Myxobacterial genomes are normally larger than nine Mbp in size, but with some exceptions of approximate five-Mbp-genomes in Anaeromyxobacter strains. The G+C contents of the sequenced myxobacterial genomes ranged from 67.4% in Enhygromyxa salina DSM 15201 to 74.9% in Anaeromyxobacter dehalogenans 2CP-C. All the sequenced myxobacterial genomes, no matter in large size or small size, contained two dnaE genes (Table S1). Phylogenetic analysis indicated that the two protein sequences encoded by the duplicate myxobacterial dnaE genes were clustered separately with determined DnaE1 and DnaE2 of other bacteria, forming the DnaE1 and DnaE2 groups (Figure S1). The single DnaE gene of E. coli was in the DnaE1 group, clearly distant from the DnaE2 group; and the two DnaE-group trees had highly similar topologies. Consistent with the previous results (Timinskas et al., 2014; Wu et al., 2014), the emergence of myxobacterial dnaE2 was derived from an early duplication event of the primordial dnaE. High conservation but great divergence suggested that the duplicate DnaE polymerases were both important for myxobacterial cells, but functioned divergently.

M. xanthus DK1622 is the model strain of myxobacteria. In the genome of DK1622, two genes, i.e., MXAN_3982 and MXAN_5844, were predicted to be dnaE, encoding for the alpha subunit of DNA polymerase III (Goldman et al., 2006). The proteins encoded by the two dnaE genes contain 1185 and 1013 amino acids (MXAN_5844 and MXAN_3982), respectively. These two proteins were phylogenetically distant, having 48% similarity and 30% identity of their amino acid sequences. According to their phylogenetic locations (Figure S1), the MXAN_3982 gene was designated as dnaE2, and the MXAN_5844 gene was dnaE1.

To evaluate their essentiality, we made deletion mutations of the dnaE1 and dnaE2 gene in M. xanthus DK1622, respectively. While dnaE2 was deletable, producing the mutant of YL1601, the dnaE1 gene could not be deleted, which was further confirmed by failed attempts to delete the dnaE1 gene in a dnaE2-overexpressing mutant (see below). Compared with that of the wild-type strain DK1622, the OD values of the dnaE2 deletion mutant YL1601 were smaller than the wild type at different time points during the exponential growth stage in CTT growth medium (two-way ANOVA, P < 0.01; Figure 1A). To confirm the function of dnaE2 in growth, we further constructed a mutant by inserting the dnaE2 gene following its own promoter at the attB site in the YL1601 genome, forming the YL1606 mutant. As expected, the complementary strain recovered the delayed growth in YL1601, showing almost the same growth curve as the wild type strain DK1622 (Figure 1A).

To confirm whether the DnaE2 protein was an error-prone enzyme, we constructed two more mutants by deleting two DNA-repair genes of mutL (MXAN_4026) and mutS (MXAN_3897) from DK1622, respectively, producing the mutants of YL1602 and YL1603. The wild type strain DK1622 and the three mutants of YL1601, YL1602 and YL1603 were cultivated on CTT growth medium supplemented with nalidixic acid (40 μg/ml) to evaluate their production abilities of resistant mutations. It is known that the inhibition of nalidixic acid on DNA repair can be reversed by the MutL and MutS enzymes of the DNA repairing system in E. coli cells, thus the deletion mutant of mutL or mutS will produce more genome mutations than the wild type strain during the chromosome replication process (Sniegowski et al., 1997). In M. xanthus, the mutation rate of DK1622 was approximately 3.69 × 10⁻⁸ per nucleotide in the presence of nalidixic acid; whereas the YL1602 and YL1603 mutants produced mutations of 5.41 × 10⁻⁷ and 1.77 × 10⁻⁶, respectively (Table 1). The mutation rates of the mutL and mutS mutants were both significantly higher than that of DK1622 (t-test, P < 0.01). This result was consistent with that reported in E. coli (Sniegowski et al., 1997), suggesting the method was also reliable in M. xanthus for the evaluation of mutation rates. As expected, the deletion of dnaE2 made the mutant, possessing only dnaE1, to have nearly 10 times lower of mutation rate than that of the wild type strain DK1622, which harbors both dnaE1 and dnaE2 (Table 1). This result confirmed the error-prone characteristics of the DnaE2 protein in M. xanthus cells.

