Functional Analysis of the Superfamily 1 DNA Helicases Encoded by Mycoplasma pneumoniae and Mycoplasma genitalium

The DNA recombination and repair machinery of Mycoplasma pneumoniae is composed of a limited set of approximately 11 proteins. Two of these proteins were predicted to be encoded by neighboring open reading frames (ORFs) MPN340 and MPN341. Both ORFs were found to have sequence similarity with genes that encode proteins belonging to the DNA helicase superfamily 1 (SF1). Interestingly, while a homolog of the MPN341 ORF is present in the genome of Mycoplasma genitalium (ORF MG244), MPN340 is an M. pneumoniae-specific ORF that is not found in other mycoplasmas. Moreover, the length of MPN340 (1590 base pairs [bp]) is considerably shorter than that of MPN341 (2148 bp). Examination of the MPN340-encoded amino acid sequence indicated that it may lack a so-called 2B subdomain, which is found in most SF1 DNA helicases. Also, the MPN340-encoded amino acid sequence was found to differ between subtype 1 strain M129 and subtype 2 strain FH at three amino acid positions. Both protein variants, which were termed PcrAs M129 and PcrAs FH, respectively, as well as the MPN341- and MG244-encoded proteins (PcrAMpn and PcrAMge, respectively), were purified, and tested for their ability to interact with DNA. While PcrAMpn and PcrAMge were found to bind preferentially to single-stranded DNA, both PcrAs M129 and PcrAs FH did not demonstrate significant DNA binding. However, all four proteins were found to have divalent cation- and ATP-dependent DNA helicase activity. The proteins displayed highest activity on partially double-stranded DNA substrates carrying 3′ single-stranded extensions.


Introduction
Mycoplasma pneumoniae and Mycoplasma genitalium are genetically closely related human pathogens that are classified within the bacterial class of Mollicutes. These bacteria represent the smallest known self-replicating organisms. It is generally accepted that the Mollicutes have evolved from a Gram-positive ancestor by a gradual, but significant, reduction in genome size and gene content. Consequently, the genomes of M. pneumoniae (strain M129) and M. genitalium (strain G37) are small in comparison to those of bacteria from other classes (816 kb and 580 kb, respectively) [1][2][3]. Also, various biochemical pathways in these bacteria are either lacking or orchestrated by a limited set of enzymes. Among the mycoplasmal pathways that appear to be significantly less convoluted than that of model bacterium Escherichia coli is the DNA recombination and repair (DRR) system. In spite of the important functions that this system may have in both replication and (antigenic) variation of pathogenic mycoplasmas (see Vink et al. for a recent review [4]), it is yet unknown how DRR is actually achieved by these 'minimal' bacteria. From in silico analyses of mycoplasma genomes it was predicted that they possess the most compact set of recombination-associated genes of all known bacteria, consisting of approximately 11 genes [5,6]. These genes have the capacity to code for proteins putatively involved in homologous DNA strand transfer (RecA and SSB), Holliday junction (HJ) branch migration and resolution (RuvA, RuvB and RecU), nucleotide excision repair (UvrA, UvrB, UvrC, and PcrA) and base excision repair (Nfo endonuclease IV). In comparison to other bacteria, however, M. pneumoniae and M. genitalium appear to lack homologs of several other enzymes and/or enzymatic pathways involved in DRR. Specifically, enzymes such as LexA and others related to the SOS response are absent. In addition, both mycoplasmas lack RecBCD, AddAB, RecQ, RecJ and RecFOR [5,6]. Nonetheless, despite the apparent limitations of their DRR machineries, homologous DNA recombination events were found to occur in both M. pneumoniae and M. genitalium [7][8][9][10][11].
To understand how DRR is executed and regulated in M. pneumoniae and M. genitalium, we have previously initiated the characterization of the putative recombination proteins from these species. As of yet, the in vitro activities have been determined of the SSB protein from M. pneumoniae [12] and the RecA [13], RecU [14,15], RuvB [16] and RuvA proteins [17,18]) from both M. pneumoniae and M. genitalium. Surprisingly, in spite of the high level of sequence conservation between these bacteria, significant differences were found in the activities of some of their orthologous recombination proteins. Most notably, M. pneumoniae was found unable to express a functional RecU protein, whereas M. genitalium RecU is a very potent Holliday junction resolvase [14]. In addition, functional differences were noted between the RuvA orthologs and RuvB orthologs from M. pneumoniae and M. genitalium [16,17].
