The C-terminal Domain of the DNA Polymerase Catalytic Subunit Regulates the Primase and Polymerase Activities of the Human DNA Polymerase α-Primase Complex*

Background: DNA polymerase α-primase synthesizes chimeric RNA/DNA primers for replicative polymerases. Results: We defined elements that modulate polα and prim activities. Conclusion: The C-terminal domain of the catalytic subunit of polymerase α and the B-subunit regulate the priming of DNA replication. Significance: We provide new information on the regulation of RNA/DNA synthesizing complex that is indispensable for replication in eukaryotes. The initiation of DNA synthesis during replication of the human genome is accomplished primarily by the DNA polymerase α-primase complex, which makes the RNA-DNA primers accessible to processive DNA pols. The structural information needed to understand the mechanism of regulation of this complex biochemical reaction is incomplete. The presence of two enzymes in one complex poses the question of how these two enzymes cooperate during priming of DNA synthesis. Yeast two-hybrid and direct pulldown assays revealed that the N-terminal domain of the large subunit of primase (p58N) directly interacts with the C-terminal domain of the catalytic subunit of polα (p180C). We found that a complex of the C-terminal domain of the catalytic subunit of polα with the second subunit (p180C-p70) stimulated primase activity, whereas the whole catalytically active heterodimer of polα (p180ΔN-p70) inhibited RNA synthesis by primase. Conversely, the polα catalytic domain without the C-terminal part (p180ΔN-core) possessed a much higher propensity to extend the RNA primer than the two-subunit polα (p180ΔN-p70), suggesting that p180C and/or p70 are involved in the negative regulation of DNA pol activity. We conclude that the interaction between p180C, p70, and p58 regulates the proper primase and polymerase function. The composition of the template DNA is another important factor determining the activity of the complex. We have found that polα activity strongly depends on the sequence of the template and that homopyrimidine runs create a strong barrier for DNA synthesis by polα.

The initiation of DNA synthesis during replication of the human genome is accomplished primarily by the DNA polymerase ␣-primase complex, which makes the RNA-DNA primers accessible to processive DNA pols. The structural information needed to understand the mechanism of regulation of this complex biochemical reaction is incomplete. The presence of two enzymes in one complex poses the question of how these two enzymes cooperate during priming of DNA synthesis. Yeast two-hybrid and direct pulldown assays revealed that the N-terminal domain of the large subunit of primase (p58N) directly interacts with the C-terminal domain of the catalytic subunit of pol␣ (p180C). We found that a complex of the C-terminal domain of the catalytic subunit of pol␣ with the second subunit (p180C-p70) stimulated primase activity, whereas the whole catalytically active heterodimer of pol␣ (p180⌬N-p70) inhibited RNA synthesis by primase. Conversely, the pol␣ catalytic domain without the C-terminal part (p180⌬N-core) possessed a much higher propensity to extend the RNA primer than the twosubunit pol␣ (p180⌬N-p70), suggesting that p180C and/or p70 are involved in the negative regulation of DNA pol activity. We conclude that the interaction between p180C, p70, and p58 regulates the proper primase and polymerase function. The composition of the template DNA is another important factor determining the activity of the complex. We have found that pol␣ activity strongly depends on the sequence of the template and that homopyrimidine runs create a strong barrier for DNA synthesis by pol␣.
DNA replication is fundamental to all living organisms. The process of DNA replication of genomic DNA in eukaryotes typ-ically depends on the RNA primers that provide a free 3Ј-hydroxyl group for the replicative DNA polymerases to bind and start elongation. There is one deviation from this established paradigm. Humans and other mammals possess a PrimPol enzyme, which can prime with either ribo-or deoxynucleotides and also serve as a translesion DNA synthesis polymerase (1). The exact role of the PrimPol is not clear, but it is limited to special situations because the PrimPol knock-out mice are viable.
The major task of accurate nuclear genome replication in eukaryotes is carried out by B-family DNA polymerases (pols) 5 (2)(3)(4)(5)(6). pol⑀ is indispensable for the assembly of the replisomes at origins (7), pol␣-primase (pol␣-prim) generates primers for the synthesis (8), pol␦ (9) and pol⑀ (10) participate in elongation and bulk DNA synthesis (11)(12)(13), and pol is a key player in the synthesis on templates that are difficult to replicate (14,15). The critical essential step of the replication is the initial primer synthesis by pol␣-prim, because pol␦, pol⑀, or pol cannot start DNA synthesis without a primer with a free 3Ј-OH end.
pol␣-prim is the only B-family polymerase complex that possesses RNA and DNA polymerase activities. pol␣ lacks proofreading exonuclease activity, and the synthetic activity of the complex is limited to the synthesis of RNA/DNA hybrids that serve as primers for the synthesis of genomic DNA. During the initiation of DNA synthesis, primase first synthesizes a short RNA primer in a two-phase reaction, the rate-limiting synthesis of dinucleotide, and the subsequent, faster, further elongation to reach around 10 ribonucleotides (16). Then, pol␣ takes over and extends the primer by ϳ20 nucleotides of DNA. The ability to determine the size of RNA and DNA parts (called "counting") is thought to be an intrinsic property of pol␣ and primase (17,18). This primer is elongated by pol␦/pol⑀ for synthesis of the bulk of chromosomal DNA. The nature of the replication on the lagging strand requires multiple initiations by the pol␣prim, which leads to the formation of 165-bp Okazaki fragments that correspond to unit size of nucleosomal repeats (19).
