The Inner Membrane Protein PilG Interacts with DNA and the Secretin PilQ in Transformation

Expression of type IV pili (Tfp), filamentous appendages emanating from the bacterial surface, is indispensable for efficient neisserial transformation. Tfp pass through the secretin pore consisting of the membrane protein PilQ. PilG is a polytopic membrane protein, conserved in Gram-positive and Gram-negative bacteria, that is required for the biogenesis of neisserial Tfp. PilG null mutants are devoid of pili and non-competent for transformation. Here, recombinant full-length, truncated and mutated variants of meningococcal PilG were overexpressed, purified and characterized. We report that meningococcal PilG directly binds DNA in vitro, detected by both an electromobility shift analysis and a solid phase overlay assay. PilG DNA binding activity was independent of the presence of the consensus DNA uptake sequence. PilG-mediated DNA binding affinity was mapped to the N-terminus and was inactivated by mutation of residues 43 to 45. Notably, reduced meningococcal transformation of DNA in vivo was observed when PilG residues 43 to 45 were substituted by alanine in situ, defining a biologically significant DNA binding domain. N-terminal PilG also interacted with the N-terminal region of PilQ, which previously was shown to bind DNA. Collectively, these data suggest that PilG and PilQ in concert bind DNA during Tfp-mediated transformation.


Introduction
Neisseria meningitidis, the meningococcus (Mc), is a human-specific opportunistic pathogen and one of the leading causative agents of meningitis and septicaemia worldwide [1].
Type IV pili (Tfp) are filamentous appendages emanating from the bacterial surface that are required for adherence of bacteria to human cells and for transformation of DNA [2,3]. Biogenesis of Tfp is not well characterized, but several proteins required for pilus assembly, extrusion and retraction have been identified [4]. In Neisseria sp., these include PilE [5],

Strains, plasmids and constructs
Strains, plasmids and constructs employed in the study are listed in Table 1. N. meningitidis strains MC58 and M1080 were grown on blood agar plates in a 5% CO 2 atmosphere at 37°C. E. coli strain ER2566 (New England Biolabs), used for plasmid propagation and recombinant protein expression, was grown in LB medium or on LB agar plates containing kanamycin (50 μg ml -1 ) at 37°C.

Bioinformatics analyses and search for signature motifs
Predictions on the meningococcal MC58 PilG (NP_273382) subcellular location was performed by using the PSORTb service [39], and the sequence was assessed for native disorder prediction by DISOPRED2 [40] and VSL1 [41]. Searches for functional domains or signature motifs in the PilG sequence were performed using the DOLOP [42], PROSITE [43] and Pfam databases [44]. Primary prediction of the PilG secondary structure and the prevalence of alpha-helical elements was performed by using the JPred service [45] and the PSIPRED service [46], while the presence and location of transmembrane helices was predicted by MEMSAT3 [47]. The presence of DNA binding motifs was first assessed by using the ExPASy site [48], whereas the electrostatic charge was calculated by using the charge program from the EMBOSS package [49]. In the search for DNA binding motifs, the PilG sequence was broken down into three parts, divided into the major groups of transmembrane helices and submitted separately to SAM-T06 for sequence-based fold recognition by use of a hidden Markov method [50]. Subsequently, the PilG sequence was submitted to the DP-Bind server [51] for additional prediction of sequence-based DNA binding sites.