The above results indicated that dnaE1 was essential for the survival of M. xanthus cells, while the error-prone DnaE2 gene was dispensable for cell survival. dnaE2 also involved in cellular growth, but in weak effects and low fidelity. We further assayed effects of the deletion of dnaE2 in the swarming and developmental processes in M. xanthus. Similar to the growth change, the swarming ability of the YL1601 mutant was weakly decreased on 0.4% CTT plate, comparing with that of DK1622 (t-test, P = 0.051), while the motility on 1.5% CTT plate showed no difference between DK1622 and YL1601 (Figure 1B).

When cultivated on the TPM developmental medium, cells of the dnaE2 deletion mutant aggregated irregularly (Figure 1C). From the phenotypes of the mutant on TPM, the dnaE2 gene probably played functions in the early stage of the aggregation in the development process. The sporulation ability of the dnaE2 mutant was decreased to approximately 70% of that of DK1622 after 5 days of incubation on TPM medium (1.49 × 10⁶ ± 9.61 × 10⁴ vs. 2.14 × 10⁶ ± 1.70 × 10⁵; Table 2). If the dnaE2 gene was complemented into the dnaE2 deletion mutant (YL1606), the irregular aggregation phenotype on TPM developmental medium was completely recovered (Figure 1C). This result suggested that the DnaE2 protein also played functions in the early development stage of M. xanthus DK1622 cells. Seemly, the function of the dnaE2 gene was more obvious in the development stage than its function in the growth stage.

Tzeng and Singer found that there is a DNA replication process during the development of fruiting bodies and sporulation in M. xanthus DK1622 (Errington and Wake, 1991; Tzeng and Singer, 2005). Accordingly, there are two processes involving the chromosome replication during the lifecycle of M. xanthus cells: one in the growth stage and the other in the development stage. To determine DnaE2 function in development, we inoculated concentrated cells of DK1622 and YL1601 on the TPM plates, and exposed culture plates with nalidixic acid solution at the final concentration of 20 μg/ml for 1 h at different cultivation time points. After 120 h of incubation, the sporulation rates of the cultures were calculated and compared with that of their respective controls without the exposure of nalidixic acid. Consistent with the previous report (Tzeng et al., 2006), the developmental progress of DK1622 was arrested by nalidixic acid in the early stage of development. After 12 h of incubation, the nalidixic acid exposure had nearly no effect on the sporulation of M. xanthus DK1622 cells (Table 2). In contrast, the sporulation ability of the dnaE2 deletion mutant YL1601 was lower than that of DK1622 after the nalidixic acid exposure at each time point. The developmental arrest in YL1601 by the exposure of nalidixic acid was released after 24 h, which was nearly two-times long of that in DK1622. This result indicated that the deletion of dnaE2 markedly prolonged the process of chromosome replication in development. Thus, together with the dnaE1 gene, the dnaE2 gene involved in the two processes of chromosome replication during the lifecycle of M. xanthus cells; its function in the development progress seemed to be more significant.

There are two possible reasons for the differentiated functions of the dnaE1 and dnaE2 genes in the growth and developmental stages: structure differences and/or expression levels. We further assayed expression levels of the dnaE1 and dnaE2 genes under either the growth or the developmental conditions. Quantitative real-time PCR amplification showed that the expression levels of dnaE1 were greatly higher than that of dnaE2 at different time points under the CTT nutritional growth conditions (Figure 2A). For example, after 12 h of incubation in CTT liquid medium, the expression of dnaE1 was more than 10 times of that of dnaE2. With the increase of incubation time, the expressed products of dnaE1 were gradually decreased and reached to less than one-tenth (8.7%) of the 12-h-level at 48 h of incubation. Similarly, the expressions of dnaE2 were also gradually decreased and were always significantly lower than that of dnaE1 at each time point (t-test, P < 0.01).

The expressions of the two dnaE-genes were in similar patterns on the TPM development plate as they were in the growth conditions (Figure 2B). In contrast, the expressions of dnaE2 in the developmental conditions were even lower than that in the growth conditions. After the deletion of dnaE2, no expression of dnaE2 was detectable in the mutant, and the expression of the dnaE1 gene was lower than that of the wild type strain at 6 h of incubation on the TPM plate (71.5%; Figure 2C). The results suggested that the deficiencies in the fruiting body formation and sporulation in the YL1601 mutant were probably resulted not only from the loss of dnaE2, but also from the decreased expression level of dnaE1.