All B-family polymerases are four-subunit proteins (the latest addition to the group was pol, which shares two accessory subunits with pol␦) (20 -24) where the largest catalytic subunit has a unique regulatory C-terminal domain binding zinc and iron ions (20,(25)(26)(27). These C-terminal domains share sequence similarities and are evolutionarily conserved from yeast to humans (20,28). Human pol␣ is composed of a 180-kDa catalytic subunit (p180) and a 70-kDa B subunit (p70) (8) (Fig. 1A). The primase part is composed of a 49-kDa catalytic subunit (p49) and a 58-kDa accessory subunit (p58) (8,29) (Fig.  1A). The two metal binding sites located on the C-terminal domain of p180 (p180C) are occupied by zinc atoms in crystal (26). When p180C was partially purified from bacteria under anaerobic conditions, it contained a detectable level of ironsulfur clusters (27), which may be a peculiarity of the heterologous expression in bacteria (20). The p58 subunit possesses an evolutionarily conserved, C-terminal four-cysteine motif that coordinates a [4Fe-4S] cluster, which is necessary for the formation and stabilization of the initial di-ribonucleotide of the RNA primer (30 -33).
Mouse and yeast homologs of the pol␣ p180C domain interact with the B subunit and primase, working as a scaffold to tether the catalytic part of p180, B subunit, and primase (34,35). The large interaction interface (4500 Å 2 ) between the C terminus of yeast pol1 and the B-subunit suggests tight contact between two polypeptides (26). The N-terminal helical domain of the B subunit of pol␣ interacts with SV40 viral helicase, which is required for the activation of the SV40 primosome (36). This structural information is consistent with the idea that p180C⅐p70 might tether the pol␣-prim to the other components of replication machinery (34,37). However, the role of p180C⅐p70 in the pol␣-primase function is not well defined.
To get a better understanding of the mechanism of RNA-DNA synthesis required for the beginning of DNA synthesis during replication, we investigated the interactions between the four subunits of the human pol␣-prim complex and the functional implications of these contacts. We confirmed that the large subunit of primase, p58, interacts with the p180C part of the catalytic subunit of pol␣ via the N-terminal domain. The primase and polymerase assays demonstrated that the p180C⅐p70 complex regulates both primase and polymerase activities. These activities are also dependent on the ratio of riboversus deoxynucleoside triphosphates and the specifics of DNA templates. DNA synthesis by pol␣ proceeds to the end of the standard templates with a scrambled sequence but halts at ϳ20 nucleotides of the homopyrimidine (poly-dT) template. This observation limits the applicability of the recent model of primer length counting by yeast pol␣ (38).
Our findings suggest that the proper replication priming is a result of balanced and sequential activities of the components of pol␣-prim complex and the availability of the proper nucle-otides. During the priming, the size of the RNA-DNA hybrid is regulated by the non-catalytic domain, p180C, and the subunit, p70, and depends on the competition between primase, pol␣, and riboversus deoxynucleotides as well as the sequence context of the template.
Cloning of Human DNA pol␣-Primase Genes and the Production of Recombinant Proteins for Pulldown Analysis-The cDNAs for the p49 (420 amino acids (aa)) and p58 (509 aa) subunits of human primase were obtained from Open Biosystems (clone IDs 3686937 and 6148494, respectively). The cDNA for the human B subunit (598 aa) was obtained from Open Biosystems (clone ID 2822514). The pcDNA3/POLA1 encoding for p180 (1462 aa) is a generous gift from Dr. Motoshi Suzuki (Nagoya University Graduate School of Medicine). To demonstrate the direct interaction between p58 and p180C, the coding sequences (cs) for p58-His 6 (cs-p58-His 6 ) and cs-SUMO-p180C were cloned together into the MCS1 and MCS2 of pETDuet-1 (vector named as pE1; Table 1), respectively, using a two-step insertion method (39). A single insertion of cs-p58-His 6 (pE2) into pETDuet-1 served as a control. Nontagged cs-p49 and cs-p58 were cloned into the MCS1 and MCS2 of pETDuet-1(pE3), respectively. The cs-p58N (encoding amino acids 1-265) or cs-p58C (aa 266 -509) were cloned into the MCS2 of the pETDuet-1 that contains cs-p49 at MCS1. Cloning of the full-length, N-terminally His-tagged human cs-p70 (p70-His 6 ) and SUMO-tagged human cs-p180C (coding for aa 1265-1462; SUMO-p180C) into the pColaDuet-1 vector was described previously (20). To obtain a pColaDuet-1 vector producing only SUMO-p180C (for negative controls), a prema- cs-SUMOp180C pColaDuet-1 ture stop codon was introduced into the cs-p70-His 6 .The pET-Duet-1 and pColaDuet-1 constructs were introduced into the Rosetta-2 (DE3) cells by transformation and selected on the LB-Amp ϩ Kan ϩ medium. Overexpression of the genes of the constructs was done as described previously (20). Methods-All experiments were repeated at least two times, but most were completed three or more times. The results are illustrated by typical gels.