To increase the understanding of the functionality of the 'minimal' DRR machinery of the pathogenic mycoplasmas, we have focused our attention on the M. pneumoniae and M. genitalium ORFs that share sequences with genes encoding PcrA/Rep/ UvrD-like DNA helicases. These enzymes belong to the superfamily 1 (SF1) of DNA helicases, and function as helicases or nucleic acid translocases in almost every aspect of the nucleic acid metabolism, such as DNA repair and the replication of specific plasmids [19][20][21][22][23][24]. While Rep and UvrD are found in Gramnegative bacteria, PcrA is the SF1 DNA helicase that is encoded by bacteria belonging to the Firmicutes and Mollicutes classes. The biological relevance of the PcrA helicases has previously been demonstrated for two Gram-positive species, i.e. Bacillus subtilis and Staphylococcus aureus. The PcrA orthologs from these species (PcrA Bsu and PcrA Sau , respectively) were both shown to be essential for cell growth [25][26][27].
Although most Firmicutes and Mollicutes species (including M. genitalium) possess a single pcrA gene, M. pneumoniae harbors two neighboring ORFs encoding PcrA homologs, i.e. MPN340 and MPN341. As only the latter ORF is conserved between M. pneumoniae and M. genitalium, MPN340 represents an M. pneumoniaespecific ORF. In this study, we have characterized and compared the in vitro activities of all PcrA(-like) proteins encoded by both M. pneumoniae and M. genitalium.

Materials and Methods
Cloning of the MPN340, MPNE_0394, MPN341 and MG244 ORFs Bacterial genomic DNA was purified from cultures of M. genitalium strain G37 (ATCCH no. 33530 TM ) and M. pneumoniae strains M129 (ATCCH no. 29342 TM ) and FH (ATCCH no. 15531 TM ), using previously described procedures [12]. Before cloning of ORFs MPN340 and MPNE_0394 from M. pneumoniae strains M129 and FH, respectively, a TGA codon within these ORFs was changed into a TGG codon using a PCR-based mutagenesis method [14]. In this procedure, the products from two separate PCRs (one with primers pET-Fw and Mutation-Rv, and another with primers Mutation-Fw and pET-Rv; supporting Table S1) were mixed, and subjected to a PCR with primers pET-Fw and pET-Rv. The resulting PCR product was digested with NdeI and BamHI (for which cleavage sites are present within the sequences of pET-Fw and pET-Rv, respectively), and ligated into NdeIand BamHI-digested vectors pET-11c and pET-16b (Novagen). The resulting plasmids were used as templates in PCRs with primers pMAL-c_Fw and pMAL-c_Rv. The amplified fragments were then digested with EcoRI and PstI and cloned into EcoRI-and PstI-digested vector pMAL-c (New England Biolabs).
Both MPN341 and MG244 were found to contain five TGA codons. These codons were changed into TGG codons in a similar fashion as described above, using a set of overlapping PCR products. These products were generated using the primers listed in supporting Table S1. For MPN341, overlapping PCR fragments were generated with the following primer pairs: (i) 341pETfw and 341Mut1rv, (ii) 341Mut1 and 341Mut2rv, (iii) 341Mut2 and 341Mut3rv, (iv) 341Mut3 and 341Mut4rv, (v) 341Mut4 and 341Mut5rv, and (vi) 341Mut5fw and 341pETrv. The outer primer pair 341pETfw and 341pETrv was employed to amplify the complete, modified MPN341 ORF. The resulting PCR product was digested with NdeI and BamHI and ligated into NdeIand BamHI-digested vectors pET-11c and pET-16b. The resulting plasmids were used as templates in PCRs with primers 341pMALcfw and 341pMALcrv. The amplified fragments were then digested with XbaI and PstI and cloned into XbaI-and PstIdigested vector pMAL-c.
ORF MG244 was modified and cloned in a similar fashion as described above for MPN341, using the MG244-specific primers listed in Table S1. Like the other three ORFs, the modified MG244 ORF was cloned in vectors pET-11c, pET-16B and pMAL-c. The pET-11c-and pET-16B-derived plasmids were employed for expression of native and poly-histidine (H 10 )-tagged proteins, respectively, in Escherichia coli strain BL21(DE3)pLysS. The pMAL-c-derived plasmids were used for the expression of maltose-binding protein (MBP)-fused proteins in E. coli strain XL1-Blue. The integrity of all DNA constructs used in this study was checked by dideoxy sequencing, as described before [13].

Generation of Plasmids Encoding Point Mutants of PcrAsM129
Expression constructs encoding point mutants of PcrA s M129 (K29R and K29A) were constructed using a mutagenesis procedure similar to that described above for modification of the TGA codons. The primers used for generation of the construct encoding mutant K29R were 340FW_K.R and 340RV_K.R (Table S1). The construct encoding K29A was generated using primers 340FW_K.A and 340RV_K.A. The final PCR products were cloned into vector pMAL-c.