Cloning and overexpression of PilG constructs
All DNA manipulations were performed according to standard techniques [52]. Full-length (FL) pilG and partial constructs were PCR-amplified from MC58 genomic DNA by using the primers listed in S1 Table. Each fragment was cloned into the vector pET28b(+) (Novagen) with a C-terminal 6X His-tag, yielding a panel of constructs encoding N-and C-terminal parts of pilG (see Table 1). The recombinant proteins were overexpressed in E. coli ER2566 (New England Biolabs). A schematic representation of the PilG constructs used in this study is given in Fig 1B, while all the PilG constructs made are summarized in S1B Fig. The results for the expression of the partial PilG constructs, also showing the one which did not yield expressed proteins, are summarized in S2 Table. (I) Membrane associated proteins. E. coli ER2566 cells harboring plasmids that encoded PilG FL , PilG 1-256 , PilG 257-410 , PilG 166-410 , PilG K3A , PilG KK12-13AA , PilG E14A , PilG KR15-16AA , Pil-G EEE39-41AAA , PilG RKK43-45AAA , PilG E41H/RKK43-45AAA , PilG Q55A or PilG SS62-63AA were grown at 37°C in LB medium containing 50 μg ml -1 kanamycin. The cultures were cooled to 18°C at OD 600 = 0.6, after 30 min induced with 0.5 mM isopropyl-D-thiogalactopyranoside (IPTG)  (II) Soluble partial protein. E. coli ER2566 cells harboring plasmids encoding PilG 1-60 , PilG 1-80 , PilG 1-140 and PilG 1-178 were grown and lysed as previously described. Unbroken cells and cell debris were removed by centrifugation twice at 20.000 × g for 20 min. The supernatant was added to a Ni-NTA column (Qiagen), washed and eluted with phosphate buffers (pH 8.0) containing increasing amounts of imidazole up to 250 mM. Fractions containing the recombinant proteins were pooled and dialyzed against phosphate buffer (pH 8.0).
(III) Purification of inclusion bodies. PilG partial protein PilG 30-80 was purified by isolation of inclusion bodies, solubilization in urea and protein purification with Ni-NTA under denaturing conditions according to protocols provided by Qiagen [54]. Urea was removed by dialysis against a phosphate buffer (pH 8.0), yielding a soluble protein. A complete list of PilG FL and partial recombinant proteins purified is available in S2 Table.

SDS-PAGE and immunoblotting
Procedures for SDS-PAGE and immunoblotting have previously been described [7,17]. The samples were mixed with an equal amount of sample buffer (20% glycerol, 5% β-mercaptoethanol, 0.05% bromophenol blue, 125 mM Tris-HCl, pH 6.8) and kept on ice for 15 min before gel electrophoresis in a Mini-PROTEAN system (Bio-Rad) at 4°C, with a constant voltage of 110 V.

Rabbit immunization and antibody production
Rabbit polyclonal antibodies were raised against the PilG FL protein as previously described [20]. Serum obtained 100 days after immunization was used to detect PilG. The anti-PilQ sera used were previously described [55].

South-western analysis
The DNA binding ability of PilG FL , partial proteins and alanine substitution mutants was assessed by a solid phase DNA overlay assay as previously described [38]. The DNA substrates used in the assay are listed in Table 2. Purified DNA glycosylase Fpg [56] and bovine serum albumin (BSA) were used as positive and negative controls, respectively. Each experiment was repeated at least three times.

Labeling of DNA substrates
Oligonucleotides were end-labeled with [γ 32 P]ATP (Perkin Elmer) using T4 polynucleotide kinase (New England Biolabs) as described by the manufacturer. Labeled oligonucleotides were separated from free nucleotides on 20% non-denaturing gels by PAGE and extracted by diffusion into water. Double-stranded labeled substrates were generated by mixing labeled oligonucleotides with an equal molar amount of complementary unlabeled oligomer, heating to 95°C for 5 min and slow cooling to room temperature. The concentration of the double-stranded DNA substrate was estimated by dot quantification on agarose plates containing ethidium bromide [52], using unlabeled DNA of known concentration as the standard.

Electrophoretic mobility shift assay (EMSA)
In the first analysis for determining the half maximal effective concentration (EC 50 ) for PilG FL and the PilG alanine substitution constructs, the binding reaction was performed for 15 min on ice in a buffer containing 25 mM NaH 2 PO 4 , 150 mM NaCl, 1.0 mM DTT, 10% [w/v] glycerol, and 0.05% [w/v] DDM at pH 8.0. In later experiments using an optimized EMSA with the PilG partial proteins PilG 1-178 , PilG 1-80 and PilG 30-80 , 4.5 fmol labeled DNA was mixed with 5 μl 2× gel shift buffer, yielding a final concentration of 50 mM HEPES, 1.0 mM DTT, 100 mM NaCl, 5 mM MgCl 2 , 10% [w/v] glycerol, 0.05% [w/v] DDM, pH 7.5, and protein in a final volume of 10 μl. These mixtures were incubated on ice for 15 min. Electrophoresis was carried out on 4% or 6% polyacrylamide gels in Tris/glycine/EDTA buffer [57]. Gels were dried, exposed in a PhosphorImager cassette, scanned in a Typhoon scanner and quantified using Image-Quant (GE Healthcare).