Yeast Two-hybrid Assay-Coding sequences of human p49, p58, p70, and p180C (encoding for aa 1,258 -1,462) (Fig. 1A) were amplified by PCR from vectors described in the previous section and ligated into the pGADT7 and pGBKT7 vectors (Clontech, Mountain View, CA) containing TRP1 and LEU2 selection markers, respectively. The yeast strain AH109 with three reporter genes, HIS3, ADE2, and LacZ, was transformed by combinations of these plasmids. The co-transformations of cs-p49/cs-p58 and cs-p180C/cs-p70 as well as the murine cs-p53/cs-SV40 large T-antigen served as positive controls. The co-transformations of each construct with an empty vector served as negative controls. The cells were allowed to grow three days on Synthetic Complete drop-out plates lacking Leu and Trp (SCϪLeuϪTrp). Positive colonies were replica-plated on SCϪLeuϪTrpϪHis, SCϪLeuϪTrp-ade, and SC-Leu-Trp-His-Ade plates. Colonies on each plate were counted after 3 days of growth at 30°C. Activation of both HIS3 and ADE2 reporters was recorded as an interaction if Ͼ80% of the colonies with prey and bait genes were His ϩ and Ade ϩ (the single HIS3 reporter produces many "false positives"; therefore, only the concomitant Ade ϩ phenotype suggests that proteins under study really interact).
Nickel-iminodiacetic acid (Ni-IDA) Pulldown Assay-After induction of gene expression in bacterial cultures, cells were harvested and kept in aliquots at Ϫ80°C. Cells were disrupted in ice-chilled water by sonication in lysis buffer containing 20 mM Tris 7.8, 0.15 M NaCl, 10 mM KH 2 PO 4 , 3% glycerol, 3 mM ␤-mercaptoethanol, 0.5 mM PMSF, and 1 g/ml leupeptin. After centrifugation, 0.2 ml of cleared lysate (corresponding to a 2.5-ml culture volume) was incubated for 1 h with rocking at 4°C with 10 l of Ni-IDA resin. The resin was washed 1 time with 0.2 ml of lysis buffer and 2 times with 0.2 ml of the lysis buffer containing 0.5 M NaCl. The bound proteins were eluted by 30 l of 0.3 M imidazole in the lysis buffer (pH 7.7). Proteins of bacterial lysates and eluted samples from Ni-IDA resin were separated by 12% SDS-PAGE and detected by Coomassie Blue (R250) staining or Western blotting using the Mini-PROTEAN Tetra cell system (Bio-Rad). Anti-His monoclonal antibody and anti-SUMO monoclonal antibodies (6G2A9 and 4G11E9, respectively; GenScript, Piscataway, NJ) were used to detect His 6 -tagged human p58 and p70 and SUMO-tagged p180C, respectively.
Purification of Human Primase⅐pol␣ Subunits and Their Complexes-Expression and purification of full-length human primase (p49⅐p58) was done as in Baranovskiy et al. (39). Expression and purification of p70⅐p180C was done as in Baranovskiy et al. (20). For expression of the p180 core (aa 335-1257, p180⌬N core) and p70⅐p180⌬N heterodimer (aa 1-598 and 291-1462, respectively) in insect cells, we cloned them into the pFastBac-1 vector (Invitrogen). In the case of the p180 core, we added cs for the N-terminal His 6 tag followed by the tobacco etch virus protease recognition site (ENLYFQ); in the case of p70 we added cs for the N-terminal His 6 -tag. Obtaining high titer baculoviruses and protein expression in insect cells was done according to the instructions of the manufacturers. The pol␣ catalytic core (aa 335-1257) with a cleavable N-terminal His 6 tag was purified to near homogeneity in four steps, including chromatography on the Ni-IDA column (Bio-Rad), overnight dialysis combined with digestion by tobacco etch virus protease, pass through the Ni-IDA column, and chromatography on Heparin HP HiTrap column (GE Healthcare).
The p70⅐p180⌬N was purified to near homogeneity in three chromatographic steps, including Ni-IDA (Bio-Rad), heparin HP HiTrap (GE Healthcare), and hydroxyapatite (Bio-Rad) columns. To obtain the tetrameric prim-pol␣ complex (p49⅐p58⅐p70⅐p180⌬N), the pure primase (109 kDa) and pol␣ (200 kDa) samples were mixed at a molar ratio of 1.2:1, incubated for 1 h at 4°C, and loaded to Superose 12 column (24 ml; GE Healthcare). For all samples used in the activity assay, the peak fractions from the last column were combined, concentrated to 2 mg/ml, and diluted 5-fold in the buffer containing 30 mM HEPES-KOH (pH 7.9), 50 mM NaCl, 1 mM DTT, and 1% glycerol and flash-frozen in aliquots. The results of analysis of the purity of our proteins are presented in Fig. 1B.