Expression and Purification of the PcrA Proteins
The proteins encoded by MPN340, MPNE_0394, MPN341 and MG244 were expressed in E. coli as native proteins, and as H 10 -and MBP-tagged proteins. The native and H 10 -tagged proteins were found to be expressed exclusively in an insoluble form (under various culturing conditions). However, the proteins could readily be expressed and purified as MBP-tagged proteins. These proteins, as well as negative control protein MBP-bgalactosidase-a (hereafter named MBP), were purified using a previously described procedure [28].

DNA Substrates
The sequences and structures of the oligonucleotide substrates that were used in the DNA binding and DNA helicase experiments are shown in Fig. 1. In each substrate, a single oligonucleotide strand was labeled at its 59 terminus with a fluorescent (6-FAM) group.

Electrophoretic Mobility Shift Assay (EMSA)
Binding of the PcrA proteins to various DNA substrates was carried out in 10-ml volumes and included 20 mM Tris-HCl pH 7.5, 1 mM DTT, 50 ng/ml BSA, 8 nM of substrate DNA and varying concentrations of protein. After incubation for 15 min at room temperature, 1 ml was added of a solution containing 40% glycerol and 0.25% bromophenol blue. Then, the reaction mixtures were electrophoresed through 6% polyacrylamide gels in 16TBE buffer (90 mM Tris, 90 mM boric acid, 2 mM EDTA). Following electrophoresis, the polyacrylamide gels were analyzed by fluorometry, using a Typhoon Trio TM 9200 Variable Mode Imager (GE Healthcare) in combination with the Typhoon Scanner Control v4.0 software (Amersham Bioscience). Images were processed using Quantity OneH 1-D Analysis Software [31].

ATPase Assay
ATPase activity was determined in the presence or absence of wX174 virion DNA (at a final concentration of 1.5 nM) using a bnicotinamide adenine dinucleotide reduced form (NADH)-coupled assay on a VersaMax Tunable Microplate Reader (Molecular Devices), as described before [13,32]. The ATP turnover rates were calculated from the equation: ATPase rate (ATP6min 21 ) = -dOD 340 /dt (OD/min)6K path 21 (mol/OD)6mol 21 PcrA protein, where K path is the molar absorption coefficient for NADH for a given optical pathlength [33]. The rates were corrected for background NADH decomposition of controls performed without protein.

M. pneumoniae and M. genitalium Encode PcrA Homologs
The genome of M. pneumoniae strain M129 contains two neighboring ORFs, MPN340 and MPN341, which were both annotated as genes encoding UvrD-like helicases [1,2]. However, these ORF differs significantly in size; while MPN340 has a length of 1,590 bp, MPN341 measures 2,148 bp. In contrast to M. pneumoniae, M. genitalium only contains a single gene putatively encoding a UvrD-like protein [3]. This gene, MG244, was suggested to represent the ortholog of M. pneumoniae MPN341. To investigate the relationship between these ORFs and similar sequences in the GenBank sequence database, their encoded amino acid sequences were subjected to protein BLAST analysis using the blastp algorithm (http://blast.ncbi.nlm.nih.gov/Blast. cgi?PAGE = Proteins). The MPN340-encoded amino acid sequence displayed the highest similarity with the sequence encoded by ORF MPNE_0394, which is the MPN340 counterpart of M. pneumoniae strain FH (99% identity; Table 1). These sequences were found to differ in only three amino acid residues. Lower similarities were found with the sequences encoded by MPN341 (47% identity) and MG244 (45% identity). Relatively high similarity scores were also found with PcrA sequences from Gram-positive bacteria, including Lactobacillus salivarius and Staphylococcus aureus. To address their relatively strong sequence similarity with PcrA(-like) proteins, the MPN340-encoded proteins from strains M129 and FH were named PcrA s M129 and PcrA s FH , respectively, in which the superscript 's' (short) indicates the relatively short size of these proteins as opposed to the MPN341encoded protein. The latter protein does not differ in sequence between strains M129 and FH and was termed PcrA Mpn . The M. genitalium ortholog of PcrA Mpn was designated PcrA Mge . The sequences of these proteins, as well as those from the PcrAs of L. salivarius and S. aureus, were included in a multiple sequence alignment (Fig. S1). The alignment demonstrated that the MPN340-, MPNE_0394-, MPN341-and MG244-encoded sequences each have features characteristic of proteins belonging to the SF1A group from the SF1 superfamily of DNA helicases [22][23][24]34]. Most notably, these features include seven conserved protein motifs (motifs I, IA and II to VI) that may be involved either in the binding and hydrolysis of ATP or in the binding of DNA. Interestingly, in contrast to PcrA Mpn and PcrA Mge , PcrA s M129 and PcrA s FH lack a counterpart of subdomain 2B, which is one of the four helicase subdomains that have previously been identified in the crystal structures of several PcrA/Rep/ UvrD-like proteins, including PcrA from Bacillus stearothermophilus (PcrA Bst ) and the Rep, UvrD and RecB proteins from E. coli [35][36][37][38][39]. As a consequence of the lack of a 2B subdomain, the theoretical molecular mass of PcrA s M129 and PcrA s FH (60.5 kDa) is significantly lower than that of PcrA Mpn (83.5 kDa) and PcrA Mge (82.0 kDa).
