Nonrepresentative PCR amplification of variable gene sequences in clinical specimens containing dilute, complex mixtures of microorganisms

(1)

JOURNALOFCLINICAL MICROBIOLOGY,Feb. 1994,p.464-468 Vol. 32, No. 2 0095-1137/94/$04.00+0

Nonrepresentative

PCR

Amplification

of Variable Gene

Sequences in Clinical

Specimens

Containing Dilute,

Complex Mixtures of Microorganisms

CHRISTINAJ.WRIGHT,1 ANN E. JERSE,2 MYRON S. COHEN,3 JANNE G.

CANNON,2

ANDH. STEVENSEIFERT1*

DepartmentofMicrobiology-Immunology, Northwestem University Medical School, Chicago, Illinois

60611,1

andDepartmentof MicrobiologyandImmunology2andDepartment ofMedicine,3 University of

North Carolina atChapel Hill, Chapel Hill, North Carolina27514

Received2September 1993/Returnedfor modification 28 October 1993/Accepted 16 November 1993

PCR amplification and DNA sequencing of the expression locus from Neisseria gonorrhoeae contained in urine sediments collected from

experimentally

infected human subjects produced two observations. First,

different pilin sequences were obtained when separate aliquots of the same sample were amplified and sequenced.Incontrast, the same pilin sequence was obtained when repeated amplifications were performed on individual colonies grown from the clinical samples. Second,mixedsequences (i.e., more than one nucleotide atvariablepositions in the pilin gene sequence)wereobserved in both the direct clinicalisolates and individual cultures grown from the isolates. These results suggest that when clinical samplesare

directly

examined by PCR amplification and sequencing, multiple amplifications may be

required

to detect sequence variants in the sample and

minority

variantsequences will not always be detected.

PCR with clinical specimens of tissue, hair, blood, and

urinefortemplates has been used to generate many valuable

diagnostictestswhichcan screenfor the presence of

infec-tious disease agents orthe genetic predisposition for many

congenital diseases. Because of the limiting amount of

template in many clinical samples, itis often necessary to

amplify target regions of DNA. Although DNA cloning procedures canprovide atemplateforgenetic studies, they aretime-consuming, andPCRtechnology has improvedthe

reliability and efficiency oftemplate amplification for diag-nosticapplications. DNAsequencinghas becomean

impor-tanttoolinbiomedical research, providing informationabout

the

molecularbasisfor manyheritableand infectious human

diseases, but a significant amount of purified template is required for DNA sequence analysis. Once sufficient

tem-plate isisolated, awide variety of methods that use either

single- or double-stranded templates canbe used to deter-mine the DNA sequence ofagene (3, 6,7).

Neisseria gonorrhoeae (gonococcus [GC]) is a

gram-negative diplococcusand thecausativeagentof thesexually

transmitted disease gonorrhea. Pili are filamentous-like structuresemanatingfrom the cell surface oftheGC,which

aidin attachment tohostepithelial cells(8, 14, 15, 18, 21).

Pili are also highly immunogenic and undergo antigenic variation to evade the human immune response and to

changethefunctionalpropertiesof thepilus(4, 5, 11, 19, 22, 24).Oneortwoexpressionloci(pilE)andmultiplesilent loci

(pilS)exist within the GC chromosome

(4, 17). pilE

contains acomplete pilingenethat is abletoproduce pilinmonomers which can be assembled into a pilus, while the multiple,

transcriptionally

silent loci are missing the 5' promoter sequences and conserved coding sequences (4, 16, 19).

* Correspondingauthor.Mailingaddress:Department of Micro-biology-Immunology, Northwestern University Medical School, Mail drop W312, 303 East Chicago Avenue, Chicago, IL 60611. Phone:(312)503-9788. Fax:(312)503-1339.Electronic mail address: [email protected].

Through homologous recombination (10), variant pilin se-quencesfrom thesilent loci aredonated into the expression locus resulting in antigenic variation (4, 5, 19). We are

studying pilinantigenicvariation in N.gonorrhoeae through a human challenge study which uses strain FA1090 to

produce clinical signsofdisease (2, 20). Pilinvariationhas

been examined inbacteria contained in urine sediments and from individual colonies grown in vitro from the clinical

samples.PCRamplificationof theexpressionlocus from the

GCchromosome allowsus tostudy the changesthatoccurin

pilinsequences duringthecolonization and infection of the

male urethra(20). PCR amplification oftarget DNAand then

cycle sequencing reactions with end-labeled primers have provento bereliable andefficientmeansofperformingthis

genetic analysis.