Primer Extension Assay-Activities of the p180⌬N-core, p70⅐p180⌬N, or p70⅐p180⌬N⅐p49⅐p58 were compared in reactions (20 l) that contained the 0.75 M poly-dT 70 template with 0.50 M 5Ј-TYE665 fluorophore-labeled poly-dA 15 DNA or poly-rA 15 RNA oligos or 73a (73b) templates pre-annealed with hetero-DNA or RNA primers. The annealing of the 73a or the 73b template with RNA or DNA primers was done by a decrease of temperature from 80 to 25°C in 0.5°C/min gradient in the thermo cycler (50 l volume, 50 mM NaCl). The activity of primase p49⅐p58 heterodimer was analyzed using close to an equal 1:1.5 ratio of enzyme to template. The activities of primase p49⅐p58⅐p70p⅐180⌬N heterotetramer was analyzed using a low 1:15 ratio of enzyme to primer/template. The low ratio of primase to template allowed us to approach single-hit conditions, whereas at the equal ratio the primase was able to synthesize longer products due to multiple binding events with substrate. This allowed us better access the inhibitory effect of dATP. Reactions were assembled on ice in the following order. The primer/template was added to the buffer containing 20 mM Tris-HCl, pH 8.0, 50 mM NaCl, 10 mM MgCl 2 , 0.2 mg/ml BSA, and 2 mM DTT followed by the addition of dATP (200 M or otherwise indicated under "Results") or 200 M dNTPs for the poly-dT70 and 73a (or 73b) template, respectively. Then, the p180⌬N-core, p70⅐p180⌬N, or p70⅐p180⌬N⅐p49⅐p58 was added into the reaction to concentrations indicated in the figure legends. Reactions were held for the indicated time points at 37°C in the thermo cycler. T4 DNA polymerase (Promega Corporation; stock concentration 25 nM) was used as a control for the primer extensions on hetero-DNA and RNA. Reaction products were mixed with formamide loading buffer (95% formamide, 0.025% Orange G, 5 mM EDTA, and 0.025% SDS) and heated at 70°C for 5 min. The reaction products were separated as described previously. Visualization of the products used the Typhoon 9410 imager (emission of fluorescence at 645 nm).

RESULTS
The Role of pol␣ in Primase Function-To study the functional relevance of the subunit contacts in pol␣-prim, we purified several variants of recombinant pol␣ with a combination of different parts of the primase functional domains (Fig. 1A): B-subunit (p70) with the catalytically proficient catalytic DNA pol subunit, p180⌬N, or C-terminal of p180 (p180C). Similar to yeast pol␣ (17), where the N terminus interacts with Ctf4 for the integration of pol␣ into the replication fork (40), the N terminus of p180 is poorly folded and has no conserved motifs necessary for DNA polymerizing activity, so it has been deleted to increase protein solubility and simplify its purification. All protein samples used for activity analysis were purified to the high level of homogeneity (Fig. 1B). We analyzed de novo RNA synthesis and the extension of these primers on the poly-dT template with radioactive [␣-32 P]ATP by the p49⅐p58 heterodimer of primase alone or in the complex with heterodimer of p70⅐p180C or p70⅐p180⌬N. The heterodimeric primase synthesized RNA primers of an average of 8 -10 nucleotides unitlength and multiples of this unit. They were extended to around 20 nucleotides and, more frequently, to around 30 nucleotides, and less efficiently, they were extended further to 50 and 60 nucleotides (Fig. 2A, lane 2). The addition of dATP dramatically inhibited the activity of primase (Fig. 2B, lanes 3-5). In the presence of p70⅐p180C the majority of products became longer, around 70 nucleotides in length ( Fig. 2A, lane 3, and titration of p70⅐p180C concentration in Fig. 2B, lanes 7-9). The explanation for the predominance of the long products could be the elevated activity/processivity of the primase and non-random initiation/extension by primase in the presence of p70⅐p180C biased toward the 3Ј-end of the template. The presence of catalytically active heterodimeric pol␣, conversely, inhibited the extension of unit-length primers when there was no dATP in the reaction (Fig. 2A, lane 4, and titration of p70⅐p180⌬N in Fig.  2B, lanes 10 -12). This observation is consistent with earlier studies (41). The addition of dATP with p70⅐p180⌬N, as expected, led to a rapid synthesis of the full-length products ( Fig. 2A, lane 5). The quantification of the product distribution and band intensity in the case of primase alone plus p70⅐p180C or p70⅐p180⌬N with or without dATP (Fig. 2C) provides support to the visual analysis of the gel shown in Fig. 2A.
The Inhibitory Effect of dATP on Primase Activity-To further study how the primase activity is inhibited by dATP (Fig.  2B, lanes 3-5), we analyzed the ability of the heterodimeric primase to extend ribonucleotide primers in the presence of riboversus deoxynucleotides or a mixture of both. We have used the high ratio of enzyme to template ("Experimental Procedures") to push the primase to its limits to detect the incorporation of dNTPs. Primase robustly extended the poly-rA 15 primers on the poly-dT 70 template with rATP. The extent of primer utilization and the length of products increased with longer reaction times, indicating the multiple cycles of binding and synthesis (Fig. 3, lanes 2-4). dATP was a poor substrate for the primase reaction; it was incorporated in the ϩ1 position quite efficiently, but further incorporations were not favored, with no products longer than ϩ4 (Fig. 3, lanes 5-7). The primase extended about half of the RNA primers with dAMP in 30 s, whereas the addition of a second dAMP was almost 10 times slower. The primer extension was also limited when both rATP and dATP were present (Fig. 3, lanes 8 -10). The primase incorporated both dATP and rATP with comparable efficiency (Fig. 3, lanes 8 -10, note that oligonucleotides containing only RNAs migrate slower than RNA primers extended with dNTPs). These results indicate that primase does not greatly discriminate against dATP, but the products with incorporated dAMP are poor substrates for the following extension.