The structural differences between the PcrA s proteins and the PcrA proteins are illustrated schematically in Fig. 2A. Clearly, the absence of the 2B subdomain from the PcrA s proteins is the most significant difference between these proteins and other PcrA(-like) proteins. Although MPN340 and MPN341 are adjacent ORFs in the M. pneumoniae M129 genome (Fig. 2B), two major observations support the notion that the latter ORF is the ortholog of M. genitalium MG244. First, the 2B subdomain is conserved between MPN341 and MG244. Second, the amino acid sequence similarity between PcrA Mpn and PcrA Mge (53% identity; Table 1) is higher than the similarity between the PcrA s proteins and PcrA Mge (45% identity). If the 2B domains are not considered in these sequence comparisons, the sequence similarity between PcrA Mpn and PcrA Mge is even higher (58% identity; Table 1), whereas the similarity between the PcrA s proteins and PcrA Mge is somewhat lower (43% identity).

Purification of the PcrA-like Proteins from M. pneumoniae and M. genitalium
To determine the characteristics of the PcrA-like proteins from both M. pneumoniae and M. genitalium, the proteins were expressed in E. coli, either in their native forms or fused to maltose-binding protein (MBP) or a poly-histidine (H 10 ) tag. The native and H 10tagged versions of the proteins were either expressed at very low levels or in a solubility state that precluded their purification. In contrast, the MBP-tagged versions of the four proteins were expressed in a soluble form and could readily be purified using the same protocol for each protein [40]. Although these proteins carry an N-terminal tag, the use of this tag has several advantages. First, the activity of the proteins can be compared to that of a negative control protein (MBP-b-galactosidase-a [MBP]), which has been purified using the same method. Second, the amylose affinitybased purification protocol that is employed for the MBP-fused proteins is both efficient and rapid, which is beneficial to the proteins' activity and stability [14][15][16][40][41][42]. Third, the activities of the purified proteins (and mutants thereof; see below) can be compared directly, and are not influenced by differences in purification procedures. Moreover, other SF1A family members, such as the PcrA proteins from Bacillus anthracis (PcrA Ban ) [43,44], Bacillus cereus (PcrA Bce ) [44], Staphylococcus aureus (PcrA Sau ) [20,45] and Streptococcus pneumoniae (PcrA Spn ) [46], as well as the Rep protein from E. coli (Rep Eco ) [47], were previously reported to be active as N-terminally tagged fusion proteins in vitro. We therefore anticipated that an N-terminal tag would not interfere in the analysis and comparison of the in vitro activities of the PcrA-like proteins from M. pneumonia and M. genitalium. The purified proteins, which will be referred to without the prefix 'MBP', were 90-95% pure (Fig. 2C). Some of the protein preparations, including those of MBP and PcrA Mge (Fig. 2C, lanes 2 and 8), contained minor products having a lower molecular mass than the full-length proteins. Such products are regularly observed for MBP fusions, particularly when these proteins are relatively large [14,40,42]. To investigate the nature of these products, the two minor species from the PcrA Mge preparation (indicated by the asterisks in Fig. 2C, lane 8), as well as the full-length protein, were excised from an SDS-polyacrylamide gel and subjected to MALDI-TOF mass spectrometry. In each of these protein species, amino acid sequences were identified that were contained within either MBP or PcrA Mge (data not shown) demonstrating that the minor, lower molecular mass species in the PcrA Mge preparation are the products of either premature translation termination or proteolytic breakdown of the MBP-fused PcrA Mge .