The presence ofmultiple antigenicvariants within urine

sedimentsand urethral swabs collected from the volunteers has created

problemns

in analyzing the changes that occur

duringinfection. We have found that mixed sequences can be observed when theminoritysequenceisaslittleas33%of the sample. In addition, when low amounts of amplifiable template encoding multiple gene sequences are present, individualamplificationsof thesamesamplecanresultin the determination ofdifferentgene sequences.Thus,when

com-plex mixtures of variant gene sequences are present in a clinical sample, multiple amplifications maybe required to

fullycharacterize amixed genepopulation.

MATERIALSANDMETHODS

Origin and collection ofclinical samples. Male volunteers between the ages of 18 and 35 yearswere inoculated with

approximately 106cells taken from 100to 200colonies ofa piliated variant of strain FA1090byusingapediatriccatheter inserted 5 cm into the urethra (2). At 2 h

postinoculation,

urine sampleswerecollected, and from then on, urine and

genitalswabsampleswere takenat three time

points daily.

In addition to urine sediments, single colonyisolateswere

464

on May 15, 2020 by guest

http://jcm.asm.org/

(2)

NONREPRESENTATIVE PCR FROM CLINICAL SPECIMENS 465

A. Expression locus

1 44

mc6 mc5 mc4 mc3

66 88 107 120

m~ -F

mc2 151

_[}-B. PCR

Amplification

i. --m

11.

C.

Cycle

Sequencing

FIG. 1. PCR amplificationandcyclesequencingof thepilin expressionlocus.(A)Schematicdepictionofthepilin expressionlocus of N.

gonorrhoeae. Listed above the drawingarethe minicassettes (minicassettes [mcl 2 to 6;the 3' minicassette 1is notlabeled)which are

proposed regions of variabilitywithin theexpressionlocus(4).The numbers above thedrawingindicatetheamino acidresidue number. The

black boxesareconservedregions,and the white boxesarevariableregionsoftheexpressionlocus.(B) Amplificationofthepilin expression locusoutof theGCchromosome. The dotted linesdepicttheproduct generated byPCRamplification.Primersareindicatedby number,and

the 5'to3' orientation is indicatedbyathinarrow.Primer 1 isPILSTART,andprimer2isSP3A;theseprimerswereused in the firstPCRs

of allsamples (Table1).Primer 3 isPILRBS,andprimer4isSMACLA1;theseprimerswereused for thenestedamplificationfrom the clinical

samples (Table 1). (C)Sequencingstrategy for the variableportionsof theexpressionlocus in GC. Thenumbers indicate the differentprimers usedfor thecycle sequencingreactions, and the thinarrowsindicate their direction.The boldarrowsshow theapproximate regionof the

expression locus,which wassequenced byeachprimer. Primer5 isCONSTF2, primer6 is CYS1R,primer7 isCYS1F, andprimer8 is

PILEND (Table 1).

passaged twice in vitro at 37°C with 5% CO2 on GCB medium (Difco, Detroit, Mich.) with Kellogg supplements (9) and VCN inhibitor (BBL, Cockeysville, Md.) before freezing.All volunteersweretreatedwithceftriaxoneatthe onset of clinical signs (purulent exudate) or at 5 days

postinoculation. All samples were stored at -80°C in 3% Trypticasesoybroth and25% glycerol.

PCR amplification of DNA templates for sequencing. In order to prepare template for PCR amplification, 5 ,ul of thawedgonococcalsamplesweremixed with5 ,ulofcolony

lysisbuffercontaining 1% Triton X-100,2 mMEDTA, and 20 mM Tris (pH 8.5). Template-lysis mixtures were then vortexed for 1min,heatedat94°Cfor 15

min,

vortexedagain for 1min,and used astemplatesforthe PCR.

Because of the lower amountsof bacteria present in the clinicalsamples,nested PCRwasusedtoamplifypilEoutof the bacterial chromosome (Fig. 1). When frozen stocks of isolated colonies were used as templates, only single PCR amplifications were necessary to produce an adequate

amount oftemplate for sequencing. However, nested PCR fromthesesamples providedidenticalsequenceinformation

assingle amplification. Theprimersused inPCRarelistedin Table 1.

