JOURNAL OF BACTERIOLOGY, Sept. 2009, p. 5613–5621 Vol. 191, No. 180021-9193/09/$08.00�0 doi:10.1128/JB.00535-09Copyright © 2009, American Society for Microbiology. All Rights Reserved.

Pilin Antigenic Variation Occurs Independently of the RecBCDPathway in Neisseria gonorrhoeae�

R. Allen Helm and H. Steven Seifert*Northwestern University, Feinberg School of Medicine, Department of Microbiology-Immunology, Chicago, Illinois 60611

Received 21 June 2009/Accepted 1 July 2009

Type IV pilus expression has been strongly implicated in the virulence of Neisseria gonorrhoeae, the causativeagent of gonorrhea. In Neisseria, these pili undergo frequent antigenic variation (Av), which is presumed toallow reinfection of high-risk groups. Pilin Av is the result of RecA-mediated recombination events between thegene encoding the major pilin subunit (pilE) and multiple silent pilin locus (pilS) copies, utilizing a RecF-likerecombination pathway. The role of RecBCD in pilin Av has been controversial. Previous studies measuringpilin Av in recB and recD mutants in two independent strains of N. gonorrhoeae (MS11 and FA1090) by indirectmethods yielded conflicting results. In addition, these two laboratory strains have been suggested to expressvery different DNA repair capabilities. We show that the FA1090 and MS11 parental strains have similarabilities to repair DNA damage via UV-induced DNA damage, nalidixic acid-induced double-strand breaks,and methyl methanesulfonate-induced alkylation and that RecB and RecD are involved in the repair of theselesions. To test the role of the RecBCD pathway in pilin Av, the rate and frequency of pilin Av were directlymeasured by sequencing the pilE locus in randomly selected piliated progeny of both MS11 and FA1090 in recBand recD mutants. Our results definitively show that recB and recD mutants undergo pilin Av at rates similarto those of the parents in both strain backgrounds, demonstrating that efficient pilin Av is neither enhancednor inhibited by the RecBCD complex.

Homologous recombination is a widely conserved processused by all organisms to repair DNA damage and to mediategenetic transfer. Usually RecA or a RecA homologue is re-quired for homologous recombination, but the other process-ing enzymes used to prepare DNA for recombination differbetween organisms. Generally, double-stranded DNA (dsDNA)breaks are repaired by the RecBCD complex in gram-negativeorganisms and the AddAB complex in gram-positive organ-isms, although exceptions exist in which AddAB has beenfound in some proteobacteria (1, 4, 29, 33, 56). Single-strandedDNA (ssDNA) gaps are typically repaired by the RecF path-way, which is ubiquitous among the Bacteria and is part of auniversal step in recombinational repair (31). While theRecBCD and RecF pathways have been extensively studied inEscherichia coli, they are less well understood in the human-specific pathogen Neisseria gonorrhoeae. N. gonorrhoeae is agram-negative bacterium that has been isolated only from in-fected humans and is believed to exist naturally only within thehuman population. Like E. coli, N. gonorrhoeae has a RecAhomologue and possesses most of the genes that define theRecBCD and RecF pathways of E. coli (27).

In E. coli, the RecBCD exonuclease begins to degrade DNAat the site of the break, and upon reaching a particular se-quence, a chi site, the RecD component is inactivated, leavingRecBC to act as a helicase, processively unwinding the DNA,yielding ssDNA, which is the substrate for RecA (reviewed inreference 7). The RecF pathway has also been extensively

studied in E. coli, where it has been shown to be involved inrepair of “daughter strand gaps,” lesions in newly synthesizedDNA, often after replication. In this pathway, single-stranded-DNA-binding proteins bind to the gap, and then RecO andRecR, acting as a complex, initiate the loading of RecA. DNAprocessing is mediated by the exonuclease RecJ and the RecQhelicase (reviewed in reference 23). Whether RecF acts incomplex with RecOR is dependent upon the availability ofdsDNA proximal to the gap (36).

The virulence of N. gonorrhoeae is strongly implied to bedependent upon the expression of type IV pili, which are long,thin, threadlike projections that extend into the extracellularmilieu. In N. gonorrhoeae, these pili are required for adherenceto the urogenital epithelium (47), efficient levels of naturalDNA transformation (21, 40, 44), twitching motility (53, 54),and symptomatic infection of the urogenital tract in volunteerstudies (5, 50). The major pilus subunit, pilin, undergoes anti-genic variation (Av) at rates higher than those of any other Avsystem currently known (6). It is thought that the high rate ofpilin Av is a factor in the ability of N. gonorrhoeae to immedi-ately reinfect a host who has recently cleared an N. gonorrhoeaeinfection, a concept that is supported by the failure of a pilus-based vaccine trial in men (2).

Pilin Av in N. gonorrhoeae is the result of recombinationbetween the gene that encodes pilin (pilE) and at least one ofthe many silent pilS copies located throughout the N. gonor-rhoeae chromosome. The pilE gene consists of a 5� constantregion, followed by a semivariable region and a hypervariable(HV) region at the 3� end, with the pilS copies possessing onlythe semivariable and HV regions (9, 11; reviewed in reference39). Each pilS copy shares homology with pilE but is missing apromoter, a Shine-Dalgarno sequence, and the initial 5� se-quence of pilE. Most of the pilS copies are located in clusters

* Corresponding author. Mailing address: Department of Microbi-ology-Immunology, Northwestern University, Feinberg School ofMedicine, S213, 303 East Chicago Avenue, Chicago, IL 60611. Phone:(312) 503-9788. Fax: (312) 503-1339. E-mail: h-seifert@northwestern.edu.

� Published ahead of print on 10 July 2009.

5613

on Septem

ber 8, 2020 by guesthttp://jb.asm

.org/D

ownloaded from

within several chromosomal pilS loci. Each pilS locus containsat least one, and often multiple, pilS copies. Immediatelydownstream of the pilE gene and each of the pilS loci is a 65-bpSma/Cla repeat (30). In strain FA1090, there are a total of 18pilS copies clustered within five pilS loci and 1 pilS copy directlyupstream of pilE for a total of 19 silent copies (12). In strainMS11, there are 17 silent copies located in five pilS loci (10).

The event responsible for pilin Av is a nonreciprocal homol-ogous-recombination event in which DNA from the pilS copyis unidirectionally transferred to pilE (9, 11). Since the pilSdonor copy remains unchanged, this recombination is classifiedas a gene conversion system (9, 48, 55). Pilin Av is RecAdependent (22) and utilizes the RecF-like pathway of recom-bination (14, 27, 37). Additionally, non-pathway-specific N.gonorrhoeae enzymes, including RecX, RdgC, the Rep heli-case, RecG, and RuvABC, have also been shown to be in-volved in pilin Av (19, 26–28, 37, 43, 45).

