Index References

Gene 1 product (gp1) of Bacillus subtilis phage φ29 is known to promote the protein primed DNA replication of φ29 during infection (Hirokawa, 1989; Meijer et al., 2001). Although its size is small (10 kDa), gp1 is reported to exhibit multiple characteristics including repression of φ29 dna genes via a post-transcriptional mechanism (Takeuchi et al., 1995; Takeuchi et al., 1998), affinity for RNA (Takeuchi et al., 1998; our unpublished data), localization to cell membranes with other φ29 DNA replication proteins (Bravo et al., 1997), and self-association into long filamentous structures (Bravo and Salas, 1998). However, it is largely unknown how these characteristics take part in φ29 DNA replication. To understand their roles, and also to gain insight into the structure of gp1, we decided to isolate a series of missense mutants of gene 1 which were affected in different characters.

During the development of screening systems to isolate functionally affected mutants, we found that growth of Escherichia coli cells was severely impaired upon induction of recombinant gp1. A similar phenomenon has been observed with various recombinant proteins, including DNA replication proteins (Minkley et al., 1984; Murray, 1983; Studier and Moffatt, 1986; Tabor and Richardson, 1985), protease (Baum et al., 1990), nuclease (Nakashima and Tamura, 2004), proteins involved in amino acid metabolism (Lin et al., 2000) and nucleotide metabolism (Cinquin et al., 2001), and membrane transporter (Donnelly et al., 2001; Mokhonova et al., 2005). In these cases, the protein's authentic functions have been shown or supposed to be responsible for the growth inhibition or toxicity. In the case of gp1, its presence even at low levels (with weak transcription promoters or even without induction) caused filamentation and lysis of the cells (our unpublished results). From these observations, we surmised that gp1 affected cellular activities via its authentic functions, rather than via its overproduction. Therefore, we attempted here to employ the growth inhibitory effect of gene 1 as a selection index, expecting that functionally affected mutants would not be able to inhibit cell growth as the wild type did.

View Details

Fig. 1. Nucleotide sequence of gene 1 in the expression plasmid pUV1-WT used for the screening of mutants. The essential part of its nucleotide sequence is shown. The primer binding sites for the error-prone PCR are shown by thick underlines (1P30 and RvHis), and the region to be mutagenized is shaded. The nucleotide sequences of the 1P30 and RvHis primers were (5'-GGTTCTGCAGTAAACGTTTGTTTAAAAGGA-GATGTTTGTAATG-3') and (5'-TGATTACGCAAGCTTAGTGAT-GGTGATGGTGATG-3'), respectively. The Shine-Dalgarno sequence, and the start (ATG) and stop (TAA) codons of gene 1 are shown by bold letters. The polyhistidine tag sequence attached to gene 1 is boxed and indicated as “His-tag”. In E. coli, minor translation which starts from the Met34 codon of gene 1 is known to produce a truncated gp1 (Prieto et al., 1989). To suppress this translation, a silent mutation was introduced into the gene 1 of pUV1-WT (A to T, underlined italic bold letters). This mutation, disrupting a putative Shine-Dalgarno sequence, was confirmed to suppress the minor translation (data not shown). In this report, we refer to the modified gene 1 in pUV1-WT as the wild type. Prior to the construction of pUV1-WT, the empty vector pUV5-RF was constructed by modification of the E. coli vector pTrc99A (Amersham Biosciences). In brief, the trc transcription promoter was changed to the lacUV5 promoter (the –35 and –10 region are double underlined), and primer binding sites for the DNA sequencing (thin arrows, Fw and Rv) and restriction enzyme sites, PstI (CTGCAG, boxed) and HindIII (AAGCTT, boxed), were inserted. To construct pUV1-WT, φ29 gene 1 was inserted between the PstI and HindIII sites of pUV5-RF. The plasmid also contained an ampicillin resistant marker (bla) and the replication origin of pBR322 plasmid (pBR ori).

For the selection by growth inhibition, we constructed an E. coli expression system of gene 1. Figure 1 shows the structure of the wild-type gene 1 cloned in the expression plasmid pUV1-WT under the control of the lacUV5 promoter. As shown in Fig. 2A, induction of the wild-type gp1 severely inhibited colony formation of the cells harboring pUV1-WT. In contrast, such inhibition was not observed when gp1 was not induced or when gene 1 was absent. Therefore, we tried to isolate mutants of gene 1 from the colonies that appeared in the presence of IPTG.