The C-terminal Domain (CTD) of the Catalytic Subunit of Human pol␣ Interacts with the N-terminal Part of the Large Subunit of Primase-Mammalian pol␣ interacts with the large subunit of primase (29). The structure of artificially made, chimeric primase-CTD of pol␣ (amino acids 1445-1462) protein suggested that the CTD of pol␣ interacts with the N terminus of the large subunit of primase (38). To examine the intersubunit interactions between primase and pol␣ in vivo we have used a yeast two-hybrid assay ("Experimental Procedures"). As expected, a strong interaction was detected between the two primase subunits as well as the C-terminal domain of pol␣ (p180C) with the second subunit of pol␣ (p70). No interaction was detected between p180C and p49. The presence of two plasmids containing cs-p58 fused to activation domains and cs-p180C (encoding for aa 1258 -1462) fused to binding domains led to a positive signal for both HIS3 and ADE2 reporters. However, there was no activation of reporters when p58 was fused to the binding domain and p180C was fused to the activation domain. We concluded that the large subunit of human primase interacts with the CTD domain of pol␣.
To get more precise information about the interaction, we explored the binding of the same proteins produced in bacteria in Rosetta-2 (DE3) cells (Fig. 4). Co-expression of the genes encoding the SUMO-tagged p180C and His 6 -tagged p58 allowed us to obtain a stable binary complex by affinity purification on the Ni-IDA resin (Fig. 4A, lane 5). The stoichiometry of the binding between p58-His 6 and SUMO-p180C was similar to the positive control (Fig. 4A, lane 4). The identities of p58-His 6 and p70-His 6 purified by Ni-IDA resin as well as the SUMO-p180C were verified by the Western blot analysis (Fig.  4B).
In the crystal structure, p58N is tightly bound to the p49 subunit of primase (38), although it lacks the C-terminal domain with the [Fe-S] cluster (amino acids 287-509). Weiner and co-workers (31,32) noted that the purification of the p58N alone is problematic because it is largely insoluble in the absence of p49. In our experiments, only a 24-kDa proteolytic product was seen (Fig. 4, A, lane 1, and B, lane 4) in the pulldown of p58-His 6 alone, which likely represents the C-terminal part of the p58 subunit. To investigate the role of the primase domain containing the [Fe-S] cluster in interaction with pol␣, plasmids encoding for two subunits of primase (p49⅐p58) and two subunits of pol␣ (SUMO-p180C⅐p70-His 6 ) were constructed, and the four-subunit complex was produced. An N-terminal SUMO tag was added to the p180C to address the problem of poor solubility of the protein in Escherichia coli. The levels of primase are typically higher than pol␣ when produced in E. coli cells. The His 6 tag, therefore, was placed in the N-terminal of the p70 subunit to obtain a quaternary complex with more even stoichiometry after pulldown. Similar to the p49⅐p58⅐SUMO-p180C⅐p70-His 6 (positive control ; Fig. 4C, lane  4), the p49⅐p58N and SUMO-p180C⅐p70-His 6 are represented by stable quaternary complexes after affinity purification on the Ni-IDA resin (Fig. 4C, lane 5). After the production of p49⅐p58C and SUMO-p180C⅐p70-His 6, only a binary complex SUMO-p180C⅐p70-His 6 was present on the gel after pulldown (Fig. 4C,  lane 6). We concluded that the pol␣ interaction with primase is through the p180C-p58N. Neither p49 nor the part containing the [Fe-S] domain of p58 plays a role in the interaction. Our data are consistent with the recent finding that the last 18 residues of the C-terminal region of the yeast pol␣ catalytic subunit are important for binding to the N terminus of the primase large subunit (34,38).