Several methods have been used over the years to detect andmeasure the frequency of pilin Av in N. gonorrhoeae. Eachassay has its own benefits and drawbacks. These methods in-clude a colony phase variation assay, which takes advantage ofthe fact that some pilin Av events yield colonies that are non-piliated (P�), allowing a visual assay in which many samplescan be screened with relative ease (48, 49). The disadvantagesof this method include the facts that events other than pilS/pilErecombination can yield P� colonies and that only a subset ofall pilin Av events yield altered colony morphologies. TwoPCR-based approaches to measure pilin Av rates have alsobeen used; a colony-based PCR assay, using primers designedto anneal to the pilE ribosome binding site and the HV regionsof two different pilS copies (28), and a reverse transcription-PCR method that uses a similar approach (42). In both ofthese techniques, a recombination event between one of theselected pilS copies and pilE yields a PCR product of a pre-dicted size. A disadvantage is that recombination events thatdo not involve one of the selected pilS copies go undetected. Aquantitative PCR-based assay was developed that detected thetransfer of the HV sequences of most of the silent copies ofstrain FA1090, but this assay could only estimate the frequencyof pilin Av and was not adaptable to other strains (34). Acoconversion assay has also been developed in which a markeris placed downstream of the pilE gene, in the Sma/Cla region,and following certain pilS/pilE recombination events, themarker is lost and replaced with the material downstream ofthe pilS copy, which is detectable by Southern blotting (15).While this approach definitely detects some pilin Av events,detection is incomplete, since not all pilE/pilS recombinationevents result in loss of the downstream marker.

It has been shown via analysis of individual recB, recC, andrecD mutants that N. gonorrhoeae utilizes the RecBCD path-way for the repair of dsDNA breaks induced by ionizing radi-ation and that all three mutants have identical phenotypes withregard to DNA repair, DNA transformation, and pilin Av (27).A study of strains MS11 and P9 concluded that RecD had norole in the repair of UV damage (3). A later study concludedthat RecB is involved in the repair of UV-induced DNA dam-age, nalidixic acid (Nal)-induced dsDNA breaks, and methylmethanesulfonate (MMS)-induced alkylation lesions (16).However, the latter study indicated that strain MS11 had amore robust DNA repair phenotype than strain FA1090. This

was an unexpected conclusion, as there are no data to suggestthat MS11 has different repair machinery than strain FA1090.Thus, the roles of the individual genes of the N. gonorrhoeaeRecBCD pathway in DNA repair are not certain.

The question of whether N. gonorrhoeae RecBCD is in-volved in pilin Av has been pursued using the assays mentionedabove. Initially, the colony-based PCR assay and the colonyphase variation assay showed no difference in pilin Av betweenrecB, recC, and recD mutants and the parent in strain FA1090(27). In contrast, a different study using the colony phase vari-ation assay concluded that a recD mutant underwent pilin Avat an increased frequency in strain MS11 (3). Another report,using the reverse transcription-PCR and coconversion assays,suggested that pilin Av is diminished in recB mutants in bothMS11 and FA1090 (16). Thus, the roles of the individualRecBCD pathway genes and the entire pathway in pilin Av arecontroversial.

As DNA sequence analysis of many samples has becomefaster, more accurate, and more cost-effective, an assay inwhich direct sequencing of the pilE genes of multiple randompiliated (P�) N. gonorrhoeae colonies has been developed andshown to have a high level of sensitivity and statistical power(6, 18). In this assay, a previously described isopropyl-�-D-thiogalactopyranoside (IPTG)-inducible recA strain (38) isgrown for a certain time on IPTG, allowing pilin Av to occur.Multiple P� progenitor colonies are picked from the IPTG-containing medium and passaged on medium without IPTG,essentially “locking in” any pilin variants that have arisen dur-ing RecA induction. P� progeny colonies arising from each ofthe progenitors are passaged in the absence of IPTG to ensurecolony clonality, and the pilE gene is sequenced from each ofthem. This assay is advantageous because it involves directanalysis of the pilE sequence and because randomly selectedP� colonies are analyzed.

Using recB and recD mutants of both MS11 and FA1090, weshow that the RecBCD pathway participates in the repair ofDNA damaged by UV, MMS, and Nal in both strains. Pilin Avoccurs at rates similar to that in the parental strain in recB andrecD mutants in both strain backgrounds. This work demon-strates conclusively that the use of the RecF-like recombina-tion pathway for pilin Av is common to different N. gonor-rhoeae isolates and that recD mutants are indistinguishablefrom recB mutants for this bacterium.

MATERIALS AND METHODS

Bacterial strains and growth conditions. Two independently isolated strains ofN. gonorrhoeae were used, FA1090 pilin variant 1-81-S2 (41) and MS11 VD300(20). In both strains, the native recA gene was replaced with the recA6 construct,which is inducible in the presence of IPTG (38). This allele generates a RecA�

phenotype in the presence of 1 mM IPTG and a RecA� phenotype in theabsence of IPTG (38). N. gonorrhoeae was grown at 37°C with 5% CO2 on Gcmedium base (GCB; Difco) plus Kellogg supplements I and II (17).

FA1090 recB::erm and recD::erm mutants were constructed via insertion of anerythromycin (Erm) resistance cassette in each gene and have been previouslydescribed (27). The FA1090 recB::erm and recD::erm mutations were trans-formed into MS11 VD300 and selected for growth on 7.5 �g/ml Erm. Theresulting transformants were twice backcrossed into MS11 VD300, and thepresence of the Erm resistance cassette was verified via PCR in each MS11VD300 mutant.

DNA damage assays. N. gonorrhoeae survival in the presence of UV, MMS,and Nal was tested by growth on solid medium with 1 mM IPTG for approxi-mately 20 generations. Bacteria were swabbed and resuspended in liquid me-

5614 HELM AND SEIFERT J. BACTERIOL.

on Septem

ber 8, 2020 by guesthttp://jb.asm

.org/D

ownloaded from

dium, and serial dilutions were plated on solid medium in the presence andabsence of UV exposure, MMS, or Nal. All plates contained 1 mM IPTG toensure RecA expression during survival assays. Samples were grown for 48 h, andthe colonies were counted. Survival was measured by dividing the number ofCFU on treated plates by the number of CFU on untreated plates. UV wasdelivered to serially diluted bacteria on plates at 20, 50, 100, and 200 J/m2 usinga UV Stratalinker 1800 (Stratagene). MMS was used at 0.0025, 0.005, and 0.01%.Because of the inherent instability of MMS, the plates were used 1 day afterbeing made. Nal was added to solid medium at 0.063, 0.125, 0.15, 0.20, 0.25, 0.5,0.75, and 1.0 �g/ml. All assays were performed in duplicate over the course of atleast three different experiments.

Pilin Av sequencing assay. The pilin Av sequencing assay was performed asdescribed previously (6, 18). Briefly, all six strains were grown on solid GCBmedium with 1 mM IPTG, allowing the expression of RecA for the timesdescribed in Results below (Table 1). Seven random progenitor colonies with aP� morphology from each strain were picked and passaged on GCB withoutIPTG, halting RecA expression and thus inactivating recombination. For eachstrain, at least 28 P� progeny colonies arising from each of the seven progenitorswith sizes and colony morphologies similar to those of the parent were pickedand passaged on GCB without IPTG two times to ensure colony clonality. Asingle colony from each sample was isolated, and the pilE gene was PCRamplified as described previously (24). The DNA sequence was determined(SeqWright, Houston, TX) and analyzed (MacVector; Symantec Corp.). In theevent a strain appeared to only have a single nucleotide substitution in pilErelative to the parent, the PCR was repeated and the DNA sequence wasanalyzed again to ensure the substitution was genuine.