View Details

Fig. 2. Expression and phenotype of gene 1 in E. coli. A. Growth inhibitory phenotype of gene 1 in E. coli. DH5α cells harboring the plasmid pUV1-WT or its derivatives with the indicated types of gene 1 were streaked on NZY-ampicillin (100 μg/ml) plates in the presence (+) or absence (–) of 1 mM IPTG. After incubation at 37°C, the colonies on these plates were photographed. B. Comparison of the productivity of wild-type gp1 and its variants after induction. E. coli DH5α cells harboring the plasmid pUV1-WT or its derivatives with the indicated types of gene 1 were grown in NZY-ampicillin broth at 37°C, and induced (indicated as +) or not induced (indicated as –) with 1 mM of IPTG for 2 h. Whole cellular protein extracts were electrophoresed on a Laemmli-type SDS polyacrylamide gel (15%) and visualized by Coomassie blue staining. The protein band corresponding to gp1 is shown by an arrow. The letters a and b indicate the positions of the size markers for 14.4 kDa and 6.5 kDa, respectively.

To introduce random mutations into the wild-type gene 1, we employed error-prone PCR in the presence of man-ganese ion (Leung et al., 1989), using the primers described in Fig. 1. To suppress multiple mutations in a clone, the manganese ion (0.5 mM) was included only in the first four PCR cycles (94°C 1 min, 70°C 1 min, 72°C 1 min). The mutagenized PCR product was then diluted 50 fold and further amplified in 25 cycles of conventional PCR in the absence of manganese ion.

For screening by growth inhibition, the mutagenized PCR product was inserted into the expression vector pUV1-WT by displacement of the PstI - HindIII fragment containing the wild-type gene 1 (Fig. 1). This library was then introduced into E. coli DH5α cells and transformants were grown on NZY-ampicillin (100 μg/ml) plates containing 1 mM IPTG. The colonies that appeared on this plate, which were resistant to both IPTG and ampicillin, were isolated as candidates of mutants. Such IPTG-resistant transformants accounted for 2%~8% of the colonies that appeared on the ampicillin-plates without IPTG. To exclude clones with undesired mutations such as frameshift or nonsense mutations, gp1s produced by the candidates were checked by SDS polyacrylamide gel electrophoresis (Fig. 2B). After selecting clones producing gp1 of normal size and amount, single missense mutants were identified by DNA sequencing using the Fw and Rv primers described in Fig. 1. We thereby obtained three clones each with a single missense mutation, designated as K8E, K33E and D36G (Table 1 and Fig. 3).

View Details

Table 1. Numbers of isolated mutants and candidates obtained at each screening step

View Details

Fig. 3. Amino acid substitutions deduced from the single missense mutations in the isolated gene 1 mutants. Names of the mutants, some with bracketed numbers, are shown at the left of the panel. Below the amino acid sequence of wild-type gp1, the deduced amino acid substitutions of the mutants are shown by one letter and the unaffected amino acids are shown by minus signs. Amino acid positions are also shown at the top. Nucleotide substitutions for missense mutations are shown at the right of the panel. These substitutions are indicated as “G5A” when guanine at the fifth nucleotide position of gene 1 was substituted with adenine, assigning the adenine of the start codon as the first position. Codons with silent mutations are also shown by +.

As shown in Fig. 2A, the growth inhibitory effect of the isolated missense mutants was obviously suppressed as compared to that of the wild type. In contrast, induction of their products was essentially unaffected by the mutations (Fig. 2B). Therefore, amino acid substitutions were suggested to have affected some properties of gp1. These facts encouraged us to continue further isolation of missense mutants, even though very few missense mutants were obtained, and the majority of the IPTG-resistant clones produced irregular-sized gp1 or could not produce a detectable amount of gp1 (Table 1). Therefore, we next tried to modify the screening conditions to improve the efficiency of selection.