DNA-binding Activity of the PcrA-like Proteins
The DNA-binding properties of the PcrA-like proteins were investigated in an electrophoretic mobility shift assay (EMSA), using four different fluorescently labeled DNA substrates, i.e. a single-stranded (ss) oligonucleotide (oligonucleotide 1 from Fig. 1A), a double-stranded (ds), blunt-ended oligonucleotide (substrate 'c' from Fig. 1B), a ds oligonucleotide with a 39 24nucleotide (nt) ss terminus (substrate 'a'), and a ds oligonucleotide with a 59 24-nucleotide (nt) ss terminus (substrate 'b'). As shown in Fig. 3A, both PcrA Mpn (lanes [11][12][13] and PcrA Mge (lanes [14][15][16] bound efficiently to the ss oligonucleotide in a protein concentration-dependent fashion. At PcrA Mpn concentrations of 22 nM (lane 11) and 67 nM (lane 12), a single protein-DNA complex (complex I) was observed. At 200 nM however, another, slower migrating complex (complex II) was seen, in addition to other complexes that were too large to enter the gel (lane 13). Although the PcrA Mge protein gave rise to similar protein-DNA complexes (lanes 14-16), these complexes were more diffuse than those formed with PcrA Mpn . Binding of PcrA Mpn and PcrA Mge was also observed to substrates carrying ss extensions ( Fig. 3C and 3D). However, while PcrA Mpn and PcrA Mge formed discrete complexes with substrate 'b' (complex III and IV, respectively, in Fig. 3D), most of the complexes that were formed with substrate 'a' did not enter the gel (lanes [11][12][13][14][15][16]. In contrast to the (partially) ss DNA substrates, the ds, blunt-ended substrate was bound very inefficiently by PcrA Mpn and PcrA Mge (Fig. 3B). Contrary to PcrA Mpn and PcrA Mge , PcrA s M129 did not demonstrate significant binding to any of the four DNA substrates. However, PcrA s FH did show some complex formation with the ssDNA substrate (Fig. 3A, lanes 8-10), albeit that this activity was considerably lower than that observed for PcrA Mpn and PcrA Mge . The negative control protein, MBP, did not show binding to any of the DNA substrates ( Fig. 3A-D, lanes 2-4). To test the putative DNA helicase activities of the purified proteins, they were incubated with DNA substrates 'a', 'b' and 'c', in the presence of ATP and Mg 2+ (Fig. 3E-G). Despite the inability of PcrA s M129 to bind to DNA in EMSA (as shown above), this protein was capable of unwinding both substrate 'a' (Fig. 3E, lane 4) and 'b' (Fig. 3F, lane 4). Similar activities were displayed by the other three PcrA-like proteins ( Fig. 3E and 3F, lanes 7-9). However, these proteins did not display significant unwinding activity on blunt-ended substrate 'c' (Fig. 3G). As expected, negative control protein MBP did not show DNA unwinding activity on any of the three substrates ( Fig. 3E-G, lane 3).
As additional (negative) control proteins, we purified two point mutants of PcrA s M129 , which carry either a Lys to Arg mutation (in mutant K29R) or Lys to Ala mutation (in mutant K29A) at position 29 of the protein (Fig. 2C, lanes 4 and 5). Lys29 is an amino acid residue that is predicted to form an invariant and essential part of conserved motif I of PcrA s M129 (Fig. S1), and may be involved in nucleotide cofactor binding. As shown in Fig. 3E-G  (lanes 5 and 6), both K29R and K29A did not display significant DNA unwinding activity on any of the DNA substrates used in this study. This result underlines the importance of amino acid residue Lys29 in the DNA unwinding activity of PcrA s M129 and also excludes the possibility that the activities observed in the DNA  We conclude that PcrA s M129 , PcrA s FH , PcrA Mpn and PcrA Mge each possess DNA helicase activity, and are capable of unwinding dsDNA substrates carrying either a 39 or 59 ss terminus. Thus, these proteins not only share sequence similarity with PcrA proteins from Gram-positive bacteria, but also in vitro DNA helicase activity.

Reaction Requirements of the DNA Helicase Activities of the PcrA Proteins
As expected, the DNA helicase activity of the four PcrA proteins was found to be temperature-, Mg 2+ -and ATP-dependent (Fig.  S2, and data not shown). Optimal activities of the proteins were observed at temperatures of 30-37uC, and at Mg 2+ and ATP concentrations of 0.5-1 mM and 0.5-2.5 mM, respectively (Fig.  S2).
The unwinding activity of PcrA s M129 on substrate 'a' (at 4 nM) could already be detected at a protein concentration of 0.8 nM, reaching optimal levels (.90% unwinding of the substrate) at concentrations $12 nM (Fig. 4A). Using 100 nM of PcrA s M129 , optimal levels of unwinding of substrate 'a' (at 4 nM) were reached within 2 min of incubation ( Fig. 4B and 5A). Similar characteristics were recorded for PcrA s FH and PcrA Mge (Fig. 5 and Fig. 6). However, PcrA Mpn was required at a ,4-fold higher concentration (,50 nM) than the three other PcrAs in order to reach ,90% unwinding of substrate 'a' within 5 min of incubation (Fig. 5C).