Determination of pilin Av frequency and rate. The pilin Av frequency wasdetermined for the progeny of each of the seven progenitors by dividing the totalnumber of pilin Av events by the number of progeny of each set, resulting inseven values for each strain. The pilin Av rate was determined by dividing thepilin Av frequency by the number of generations each sample was grown in thepresence of IPTG as determined by a CFU/colony assay at the time of harvest.Determination of statistical significance using the Wilcoxon rank sum test wasperformed as described previously (18). A pilin Av event was defined as thenumber of pilS copies donated in a single pilE variant. For FA1090, in which all19 pilS copies are sequenced and annotated, the vast majority of variant pilEsequences are the result of the incorporation of a single pilS donor copy, withonly 5 samples being the result of the incorporation of two pilS donor copies outof a total of 86 samples with a variant pilE (data not shown). Since the full arrayof MS11 pilS copies has not yet been sequenced and annotated, this analysis isnot possible, and therefore, to calculate the MS11 pilin Av frequencies, eachvariant pilE was assumed to be the result of a single pilin Av event.

RESULTS

Survival of recB and recD mutants in FA1090 and MS11.Based on UV survival phenotypes, it has been concluded thatN. gonorrhoeae RecD is not involved in the repair of DNA

damage (3). It has also been proposed, based on MMS and Nalsurvival assays, that strain MS11 possesses a more robust DNArepair system than strain FA1090 (16). We reexamined theDNA repair phenotypes of strains MS11 and FA1090 and recBor recD mutants using UV light, MMS, and Nal as DNA-damaging agents (Fig. 1). UV irradiation induces pyrimidinesto dimerize, generating deleterious lesions (reviewed in refer-ence 8). MMS acts as a DNA-alkylating agent and is thought tocause stalled replication forks (25). Nal is a broad-spectrumantibiotic and the parent of the fluoroquinolone family; itcauses accumulation of dsDNA breaks by inhibiting gyrasefrom resealing the breaks and can be used to measure dsDNAbreak repair capabilities (reviewed in reference 13). Thus,these three treatments produce different types of damage thatshould reveal divergent repair capabilities.

The results from the UV survival assay showed an approxi-mately 10-fold difference in survival between the rec mutantsand the parent strains at 50 and 100 J/m2, with the rec mutantshaving nearly no viability at 200 J/m2, clearly indicating thatboth RecB and RecD are involved in repair of UV-induceddamage (Fig. 1A). Furthermore, the UV assay showed thatthere is no difference in survival between the parental FA1090and MS11, a conclusion that has been reported previously (16).Interestingly, at 50 J/m2, the MS11 recB and recD mutants hada statistically higher survival rate than the FA1090 recB andrecD mutants, indicating that there may be a secondary path-way involved in the repair of UV-induced DNA damage thatallows increased MS11 survival in the absence of RecBCD(Fig. 1A).

The MMS survival assay showed no difference in survival ofthe parental MS11 strain relative to the parental FA1090 strainin all three MMS concentrations tested (Fig. 1B). As shownpreviously, the recB and recD mutants both showed greatlyreduced survivability relative to the parent strain in both theMS11 and FA1090 backgrounds (Fig. 1B) (16). It is notewor-thy, however, that at 0.0025% and 0.005% MMS the FA1090recB and recD mutants had a statistically higher rate of survivalthan the corresponding mutants in MS11, indicating that theremay be a secondary pathway involved in the repair of MMS-induced DNA damage that allows FA1090 to survive betterthan MS11 in the absence of RecBCD (Fig. 1B). Taken to-gether, these results show that the RecBCD pathway is re-quired for N. gonorrhoeae to survive both UV light and MMSdamage and that the MS11 and FA1090 parent strains havesimilar survivabilities in the presence of both DNA-damagingagents.

Since the RecBCD recombination pathway is well estab-lished to be critical for the repair of double-strand breaks, anda differential repair capability between FA1090 and MS11 hasbeen reported, the effect of Nal on survival was a criticalmeasurement. However the measurement of Nal sensitivitywas complicated by different MICs of Nal for each strain. ForFA1090, the survival rate of the parent strain was approxi-mately 5 orders of magnitude higher than those of the recB andrecD mutants at the maximum reported Nal concentration of0.25 �g/ml (Fig. 1C). The survival of the FA1090 rec mutantsat the next-highest Nal concentration tested (0.5 �g/ml) wasundetectable (data not shown). For MS11, the survival rate ofthe parent strain was 4 to 5 orders of magnitude higher thanthose of the recB and recD mutants at the maximum reported

TABLE 1. Growth time and total number of generations forwhich each sample was grown in the presence of IPTG

as determined by CFU/colony assay

Strain Time grownon IPTG (h)

No. ofCFU/colony

at timeof harvest

Calculated no.of generations

on IPTG

FA1090Parental 22 1.25 � 106 20recB 40 1.18 � 106 20

22 1.30 � 104 14recD 40 1.14 � 106 20

22 5.70 � 104 16MS11

Parental 24 8.10 � 105 20recB 40 8.05 � 105 20

24 1.05 � 104 13recD 40 1.14 � 106 20

24 7.20 � 104 16

VOL. 191, 2009 GONOCOCCAL RecBCD 5615

on Septem

ber 8, 2020 by guesthttp://jb.asm

.org/D

ownloaded from

concentration of 0.75 �g/ml (Fig. 1D). The survival of theMS11 rec mutants at the next-highest Nal concentration tested(1 �g/ml) was undetectable (data not shown). The difference insurvival between MS11 and FA1090 on Nal corresponds to thedifference in the MICs of Nal for the strains, with the MIC forFA1090 being approximately threefold lower than the MIC forMS11 (data not shown). It is noteworthy that FA1090 is alsomore sensitive than MS11 to chloramphenicol, Erm, tetracy-cline, rifampin (rifampicin), and ampicillin, all of which utilizemechanisms of action that are independent of DNA damage(data not shown). The difference in antibiotic sensitivity be-tween the two strains is partially the result of differential in-duction of the MtrCDE efflux pump between the two strains.In MS11, the efflux pump is upregulated relative to otherstrains of N. gonorrhoeae (52), whereas in FA1090, the pump isdownregulated due to a mutation in the inducer, mtrA (35),resulting in different sensitivities of the strains to a variety ofantimicrobials.

Determination of pilin Av rates using the sequencing assay.To assess the pilin Av rates of recB and recD mutants in MS11and FA1090, a previously described sequencing assay was per-formed in which RecA expression was induced in N. gonor-rhoeae for a given period of time and the pilE gene was se-quenced from each of the progeny colonies (6, 18). Subsequentto RecA induction, seven P� colonies were picked at randomand grown in the absence of IPTG to lock in the pilE sequence.The pilE sequences from at least 27 progeny arising from eachof the seven progenitor colonies were analyzed, and the pilinAv frequency was determined by dividing the total number of

detected Av events by the total number of progeny coloniesanalyzed for each of the seven progenitors. The pilin Av ratewas then determined by dividing each frequency by the numberof generations of RecA induction. Therefore, each strain hadseven frequencies and seven rates reported, corresponding toone frequency and one rate for each progenitor.