Endoproteinase cleavage of N-terminal PilG
In order to map the location of the DNA binding domain of PilG and the site mediating protein-protein interactions, the N-terminal recombinant proteins PilG 1-80 and PilG 1-178 were cleaved with endoproteinase Asp-N enzyme (Roche Applied Science). Briefly, 400 μg of purified protein was mixed with 2 μg of lyophilized endoproteinase Asp-N sequencing grade enzyme. 1 M urea and 0.01% SDS were added to the proteolysis reaction in order to resolve protein folding, which might shield enzyme-specific cleavage sites. The reaction was conducted at 37°C and aliquots were collected at selected time points for a period of 24 h. All samples were separated by SDS-PAGE and DNA binding activity of the cleavage products was determined by South-western analysis. The most predominant Asp-N cleavage products were excised from the Coomassie Blue-stained gel, further in-gel digested with trypsin, and the obtained peptides were identified by mass spectrometry (MS) analysis.

Peptide identification by mass spectrometry
PilG fragments selected from the endoproteinase cleavage assay were identified by peptide mass fingerprinting/MS as previously described [58]. In brief, tryptic peptides obtained from in-gel digestion were desalted and concentrated using C18 3M Empore Extraction Disks (Varian) placed in GELoader tips (Eppendorf). The peptides retained were eluted onto a stainless steel target plate with a solution containing 70% acetonitrile, 0.1% trifluoroacetic acid and 10 Table 2. DNA substrates employed in the study.

Expression of mutant pilG in vivo
To analyze the biological significance of the amino acids involved in DNA binding the mutations pilG EEE39-41AAA and pilG RKK43-45AAA were introduced into Mc. For this, the pilG was amplified from Mc with the primers SF101 and SF134, the aph was amplified form pUP6 with the primers OHA_AphEcoRI_REV and SAF-Tn5-aph-for, and the 5' pgi with the primers SF135 and SF136. The PCR products were digested with the appropriate restriction enzymes (S2 Table) and as a concatenate ligated into pBluescript II SK(+) (Stratagene) giving the vector pSAF67. The mutations to generate pilG EEE39-41AAA and pilG RKK43-45AAA were introduced into pSAF67 giving the vectors pSAF68B and pSAF69, respectively ( Table 1). The wild-type and mutant pilG genes were transformed into the Mc strain MC58. Positive clones were selected for by kanamycin resistance and the mutations in pilG confirmed by DNA sequencing and mass spectrometry of the expressed proteins. Pilus expression and PilG expression were confirmed by immuno blot (S10 Fig). The mutant strains were tested by quantitative competence screening as described below.
Phenotypic analysis of N. meningitidis PilG site-directed mutants N. meningitidis site-directed mutants were compared to wild type strains in phenotypic analysis.
(I) Colony morphology. N. meningitidis strains cultured on clear GC plates were assessed by stereo microscopy to define whether they had a piliated (P+) or non-piliated (P-) colonial morphology [60].
(II) Purification of type IV pilus fibers. Type IV pili were purified following the short method as described by Brinton [61]. The pilus preparations were separated by SDS-PAGE, blotted onto nitrocellulose membrane and immunoblot was performed with a pilin specific antiserum.
(III) Competence screening. Competence for the transformation of wild type and mutant strains was performed using the plasmid pOHA-D4 as donor DNA (Table 1). Wild type and mutant strains were harvested in a CO 2 -saturated liquid GC medium containing 7 mM MgCl 2 and 1× IsoVitaleX. The bacteria were exposed to either plasmid pOHA-D4 or distilled water (negative control) for 45 min before adding 10 volumes of liquid GC medium. The bacterial solutions were incubated with tumbling at 37°C for 4.5 h and subsequently diluted and plated on both plain GC medium and GC medium containing erythromycin (Erm) [8 μg ml -1 ]. The transformation rate was calculated by dividing the number of Erm-resistant colony forming units (c.f.u.) by the total number of c.f.u. The assay was repeated at least three times for each null mutant.