DNA Sequence Dependence of the Extent of Synthesis by pol␣-To examine the parameters of the primer extension on different templates, we utilized a version of an enzyme with a core polymerase domain of human pol␣ (p180⌬N-core; aa 335-1257). This fully catalytically proficient protein lacks both the N-terminal part of p180 and the C-terminal domain of p180 with two metal binding sites necessary for the interaction with p70 (Fig. 1A). In the same time of reaction on the poly-dT 70 template, the efficiency of the extension of the 15-mer DNA primer by p180⌬N-core is much lower in comparison to the 15-mer RNA primer ( Fig. 5A; note the different electrophoretic mobility of DNA versus RNA). The efficiency of this reaction with both types of primers is increased when the concentration of dATP goes up (Fig. 5B). However, most products remain short, about 20 nucleotides of newly synthesized DNA, despite overall stimulation. It appears that the addition of 20 nucleotides of DNA attached to 15-mer DNA or RNA is around the upper limit of the synthetic potential of the p180⌬N-core on this template (Fig. 5B, lanes 7 and 8). The result is in agreement with the previous findings for yeast pol␣ with the same poly-dT template (17). To explore whether this "20-mer rule" is a general or template-specific property of pol␣, we examined DNA synthesis by the same core enzyme in the extension of RNA and DNA 16-mer primers with identical heterogeneous sequences. We designed two different templates with a heterogeneous sequence and a single, fixed site of primer annealing ("Materials," under "Experimental Procedures"). The first template, 73a, included a 35 homopolymeric-dT run at the 5Ј-end. The second template, 73b, contained all heterogeneous sequences with the 3Ј 38-mer sequence identical to the template 73a. Surprisingly, the efficiency of the extension of DNA and RNA primers by the p180⌬N-core was identical in the matched primer/template pair (Fig. 6). Interestingly, the synthesis by the p180⌬N-core on template 73a was prone to termination at positions around 20 nucleotides deep into the 35 homo-dT run. No such pausing was detected on the heteropolymeric 73b template, and pol␣ reached the end of the template, similar to T4 DNA polymerase. We concluded that pol␣ has a limited capacity for DNA synthesis only on the poly-dT template.
The C-terminal Domain of pol␣ Regulates DNA Polymerase Activity-In comparison to the p180⌬N core, the primer extension activity of the dimeric pol␣ complex (p180⌬N⅐p70) on the poly-dT 70 template with RNA primer is much lower (note a 2-fold higher enzyme concentration of p180⌬N⅐p70; Fig. 7A). It is interesting that the average length of the synthesized fragments is also decreased to ϳ10 nucleotides. The increase of concentrations of dATP in reaction with the RNA primer on the poly-dT70 template did not result in a stimulation of the primer usage. The slight increase in the product size (Fig. 7B) is negligible in comparison to the 10-fold stimulation in the p180⌬N-core by dATP (Fig. 5B, lanes 5-8). The presence of CTD with metal binding sites rendered the dimeric enzyme sensitive to DTT (Fig. 7C), whereas DDT did not modulate the activity of the p180⌬N-core (Fig. 7D). With an increased reaction time and greater than a 5-fold excess of the enzyme, we obtained a comparable activity of the p70⅐p180⌬N and p180⌬N-core on RNA 15 /poly-dT 70 . The synthesis by the dimeric pol␣ was still terminated a few nucleotides earlier than the synthesis by the p180⌬N-core (Fig. 7E). The lower activity of the p70⅐p180⌬N dimer compared with the p180⌬N-core was also observed on the hybrid homo-heteropolymeric 73a template with the RNA primer (Fig. 7F).
Primase and pol␣ Activities of Quaternary pol␣-prim Complex-During the initiation of DNA replication, the utilization of ribonucleotides and deoxynucleotides by primase⅐pol␣ should happen in a sequential order. Primase first synthesizes RNA, then pol␣ synthesizes the DNA part of the primer. To study the primase activity in the pol␣-prim complex, we analyzed the activities of primase and pol␣ within the four-subunit complex in the primer extension assay with riboversus deoxynucleotides or a mixture of both. In these experiments we used a low 1:15 ratio of enzyme to template to bring reactions closer to single-hit conditions. Primase p49⅐p58 and pol␣ p70⅐p180⌬N were used as internal controls (Fig. 8). The primase is a non-processive enzyme and was able to incorporate only a few nucleotides (up to 10) (lanes 2-4). Oligonucleotides with a higher number of incorporations seen after 4 min of reaction most likely represent rebinding of the enzyme (lane 4). The heterodimer of pol␣ processively added substantially more nucleotides with the median around position ϩ15. In complex with pol␣, primase activity was less than the primase alone (Fig.  8, compare lanes 8 -10 and 2-4). This was likely due to the handover of primers from primase to pol␣. pol␣ robustly extended the primer in complex with primase with only dATP (Fig. 8, lanes 11-13). In comparison to p70⅐p180⌬N alone (Fig.  8, lanes 5-7), there is a slight increase of primer usage and product length. In the reactions with both rATP and dATP, only a few rAMPs were attached by primase to the rA 15 (position ϩ1), and pol␣ rapidly outcompeted primase for the primer 3Ј-end and robustly extended it (Fig. 8, lanes 14 -16).

DISCUSSION
The eukaryotic primases involved in the bulk replication of large genomes are embedded into a four-subunit protein complex with the DNA pol␣ (Fig. 1A). The understanding of the mechanism of regulation of the synthesis of the RNA-DNA hybrid primer during replication is fragmentary, in part due to the lack of knowledge of the protein-protein interaction within the four-subunit complex. One of the key parameters of the pol␣-prim reaction is the regulation of the length of the RNA and DNA parts of the primers made available for major repli- cative DNA polymerases. The limitation of the current models of counting is that they are based on structures of either the primase or the pol␣ parts of the complex. Several x-ray crystal structures of the parts of the eukaryotic primase are available. The most complete is a structure of the human dimeric primase lacking the CTD of the large subunit (38). The analysis suggests that the same set of residues in the small catalytic subunit is responsible for the initiation and extension step of primer synthesis. The role of the highly conserved CTD of the large subunit, necessary for the initial dinucleotide synthesis, could only be indirectly inferred from this structure. The x-ray crystal structures of the p58C and its analog from yeast are available (33,42,43). The high structural similarity to the DNA photolyase/cryptochrome family of flavoproteins that binds flavin adenine dinucleotide (FAD) (43) suggests that this domain may be involved in the binding and stabilizing of the initial dinucleotide synthesized by primase to the template for further elongation (38). The folding of the region that binds single-stranded DNA indicates that the N terminus of the p58C can adapt to become ␣-helical or ␤-sheet structure, which suggests possible conformational change during priming (42). Based on our data that p58C does not interact with the other parts of the pol␣-prim complex (Fig. 4C), it is possible that the p58C domain is flexible by itself and with respect to other parts of the complex. This domain may act as an extra finger domain that binds to the first 5Ј nucleotide for the critical rate-limiting step of dinucleotide synthesis.