Allamplificationsweredone in100-,ulvolumescontaining

50 pmol of each primer, 0.2 mM (each) deoxynucleoside triphosphates (Promega, Madison, Wis.), 50 mM KCl, 10 mMTris-HCl,0.1% TritonX-100,3 mMMgCl2, and 2U of TaqDNApolymerase (Promega). Evaporationof the

reac-tion mixturewaspreventedbyusing paraffinbeads(25).The

thermocycling profileused for the outerprimerpairwas 30 cyclesof94°Cfor 1min, 60°Cfor 1min,and72°Cfor2min,

producingaproductofapproximately780bp (PTC-100,MJ Research, Watertown, Mass.). The thermocycling profile used for the internalprimer pairwas30cyclesof94°Cfor 1 min, 55°Cfor 1min,and72°Cfor 2min, producingaproduct

ofapproximately 630bp.

DNA sequence analysis. Prior to sequencing, the entire PCRamplificationmixturewaspassedthroughaSepharose

TABLE 1. Oligonucleotidesused for PCR andsequencingstrategy

Primer DNAsequencea Description

PILSTART GAGATAAACGCATAAAATTTCACC Sequencein 5' conserved promoterregion ofpilEl used withSP3A SP3A CCGGAACGGACGACCCCG Sequencein3' conserved untranslatedregionof allpilloci PILRBS GGCTTTCCCCTTTCAATTAGGAG Sequenceatribosome-bindingsite ofpilEused withSMACLA1 SMACLA1 CAAACCCTTAAAAGACAGC Sequencein 3' untranslatedregionofallpilloci

CONSTF2 TACCAAGACTACACCGCCCG Sequencein5'conservedregionof allpilloci used forsequencing CYSlR GTCCGCAGAACCATTTTACCG Sequencein conserved cyslregionof allpil lociusedforsequencing CYSlF CGGTAAAATGGTTCTGCGGAC InversecomplementofCYSlR used forsequencing

PILEND CGCTTGATTTATTTAAAATTTAAGG Sequencein 3' untranslatedregionofsomepillociusedforsequencing

a All DNAsequencesarelisted5'to3'.

VOL. 32,1994

on May 15, 2020 by guest

http://jcm.asm.org/

(3)

466 WRIGHT ET AL.

CL-6B (Sigma, St. Louis, Mo.) spin column containing 15 timesthe sample volume ofa60% slurry of resin in TE (10

mM Tris [pH 8.0], 1 mM EDTA). This removed both residual oligonucleotide primers and buffer components fromthe PCR amplification mixture.Approximately 10to50

ngofamplified templatefromthespin-dialyzed PCR mixture was then used for sequencing by using [y-32P]ATP-end-labeled primers and the

finole

DNA Sequencing System (Promega). The end-labeled primers corresponded to

con-served regions within the pilE and generated complete double-stranded sequence data for the variable portions of

each pilin gene (Table 1 and Fig. 1). The thermocycling

profile used with these primers was initial denaturation at

95°C for2min andthen30 cycles of95°C for 30s, 60°C for

30s,and70°Cfor 1min.Thesampleswereheatedat75°C for 2 min and run on a 5% Long Ranger denaturing gel (J. T.

Baker, Phillipsburg, N.J.)at60 W, driedat80°C for 40min, andexposedtoX-OMAT-ARfilm(Kodak, Rochester, N.Y.) overnight.

Sequencingofmixed templates. Purified PCRproducts of two different pilinvariantswere diluted, and their

concen-trations were equalized by matching band intensities on

ethidiumbromide-stainedagarosegels. Thesetwotemplates

were thenmixed at the following ratios: 1:1, 2:1, 4:1, and 8:1.Onetemplatewaskeptat10ng,andtheconcentration of the othertemplate wasvaried by diluting at twofold incre-ments, from 10 to 1.25 ng. Once the templates were

com-bined, the mixture was used as the starting template for

sequencing reactions following the protocol described above.Toanalyze mixedsequencesafterPCRamplification,

serial dilutions of the twostarting templateswere analyzed

byPCR with the nested primers until the minimal

concen-tration that produced a visible product on an ethidium bromide-stained agarose gelwasfound. Approximately 1.6

pgof starting template was consistently ableto produce a

visible PCR product. One templatewas usedat 1.6pg,and

the other templatewas diluted at twofold increments from 1.6to0.2pg.Templateswerethenmixedatratiosof1:1,1:2, 1:4, and 1:8 and amplified by PCR; this was followed by DNAsequencing.