Separation of outer and inner membranes by sucrose density gradient
Mc outer and inner membranes were separated by sucrose density gradient as previously described [11]. In brief, Mc M1080 cells were washed with phosphate buffered saline (PBS), resuspended in 50 mM Tris buffer, pH 8.0, containing 50 μg RNase and DNase (Sigma) and processed twice through a French press (103.500 kPa, Thermo Electron). The debris was removed by centrifugation at 10.000 × g for 10 min. Sucrose gradient centrifugation was carried out in water with 3 mM EDTA, pH 8.0 [62]. The sample was transferred onto two layers of sucrose consisting of 3 ml of 55% [w/v] and 4 ml 15% sucrose, and centrifuged at 217.000 × g and 4°C for 5 h in a SW40Ti rotor (Beckman). The membrane fraction positioned at the interface was collected and diluted down to 30% sucrose, applied to a discontinuous sucrose gradient consisting of 3 ml of 50, 45, 40 and 35% sucrose, and centrifuged in a SW40Ti rotor at 180.000 × g and 4°C for 35 h. After fractionation, 10 μl samples were analyzed by SDS-PAGE, followed by Coomassie Blue-staining or immunoblotting.

PilG is predicted to have a putative DNA binding domain
Alignments of the predicted amino acid (aa) sequences of PilG and orthologs revealed sequence conservation around residues 70 to 100, 120 to 220 and 270 to 400, while there is only little conservation at residues 1 to 60 (S1 Fig). To search for DNA binding motifs, we focused on three regions of the PilG aa sequence, residues 1 to 185, 200 to 225 and 250 to 377, and excluded predicted transmembrane helices between these segments. This in silico analysis suggested that a putative PilG DNA binding domain localizes to residues 40 to 130 (Fig 1A). Bioinformatics analysis of the PilG aa sequence showed no other commonly recognized/conserved functional domains or signature motifs.

Estimating the DNA binding capacity of PilG
Our previous study showed that full-length PilG (PilG FL ) binds single-stranded DNA (ssDNA) and double-stranded DNA (dsDNA) without DUS specificity [38]. In this study, we further extended the analysis to quantify the DNA binding activity of PilG at low DNA concentration (0.45 nM) and high PilG concentration (94 nM) using an EMSA. These conditions were selected as the protein concentration required to bind half of the input DNA approximates the dissociation constant, K d , of the protein-DNA complex [63]. In this preliminary EMSA assay using a 52 base pair (bp) dsDNA substrate, the estimated apparent PilG K d value for PilG FL was 0.3 μM (S3 Fig).

