Crystal structure of an archaeal DNA sliding clamp: Proliferating cell nuclear antigen from Pyrococcus furiosus

Mutation for Met73 to Leu

We made an overexpression system for the PfuPCNA gene in E. coli. The PfuPCNA protein was purified with ease by heating the crude cell extract to denature most of the E. coli proteins, followed by anion exchange chromatography (Cann et al. 1999). However, as we described previously, an extensive amount of another thermostable protein with a molecular mass of 20 kD was produced, as determined by SDS-PAGE analysis. The N-terminal amino acid sequence suggested that the protein was the product of an unexpected translational initiation at the ATG codon for Met73 in E. coli cells. This 20-kD product was not detected in the P. furiosus cells from the Western blot analysis using anti-PfuPCNA antiserum. It was predicted that the truncated protein would interfere with the formation of homogeneous crystals appropriate for structure analysis. To avoid this inconvenience, we disrupted the internal ATG codon of the structural gene for PfuPCNA. The sequence alignment of the P. furiosus, yeast, and human PCNAs showed that the yeast and human PCNAs have Leu at the position corresponding to Met73 of PfuPCNA (Fig. 1 ▶). Therefore, we changed the ATG to CTG at the corresponding site by site-directed mutagenesis, as described in Materials and Methods. The truncated product was not detected in the process of expression and purification of the mutant protein PfuPCNA(M73L). The purified PfuPCNA(M73L) stimulated the primer extension activity of P. furiosus Pol I, as in the case of the wild-type PfuPCNA.

Overall structure of Pfu PCNA(M73L)

The crystal structure of PfuPCNA(M73L) was determined by the molecular replacement method, using the structure of yeast PCNA (Krishna et al. 1994) as a search model. As predicted from the sequence alignment (Cann et al. 1999), PfuPCNA(M73L) has two β–α–β–β–β–β–β–α–β–β–β structural domains linked to each other with an interdomain connecting loop (Fig. 2 ▶). However, the three-dimensional structure of the connecting loop could not be modeled in this study because of poor electron density in this region.

PfuPCNA(M73L) forms a trimer around the crystallographic threefold axis, which adopts a pseudo-sixfold, ring-shaped architecture similar to the eukaryotic PCNA and the bacterial β subunit (Fig. 3 ▶).

Structural basis for thermal stability

P. furiosus is a hyperthermophilic archaeon that grows optimally at 100°C (Fiala and Stetter 1986). There have been many explanations for the stability of proteins from thermophiles (Vogt and Argos 1997; Jaenicke and Böhm 1998), and PfuPCNA adopts some of the features found in those surveys.

The number of amino acids constituting PfuPCNA is 249, which is smaller than the number constituting the PCNAs of yeast (258) and human (261). PfuPCNA has shorter loops between βH₁–βI₁ and βD₂–βE₂, and at the C terminus (Fig. 1 ▶). A decrease in loop length is also observed in P. furiosus citrate synthase (Russell et al. 1997) and may contribute to the enhancement of the thermal stability.

PfuPCNA contains a 4%–6% higher percentage of charged residues and a 5%–10% lower percentage of polar/uncharged residues compared with the yeast and human PCNAs (Table 1). This fact agrees with the trend observed for thermophilic proteins (Deckert et al. 1998), although this statistical ``traffic rule'' is not necessarily applicable to individual proteins (Böhm and Jaenicke 1994).

The enhancement of electrostatic interactions through ion pairs is thought to be one of the main causes of the increased thermostability of proteins (Vogt and Argos 1997). PfuPCNA has 27 short ion pairs (distance ≤4.0 Å) per molecule between oppositely charged groups, whereas yeast PCNA (Krishna et al. 1994) has 9 per molecule, and human PCNA (Gulbis et al. 1996) has 39 per trimer. In the case of PfuPCNA, 10 out of 27 ion pairs are intermolecular, and they are localized at the trimer-forming interface. On the other hand, yeast PCNA has 2 and human PCNA has 5 (per trimer) intermolecular ion pairs at the interface.

Intermolecular interface

The PCNA trimer is formed through intermolecular main chain amide-to-carbonyl hydrogen bonds between the antiparallel β strands βI₁ and βD₂. PfuPCNA(M73L) possesses four hydrogen bonds (N···O ≤ 3.3 Å), whereas seven hydrogen bonds were observed in yeast PCNA and eight in human PCNA (Table 2; Fig. 4 ▶). The reduction of the amino acids in PfuPCNA occurs at the loops adjacent to βI₁ and βD₂, and these differences in the loop lengths may be the cause of the distinct intermolecular hydrogen bond patterns.