RESULTS ANDDISCUSSION

During our studies of gonococcal pilin variation, we

collected urine sediments containing various numbers of viable bacteria(Table 2). Inaddition, bacteriawereisolated

fromgenitalswabspecimensfromthe time that clinicalsigns appeared. ThepilE locus of the bacteriawas amplified by

PCR by usingeither nested or single rounds ofPCR. The resultant PCR products were then sequenced byusing the

finole DNASequencingkit(Promega)and four oligonucleo-tidesthatproducedthe DNAsequenceof both strands of the variablegenesequences.

Priortothesestudies,thenumber of different variantgene sequences that could be expressed at anyone time during infection was unknown. We assumed that for any one

sample one or two predominant sequences would be

ex-pressed in the population. Our initialobservations showed that whenamplificationwasdone from the urinesediments,

one ortwopilinvariantsequenceswereobserved.However,

whenwereturnedto anidenticalurinesample and

reampli-fiedittocreateanew sequencing template,genesequences

different fromthose found originallywere observed (Table

2). Therefore, it appears that the urine sediments contain mixed populations of bacteria expressing different pilin

genes. Theinitial uniform gene sequences gained from the

TABLE 2. Number of sequences generated from repeated amplifications of directclinical versus colony isolate samples

No.of:

Isolate Viable Different

bacteria Amplifications sequences

Colony isolates

Inoculuma >106 3 1

1-81-S2b >106 2 1

Clinicalsamples

1-2-BU 1.9 x 103/ml 3 3

1-57-BU 1.6 x 102/ml 2 2

2-2-BU 4.7x 102/ml 2 2

aThe FA1090inoculumwasthe sampleused toinitiate infection in the volunteers(20).

bThenumberingsystemfor therestof thesamplesmaybe readasfollows: thefirst number indicates the volunteer fromwhomthe samplewascollected, the second number indicates the time postinoculation (in hours) that the samplewascollected, Bindicatesadirectclinical specimen,Uindicatesa urine sample, and S indicatesasample fromagenitalswab.

urine samples presumably represent one or two variant genes in the population that were amplified early in the nested PCRs. The different sequences found were not a resultof errors produced by the Taq polymerase since the changesoccurred onlyin variablepositions of the pilin gene

(20). Additionally, the number of differences was much greater than that which could be produced by polymerase

misincorporation (20).Thus,eachtime the sample was used toproduceanewtemplate,a newmember of thepopulation was amplified early and produced a different sequence.

Multiple amplifications and sequencing reactions may be

requiredto obtainabetterrepresentation of thepopulation presentwithin any mixed clinical sample.

In addition to the problem of clinical samples yielding different sequence information upon reamplification, se-quenceswerealsomixed because of the presence oftwoor more variant templates that contributed to the sequence

information. To determine the percentage of the total

tem-plate atwhich aparticular pilinvariant mustbe present to provide mixed sequences, we performed template mixing experiments. Low levels of two different templates were

equalized on an ethidium bromide-stained agarose gel and werethenmixedatratios of1:1,2:1, 4:1, and 8:1. The mixed templateswerethenused forcyclesequencing reactions;the

minority sequence became visible at template ratios of 4:1

(Fig. 2B). Inotherwords, thetemplate producing the

sec-ondarysequencemustbe presentatlevels of20%of the total

templateorgreaterin ordertoproduceavisiblebandonthe

autoradiogram. However,itwasnotuntil the 2:1 ratio

(33%

oftotal

template)

that thesecondarysequencebegantorise

tolevels above whatwewould consider normalbackground

levels(Fig. 2B). Template mixing priortoPCRamplification

was done in twoways, keepingone template constant and diluting the other template. Both mixing experiments pro-duced similar results. When the different templates were diluted to the minimal level that produced a detectable

amplification product and were mixed prior to both PCR

amplificationand DNAsequencing,thesecondarysequence wasagainobservableatthe 1:4 ratio oftemplatemixingand abovebackgroundlevelsat the 1:2 ratio(Fig.

2A).

Second-ary sequencesresultingfromtemplatemixingpriorto PCR

amplification were less intense than those resulting from

template mixing andthen direct sequence

analysis.

