM NaCl, 0.5 mM PMSF, 0.5 mM Benzamidine hydrochloride, 10 mM-mercaptoethanol and protease inhibitor cocktail). Cells were lysed with a high pressure homogenizer at 10K psi (STANSTED). The lysate was cleared after centrifugation at 10K rpm for 10 min.
Further the supernatant was centrifuged at 30K rpm for 1 hour in P70AT rotor (Hitachi). Rest of the steps used for purification was same as described before (39). All the proteins were stored in the buffer containing final concentration of 50 mM Tris-HCl, pH 7.5, 150 mM NaCl, 10% glycerol, 5 mM dithiothreitol (DTT) and 0.01% NP-40. The purity of the protein was confirmed after resolving on 12 % SDS-PAGE and stained by Coomassie Blue.
However, for human Pol purification, bacterial strain BLR (DE3) was co-transformed with pLacRARE2 plasmid from Rosetta2 strain (Novagen), GST-p125 and pCOLA-hPold234; and colonies were selected on LB agar plate containing ampicillin (50g/ml), kanamycin (30g/ml) and chloramphenicol (35g/ml). About 60 ml overnight grown pre-culture was inoculated into 6 L LB with mentioned antibiotics and grown at 37 °C to an OD600 of 0.6;followed by induction with 1 mM IPTG and further growth was continued for 15 hrs at 16 °C. The cells were harvested and stored at -80°C until use.
The cell breaking condition and other purification steps were followed as mentioned. Taking advantage of strategically located PreScision protease site, cleaved Pol4 (p125-p50-p68-p12) was obtained in which only p12 subunit was amino-terminally FLAG tagged. Similarly, p12 protein was also purified by using bacterial GST-p12 construct. followed by rinsing with DPBS, pH7.4, and then slide were prepared using antifade as mounting agent.

Size exclusion Chromatography
Images were taken using Leica TCS SP5 at 63 X objective. In silico analysis of p12 structures: p12 RKR (1-MGRKRLITDSYPVK-14) and PIP (92-GDPRFQCSLWHLYPL-106) domains were used for peptide structure prediction by using PEP-FOLD3 server (http://bioserv.rpbs.univ-paris-diderot.fr/services/PEP-FOLD3/). The models generated were validated by SAVES and Ramachandran plot showed all residues in allowed regions, validating both the models. Further, the generated structural models were aligned with PIP peptide sequences from p21 (1AXC) and p68 PIP (1U76).  Circles and lines denote the raw measurements and the fitting to a one set of identical sites.