From the time series experiment shown in Fig. 6, we estimated the time taken by the PcrA proteins to displace 50% of the DNA substrates. These data were subsequently converted to relative rates of DNA unwinding in a similar fashion as described by Soultanas and coworkers [48]. While PcrA s FH was found to have the highest rate of DNA unwinding, considerably lower rates were observed for PcrA Mpn and PcrA s M129 (Table 2). It is also evident from Table 2 and Fig. 5 that each of the four proteins displayed a higher rate of unwinding of substrate 'a' than of substrate 'b'. This indicated that substrates with a 39 ss protruding end are more efficiently unwound by the PcrA proteins than substrates with a 59 ss extension. This finding was not influenced by the position of the fluorescent label on the substrates, as substrates carrying a 39 6-FAM label where unwound with similar efficiencies as their 59 6-FAM-labeled counterparts (Fig. 7A).
The notion that DNA substrates with 39 ss extensions are more efficiently unwound by the mycoplasmal PcrA proteins than substrates with either 59 ss extensions or blunt ends, was corroborated by experiments in which additional oligonucleotide substrates were included. Specifically, a three-armed DNA substrate with a 39 ss extension was unwound more efficiently than a similar substrate with a 59 ss extension (Fig. 7B, compare  lanes 2 and 4). Moreover, a branched DNA substrate carrying four ds ends was not detectably unwound by PcrA s FH (Fig. 7B, lane 6), similar to what was reported above for the linear, blunt-ended substrate 'c' (Fig. 3G).
The DNA Helicase Activity of the PcrA Proteins is Dependent on ATP Hydrolysis As described above, the DNA unwinding activity of the four PcrA proteins was dependent on the presence of ATP in the reaction; in the absence of ATP, or in the presence of ATPcS, the proteins were inactive. Nevertheless, dATP could efficiently replace ATP as an essential nucleotide cofactor (Fig. S2). To investigate the ability of the PcrA proteins to consume ATP, an NADH-coupled ATPase assay was performed. As shown in Table 3, all four PcrA proteins were found to hydrolyze ATP. Interestingly, PcrA Mpn exhibited the highest ATPase rate of all four proteins (783.4635.8 min 21 in the presence of ssDNA). Importantly, the ATPase rates were strongly induced by the presence of ssDNA in the reaction. While the basic ATPase rate of PcrA Mpn was stimulated ,75-fold, the basic ATPase rates of the PcrA s proteins were stimulated ,30-fold by ssDNA. Only a ,12-fold ssDNA-induced increase was observed in the ATPase rate of PcrA Mge . However, this protein displayed the highest basic rate of ATP hydrolysis (29.062.3 min 21 ) of all four proteins.

Discussion
The representatives of the Mollicutes class of bacteria have very compact genomes. Likewise, the set of genes encoding enzymes and proteins involved in the replication, recombination and repair of DNA is small in these organisms [4][5][6]. It was therefore surprising to find that the genome of M. pneumoniae harbors two consecutive genes that have the potential to encode PcrA-like DNA helicases, whereas M. genitalium only possesses one of such  The data in Fig. 6 were fitted to estimate the time required to displace 50% of either substrate 'a' or substrate 'b'. The results were then expressed as relative rates by comparison with the results obtained from the most efficient DNA helicase reaction, which included PcrA s FH and substrate 'a'. The procedure used to calculate the relative rates was described by Soultanas et al. [48]. doi:10.1371/journal.pone.0070870.t002 genes. In this study, we showed that both putative helicases of M. pneumoniae, i.e. PcrA s M129 (or PcrA s FH in subtype 2 strains) and PcrA Mpn , possess Mg 2+ -and ATP-dependent DNA helicase activity. A similar activity could be attributed to the PcrA Mge protein from M. genitalium. Based on primary structure analysis as well as protein (sub)domain predictions, we proposed that PcrA Mge represents the ortholog of PcrA Mpn . SF1 DNA helicases can be classified as SF1A or SF1B helicases [24,34,49]. Proteins belonging to the first group primarily have a 39R59 (or 'type A') polarity, whereas SF1B helicases have a 59R39 ('type B') polarity. A protein with a clear SF1A signature is the PcrA protein from B. stearothermophilus [50]. Other PcrA proteins, however, were shown to have a bipolar nature, by displaying similar helicase activities in the 39R59 and 59R39 directions. These proteins include PcrA Sau [45], PcrA Ban [43] and PcrA Spn [46]. The four PcrA proteins from M. pneumoniae and M. genitalium were also found to have bipolar activities, albeit that substrates with a 39 ss extension were unwound somewhat more efficiently by these proteins than substrates with a 59 ss extension.