The buried surface area per molecule at intermolecular interface is 1.44 × 10³ Ų, which is 7% smaller than the 1.55 × 10³ Ų of the yeast PCNA trimer. This value additionally supports a weaker intermolecular interaction within PfuPCNA.

Site-directed mutagenesis for Met73 to Leu

Polymerase chain reaction (PCR)–mediated mutagenesis was performed to change the ATG codon for Met73 to CTG. The plasmid pTPCNA (the structural gene for PfuPCNA is inserted into the pET21a vector [Novagen, Madison, WI]), which was made previously (Cann et al. 1999), was directly amplified by PCR using the two primers M73LF (5′-CAATTGGAGTTAACCTGGACCACC TAAAG-3′) and M73LR (5′-CTTTAGGTGGTCCAGGTTAACT CCAATTG-3′), which bind to the region containing the target codon (underlined). The PCR conditions involved 30 cycles of denaturation for 30 sec at 95°C, annealing for 1 min at 55°C, and extension for 10 min at 68°C with Pyrobest DNA polymerase (Takara Shuzo, Kyoto, Japan). The amplified plasmid was purified by agarose gel electrophoresis and used for transformation. The plasmids isolated from several independent transformants were sequenced. The plasmid that contained the gene with only one mutation at the target codon was selected and designated as pTPCNAL73. The production and purification of PfuPCNA(M73L) from E. coli BL21(DE3)/pTPCNAL73 were performed as described for PfuPCNA (Cann et al. 1999). The purified PfuPCNA(M73L) solution was concentrated to 20 mg/mL by ultrafiltration, using Microcon YM-30 (Millipore, Bedford, MA), and was stored at 4°C.

Crystallization and data collection

Single crystals of PfuPCNA were prepared by the hanging drop vapor diffusion method at 20°C. The initial condition search was performed with Grid Screen ammonium sulfate (Hampton Research, Laguna Niguel, CA) and was optimized with a finer grid of pH, precipitant concentration, and additives. The optimal composition of the precipitation solution was 100 mM sodium citrate (pH 5.5), 2.4 M ammonium sulfate, 5% (v/v) 1,4-dioxane. Then, 1.0 μL of 20 mg/mL protein solution and 1.0 μL of precipitation solution were mixed and placed over 500 μL of precipitation solution. Colorless hexagonal needles of ∼1.0 × 0.2 × 0.2 mm were obtained in a week. The PfuPCNA crystal was mounted in a glass capillary for X-ray diffraction data collection.

Diffraction data were collected at 296 K on the BL-6A beamline at Photon Factory, using 1.0000 Å radiation with an ADSC Quantum 4R CCD detector (Area Detector Systems, Poway, CA). The data were processed using DPS/MOSFLM (Rossmann and van Beek 1999), scaled, and converted for structure determination using SCALA and TRUNCATE in the CCP4 suite (Collaborative Computational Project, Number 4 1994).

Structure determination

The initial crystal structure of PfuPCNA was determined by the molecular replacement method using AMoRe (Navaza 1994) with yeast PCNA (Krishna et al. 1994, PDB code 1PLQ) as the search model. The solution that gave the lowest R value of 0.559 after rotation and translation search using 25.0–3.5 Å data was used for the structure refinement. The molecular model was built using O (Jones et al. 1991) and was refined using CNS software (Brünger et al. 1998). Met1, Val118-Leu125, and Glu248-Glu249 were not modeled because of poor electron density at these regions. The results of the structure analysis are summarized in Table 3. The coordinates and structural factors have been deposited to the Protein Data Bank under the accession code 1GE8.

Comparison of three-dimensional structures

The multiple sequence alignment of the P. furiosus, yeast, and human PCNAs was performed using CLUSTAL X V1.8 (Thompson et al. 1994) with a slight modification by hand. Secondary structures were assigned using PROCHECK (Laskowski et al. 1993) based on the three-dimensional structures of PfuPCNA, yeast PCNA (Krishna et al. 1994, PDB code 1PLQ), and human PCNA (Gulbis et al. 1996, PDB code 1AXC). Hydrogen bond and ion pair distances were calculated using ACT in CCP4. Buried surface areas were calculated using SURFACE in CCP4, with a probe radius of 1.40 Å. The figures of the molecules were prepared using RasMol (Sayle and Milner-White 1995).