N-terminal PilG binds DNA
Based on the bioinformatics analyses, constructs for the expression of truncated PilG proteins were made (Fig 1B and S2B Fig). The truncated variants of recombinant PilG (Fig 1B) were also tested for DNA binding activity using South-western analysis (Fig 2 and Table 3). This analysis mapped the DNA binding activity of PilG to the N-terminal 80 amino acids. In particular, PilG 1-80 strongly bound DNA (Fig 2, lane 6) while the even shorter partial PilG proteins PilG 1-60 and PilG 30-80 did not bind DNA (Fig 2, lanes 4 and 8, respectively). The Southwestern analysis also yielded no difference in PilG affinity for ssDNA and dsDNA binding, and there was no difference in affinity for DNA with or without the DUS (data not shown), which corresponds to our previous findings for PilG FL [38]. In order to define in more detail which PilG residues contribute to DNA binding, PilG 1-178 was cleaved with endoproteinase and the cleavage products were evaluated by South-western analysis and mass spectrometry (MS). The cleavage products with DNA binding activity obtained corresponded to residues 1 to 154 and 27 to 178 (Fig 3). When PilG 1-80 was subjected to endoproteinase cleavage and South-western analysis, residues 1 to 70 was the smallest truncated form of PilG with DNA binding activity (S5 Fig). EMSA assays with dsDNA confirmed that PilG 1-178 and PilG 1-80 bound DNA with an affinity similar to that of PilG FL , whereas no DNA binding was seen with PilG 30-80 at the same protein concentration (Table 4). Based on these findings, further investigations mainly focused on the PilG amino acids 1-80, which were required and sufficient for the DNA binding activity.
DNA binding sites tend to have positive electrostatic charge and that charged and polar amino acids within DNA binding sites often play a direct role in protein-DNA interactions [64,65]. Therefore, we expected that alanine substitutions of lysine, arginine, serine and glutamine residues in the N-terminal 80 residues domain of PilG might reduce or abolish DNA binding. Alanine substitutions were introduced into the full-length PilG protein at selected sites (Table 1), including a motif with three consecutive glutamate residues (EEE39-41) and a motif with three consecutive basic residues (RKK43-45) (S1 and S6 Figs). The mutant PilG variants were tested for DNA binding affinity in South-western and EMSA assays (Fig 4 and Table 3). Two of the eight PilG mutants exhibited changes in DNA binding when measured by South-western analysis (Fig 4B). The substitution RKK43-45AAA (with and without the additional substitution E41H) decreased DNA binding affinity (Fig 4B, lane 6, and S7 Fig), while alanine substitution of the glutamic acid residues 39 to 41 increased DNA binding affinity ( Fig  4B lane 9). South-western assays using increasing amounts of protein demonstrated that the protein PilG EEE39-41AAA bound DNA with significantly higher affinity than PilG FL , while PilG RKK43-45AAA bound DNA with a significantly lower affinity than PilG FL (Fig 4C). The DNA binding affinity of PilG K3A , PilG KK12-13AA , PilG E14A and PilG KR15-16AA was slightly reduced, while the PilG Q55A and PilG SS62-63AA affinity was the same as of PilG FL (data not shown). To further quantify the DNA binding, the EMSA was optimized (see Methods and compare S3 Fig and S8 Fig). In this shift assay, the PilG EEE39-41AAA and PilG RKK43-45AAA , which demonstrated significant differences to wild type PilG FL in South-western analysis, showed slightly different results (Fig 5). Both mutant proteins showed reduced binding to dsDNA as well as ssDNA compared to the wild type protein PilG FL . Notably, while the PilG EEE39-41AAA required only a 50% increased value in the protein concentration (yealding half activity, EC 50 ), the PilG RKK43-45AAA protein was calculated to need a several hundred-fold increase in protein concentration to reach 50% binding (Table 5), although at the highest protein concentration used (5μM) about 50% binding was seen (Fig 5).
Based on this data, neisserial PilG residues 1 to 30, the RKK motif at aa 43 to 45, and the residues 60 to 80 were predicted to play critical roles in binding of DNA. The results are in line with the structural modeling predicting that these regions are located close to each other in the native PilG (Fig 6). To test the significance of the PilG DNA binding in transformation, two alanine substitutions were introduced in situ into Mc host cells. Interestingly, DNA transforming activity was significantly reduced in Mc expressing PilG RKK43-45AAA when compared to the wild type (32 times) and the control (38 times) carrying only the selective marker (Fig 7). The efficiency of transformation was however not significantly reduced with the mutation PilG EEE39-41AAA (Fig 7). These in vivo findings corroborate the results obtained in vitro by EMSA showing the strong negative effect of the PilG RKK43-45AAA mutation on DNA binding.