pol␣ is composed of a large catalytic subunit and a smaller B-subunit. Only the catalytic domain of the large subunit is necessary for the DNA synthesis, whereas its CTD and B-subunit are required for the replication and cell viability (44,45). The contact between the B-subunit and the CTD with primase is thought to tether primase to the replisomes and origins (26,35,46,47). The exact role of the pol␣ CTD and B-subunit in primer synthesis is unknown. In the SV40 replication system, pol␣ synthesizes the whole viral genome (5.2 kb) in the presence of the viral helicase (large-T antigen) through the helicase interaction with p49, p70, and p180 (37,48,49). The most critical contacts seem to be with the N terminus of p70 (37). We have found that the presence of p70⅐p180C sharply increases  the lengths of the reaction products, most likely affecting the primase processivity (Fig. 2). The binding to p180C can stabilize the p58N domain, which is proposed to be involved in the interaction with the DNA-RNA hybrid duplex (50). Furthermore, p180C by itself may participate in stabilizing the primase complex with the RNA-primed template because it is located near p58N. Yeast p70⅐p180C heterodimer has an oligonucleotide/ oligosaccharide-binding fold and zinc finger domains and possesses a micromolar affinity to dsDNA (26).
The deoxynucleotides act as natural competitive inhibitors of primase during the primer extension reaction (Fig. 2B, lanes  3-5). We observed that primase inserts first dNTP with the efficiency close to that of rNTPs but slowly extends from them (Fig. 3). We concluded that primase lacks significant selectivity between ribo-and deoxynucleotides at insertion step, and the inhibitory effect of deoxyribonucleotides is mediated by an inefficient extension. Primase has two NTP binding sites: socalled initiation and elongation sites. In the course of dinucleotide synthesis the "initiation" site binds NTP, which becomes the 5Ј-terminal nucleotide of the primer. Biochemical and structural data accumulated so far indicate that primase uses the same set of functional residues for dinucleotide synthesis and its extension, which means that the 3Ј-terminal nucleotide of primer is located in the initiation site (38,51,52). Our data indicate that the initiation site is more selective for rNTPs than the elongation site. Primase is known to synthesize the RNA primer 8 to 10 nucleotides long, which is then extended by pol␣ to 30 nucleotides (2,16,53). In vitro biochemical evidence from our study and others suggests that the heterodimeric primase is a low processivity enzyme, but it is capable of synthesis far beyond 10 nucleotides by iterative cycles of synthesis (30,32). We confirmed that the interaction of primase in a tetrameric complex with pol␣ is mediated by binding of pri2N to p180C (Fig. 4C). This stabilizes the complex and enhances primase activity (Fig. 2). The presence of pol␣ inhibited the primase activity in the reactions containing rNTPs but no dNTPs. This suggests that polymerase and primase activities of the complex are competitive in vitro. In the crystal structure, pol␣ establishes an elaborated network of contacts with 9 bp of DNA-RNA duplex (17) and should effectively capture an RNA- primed template when the primase synthesizes the unit-length primer. This reduces the effective amount of primers available for the primase.
Ribonucleotides at high concentrations inhibit B-family pols in a step preceding base pairing (54). pol␣ strongly prefers dNTPs over NTPs (55); therefore, in the absence of deoxynucleotides it does not participate significantly in primer extension. We also observed a limited RNA and robust DNA pols activity of the tetrameric complex in reaction with a mixed pool of nucleotides (Fig. 8). These results demonstrate that the primase activity is further suppressed in the pol␣-prim complex.
In the presence of primase in the pol␣ complex, the product length increased (Fig. 8, compare lanes 5-7 to 11-13). pol␣primase complex has two active sites for the primed DNA template, which increases the probability of substrate binding. After initial binding in the primase active site and after extension, the substrate might be translocated to the pol␣ active site without dissociation from the complex (29).