During handling of the isolated mutants, we noticed that the inhibitory effect of K8E was not completely suppressed. The cells expressing K8E grew slower and their colony size was smaller than that of the cells expressing K33E or D36G or of the cells without gene 1 (Fig. 2A). We therefore selected small colonies on the plate with IPTG which appeared to retain some inhibitory phenotype, assuming that a single missense mutation would not always completely abolish the inhibitory phenotype. In addition, for efficient detection of such small or slowly growing colonies, we reduced the concentration of IPTG to 0.3 mM. The inhibitory phenotype of the wild type in this condition was practically indistinguishable from that observed at 1 mM IPTG (data not shown). As shown in Table 1, the modification resulted in a significant increase of IPTG-resistant clones which could produce normal-sized gp1s. Surprisingly, the proportion of the finally obtained single missense mutants among the candidates after the first screening increased 26 fold (Table 1). These results demonstrated that selection by a moderate change of phenotype was effective, at least in the case of gene 1, for isolation of single missense mutants. As depicted in Fig. 3, the 39 independent clones thus obtained had 31 different single amino acid substitutions at 26 positions along the entire region of gp1. Thus, we could finally obtain a series of missense mutants which were expected to affect various functions of gp1.

Recently, we have purified several products of the isolated mutants and shown that they were indeed affected in the known properties of gp1 (e.g., self-association and affinity for RNA; Hashiyama et al., 2005a, 2005b). These results proved the utility of the isolated mutants for investigating the functions and structure of gp1. Thus, for further detailed studies, we inquired into the amino acid sequence of gp1 to evaluate the significance of all the substituted amino acid residues described here. First, we examined whether the substituted residues were conserved among the orthologs of gp1 encoded by φ29-related phages. In Fig. 4, an alignment of the deduced amino acid sequences of these gp1 orthologs is shown with the amino acid substitutions presented in Fig. 3. Interestingly, 85% of the substitutions were mapped to strongly conserved residues, while the overall conservation of φ29 gp1 residues was 52%. This fact supported the functional or structural importance of the amino acid residues substituted by the mutations.

View Details

Fig. 4. Representation of the amino acid substitutions of the mutants on an alignment of the sequences of gp1 orthologs with sequence motifs conserved among them. Deduced amino acid sequences of the orthologs of gp1 from phages φ29 (ACC: P03679), PZA (ACC: ERBP1Z), M2 (ACC: AB218446), and B103 (ACC: BB103G, nt location 645 to 845) were aligned with the number of the amino acid residues of φ29 gp1 indicated at the top. Identical residues among the four orthologs are highlighted in black, and strongly conserved residues according to Gonnet Pam250 matrix (score > 0.5) are shown by shaded letters (Gonnet et al., 1992). The residues substituted in the isolated mutants (Fig. 3) are indicated above the alignment, and highlighted in bold letters when the substitutions occurred at conserved residues. The predicted coiled-coil (Lupas et al., 1991) and strongly hydrophobic regions (Kyte and Doolittle, 1982) in the four orthologs are indicated by lines (Asp31 to Asn66 and Trp71 to Gly78, respectively) below the alignment. In the region extended with a dotted line (Asp31 to Asp36), the probability of a coiled-coil structure differed among the four orthologs, with φ29 gp1 having the highest probability. Note that Bravo et al (2001) described the coiled-coil region as being from Glu38 to Asn65 and that Serrano-Heras et al. (2003) described the hydrophobic region as being from Tyr68 to Ala84. Phage M2 is a variant of phage Nf (Matsumoto et al., 1981; Mizukami et al., 1986), and its nucleotide sequence of the gene 1 homolog was determined in this work (ACC: AB218446). In the cases of phage PZA and B103, we used deduced amino acid sequences of the ORFs previously assigned as 1c (Pačes et al., 1985; Pečenková et al., 1997). All of these ORFs from M2, PZA and B103 were located in similar regions of the genomes as in φ29, and no other ORFs were found to have significant homology with φ29 gene 1.

Some specific regions of gp1 have been shown to be involved in specific characteristics of gp1. Thus we next examined whether these regions were conserved, and at the same time, whether the mutations affected structural or functional features of these regions. The region involved in the membrane localization of gp1 in B. subtilis cells has been mapped in the carboxyl terminal half (Met⁴³ to Lys⁸⁵) by deletion analyses (Bravo and Salas, 1997). In addition, it has been pointed out that the hydrophobicity of the carboxyl terminal region spanning Tyr⁶⁸ to Ala⁸⁴ might be related to the amphipathic nature of gp1 (Serrano-Heras et al., 2003). As shown in Fig. 4, alignment of gp1 orthologs revealed a cluster of amino acids which were strongly conserved and hydrophobic (Trp⁷¹ to Gly⁷⁸). Interestingly, all the mutations except V75A in this cluster (W71R, W71R (64), K74E, F77S, F77S (50), and G78R) substituted hydrophilic residues for hydrophobic residues. Thus, we propose that the membrane localization of gp1 is accomplished via direct hydrophobic interaction between the conserved cluster and lipid bilayer, and that presence of gp1 on the membrane is essential for the growth inhibition of E. coli cells. Consistent with this view, we have observed that the products of W71R and K74E had increased solubility upon extraction from E. coli cells, whereas the wild-type gp1 was insoluble (Hashiyama et al., 2005a, 2005b). This might imply that they would have lower affinity for the cellular membrane. All the above considerations indicate that the mutants with substitutions in the conserved hydrophobic cluster would be useful for studying the membrane localization of gp1.