N-terminal PilG interacts with N-terminal PilQ
Once the localization of PilG in the inner membrane [38] was confirmed (S9 Fig) a solid phase overlay Far-western analysis was employed to assess the interaction between PilG and other pilus biogenesis proteins. Interaction between PilG and PilQ was detected but there was no interaction with PilN, PilO, PilP, PilF, PilT, ComP, ComL or the pilus itself. PilG FL , PilG 1-256 and PilG 1-178 interacted with PilQ FL and PilQ fragments with PilQ 25-132 being the shortest fragment showing PilG binding ( Table 6 and Fig 8). This suggests that PilG residues 1 to 178 mediate the interaction with N-terminal PilQ. Far-western analysis was also carried out on endoproteinase cleavage products of PilG 1-178 , using PilQ 25-354 as a probe. MS analysis identified interacting peptides as PilG  and PilG 1-154 , further refining the interaction with PilQ to amino acids 27 to 154 (data not shown). We did not detect a direct interaction between PilG Table 3. Summary of the results of the oligo nucleotide binding tests using recombinant PilG proteins. The "+" and "-"refer to the DNA binding activity of the proteins when analyzed by South-western.    (Table 6), although a previous study suggests a functional relationship between neisserial PilG and PilT [4].

Discussion
DNA uptake through natural transformation is a dynamic process that mediates allelic replacement of genes into the genome in many bacteria [66,67]. Transformation is an important process influencing bacterial fitness and survival, and may also increase the spread of antibiotic resistance among bacterial species [68]. In neisserial species, transformation is dependent on type IV pili. Hence, in order to better understand the mechanism by which DNA is taken into neisserial cells, we studied the DNA binding and protein interaction properties of the pilus biogenesis protein PilG. The fact that PilG null mutants are non-piliated emphasizes the role of PilG in pilus biogenesis [20]. We have previously shown that PilG co-purifies with the inner membrane, suggesting that it is in direct contact with both the cytosol and the periplasm. As indicated by Derrick and co-workers in their PilG structural analysis [19], N-terminal PilG is predicted to be oriented into the cytoplasm. Here, we demonstrate that the N-terminal region of PilG binds DNA in a DUS independent manner, indicating that PilG is not a selective factor in the DNA transport in Neisseria sp. In addition, N-terminal PilG directly interacts with N-terminal PilQ.
Our results demonstrated that PilG N-terminal residues 1 to 30 and 60 to 80 were required for DNA binding activity. Because the isoelectric points of PilG 1-80 and PilG 30-80 are 10.58 and 10.59, respectively, DNA binding by PilG 1-80 cannot be explained solely by overall electrostatic charge. By analyzing PilG alanine substitution mutants in vitro, we determined that PilG residues EEE39-41 and RKK43-45 exert negatively and positively modulating effects on PilG DNA binding, respectively. Structural modeling studies predicted that the residues 1 to 30 and 60 to 80 are located close to each other in native PilG (Fig 6). We propose that these two PilG regions cooperatively form a DNA binding site.
PilG DNA binding affinity, as monitored by EMSA (Fig 5 and Table 5), was rather low (K half 1.16 μM for ssDNA and K half 1.07 μM for dsDNA). Nonetheless, the competence protein ComEA of Bacillus subtilis has similarly low affinity for a 112 bp dsDNA substrate (K d 0.5 μM) [69]; thus, the PilG-mediated DNA binding activity is comparable to that of other validated DNA binding proteins involved in DNA transformation. Importantly, we observed that PilG EEE39-41AAA has only a slightly reduced DNA binding activity (Table 5) and its expression has no significant negative effect on DNA transformation, while PilG RKK43-45AAA has a strongly reduced DNA binding activity and in vivo significantly reduces DNA transformation  Table 4. DNA binding abilities of PilG partial proteins in EMSA. The protein concentration required to achieve a shift with 50% of the DNA substrate is indicated and can be used to estimate the dissociation constant, K d , of the protein-DNA complex [63].   Table 2 Table 3). The positions of the protein size standards (kDa) are shown on the left. (C) Relative level of bound DNA substrate South-western at different protein concentrations of recombinant PilG FL (WT), PilG RKK43-45AAA (RKK), and PilG EEE39-41AAA (EEE) measured by densitometry and plotted as relative units with the average value for 1μg PilG FL set to 1.