Two of the mycoplasma PcrA proteins, i.e. PcrA Mpn and PcrA Mge , were found to resemble PcrA Bst also with respect to DNA binding characteristics; these proteins each prefer to bind to substrates containing ssDNA [50]. In contrast, PcrA Spn , PcrA Sau and PcrA Ban are unable to stably interact with ssDNA, and prefer to bind to substrates containing hairpins and/or partially ds regions [43,45,46,50].
The most notable observations from this study, however, concern the characteristics of the PcrA s proteins, which may be regarded as the first naturally occurring representatives of the SF1 family that lack an entire 2B subdomain. While the 2B subdomain sequences can vary significantly in size and sequence among SF1 helicases, the smallest 2B subdomains reported thus far are those of the HelD proteins from E. coli and B. subtilis (with lengths of 79 and 89 amino acids, respectively) [24,34,49,51,52]. Despite this apparent structural deficiency, both PcrA s proteins were found to be highly active DNA helicases. However, this finding is not without precedent, because the dispensability of the 2B subdomain has previously been shown for another member of the SF1A protein family, Rep Eco [47]. In fact, a Rep Eco mutant deleted of the 2B region (RepD2B) displayed a faster rate of DNA unwinding than did the WT protein [47]. Based on this observation, it was suggested that the 2B subdomain might play a role in (i) (auto)regulation of the DNA helicase activity of Rep Eco , and (ii) the interaction with other, regulatory proteins [47,53]. If this notion would also apply to the PcrA proteins from M. pneumoniae, this species would express one PcrA protein of which the activity can be regulated (PcrA Mpn ), either intra-or intermolecularly, and a second PcrA protein that displays constitutive activity (PcrA s M129 or PcrA s FH ). Another unique property of the PcrA s proteins as opposed to other PcrA proteins (including PcrA Mpn and PcrA Mge ) was the inability to form stable protein-DNA complexes in EMSA. It is possible that this deficiency is the consequence of the lack of a 2B subdomain, and that this subdomain plays a role in stable DNA binding. In agreement with this notion, the 2B subdomain from PcrA Bst was found to interact with the ds part of a small DNA substrate with a 39 ss tail in PcrA Bst -DNA crystal structures [36]. The introduction of specific point mutations in the 2B region of PcrA Bst resulted in proteins (K419A, T426A and K456A) that were defective in dsDNA binding and helicase activity [48].
A crucial question that remains to be addressed is the in vivo role of the mycoplasmal PcrA DNA helicases. As mentioned above, the PcrA proteins from B. subtilis and S. aureus are essential for cell growth and viability [25][26][27]. Moreover, PcrA Bsu was shown to restore UV resistance in a uvrD mutant of E. coli, and to play a role in the resolution of stalled replication forks [25,26].
Interestingly, the lethality of a pcrA null mutation in B. subtilis could be suppressed by additional mutations in genes recF, recL, recO and recR, which belong to the same complementation group [26]. While the function of RecL is unknown, the RecF, RecO and RecR (RecFOR) proteins assist the major recombinase RecA in binding to ssDNA, and thereby initiate and catalyze homologous DNA recombination. Thus, the lethality of the pcrA null mutant is an indirect phenomenon, which can be overcome by inactivation of the RecFOR-dependent recombination pathway. It was suggested that RecFOR induces an unusually high (and thereby toxic) level of recombination when PcrA is absent. This notion was supported by the observation that B. subtilis strains become hyperrecombinogenic (,15 times higher than the WT strain) when the amount of PcrA is reduced by a factor of 10 [26]. This high level of recombination was dependent on both RecA and the RecFOR pathway [26]. Thus, PcrA Bsu appears to have an anti-recombinogenic effect.