To further investigate effects of the dnaE2 expression, we introduced a second copy of the gene into the attB site of DK1622 chromosome by the Mx8 integrase in pSWU19 (Wu and Kaiser, 1995), producing the YL1605 mutant (containing a local dnaE2 and an introduced dnaE2). The introduced dnaE2 gene was designed to follow the pilA promoter, a high efficient promoter in M. xanthus cells. Meanwhile, a mutant containing the empty pSWU19 plasmid was also constructed into the attB site of DK1622 chromosome as control (the mutant YL1604). The total expression levels of the two dnaE2 genes were approximately 10 times higher than that of dnaE1 in the YL1605 mutant under either the nutritional (Figure 3A; at 24 h of incubation in CTT; t-test, P < 0.01) or the developmental (Figure 3B; at 6 h of incubation on TPM; t-test, P < 0.01) conditions. In contrast, the expressions of dnaE2 in YL1605 were approximately 500 times and 1500 times higher than that in YL1604 under the nutritional and the developmental conditions, respectively. These results suggested that the overexpression of dnaE2 was mostly resulted from the introduced dnaE2 gene. When cultivated in CTT liquid medium, the OD values of the dnaE2 overexpression mutant YL1605 were higher than the control mutant YL1604 during the early exponential growth stage (Figure 3C). The growth differences were statistically significant (two-way ANOVA, P < 0.01). This result indicated that while the deletion of dnaE2 led to weakly delayed growth, overexpression of the gene could weakly increase cellular growth.

Interestingly, the overexpression of dnaE2 made the YL1605 mutant to have nearly three times higher of mutation rate than that of YL1604 cultivated on CTT plates supplemented with high concentrations of nalidixic acid (1.36 × 10⁻⁷ vs. 4.28 × 10⁻⁸, referred to Table 1). The difference was statistically significant (t-test, P < 0.01). In order to verify whether the overexpression changed the functions of dnaE2 in the development progress, we exposed the TPM cultures of YL1605 and YL1604 mutants with nalidixic acid solution for 1 h at the same time points described above (Table 3). The results showed that the dnaE2-overexpressing mutant also had higher sporulation ability than the control when cells were treated with nalidixic acid in the early stage of development (before 12 h of incubation). Thus, overexpression of the dnaE2 gene was not only able to increase the genome mutation rate, but also speed up the development progress. However, compared with effects of the dnaE2 deletion, overexpression of dnaE2 produced rather weak effects. The dnaE1 gene still played major roles in the chromosome replication.

Imbalanced effects of the deletion and overexpression of dnaE2 suggested that the structural differences of the two DnaE proteins played more important roles for function than their expressions. Structural differences between DnaE2 and DnaE1 have been thoroughly investigated (Warner et al., 2010; McHenry, 2011a,b; Timinskas et al., 2014). To determine whether the essentiality of dnaE1 was due to high expressions in DK1622, we retried to knock out the dnaE1 gene in the dnaE2-overexpressing mutant (YL1605) using the pBJ113Cm-5844 plasmid but failed. This result indicated that dnaE2 could not replace dnaE1 even if the dnaE2 gene was overexpressed, which confirmed that the non-fungibility of DnaE1 was due to the structural differences between DnaE1 and DnaE2. We modeled three-dimensional structures of DK1622 DnaE1 and DnaE2 (Figure 4A), using the threading approach (Yang et al., 2015). The estimated TM-scores are 0.84 ± 0.08 and 0.87 ± 0.07 for the modeled structures of DnaE1 and DnaE2, respectively. In general, the modeled protein structure with a TM-score > 0.5 is acceptable with the correct topology (Yang et al., 2015). The DnaE1 and DnaE2 proteins of M. xanthus DK1622 had highly similar three-dimensional structures, even though their sequences were highly different. The active sites and some of the DNA binding surfaces were also highly conserved in the DnaE1 and DnaE2 polymerases. In addition, the two DnaE proteins were also considerably conserved of their surface amino acid positions (Figure 4A, right panels). However, the DnaE1 polymerase contained five domains, including PHP, Pol3, HhH, OB, and CTD domains, while DnaE2 had four domains, lacking the CTD domain (Figure 4A, left panels), which is important for binding to Pol III τ-subunit (Liu et al., 2013).