The results of at least three experiments using ssDNA and dsDNA are shown. No significant substrate specificity was detected.
doi:10.1371/journal.pone.0134954.g004 (Fig 7). This result shows that the DNA binding activity of PilG plays a biologically significant role, facilitating DNA uptake during transformation of neisserial species. While PilG residues RKK43-45, together with residues 1 to 30 and 60 to 80, appear to be of prime importance for DNA binding, the more C-terminally located part of N-terminal PilG may contribute to protein-protein interactions that are also important during DNA uptake and transformation. As documented by our combined endoproteinase cleavage and Far-western analysis, as well as Far-western analysis with partial PilG proteins, PilG residues 80 to 154 contribute significantly to the interaction with N-terminal PilQ. We propose that highly conserved lysine and arginine residues in PilG 1-30 and PilG 60-80 , forming a positively-charged protein region, may facilitate both protein-DNA and protein-protein interactions. Positive electrostatic surfaces are commonly found in DNA binding sites and can also promote binding to negatively-charged membranes, receptors or other proteins [70]. However, differences between protein-DNA and protein-protein interaction sites in terms of polarity and charge have been described [46].
The observation that N-terminal PilG binds to N-terminal PilQ underscores the importance of PilG in pilus biogenesis and could be mediated by the short predicted periplasmic loop of Table 5. Best-fit values for k half (= EC 50 , in [μM]) for the EMSA analysis of wild type and mutant PilG proteins using DUS containing substrates. The k prime and h values were calculated using GraphPad Prism 5 [75] and converted into k half values. The goodness of fit is given as R 2 values in parentheses. Due to the conditions used in the EMSA k half is approximately equal to kd [63].  PilG [19]. Previous studies have shown that the 900 kDa PilQ complex spans both the outer and inner membranes [11], undergoes structural conformation changes upon binding to type IV pili [17] and exhibits DNA binding activity [10]. In addition, the finding that N-terminal PilQ directly interacts with PilG is consistent with previous findings that N-terminal PilQ binds DNA [10] and is directed towards the inner membrane [55]. Structural modeling of these interactions was previously reported [27].  Table 6. Summary of Far-western analysis assessing the interaction between PilG and other pilus biogenesis proteins. The "+" refers to positive and the "-"refers to no protein-protein interaction detected.
a results confirmed by testing both components alternating in liquid and bound to solid phase. Exogenous DNA is thought to pass through the PilQ pore during DNA uptake [10]. Subsequently, PilG could guide the DNA into the cytoplasm, acting as an intermediate chaperone in the vicinity of the inner membrane. Since N-terminal PilQ also binds DNA [10], the PilG DNA binding site could potentially interact with PilQ and transforming DNA in a coordinated or sequential way. This interaction of the cytoplasmic N-terminal part of PilG could be allowed by the linking of the cytoplasm and periplasm through the PilG multimer platform structure which was suggested earlier [19]. It is also possible that PilG and the ComA pore might cooperatively facilitate DNA uptake. The observation that B. subtilis RecA co-localizes with the competence machinery indicates that DNA uptake and recombination are closely linked in space and time [71]. Therefore, PilG could potentially play direct roles in both competence and DNA recombination.
Taken together, our data suggest that the N-terminal domain of PilG is a DNA binding site that potentially operates in conjunction with N-terminal PilQ. These findings point to the biological significance of PilG/PilQ-mediated DNA binding and more comprehensive functional and structural studies of PilG are needed. These should address questions about PilG multimerization, topology, interactions with pilus biogenesis proteins and DNA, and, in particular, more precise identification of the residues and motifs that enable these specific interactions. More complete elucidation of the biological function(s) of PilG will likely improve our understanding of neisserial transformation and hence antigenic variation and antibiotic resistance.  Table. Oligonucleotides. Primers employed in PilG recombinant protein construction and alanine substitution mutants. (DOCX) S2 Table. Recombinant proteins. Expression and purification of PilG recombinant proteins. All constructs contain a C-terminal 6×His-tag predicted to be located in the cytoplasm. (DOCX)