In contrast to the situation in Gram-positive bacteria, the PcrA proteins are not essential in M. pneumoniae and M. genitalium. By  using a global transposon mutagenesis protocol, several pcrA transposon insertion mutants of M. genitalium were obtained [54]. In addition, we recently identified M. pneumoniae M129 mutant strains carrying transposon insertions in either ORF MPN340 or MPN341 (E. Spuesens, C. Vink, J. Stülke, unpublished data). The non-essential nature of these genes in M. pneumoniae and M. genitalium is not surprising, however, because (i) the lethality of the pcrA genes in Gram-positive bacteria is dependent on a functional RecFOR pathway, and (ii) the RecFOR pathway is absent in M. pneumoniae and M. genitalium [6,17]. This raises the question why PcrA function is maintained in these mycoplasmas during evolution, whereas RecFOR has been lost. Another crucial question is whether the PcrA proteins from M. pneumoniae and M. genitalium have a similar anti-recombinogenic function as do their gram-positive counterparts, despite the absence of a RecFOR pathway in the mycoplasmas. In this regard, it is tempting to speculate on a putative role of the PcrA proteins in homologous recombination between repetitive DNA elements in M. pneumoniae and M. genitalium. These recombination processes were found to induce antigenic variation of major bacterial surface proteins (for a review, see [4]). It was shown for M. genitalium that these events depend upon the function of the RecA protein [55]. Interestingly, the frequency of recombination between repetitive DNA elements is higher in M. genitalium than in M. pneumoniae. This difference was previously hypothesized to be caused by differences in the specific activities of the RuvA, RuvB and RecU proteins from these species [4,14,16,17]. However, it is also possible that the M. pneumoniae PcrA s proteins, which do not have an ortholog in M. genitalium, play an inhibitory role in the recombination between repetitive DNA elements. This notion is currently being tested by monitoring (the changes in) the sequences of the repetitive elements during propagation of the M. pneumoniae MPN340 null mutant in culture.
The involvement of an SF1 family member in DNA recombination-induced antigenic variation is not unprecedented. In Gram-negative bacterium Neisseria gonorrhoeae, a system of antigenic variation is operational that is similar to that in M. genitalium and M. pneumoniae. This system, which is termed pilin antigenic variation, depends on the function of a set of proteins that not only includes RecA, RuvA, RuvB, and RuvC, but also Rep [4,56]. However, in contrast to the suggested negative effect of PcrA s on homologous DNA recombination in M. pneumoniae, the Rep protein of N. gonorrhoeae was found to have a positive influence on the overall efficiency of pilin antigenic variation [56].
Finally, it is important to consider that this study was performed exclusively with MBP-fused proteins. We were constrained to use these fusion proteins because the native versions of the PcrA proteins were either expressed at very low levels, or could not be purified in their native state due to solubility problems. While the MBP tag may theoretically influence the activity of the attached protein, numerous studies are available showing that this tag is functionally inert, in particular concerning the function of DNAinteracting fusion partners, such as the integrase proteins from HIV-1, HIV-2 and feline immunodeficiency virus [28,57], the M. genitalium HJ resolvase RecU Mge [14,15], the E. coli HJ resolvase RuvC [58] and the RuvB helicases from M. pneumoniae and M. genitalium [16]. Moreover, several other SF1 proteins have previously been shown to be fully active as variants containing an N-terminal polyhistidine-tag. These proteins include PcrA Ban [43,44], PcrA Bce [44], PcrA Sau [20,45,59] and PcrA Spn [46]. Also, it was shown that the activities of a Rep Eco mutant, RepD2B, differed only marginally (in efficiency) from those of an N-terminally tagged variant of this protein (+HRepD2B) [47]. Nonetheless, attempts to obtain non-tagged variants of the mycoplasma PcrA proteins are ongoing in our laboratory.
In conclusion, we have determined the in vitro activities of the SF1 proteins encoded by M. pneumoniae and M. genitalium, and found each of these proteins to act as DNA helicases in vitro. The main challenge of future studies will be to determine the in vivo roles of these proteins, in particular in light of the lack of a RecFOR pathway in both M. pneumoniae and M. genitalium. We will also aim to address the question if, and how, the two different PcrA proteins from M. pneumoniae interact, either physically or functionally. Clearly, the answers to these questions will shed further light on the functionalities of the DNA repair and recombination pathways in bacteria with a strongly reduced, or 'minimal' [54], genome. Predicted domains and motifs of the PcrA(-like) proteins are indicated above and below the alignment and are predominantly based on the crystal structure of the PcrA protein from Bacillus stearothermophilus [35]. The multiple alignment was performed using Clustal W (http://www.ebi.ac.uk/Tools/msa/clustalw2/). The program BOXSHADE 3.21 (http://www.ch.embnet.org/ software/BOX_form.html) was used to produce white letters on black boxes (for amino acid residues that are identical in at least three out of six sequences) and white letters on grey boxes (for similar residues). The three residues that differ between PcrA s M129 and PcrA s FH are indicated by red dots above the sequences.  S1 Oligonucleotide primers used for the cloning of ORFs encoding the PcrA-like helicases from M. pneumoniae and M. genitalium. a The different ORFs are from M. pneumoniae strains M129 (MPN340 and MPN341) and FH (MPNE_0394), and from M. genitalium strain G37 (MG244). b Restriction endonuclease recognition sites that were incorporated in the primer sequences for cloning purposes are indicated in italics. The TGG codons (or the complementary sequences CCA) that were incorporated in the oligonucleotides in order to modify the TGA codons within the native ORFs, are underlined. (DOC)