In order to get insight into the domain conservation, we estimated the dN/dS values, an indicator of the selective pressure acting on a protein-coding gene, for the duplicated dnaE genes of those sequenced myxobacteria. The dN/dS values of the complete dnaE1 and dnaE2 sequences were 0.125 (95% confidence interval: 0.121–0.129) and 0.172 (95% confidence interval: 0.167–0.178), respectively, suggesting both polymerases were subjected to a strong negative selection pressure for their strict functional constraint. The dN/dS values were close between the Pol3, HhH, and OB domain sequences of the dnaE1 and dnaE2 genes, but had a great difference in the PHP sequence (t-test, P < 0.01, Figure 4B), which has a binding site of the proofreading ε-subunit (Wieczorek and McHenry, 2006; Ozawa et al., 2013). The dN/dS value of the dnaE2 PHP domain was more than twice than that of dnaE1. Accordingly, dnaE1 is more evolutionarily conservative than dnaE2 in myxobacteria. Stano et al. reported that the PHP domain of DnaE-type polymerases had 3′-5′ exonuclease activity (Stano et al., 2006). Similarly, the PHP domain of Mtb DnaE1 retains all amino acids associated with exonuclease activity in M. tuberculosis, but the DnaE2 PHP lost some key residues (Wieczorek and McHenry, 2006; Baños et al., 2008). It was speculated that the loss of conserved residues in the DnaE2 PHP domain affected relative fidelity. The DNA polymerase III complexes constructed with the two DnaE proteins thus probably had divergent functions.

The bacterial strains used in this study are listed in Table S2. The M. xanthus strains were cultivated in Casitone-based CTT medium (Hodgkin and Kaiser, 1977) for growth assays and on TPM agar plate (Kearns et al., 2000) for developmental assays. E. coli strains were routinely grown on Luria-Bertani (LB) agar or in LB liquid broth. M. xanthus strains were incubated at 30°C, and E. coli strains were at 37°C. When required, a final concentration of 40 μg/ml of kanamycin (Km) was added to the solid or liquid media.

We performed the in-frame deletions of the dnaE1, dnaE2, mutS, and mutL genes in M. xanthus strains using pBJ113 plasmid, respectively. The plasmid contains a kanamycin resistant cassette for the first round of screening and a galK gene for the negative screening (Ueki et al., 1996). Briefly, homologous arms of dnaE1, dnaE2, mutS, or mutL were cloned with primers (listed in Table S3) and the products were inserted into the EcoRI/XbaI site of pBJ113. The resulting plasmid were introduced into M. xanthus DK1622 strains via electroporation (1.25 kV, 400 W, 25 mF, 0.1-cm cuvette gap), respectively. The second round of screening was then performed on CTT plates containing 1% galactose (Sigma). The deletion mutants that grew on galactose but were sensitive to kanamycin were identified and verified by PCR amplification and sequencing. Restriction enzymes, DNA ligase, and other DNA enzymes were used according to the manufacturers' recommendations. All fragments were validated by Sanger 3730 sequencing.

The dnaE2 gene and upstream 500 bp sequence that contains the whole promoter of dnaE2 gene was cloned using the primer pair of MXAN_3982_NativeF and MXAN_3982_NativeR. The fragment and pSWU19 plasmid (replacing the tet resistant cassette of pSWU30 with a kanamycin resistant cassette) were digested with the XbaI/EcoRI digestion solution. The two fragments were ligated with T4 DNA ligase, producing the recombinant plasmid pSW3982. The promoter of the pilA gene was cloned using the primer pair of MXAN_3982P_UF and MXAN_3982P_UR, and fused with the dnaE2 gene. The fusion product was inserted into XbaI/EcoRI sites of pSWU19, producing the plasmid pSWp3982.

The dnaE2 fusion construct was transferred by electroporation into M. xanthus DK1622. Individual kanamycin-resistant clones were selected, and the mutants of the dnaE2 transcription fusion were verified by PCR amplification (primers are listed in Table S3) and sequencing.

The mutation rate assay was conducted by screening the nalidixic acid resistant strains according to previous report (Tzeng et al., 2006). Briefly, approximately 5 × 10⁹ cells of M. xanthus strains were placed on CTT agar containing 40 μg/ml nalidixic acid. Nalidixic acid- resistant candidates were counted and mutation rate was calculated using the following formula, r = N_m/N₀, where r is the mutation rate, N_m represents the count of nalidixic acid resistant mutant, and N₀ is the count of cells placed on the screening plate.