Comparison of pilin Av rates in FA1090 recB and recDmutants relative to the parent strain. Pilin Av frequencieswere determined for FA1090 pilin variant 1-81-S2 and recBand recD mutants in this strain background after they weregrown for approximately 20 generations on IPTG (Table 1).Figure 2A shows the pilin Av frequencies obtained for theprogeny of each of the seven progenitor starter colonies. Thefrequencies of the seven progenitors from each sample rangedfrom 0.067 to 0.23, 0.033 to 0.30, and 0.067 to 0.36 events/cellfor the parent strain and the recB and recD mutants, respec-tively (Fig. 2A). Each of the corresponding frequencies wasdivided by the number of generations grown in the presence ofIPTG (Table 1) to determine the pilin Av rates. The medianrates for the parent and the recB and recD mutants were6.42 �10�3, 1.03 � 10�2, and 4.97 � 10�3 events/cell/genera-tion, respectively (Fig. 2B). The pilin Av rates for all threesamples were indistinguishable by the Wilcoxon rank sum test,with a P value of �0.05. Therefore, in strain FA1090, RecB isnot required to achieve a parental rate of pilin Av after growthfor 20 generations. Furthermore, these results definitivelyshow that a recD mutation neither diminishes nor enhancespilin Av in strain FA1090. This finding confirms previous databased on a colony phase variation assay and a colony PCR

FIG. 1. FA1090 and MS11 survival in the presence of UV, MMS, and Nal. (A) Survival of FA1090 and MS11, and recB and recD mutants ineach background, in the presence of four different doses of UV irradiation. *, MS11 recB and recD had increased survival relative to thecorresponding mutants in FA1090 (P � 0.05 at 50 J/m2 by Student’s t test). (B) Survival of FA1090 and MS11, and recB and recD mutants in eachbackground, in the presence of three different concentrations of MMS. **, FA1090 recB and recD had increased survival relative to thecorresponding mutants in MS11 (P � 0.05 at 0.0025 and 0.005% MMS by Student’s t test). (C) Survival of the FA1090 parent and recB and recDmutants in the presence of five different concentrations of Nal. (D) Survival of the MS11 parent and recB and recD mutants in the presence of fourdifferent concentrations of Nal. The error bars indicate standard errors of the means.

5616 HELM AND SEIFERT J. BACTERIOL.

on Septem

ber 8, 2020 by guesthttp://jb.asm

.org/D

ownloaded from

approach to determine the effect, if any, of RecBCD in thisstrain of N. gonorrhoeae (27).

As has been described previously, recB and recD N. gonor-rhoeae mutants have a significant growth defect relative to theparent strain (16, 27). For this reason, there was a disparity inthe amount of time the FA1090 mutants were grown in thepresence of IPTG relative to the parent strain to achieve ap-proximately 20 generations of growth (Table 1). To ensure thatthe difference in RecA induction times between the parent andthe mutants did not have an impact on the pilin Av rates, werepeated the assay of the recB and recD mutants after only 22 hof growth on IPTG. Because 22 h of growth reduced thenumber of generations from 20 to 14 for recB and 16 for recD(Table 1), more progeny samples were sequenced to increasethe probability that detectable frequencies would be obtained(Fig. 3A).

Predictably, the pilin Av frequencies for the recB and recDmutants were generally reduced compared to the 20-genera-

tion experiment, with ranges from 0.021 to 0.17 and from 0 to0.29 events/cell for the recB and recD mutants, respectively.However, the pilin Av rates remained indistinguishable fromthose of the parent strain grown for 20 generations, with me-dian rates for recB and recD of 4.65 � 10�3 and 3.52 � 10�3

events/cell/generation, respectively. These pilin Av rates areindistinguishable from each other and from that of the parentstrain, with a P value of �0.05 by the Wilcoxon rank sum test,showing that the rate of pilin Av in FA1090 is not dependentupon the number of generations for which RecA is induced(Fig. 3B).

Comparison of pilin Av rates in MS11 recB and recD mu-tants relative to the parent strain. The pilin Av frequencies ofrecB and recD mutants in MS11 pilin variant VD300 weredetermined relative to the parent strain for seven progenitorsof each sample after growth on IPTG for approximately 20generations. The pilin Av frequencies for the parent strain andthe recB and recD mutants ranged from 0 to 0.20, 0 to 0.071,and 0.035 to 0.5 events/cell, respectively (Fig. 4A). The pilin Avrates for each progenitor were determined for the parent andthe recB and recD mutants by dividing the frequencies by thenumber of generations each was grown in the presence of

FIG. 2. Pilin Av frequencies and rates of seven progenitor coloniesof the FA1090 parent and recB and recD mutants after approximately20 generations of RecA induction. (A) Pilin Av frequencies. (B) PilinAv rates. Each data point represents the pilin Av rate of one of sevenprogenitors for the parent (diamonds), recB (crosses), and recD (tri-angles). The median pilin Av rates for the parent and recB and recDmutants were 6.42 � 10�3, 1.03 � 10�2, and 4.97 �10�3 pilin Avevents/cell/generation, respectively, and are denoted with a horizontalbar in each column. The Av rates for all strains were indistinguishable(P � 0.05 by the Wilcoxon rank sum test).

FIG. 3. Pilin Av frequencies and rates of seven progenitor coloniesof FA1090 recB and recD mutants after 22 h of RecA induction.(A) Pilin Av frequencies. (B) Pilin Av rates of FA1090 mutants com-pared to the parent strain grown for 20 generations. Each data pointrepresents the pilin Av rate of one of seven progenitors for the parentstrain (diamonds), recB (crosses), and recD (triangles). The medianrates for the parent and recB and recD mutants were 6.42 � 10�3 (fromFig. 2B), 4.65 � 10�3, and 3.52 � 10�3 pilin Av events/cell/generation,respectively, and are denoted with a horizontal bar in each column.The Av rates for all strains were indistinguishable (P � 0.05 by theWilcoxon rank sum test).

VOL. 191, 2009 GONOCOCCAL RecBCD 5617

on Septem

ber 8, 2020 by guesthttp://jb.asm

.org/D

ownloaded from

IPTG (Table 1). The median pilin Av rates for the parentstrain and the recB and recD mutants were 1.70 � 10�3, 1.73 �10�3, and 3.15 � 10�3 events/cell/generation, respectively (Fig.4B). The pilin Av rates of the mutants and the parent wereindistinguishable by the Wilcoxon rank sum test, with a P valueof �0.05. As with FA1090, these data conclusively show thatpilin Av occurs in recB and recD mutants at a rate similar tothat in the parent strain in MS11 when RecA is induced for 20generations of growth.

