Evolution of Antigenic VariationSteven A.Fra - PDF document

Evolution of Antigenic VariationSteven A.FrankDepartment of Ecology and Evolutionary Biology,University of California,Irvine,CA 92697Ð2525,USAprocesses that shape genetic variability and evolutionarychange.The causes of variability and change provide the basisfor understanding why simple vaccines work well againstsome viruses,whereas complex vaccine strategies achieve Encyclopedia of Infectious Diseases:Modern Methodologies,by M.TibayrencCopyright © 2007 John Wiley & Sons,Inc..15.1INTRODUCTIONInfectious disease remains a major cause of morbidity andmortality.Consequently,great research effort has been devotedto parasites and to host immune responses that Þght parasites. 1 shape the landscape in which parasite variants succeed orfail.These coevolutionary processes determine the naturalselection of antigenic variants and the course of evolutionin the parasite population.I discuss several methods thatcan be used to infer the evolutionary processes that shapeantigenic variation.15.2WHY DO PARASITES VARY?In this section,I describe the beneÞts that antigenic variationprovides to parasites.These beneÞts help to explain why par-asites vary in certain ways.15.2.1Extend Length of InfectionTrypanosoma brucei,the protozoan parasite responsible forhuman African trypanosomiasis (formerly sleeping sickness),changes its dominant antigenic surface glycoprotein at a rateof 10per cell division [138].The trypanosomechanges to another surface coat by altering expressionbetween different genes already present in the genome.Infections lead to successive waves of parasitemia and clear-ance as novel antigenic types spread and are then checked byspeciÞc immunity.Some viruses,such as HIV,escape immune attack bymutating their dominant epitopes [86].Mutational changesto new,successful epitopes may be rare in each replication ofthe virus.But the very large population size of viruses with-in a host means that mutations,rare in each replication,oftenFor parasites that produce antigenic variants within hosts,the infection continues until the host controls all variants,raises an immune response against a non-varying epitope,orclears the parasite by nonspeciÞc defenses.Antigenic varia-tion can extend the total time before clearance [36,47,89].Extended infection beneÞts the parasite by increasing thechances for transmission to new hosts.15.2.2Infect Hosts with Prior ExposureHost immune memory recognizes and mounts a rapidresponse against previously encountered antigens.Antigenicvariants that differ from a hostÕs previous infections escapethat hostÕs memory response.The distribution of immunememory proÞles between hosts determines the success ofeach parasite variant.In the simplest case,each antigenic type acts like a sepa-rate parasite that does not cross-react with other variants.Ashost individuals age,they become infected by and recoverfrom different antigenic variants.Thus,the host populationcan be classiÞed by resistance proÞles based on the past infec-tion and recovery of each individual [7].Two extreme cases deÞne the range of outcomes.On theone hand,each variant may occasionally spread epidemicallythrough the host population.This leaves a large fraction ofthe hosts resistant upon recovery,driving that particular vari-ant down in frequency because it has few hosts it can infect.The variant can spread again only after many resistant hostsdie and are replaced by young hosts without prior exposureto that antigen.Variants may,on the other hand,be main-tained endemically in the host population.This requires a bal-ance between the rate at which infections lead to host deathor recovery and the rate at which new susceptible hosts enterthe population.The parasite population maintains as manyvariants as arise and do not cross-react,subject toÒbirthÐdeathÓprocesses governing the stochastic origin ofnew variants and the loss of existing variants.15.2.3Infect Hosts with Genetically VariableHost genotype can inßuence susceptibility to different para-site variants.For example,MHC genotype determines thehostÕs efficiency in presenting particular epitopes to T cells.From the parasiteÕs point of view,a particular antigenic vari-ant may be able to attack some host genotypes but notothers.Hill [55] pointed out that hepatitis B virus provides agood model for studying the interaction between MHC andparasite epitopes.Preliminary reports found associationsbetween MHC genotype and whether infections werecleared or became persistent [6,57,136].The hepatitis B virusgenome is very small (about 3000 base pairs,or bp),whichshould allow