Developmental experiments were performed on TPM agar plates (10 mM Tris [pH 7.6], 8 mM MgSO₄, and 1 mM KH₂PO₄ containing 1.5% agar), as described below. Cells were allowed to develop in a humidity chamber at 30°C. When indicated, nalidixic acid solution of the 20 μg/ml concentration was added onto different plates at the time of 0, 6, 12, 18, or 24 h for 1 h, and then the extra solution was poured out. After 120 h cultivation at 30°C. Sporulation was conducted as follows: five colonies were collected and resuspended in liquid CTT media. Sonication was conducted at 200 W for 4 s twice to release myxospores from fruiting bodies. The myxospore suspensions were incubated at 55°C for 2 h to kill the vegetative cells. Then the suspensions were serially diluted and aliquots of 50 μL were used to plate. The sporulation rate was counted after 5 days of cultivation on the CTT plate.

The M. xanthus DK1622 cells were concentrated to approximately 1.75 × 10¹⁰ cells per milliliter. The cell suspension was washed three times using TPM buffer. RNA was extracted immediately with a BIOZOL total RNA extraction reagent (BioFlux) following the manufacturer's instructions. Genomic DNA contamination was removed by using DNA eraser supplied in the PrimeScript RT Reagent kit with gDNA Eraser (TaKaRa). The purified RNA extracts were transcribed reversely to cDNA and stored in aliquots at −80°C. Quantitative real-time PCR was performed in a total reaction volume of 25 μl, containing 250 nM primers, 12.5 μl of SYBR Premix Ex Taq GC mix (TaKaRa), 10.5 μl of RNase-free water (TaKaRa), and 1 μl of a 10-fold-diluted cDNA template. PCR was conducted in a Roche LightCycler 480 sequence detection system, following the program: 3 min at 95°C, followed by 40 cycles of 30 s at 95°C, 30 s at 55°C, and 15 s at 72°C. The gapA gene, encoding for glyceraldehyde-3-phosphate dehydrogenase, was used as the normalization signal. Calibration curves of gapA, MXAN_3982, and MXAN_5844 were generated from a 10-fold dilution of M. xanthus DK1622 genomic DNA. The primer pairs used for each gene are listed in Table S3.

The dnaE1 and dnaE2 genes were extracted from the genome of M. xanthus DK1622 strain. All the dnaE genes used for phylogenetic analysis were extracted according to the Genomic Encyclopedia of Bacteria and Archaea (Wu et al., 2009). These genes were translated into amino acid sequences according to the bacterial standard codon. A multiple sequence alignment of the DnaE1 and DnaE2 proteins was established using the MAFFT program (Katoh and Standley, 2013). The phylogenetic tree was constructed using the PhyML program with the LG substitution model and the four substitution rate categories (Guindon et al., 2010). LG model was regarded as the best substitution model by automatic model selection. Branch support was calculated using the approximate likelihood ratio tests (aLRT SH-like) (Anisimova and Gascuel, 2006). The phylogenetic tree was visualized by iTOL (Letunic and Bork, 2016).

The structures of the DnaE1 and DnaE2 were modeled using the I-TASSER program based on a threading approach (Yang et al., 2015). The evolutionary conservation of amino acid positions in the dnaE1 and dnaE2 sequences was estimated by using the ConSurf algorithm (Ashkenazy et al., 2010). The JTT substitution matrix was used and the computation was based on the empirical Bayesian paradigm. The conservation scale was defined from the most variable amino acid positions (grade 1), which were considered to be evolved rapidly, to the most conservative positions (grade 9), which were considered to be evolved slowly. The sequences and modeled structures of the DnaE1 and DnaE2 were shown in nine-color conservation grades.

The number ratio of nonsynonymous substitutions per nonsynonymous site (dN) to synonymous substitutions per synonymous site (dS), dN/dS, is an indicator of the selective pressure acting on a protein-coding gene. The dN/dS calculation of the dnaE1 and dnaE2 locus was set at the 95% confidence intervals, using the Datamonkey web server (Delport et al., 2010). For the calculation, a multiple sequence alignment of the nucleic acid sequences of myxobacterial dnaE1 and dnaE2 genes, based on codon, was established with TranslatorX (Abascal et al., 2010). Domains of DnaE1 and DnaE2 proteins were extracted by blasting against the PFAM database (Finn et al., 2014).

The statistical analysis was conducted using IBM SPSS Statistics.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.