During primer synthesis, the RNA synthesized by primase is extended by pol␣. The RNA primer (poly-rA 15 ) and singlestranded DNA template (poly-dT 70 ) form an unusual A-form RNA/DNA duplex, which is extended by yeast pol␣ more effi-  [2][3][4][5] or by the p180⌬N-core (enzyme to primer/template ratio ϭ 1:100; lanes 6 -9). nt, nucleotides. B, extension of the poly-rA15 primers by p70⅐p180⌬N on the poly-dT70 template (enzyme to primer/template ratio ϭ 1:50, reaction time was 5 min). The black triangle indicates that the reactions were carried out with the increase of dATP concentrations of 0.01, 0.1, and 1 mM, respectively. C and D, effect of DTT on the ability of pol␣ to extend the poly-rA15 primer. C, pol␣ p70⅐p180⌬N (enzyme to primer/template ratio ϭ 1:50, reaction time 10 in). D, p180⌬N-core (enzyme to primer/template ratio ϭ 1:100, reaction time 5 min). The black triangle indicates that the reactions were carried out with the increase of DTT concentrations of 0, 0.5, 1.0, 2.0 mM DTT. E, smaller fragments in primer extension by excess of dimeric pol␣ (p70⅐p180⌬N) in comparison to the pol␣ core. Extension of the poly-dA15 and poly-rA15 primers in long reactions by the excess of p70⅐p180⌬N (enzyme to primer/template ratio ϭ 1:50, reaction time 15 min) or by the p180⌬N-core (enzyme to primer/template ratio ϭ 1:250, reaction time 2 min) with 0.2 mM dATP. F, lower activity of dimeric pol␣ in comparison to the pol␣ core on hybrid homo-heteropolymeric DNA template. Extension of the RNA primers on the DNA template 73a by p70⅐p180⌬N (0.025 M) or by p180⌬N-core (enzyme to primer/template ratio ϭ 1:50) for the indicated time points is shown. As a control, T4 DNA polymerase (ϳ1:2500, reaction time 2 min) was used to obtain a completely extended, full-length product.
ciently than the DNA primer (poly-dA 15 ), which forms a B-form duplex with the template (17). The extension by pol␣ comes to a stop when synthesis reaches 20 nucleotides long. This effect was elegantly explained and thought to be the intrinsic property of yeast pol␣ undergoing conformational change when it reaches a certain distance from the A-form of DNA (17). Our results on the extension of the 5Ј-end labeled fluorescent DNA (poly-dA) and RNA (poly-rA) primers by human pol␣ agree with this observation (Fig. 5). The increase of the dATP concentration increased the efficiency of these reactions, but a termination zone around 20 nucleotides still existed. Surprisingly, the extension assay using RNA and DNA primers with a random sequence as well as a template with a combination of terminal 5Ј-poly-dT and heterogeneous sequences provided a completely different overall picture (Fig. 6). First, there was no difference for RNA versus DNA primers for the p180⌬N-core to extend. Second, no termination zone was seen on these templates until the synthesis proceeded into the polyT stretch. Obviously, a sharp contrast between RNA and DNA primers and the "20-nucleotide rule" could only be seen for the templates with poly-dT homopolymeric runs. The DNA strands of such runs are known to form an atypical structure (i.e. bending, H-form variants) in the neutral pH condition and in the presence of divalent metal ions (i.e. Mg 2ϩ ) (56 -58). We conclude that pol␣ typically utilizes RNA and DNA primers with equal efficiency and the homopyrimidine (dT) run in duplex with homopurine (dA) forms an unusual structure, which prevents the moving of pol␣ above a certain distance. This specific property of pol␣ may apply to other types of polypyrimidine tracts (59). Our data suggest that counting by pol␣ itself is an exception rather than the rule, and there are additional external signals in vivo that regulate the length of primers synthesized by pol␣.
The comparison of the catalytic activity of the pol␣ p180⌬Ncore and p70⅐p180⌬N indicates that the catalytic activity of pol␣ is negatively regulated by p70⅐p180C. The active conformation of the yeast pol␣ core appears to be in tighter contact with the DNA-RNA duplex (17) in comparison to RB69 polymerase (gp43), a prototype of the eukaryotic B-family pols ␣, ␦, ⑀, and , which lacks the distinctive C-terminal domain and B-subunit (60,61). The C-terminal domain and the B-subunit of pol␣ are not required for its catalytic activity. It is currently unknown how these additional elements fit into the tertiary complex. We propose that the p70⅐p180C in the pol␣ complex interferes with the formation of the ternary complex including the primer/template duplex and deoxynucleotides. It is possible that the p70⅐p180C is positioned in the complex in a way that lowers accessibility of the pol active site.
Additional evidence that supports that the C-terminal domain and B-subunit plays a role in regulation of pol␣ activity is the suppression of the inhibitory effect of p70⅐p180C on pol activity by the optimal concentration of DTT (Fig. 7, C and D; note that the p180⌬N-core works completely independently of the DTT concentration). It is noticeable that certain concentrations of DTT also result in a change of the distribution of sizes of DNA fragments synthesized by p70⅐p180⌬N. DTT is a well known agent modulating DNA polymerases activities in vitro (49). It is thought to reduce the intramolecular disulfide linkages that may affect p180C folding (62). The p180C contains two highly conserved four-Cys motifs, which bind either two zinc atoms or one zinc and one [Fe-S] cluster (17,20,27). At the present time there is not enough structural information on the folding of the pol␣ complex and the effects of oxidation on this folding to explain the effect of DDT. It is possible that the effect of Cys oxidation that inhibits pol␣ and the natural regulatory effect of the C-terminal domain/B-subunit to pol␣ activity represent separate mechanisms.