As in strain FA1090, recB and recD mutants have growthdefects in strain MS11 (3, 16). For this reason, there was a lackof correspondence in the amount of time the parent strain wasgrown on IPTG relative to the recB and recD mutants toachieve approximately 20 generations, with the parent and recmutants growing for 24 and 40 h, respectively (Table 1). Toensure that the rates determined for the recB and recD mutants

were not the result of the difference in the duration of RecAinduction, the sequencing assay was repeated, with each mu-tant grown on IPTG for only 24 h. The pilin Av frequencies ofeach of the seven progenitors for the recB and recD mutantsranged from 0 to 0.064 and 0 to 0.032 events/cell, respectively(Fig. 5A). Each of the frequencies was divided by the numberof generations for which each mutant was grown in the pres-ence of IPTG to obtain the pilin Av rates (Table 1). Themedian rates for the recB and recD mutants grown for 24 hwere 0 and 1.43 � 10�3 pilin Av events/cell/generation, respec-tively (Fig. 5B). The rates obtained were indistinguishablefrom each other and from that of the parental MS11 straingrown for 20 generations, with a P value of �0.05 by theWilcoxon rank sum test.

Nonrandom incorporation of pilS donors in FA1090. A pre-vious study using a similar sequencing assay in FA1090 re-ported that there was nonrandom incorporation of pilS se-quences involved in pilin Av in P� colonies of FA1090 pilinvariant 1-81-S2 (6). When FA1090 variant pilE sequences fromthe 20-generation assay were aligned with the 19 FA1090 pilScopies and the donor copy was identified, similar results wereobtained in this study (Table 2). Consistent with the previousresults, there was clearly nonrandom incorporation of pilS cop-

FIG. 4. Pilin Av frequencies and rates of seven progenitor coloniesof the MS11 parent and recB and recD mutants after approximately 20generations of RecA induction. (A) Pilin Av frequencies. (B) Pilin Avrates. Each data point represents the pilin Av rate for one of sevenprogenitor colonies for the parent strain (diamonds), recB (crosses),and recD (triangles). The median rates of the parent and recB and recDmutants were 1.70 � 10�3, 1.73 � 10�3, and 3.15 � 10�3 pilin Avevents/cell/generation, respectively, and are denoted with a horizontalbar in each column. The pilin Av rates for all strains were indistin-guishable (P � 0.05 by the Wilcoxon rank sum test).

FIG. 5. Pilin Av frequencies and rates of seven progenitor coloniesof MS11 recB and recD mutants after 24 h of RecA induction. (A) PilinAv frequencies. (B) Pilin Av rates of MS11 mutants compared to theparent strain grown for 20 generations. Each data point represents thepilin Av rate for one of seven progenitor colonies for the parent strain(diamonds), recB (crosses), and recD (triangles). The median rates ofthe parent and recB and recD mutants were 1.70 � 10�3 (from Fig. 4B),0, and 1.43 � 10�3 pilin Av events/cell/generation, respectively, andare denoted with a horizontal bar in each column. The Av rates for allstrains were indistinguishable (P � 0.05 by the Wilcoxon rank sumtest).

5618 HELM AND SEIFERT J. BACTERIOL.

on Septem

ber 8, 2020 by guesthttp://jb.asm

.org/D

ownloaded from

ies, with only 8 of the 19 available pilS copies contributing tothe resulting pilin variants (Table 2). As with the previousstudy, pilS6c1 or pilS2c1 was the most prevalent donor (thesection of sequence incorporated is the same in both of thesepilS copies, so it is impossible to determine which of the two isthe actual donor), contributing to 40.7% of detectable pilin Avevents, and pilS3c1 was the second most prevalent donor at18.5% (Table 2). pilS3c3, pilS2c4, pilS6c3, and pilS2c5 weredetected at low frequencies of �3.3% in P� variants in theprevious study but were not found in any variants of the pa-rental strain in the current work (6) (Table 2).

The recB mutant grown for approximately 20 generations(Table 1) under RecA induction also showed nonrandom in-corporation of pilS copies, although with a slightly differentprofile than the parental strain. The most common donor copywas either pilS1c4 or pilS2c4 at 20% of detectable events (thedonated sequence is identical in the two copies, so it is notpossible to discriminate between them). The next most preva-lent donated copies were pilS3c1, occurring at 13.3% of de-tectable events, and either pilS2c1 or pilS6c1, occurring at16.7% of the observed pilin Av events (Table 2).

The recB mutant grown for 22 h (13.72 generations) onIPTG also maintained a nonrandom donor pilS distribution,with the most prevalent donor copies being pilS2c1 or pilS6c1at 26.3% and pilS3c1 at 26.3% of the observed pilin Av events(Table 2). Although the profile is slightly different from that ofthe recB mutant grown for 20 generations on IPTG, it is note-worthy that either pilS1c4 or pilS2c4 was a donor in the recB

mutant at both times of RecA induction, and neither of thesecopies was found to be a pilin Av donor in the parent strain inthis study (Table 2) and at only 1.1% in the previous work (6)(Table 2).

As with the parent strain and the recB mutant, there was abias in pilS donor copy preferences involved in pilin Av in therecD mutant. In the sample grown for approximately 20 gen-erations under RecA induction, there was a preference foreither pilS2c1 or pilS6c1 and pilS3c1, occurring at 23.3% and30% of the observed pilin Av events, respectively (Table 2).For the recD mutant grown for only 22 h (15.8 generations) inthe presence of IPTG, the most prevalent donor copy waseither pilS2c1 or pilS6c1, occurring at 56.5% of the observedpilin Av events (Table 2). As with the recB mutant, the recDdata represent a potential shift in donor pilS bias relative to theparent strain, although further, more in-depth studies analyz-ing more samples would be required to definitely prove thishypothesis. Taken together, these data show that regardless ofwhich rec mutant is used and regardless of the duration ofgrowth in the presence of IPTG, there remains a bias shifttoward certain pilS donor copies, and therefore, not all avail-able pilin donor copies are utilized with equal efficiency in N.gonorrhoeae pilin Av.

DISCUSSION

In bacteria, DNA recombination is used for genetic ex-change, DNA repair, and, in the case of N. gonorrhoeae, toallow frequent gene conversion between pilS copies and thepilE locus. We demonstrated here that the rate of pilin Av doesnot depend on the RecBCD pathway of recombination, thatthe recB and recD mutants have similar phenotypes for DNArepair and pilin Av, and that the survival rates of two standardlaboratory strains of N. gonorrhoeae in the presence of DNA-damaging agents are similar.

We tested for different roles for RecB and RecD in therepair of UV-induced pyrimidine dimers, Nal-induced dsDNAbreaks, and MMS-induced alkylation lesions in two distinct,independently isolated strains of N. gonorrhoeae (32, 46). Aprevious study using MS11 suggested that RecD was not in-volved in the repair of UV-induced DNA damage (3), while alater study by the same group showed that RecB was involvedin the repair of UV damage in both MS11 and FA1090 (16),suggesting different roles for RecB and RecD in UV repair.However, the results from the current study show that bothRecB and RecD are involved in the repair of UV-inducedDNA damage in both MS11 and FA1090, although a second-ary pathway that enhances the survival of MS11 relative toFA1090 in the absence of RecBCD may be involved (Fig. 1A).