Similar

results from template mixing experiments were seen by

J.CLIN. MICROBIOL.

on May 15, 2020 by guest

http://jcm.asm.org/

(4)

NONREPRESENTATIVE PCR FROM CLINICAL SPECIMENS 467

1: 1:2 1:4 1:8

\ t (.I -4((;,TI Ax (1CGT ( C

when a variable gene sequence is cloned from a

PCR-amplified sample. -m -O -B. U--s -:-.mp wi . "r-M- A i--

-",

-0

--.W

-

w

or A

( A:'

or(i

1:1 2:1 4:1 8:1 \((.1 -\ ( (. 1 C C (. 1. \{ (

mc.

Us-~

4o

A a

A..._..

C(GCCor(GIAAA

AorT

AorC dwk. X-"q < Z. 4, ";|-b .J .PI

FIG. 2. Templatemixingandits effectonPCRamplificationand DNA sequencing. (A) Templates were mixed at the ratios listed above thepanel,withtemplateBat1.6pgand templateAattwofold dilutions from 1.6to0.2 pg. Points ofsequencevariation between

thetwogenes areindicatedontheright.Mixedtemplateswereused

for PCR amplification and then DNA sequencing, generating the data shown.(B) By usingthesamestock oftemplateas wasused in panel A, templateAremainedat 10ngandtemplateBwasdiluted

at twofold dilutions from 10 to 1.25 ng. Mixed templates were

sequenced directly, and points ofsequencevariationare indicated ontheright.

Leitner et al. (12), who used solid-phase purification of single-stranded template andfluorescence-based automated sequencing. Detection ofsecondary sequences in this

sys-tem was more sensitive, with 10 to 25% oftotal template beingdetectable (12).

When PCR amplification is used to produce sequencing templatefromauniformsource,apureculture,or acellline,

thepossibilityof mixedsequences beingpresentisremote. Alternatively, when clinical or environmental samples are

used to generate sequencing templates, it is oftenpossible thatmore thanonegene sequenceispresent in thesample. The gonococcal pilin gene is unusual in that it is highly

variable, with multiplesequencevariantspossible inasingle

strain. The observations reported here wouldalso relate to

sequence analysis of othervariable genes such asBorrelia

variable major proteins (1), Trypanosoma variable surface glycoproteins (13), or rabbit or chicken immunoglobulin

genes(23).Inaddition,mixedpopulations of different strains of a microorganism would create a similar situation when

more than one version of a gene sequence was present. These observations may also be pertinent when a gene

familyisstudied in aclonalcell population. Ourdata show that ifsingle amplificationsare performed, onlyone of the gene sequences might be recorded and others might be missed. Therefore, inordertoobtain an accurate

represen-tation ofvariantgene sequences expressed in populations, multiplePCRandsequenceanalysesmaybe required. While

this type of selective amplification is an important factor

when samples are sequenced directly, it could also occur

ACKNOWLEDGMENTS

The research described here was supported by the following

grants: R29AI27195, M01R0046, U01AI31494, andU01AI31496. H.S.S. isarecipient ofaJunior Faculty award from the American

Cancer Society.

REFERENCES

1. Barbour, A. G. 1990. Antigenic variation ofa relapsing fever

Borreliaspecies. Annu. Rev. Microbiol. 44:155-171.

2. Cohen, M. S., J. G. Cannon, A. E. Jerse, L.Charniga, S. Isbey, and L.Whicker. J. Infect. Dis., inpress.

3. Gyllensten, U. B.,H. A.Erlich,H. H.Li,X. F.Cui,R. K.Saiki,

and N. Arnheim.1988. Generation ofsingle-stranded DNA by the polymerase chain reaction and its application to direct sequencing of the HLA-DQA locus: amplification and analysis of DNAsequences in single human sperm and diploid cells. Proc. Natl. Acad. Sci. USA 335:414-417.

4. Haas, R., and T. F. Meyer. 1986.Therepertoire of silent pilus

genesinNeisseriagonorrhoeae:evidence forgeneconversion. Cell44:107-115.

5. Hagblom, P., E. Segal, E.Billyard,andM. So.1985.Intragenic recombination leads to pilus antigenicvariation in Neisseria

gononhoeae. Nature(London) 315:156-158.

6. Higuchi, R. G., and H. Ochman. 1989. Production of single-strandedDNAtemplates by exonuclease digestion following the polymerase chain reaction. Nucleic Acids Res. 17:5865. 7. Hultman, T., S. Stahl, E. Hornes, and M. Uhlen. 1989. Direct

solid phase sequencing of genomic and plasmid DNA using magneticbeadsas asolidsupport.Nucleic Acids Res. 17:4937-4946.