Besides being localized in the cell membrane, gp1 has been reported to self-associate in vitro, and a part of the hydrophobic region near the carboxyl terminus (Val⁷⁵ to Lys⁸⁵) was shown to be required for the self-association by deletion analyses (Bravo and Salas, 1998; Bravo et al., 2000). According to our recent observations, substitutions in the conserved hydrophobic cluster (Trp⁷¹ to Gly⁷⁸) by hydrophilic residues (W71R, K74E and F77S) severely affected the self-association property (Hashiyama et al., 2005b). Therefore, we assume that the conserved hydrophobic cluster is the core region essential for both the self-association and membrane localization of gp1. Furthermore, we propose that both of these properties are requi-red for the malignant effect of gp1 on E. coli growth. Use of missense mutants with substitutions in the conserved cluster might enable the dissection of these properties, while deletions would disrupt both of them together.

Another known functional region is the coiled-coil motif which is also essential for the self-association of gp1 (Bravo et al., 2001). As shown in Fig. 4, this region (Asp³¹ to Asn⁶⁶) was well conserved and 21 mutations were found to be located in this region. Among them, five mutations (D36G, E49G, E59G, E59G (33), L60P) resulted in the substitution of residues by proline or glycine which are known as α-helix destroyers, and four (M53T, L57S, L57V, L60P) resulted in the substitution of residues at the a or d position of the heptad repeat, which is known to be important for interaction between the α-helices (Bravo et al., 2001; Lupas, 1996). These mutants would be useful to confirm the presence of the coiled-coil structure with full-length gp1; this structure was suggested by previous studies employing truncated gp1s (Bravo et al., 2001). Interestingly, the substitution of Lys³³ or Lys⁴¹ by Glu (K33E or K41E) has been shown to decrease the self-association of gp1 (Hashiyama et al., 2005b), though the substitution is not predicted to disrupt α-helices or located at the a or d position. Thus, exami-nation of other substitutions found in this region might also identify novel and important residues which participate in the self-association. The occurrence of many substitutions in the coiled-coil motif might also support our view that the growth inhibition of E. coli was correlated with the self-association of gp1.

While many of the substituted residues belonged to the predicted coiled-coil and the hydrophobic region, substitutions were also found in the amino terminal region (Fig. 4). A truncated gp1 retaining its amino terminal half spanning Gly¹ to Lys⁴² interacts with the primer protein of φ29 (the product of gene 3) in vitro (Bravo et al., 2000). However, neither its precise functional region nor any distinctive feature in the amino acid sequence has been identified yet. As shown in Fig. 4, alignment of gp1 orthologs revealed a short conserved region spanning Gly¹ to Lys⁸, and seven mutations were mapped in this region. Interestingly, the substitutions caused by these mutations, except K2N, involved changes in electrostatic character. It would be of interest to examine whether this short amino terminal conserved region is involved in the interaction of gp1 with primer protein, or other unknown factors.

In conclusion, we isolated a series of missense mutants of gene 1 utilizing its growth inhibitory effect in E. coli as a selection index. Although the inhibitory effect was an artificial phenomenon, most of the isolated mutants had substitutions of amino acid residues which were strongly conserved among the orthologs of gp1. Moreover, many of these substitutions affected features of the sequences which were conserved and/or had been assumed to be involved in known properties of gp1. From these observations, the substitutions were strongly expected to affect authentic functions of gp1. The isolated mutants would be useful for investigating the properties of gp1, including self-association and interaction with other factors (e.g., the cellular membrane, φ29 DNA replication proteins, unknown host factors, and mRNAs).

We thank Ms Yaoko Ikeda and Prof. Hideo Hirokawa, for providing some draft nucleotide sequences of phage M2 genes. This study was partially supported by a Grant-in-Aid for Scientific Research (C) 10660098 (to O. M.) from the Ministry of Education, Culture, Sports, Science and Technology, Japan.