It has been posited that MS11 has a more robust DNArepair phenotype than FA1090 based on the observation thatMS11 recB mutants, and to a degree the MS11 parent strain,survived Nal-induced dsDNA breaks and MMS-induced alkyl-ation lesions better than FA1090 (16). In contrast to the pre-vious study, which used a disc diffusion survival assay, we testedMMS sensitivity by measuring the efficiency of plating on threeconcentrations of MMS. As reported previously, recB and recDmutants in both strain backgrounds had greatly diminishedsurvival relative to the parent strains in a dose-dependentmanner (Fig. 1B). However, contrary to the UV results, sur-

TABLE 2. Distribution of donated pilS copies for FA1090 pilinvariant 1-81-S2 P� colonies

DonatedpilS copya

Distribution (%) for no. of generations:

Parent recB recD

–b �20 �20 �14 �20 �16

2c1 or 6c1c 34.8 40.7 13.3 26.3 23.3 56.53c1 21.7 18.5 16.7 26.3 30.0 8.73c2 17.4 7.4 6.7 6.7 4.31c1 8.3 11.1 15.8 10.0 13.02c4 or 6cc 4.33c3 3.3 10 5.31c5 3.3 7.4 6.7 10.5 8.72c4 1.11c4 1.17c1 1.1 3.76c3 1.12c5 1.16c1 1.1 13.3 10.01c2 3.3 5.32c6 6.72c2 4.31c1 or 3c1c 3.71c4 or 2c4c 20.0 10.52c3 10.02c1d 4.9 6.7 4.36c2d 1.6 3.31c3d 1.6 7.4 3.3

a Refers to the 19 distinct pilS copies available as donors in pilin Av, e.g., “2c1”refers to pilS2c1 and “6c1” refers to pilS6c1.

b Previous data from reference 6 using a larger number of sequence colonies.c The sequence of the DNA transferred from pilS to pilE makes it impossible

to discern which of these two copies is the actual donor.d Criss et al. (6) reported these pilS donors only in P� colonies.

VOL. 191, 2009 GONOCOCCAL RecBCD 5619

on Septem

ber 8, 2020 by guesthttp://jb.asm

.org/D

ownloaded from

vival was enhanced in the FA1090 rec mutants relative to theMS11 rec mutants, suggesting a pathway for UV-induced DNAdamage that enhances FA1090 survival in the absence ofRecBCD.

Since the MIC of Nal for MS11 is over threefold higher thanthat for FA1090, we adjusted the Nal concentrations to accu-rately measure the sensitivities of MS11 and FA1090 todsDNA breaks (Fig. 1C and D). With these adjusted levels ofNal, there were no differences in the survival of MS11 orFA1090, or the recB or recD mutant. Taken together, thesefindings show that there is no difference in DNA repair effi-ciencies between the two parental strains and confirm that bothRecB and RecD are required for repair. It is noteworthy thatFA1090 RecB/RecD has 99% amino acid similarity to MS11RecB/RecD. Also, unlike those of E. coli, N. gonorrhoeae recB,recC, and recD do not exist in an operon. Immediately down-stream of recB is pspE, which encodes a phage shock proteinprecursor, and immediately downstream of recD is Ngo772,which encodes a hypothetical conserved protein that is tran-scribed in the opposite orientation to recD. Therefore, muta-tions in either recB or recD are not predicted to have polareffects on genes involved in recombination.

By utilizing the previously described sequencing assay forpilin Av (6, 18), we conclusively demonstrated that recB andrecD mutants show the same rate of pilin Av regardless of thelength of RecA induction (Fig. 2 to 5). It was previously es-tablished that the RecF-like pathway is required for efficientpilin Av in N. gonorrhoeae (14, 27), which suggests that pilin Avis mediated by an ssDNA gap (27) and not a dsDNA break. Insupport of a gapped recombination substrate model, a numberof nicks that are processed by some RecF-like pathway en-zymes have been detected near the pilE locus in N. gonorrhoeae(L. Cahoon, personal communication). It is possible that thepolyploid nature of N. gonorrhoeae may allow high-frequencygene conversion by the RecF-like pathway and may also influ-ence how DNA exchange is mediated (51). However, we haveno direct evidence of how the polyploidy influences these pro-cesses.

This work shows that E. coli cannot be used as a strict modelfor recombination processes in all gram-negative bacteria. Forexample, E. coli recD mutants are hyperrecombinant and haveno growth defect, while N. gonorrhoeae recD mutants are hy-porecombinant and have a severe growth defect (16, 27). TheN. gonorrhoeae RecF-like pathway appears to repair dsDNAbreaks, but similar mutants in E. coli show no repair phenotype(27). For these reasons, further analysis of the RecF-like path-way and the RecBCD system in N. gonorrhoeae is warranted tomore clearly elucidate the mechanisms involved in their func-tion in neisserial DNA recombination and repair.

Our data suggest why some previous studies have identifieda role of the RecBCD pathway in pilin Av. As shown here andin previous publications, pilS donor copies are not incorpo-rated randomly during pilin Av in FA1090 pilin variant 1-81-S2, and there is a bias toward either pilS2c1 or pilS6c1 as themain donor copy (6) (Table 2). However, in the recB and recDmutants in this strain, the repertoire of silent donor copies wasaltered. For example, pilS2c4 and pilS1c4 were frequent donorcopies in the recB mutant in two different experiments, butneither of these copies was found to be a donor in any of thepilin variants isolated in either the parent strain or the recD

mutant (Table 2). If in fact there is a shift in pilS donor bias inthe recB or recD mutants, it may be responsible for differentialobservations when indirect methods are used to detect pilinAv. First, in the PCR-based assays in which primers are de-signed to anneal to certain pilS donor copies but not others, arec mutant that shifts a preference from one pilS donor copy toanother may result in either an increase or a decrease in thedetected pilin Av rate. A similar caveat exists for the cocon-version assay, as recombination with only the most 3� pilS copyin a pilS locus results in the loss of the marker downstream ofpilE, and a shift in the pilS donor preference could alter theseresults, as well. Additionally, the colony phase variation assaycould also be affected by a shift in the pilS donor copy profile.It has been shown that some pilS donor copies are more pre-dominant in P� colonies and some are more likely to be foundin P� colonies (6, 49). A shift in the pilS donor copy preferencecould definitively alter the results of a colony morphology assaywithout actually altering the frequency or rate of pilin Av in recmutants. While it is clear that RecBCD is not involved in pilinAv, the possibility that these mutants have a different pilSdonor profile than the parent strain is intriguing. One reasonfor this potential difference is the severe growth defect of thesemutants compared to the parent strain. It is possible that aslower-growing cell with a reduced pilin Av frequency (albeitwith a similar pilin Av rate) may incorporate different pilScopies than the faster-growing parent. Therefore, pilin Av as-says that do not allow analysis of the full spectrum of possiblepilS donors may lead to erroneous conclusions.