8. Johnson, A. P., D.Taylor-Robinson, and Z. A. McGee. 1977. Speciesspecificityof attachment anddamagetooviductmucosa

by Neisseriagonorrhoeae. Infect.Immun. 18:833-839. 9. Kellogg,D. S., Jr., W. L. Peacock, W. E. Deacon, L. Brown, and

C.I. Pirkle.1963. Neisseriagonorrhoeae. I. Virulence

geneti-callylinkedtocolonial variation. J. Bacteriol.85:1274-1279. 10. Koomey, M., E. C. Gotschlich, K Robbins, S.Bergstrom, and J.

Swanson. 1987. Effects ofrecA mutations on pilus antigenic

variation and phase transitions in Neisseria gonorrhoeae. Ge-netics117:391-398.

11. Lambden,P.R., J. E. Heckels, and P. J. Watt. 1982. Effect of anti-pilus antibodiesonsurvival ofgonococci within guinea pig subcutaneouschambers. Infect.Immun.38:27-30.

12. Leitner, T., E. Halapi, G. Scarlatti, P. Rossi, J. Albert, E.-M. Fenyo, and M. Uhlen. 1993. Analysis of heterogeneous viral populations by directDNAsequencing.BioTechniques

15:120-127.

13. Lex, H.T., V. derPloeg, K. Gottesdiener, and M. G. S. Lee.

1992. Antigenic variation in African trypanosomes. Trends

Genet.8:452-456.

14. McGee, Z.A., A. P. Johnson, and D. Taylor-Robinson. 1981. Pathogenic mechanisms of Neisseria gonorrhoeae:

observa-tionsondamage tohumanfallopian tubes inorganculture by

gonococci of colonytype1ortype4.J.Infect. Dis.143:413-422. 15. Melly,M. A.,C. R. Gregg, and Z. A. McGee. 1981. Studies of toxicity of Neisseria gonorrhoeae for human fallopian tube mucosa.J.Infect. Dis.143:423-431.

16. Meyer, T. F., E. Billyard,R. Haas, S. Storzbach, and M. So. 1984. Pilus genesof Neisseriagonorrhoeae: chromosomal

or-ganization and DNA sequence. Proc. Natl. Acad. Sci. USA

81:6110-6114.

17. Meyer, T. F.,N.Mlawer, and M. So. 1982. Pilus expression in Neisseriagonorrhoeae involves chromosomal rearrangement.

Cell 30:45-52.

18. Rudel, T., J. P. M.vanPutten,C. P. Gibbs, R. Haas, and T. F.

Meyer.1992.Interaction oftwovariableproteins (PilE and PilC) required for pilus-mediated adherence of Neisseria gonorrhoeae

tohumanepithelialcells. Mol.Microbiol.6:3439-3450.

19. Segal, E.,P.Hagblom, H.S.Seifert, and M. So. 1986. Antigenic

A.

VOL.32, 1994

a

S.-:0-w mom

ZL

as as 490

I

Jo

4mo

on May 15, 2020 by guest

http://jcm.asm.org/

(5)

J. CLIN. MICROBIOL.

variation of gonococcal pilus involves assembly of separated silentgenesegments.Proc.Natl. Acad. Sci. USA 83:2177-2181. 20. Seifert, H. S., C. Wright, A. Jerse, M. Cohen, and J. G. Cannon.

Submitted forpublication.

21. Stephens, D. S., J. W. Krebs, and Z. A. McGee. 1984. Loss of pili and decreased attachment to human cells by Neisseria meningitidis and Neisseriagonorrhoeae exposedto

subinhibi-toryconcentrations of antibiotics. Infect. Immun.46:507-513. 22. Swanson, J., S. Morrison,0.Barrera,andS. Hill.1990. Piliation

changes in transformation-defective gonococci. J. Exp. Med.

171:2131-2139.

23. Thompson, C. B. 1992. Creation of immunoglobulin diversityby intrachromosomalgeneconversion. Trends Genet. 8:416-421. 24. Virji, M., J. S. Everson, and P. R. Lambden. 1982. Effect of

anti-pilus antiseraonvirulence of variants ofNeisseria

gonor-rhoeae for cultured epithelial cells. J. Gen. Microbiol. 128:1095-1100.

25. Wainwright, L.A.,and H. S. Seifert. 1993. Paraffin beadscan

replace mineral oilasanevaporationbarrier in PCR.

BioTech-niques 14:34-36. 468 WRIGHT ETAL.