It is also noteworthy that the pilin Av rate for MS11 in thisstudy was reduced relative to that of FA1090. There are severalpossibilities to account for this observation. The difference inthe pilin Av rates could be due to the fact that each strain hasa different starting pilE sequence (variant VD300 in MS11 andvariant 1-81-S2 in FA1090). It has been demonstrated that thestarting pilE sequence plays an important role in directing thepilin Av frequency and rate (34). There may also be other,non-pilE/pilS strain differences that influence the rate of pilinAv that may indicate rate-limiting steps controlling this spe-cialized recombination system.

ACKNOWLEDGMENTS

We are grateful to Alison Criss for technical advice, to Ben Jones forassistance with statistical analysis, and to Paul Duffin and AdrienneChen for critical reading of the manuscript.

This work was supported by grants R01 AI044239, R01 055977, andR37 AI033493 to H.S.S. R.A.H. was partially supported by F32AI065091.

REFERENCES

1. Amundsen, S. K., J. Fero, L. M. Hansen, G. A. Cromie, J. V. Solnick, G. R.Smith, and N. R. Salama. 2008. Helicobacter pylori AddAB helicase-nucleaseand RecA promote recombination-related DNA repair and survival duringstomach colonization. Mol. Microbiol. 69:994–1007.

2. Boslego, J. W., E. C. Tramont, R. C. Chung, D. G. McChesney, J. Ciak, J. C.Sadoff, M. V. Piziak, J. D. Brown, C. C. Brinton, Jr., S. W. Wood, et al. 1991.Efficacy trial of a parenteral gonococcal pilus vaccine in men. Vaccine 9:154–162.

3. Chaussee, M. S., J. Wilson, and S. A. Hill. 1999. Characterization of the recDgene of Neisseria gonorrhoeae MS11 and the effect of recD inactivation onpilin variation and DNA transformation. Microbiology 145:389–400.

4. Chedin, F., and S. C. Kowalczykowski. 2002. A novel family of regulatedhelicases/nucleases from Gram-positive bacteria: insights into the initiationof DNA recombination. Mol. Microbiol. 43:823–834.

5. Cohen, M. S., and J. G. Cannon. 1999. Human experimentation with Neis-seria gonorrhoeae: progress and goals. J. Infect. Dis. 179(Suppl 2):S375–S379.

5620 HELM AND SEIFERT J. BACTERIOL.

on Septem

ber 8, 2020 by guesthttp://jb.asm

.org/D

ownloaded from

6. Criss, A. K., K. A. Kline, and H. S. Seifert. 2005. The frequency and rate ofpilin antigenic variation in Neisseria gonorrhoeae. Mol. Microbiol. 58:510–519.

7. Dillingham, M. S., and S. C. Kowalczykowski. 2008. RecBCD enzyme andthe repair of double-stranded DNA breaks. Microbiol. Mol. Biol. Rev. 72:642–671.

8. Goodman, M. F. 2002. Error-prone repair DNA polymerases in prokaryotesand eukaryotes. Annu. Rev. Biochem. 71:17–50.

9. Haas, R., and T. F. Meyer. 1986. The repertoire of silent pilus genes inNeisseria gonorrhoeae: evidence for gene conversion. Cell 44:107–115.

10. Haas, R., S. Veit, and T. F. Meyer. 1992. Silent pilin genes of Neisseriagonorrhoeae MS11 and the occurrence of related hypervariant sequencesamong other gonococcal isolates. Mol. Microbiol. 6:197–208.

11. Hagblom, P., E. Segal, E. Billyard, and M. So. 1985. Intragenic recombina-tion leads to pilus antigenic variation in Neisseria gonorrhoeae. Nature 315:156–158.

12. Hamrick, T. S., J. A. Dempsey, M. S. Cohen, and J. G. Cannon. 2001.Antigenic variation of gonococcal pilin expression in vivo: analysis of thestrain FA1090 pilin repertoire and identification of the pilS gene copiesrecombining with pilE during experimental human infection. Microbiology147:839–849.

13. Higgins, P. G., A. C. Fluit, and F. J. Schmitz. 2003. Fluoroquinolones:structure and target sites. Curr. Drug Targets 4:181–190.

14. Hill, S. A. 2000. Neisseria gonorrhoeae recJ mutants show defects in recom-binational repair of alkylated bases and UV-induced pyrimidine dimers. Mol.Gen. Genet. 264:268–275.

15. Hill, S. A., and C. C. Grant. 2002. Recombinational error and deletionformation in Neisseria gonorrhoeae: a role for RecJ in the production of pilE(L) deletions. Mol. Genet. Genomics 266:962–972.

16. Hill, S. A., T. Woodward, A. Reger, R. Baker, and T. Dinse. 2007. Role forthe RecBCD recombination pathway for pilE gene variation in repair-pro-ficient Neisseria gonorrhoeae. J. Bacteriol. 189:7983–7990.

17. Kellogg, D. S., Jr., W. L. Peacock, W. E. Deacon, L. Brown, and C. I. Pirkle.1963. Neisseria gonorrhoeae. I. Virulence genetically linked to clonal varia-tion. J. Bacteriol. 85:1274–1279.

18. Kline, K. A., A. K. Criss, A. Wallace, and H. S. Seifert. 2007. Transposonmutagenesis identifies sites upstream of the Neisseria gonorrhoeae pilE genethat modulate pilin antigenic variation. J. Bacteriol. 189:3462–3470.

19. Kline, K. A., and H. S. Seifert. 2005. Role of the Rep helicase gene inhomologous recombination in Neisseria gonorrhoeae. J. Bacteriol. 187:2903–2907.

20. Koomey, J. M., and S. Falkow. 1987. Cloning of the recA gene of Neisseriagonorrhoeae and construction of gonococcal recA mutants. J. Bacteriol. 169:790–795.

21. Koomey, M. 1998. Competence for natural transformation in Neisseria gonor-rhoeae: a model system for studies of horizontal gene transfer. APMIS Suppl.84:56–61.

22. Koomey, M., E. C. Gotschlich, K. Robbins, S. Bergstrom, and J. Swanson.1987. Effects of recA mutations on pilus antigenic variation and phase tran-sitions in Neisseria gonorrhoeae. Genetics 117:391–398.

23. Kuzminov, A. 1999. Recombinational repair of DNA damage in Escherichiacoli and bacteriophage lambda. Microbiol. Mol. Biol. Rev. 63:751–813.

24. Long, C. D., D. M. Tobiason, M. P. Lazio, K. A. Kline, and H. S. Seifert.2003. Low-level pilin expression allows for substantial DNA transformationcompetence in Neisseria gonorrhoeae. Infect. Immun. 71:6279–6291.

25. Lundin, C., M. North, K. Erixon, K. Walters, D. Jenssen, A. S. Goldman, andT. Helleday. 2005. Methyl methanesulfonate (MMS) produces heat-labileDNA damage but no detectable in vivo DNA double-strand breaks. NucleicAcids Res. 33:3799–3811.

26. Mehr, I. J., C. D. Long, C. D. Serkin, and H. S. Seifert. 2000. A homologueof the recombination-dependent growth gene, rdgC, is involved in gonococ-cal pilin antigenic variation. Genetics 154:523–532.

27. Mehr, I. J., and H. S. Seifert. 1998. Differential roles of homologous recom-bination pathways in Neisseria gonorrhoeae pilin antigenic variation, DNAtransformation and DNA repair. Mol. Microbiol. 30:697–710.

28. Mehr, I. J., and H. S. Seifert. 1997. Random shuttle mutagenesis: gonococcalmutants deficient in pilin antigenic variation. Mol. Microbiol. 23:1121–1131.

29. Mertens, K., L. Lantsheer, D. G. Ennis, and J. E. Samuel. 2008. ConstitutiveSOS expression and damage-inducible AddAB-mediated recombinationalrepair systems for Coxiella burnetii as potential adaptations for survivalwithin macrophages. Mol. Microbiol. 69:1411–1426.

30. Meyer, T. F., E. Billyard, R. Haas, S. Storzbach, and M. So. 1984. Pilus genesof Neisseria gonorrhoeae: chromosomal organization and DNA sequence.Proc. Natl. Acad. Sci. USA 81:6110–6114.

31. Morimatsu, K., and S. C. Kowalczykowski. 2003. RecFOR proteins loadRecA protein onto gapped DNA to accelerate DNA strand exchange. Auniversal step of recombinational repair. Mol. Cell 11:1337–1347.

32. Nachamkin, I., J. G. Cannon, and R. S. Mittler. 1981. Monoclonal antibodiesagainst Neisseria gonorrhoeae: production of antibodies directed against astrain-specific cell surface antigen. Infect. Immun. 32:641–648.

33. Rocha, E. P., E. Cornet, and B. Michel. 2005. Comparative and evolutionaryanalysis of the bacterial homologous recombination systems. PLoS Genet.1:e15.

34. Rohrer, M. S., M. P. Lazio, and H. S. Seifert. 2005. A real-time semi-quantitative RT-PCR assay demonstrates that the pilE sequence dictates thefrequency and characteristics of pilin antigenic variation in Neisseria gonor-rhoeae. Nucleic Acids Res. 33:3363–3371.

35. Rouquette, C., J. B. Harmon, and W. M. Shafer. 1999. Induction of themtrCDE-encoded efflux pump system of Neisseria gonorrhoeae requiresMtrA, an AraC-like protein. Mol. Microbiol. 33:651–658.

36. Sakai, A., and M. M. Cox. 2009. RecFOR and RecOR as distinct RecAloading pathways. J. Biol. Chem. 284:3264–3272.

37. Sechman, E. V., M. S. Rohrer, and H. S. Seifert. 2005. A genetic screenidentifies genes and sites involved in pilin antigenic variation in Neisseriagonorrhoeae. Mol. Microbiol. 57:468–483.

38. Seifert, H. S. 1997. Insertionally inactivated and inducible recA alleles for usein Neisseria. Gene 188:215–220.

39. Seifert, H. S. 1996. Questions about gonococcal pilus phase- and antigenicvariation. Mol. Microbiol. 21:433–440.

40. Seifert, H. S., R. S. Ajioka, D. Paruchuri, F. Heffron, and M. So. 1990.Shuttle mutagenesis of Neisseria gonorrhoeae: pilin null mutations lowerDNA transformation competence. J. Bacteriol. 172:40–46.

41. Seifert, H. S., C. J. Wright, A. E. Jerse, M. S. Cohen, and J. G. Cannon. 1994.Multiple gonococcal pilin antigenic variants are produced during experimen-tal human infections. J. Clin. Investig. 93:2744–2749.

42. Serkin, C. D., and H. S. Seifert. 1998. Frequency of pilin antigenic variationin Neisseria gonorrhoeae. J. Bacteriol. 180:1955–1958.

43. Skaar, E. P., M. P. Lazio, and H. S. Seifert. 2002. Roles of the recJ and recNgenes in homologous recombination and DNA repair pathways of Neisseriagonorrhoeae. J. Bacteriol. 184:919–927.

44. Sparling, P. F. 1966. Genetic transformation of Neisseria gonorrhoeae tostreptomycin resistance. J. Bacteriol. 92:1364–1371.

45. Stohl, E. A., and H. S. Seifert. 2001. The recX gene potentiates homologousrecombination in Neisseria gonorrhoeae. Mol. Microbiol. 40:1301–1310.

46. Swanson, J. 1972. Studies on gonococcus infection. II. Freeze-fracture,freeze-etch studies on gonocci. J. Exp. Med. 136:1258–1271.

47. Swanson, J. 1973. Studies on gonococcus infection. IV. Pili: their role inattachment of gonococci to tissue culture cells. J. Exp. Med. 137:571–589.

48. Swanson, J., S. Bergstrom, K. Robbins, O. Barrera, D. Corwin, and J. M.Koomey. 1986. Gene conversion involving the pilin structural gene correlateswith pilus� in equilibrium with pilus-changes in Neisseria gonorrhoeae. Cell47:267–276.

49. Swanson, J., S. Morrison, O. Barrera, and S. Hill. 1990. Piliation changes intransformation-defective gonococci. J. Exp. Med. 171:2131–2139.

50. Swanson, J., K. Robbins, O. Barrera, D. Corwin, J. Boslego, J. Ciak, M.Blake, and J. M. Koomey. 1987. Gonococcal pilin variants in experimentalgonorrhea. J. Exp. Med. 165:1344–1357.

51. Tobiason, D. M., and H. S. Seifert. 2006. The obligate human pathogen,Neisseria gonorrhoeae, is polyploid. PLoS Biol. 4:e185.

52. Warner, D. M., W. M. Shafer, and A. E. Jerse. 2008. Clinically relevantmutations that cause derepression of the Neisseria gonorrhoeae MtrC-MtrD-MtrE efflux pump system confer different levels of antimicrobial resistanceand in vivo fitness. Mol. Microbiol. 70:462–478.

53. Wolfgang, M., P. Lauer, H.-S. Park, L. Brossay, J. Hebert, and M. Koomey.1998. PilT mutations lead to simultaneous defects in competence for naturaltransformation and twitching motility in piliated Neisseria gonorrhoeae. Mol.Microbiol. 29:312–330.

54. Wolfgang, M., H. S. Park, S. F. Hayes, J. P. van Putten, and M. Koomey.1998. Suppression of an absolute defect in type IV pilus biogenesis byloss-of-function mutations in pilT, a twitching motility gene in Neisseriagonorrhoeae. Proc. Natl. Acad. Sci. USA 95:14973–14978.

55. Zhang, Q. Y., D. DeRyckere, P. Lauer, and M. Koomey. 1992. Gene conver-sion in Neisseria gonorrhoeae: evidence for its role in pilus antigenic variation.Proc. Natl. Acad. Sci. USA 89:5366–5370.

56. Zuniga-Castillo, J., D. Romero, and J. M. Martinez-Salazar. 2004. Therecombination genes addAB are not restricted to gram-positive bacteria:genetic analysis of the recombination initiation enzymes RecF and AddABin Rhizobium etli. J. Bacteriol. 186:7905–7913.

VOL. 191, 2009 GONOCOCCAL RecBCD 5621

on Septem

ber 8, 2020 by guesthttp://jb.asm

.org/D

ownloaded from