direct study of how variation in viral epitopesinteracts with the hostÕs MHC genotype.Host genotype canalso affect the structure of the cellular receptors to which par-asites attach.For example,the human CCR5 gene encodes acoreceptor required for HIV-1 to enter macrophages.A 32 bpdeletion of this gene occurs at a frequency of 0.1 in Europeanpopulations.This deletion prevents the virus from enteringmacrophages [81,96,122].It is not clear whether minor vari-ants of cellular receptors occur sufficiently frequently to favorwidespread matching variation of parasite surface antigens.Several cases of this sort may eventually be found,but in ver-tebrate hosts genetic variation of cellular receptors may be arelatively minor cause of parasite diversity.15.2.4Vary Attachment CharactersParasite surface antigens often play a role in attachment andentry into host cells or attachment to particular types of hosttissue.Varying these attachment characters allows attack ofdifferent cell types or adhesion to various tissues.Such vari-ability can provide the parasite with additional resources orprotection from host defenses.Protozoan parasites of thegenus cause malaria in a variety of vertebratehosts.Several species switch antigenic type [22].Switching has been studied most extensively in falciparum[109].Programmed mechanisms of gene expressionchoose a single gene from among many archival geneticP.falciparumerythrocyte membrane protein 1(PfEMP1) [30].As its name implies,the parasite expresses thisantigen on the surface of infected erythrocytes.PfEMP1induces an antibody response,which likely plays a role in thehostÕs ability to control infection [109].PfEMP1 inßuences cytoadherence of infected erythrocytes to capillary endothe-lia [109].This adherence may help the parasite to avoid clear-ance in the spleen.Thus,antigenic variants can inßuence thecourse of infection by escaping speciÞc recognition and byhiding from host defenses [109].Full understanding of theforces that have shaped the archival repertoire,switchingprocess,and course of infection requires study of both specif-ic immune recognition and cytoadherence properties of thedifferent antigenic variants.HIV provides another example.This virus links its surfaceprotein gp120 to two host-cell receptors before it enters thecell [96].One host-cell receptor,CD4,appears to berequired by most HIV variants (but see [117]).The secondhost-cell receptor can be CCR5 or CXCR4.Macrophagesexpress CCR5.A host that lacks functional CCR5 proteinsapparently can avoid infection by HIV,suggesting that theinitial invasion requires infection of macrophages.HIV iso-lates with tropism for CCR5 can be found throughout theinfection;this HIV variant is probably the transmissive formthat infects new hosts.As an infection proceeds within ahost,HIV variants with tropism for CXCR4 emerge [96].This host-cell receptor occurs on the surface of the CD4(helper) T lymphocytes.The emergence of viral variantswith tropism for CXCR4 coincides with a drop in CD4cells and onset of the immunosuppression that characterizesAIDS.These examples show that variable surface antigensmay sometimes occur because they provide alternative cellor tissue tropisms rather than,or in addition to,escape fromimmune recognition.15.3MECHANISMS THAT GENERATEVARIATIONIn this section,I summarize the different ways in which par-asites generate antigenic variants.The amount of new varia-tion and the kinds of new variants inßuence antigenic poly-morphism and the pace of evolutionary change [36,47,89].15.3.1Mutation and HypermutationRNA virus populations typically have high frequencies ofmutants and often evolve rapidly.However,few studies haveprovided direct estimates of mutation rates.The limited datasuggest relatively high mutation rates on the order of 10per base per replication [31,38,39,106].Drake et al.[38] summarized mutation rates for variousmicrobes with DNA chromosomes.They found an amazinglyconsistent value of approximately 0.003 mutations pergenome per generation.This value holds over genomes thatvary in total size by from 6 bp;conse-quently,the per base mutation rates also vary over four ordersof magnitude from 7 None of the microbes summarized in Drake et al.faceintense,constant selective pressure on antigens imposedby vertebrate immunity Ð for example,it is unlikely thatEscherichia colidepends on antigenic variation to avoidclearance from its hosts.It would be interesting to know ifpathogens under very intense selection by host immunityhave higher baseline mutation rates than related microbesunder less intense immune pressure.High genome-widemutation rates arise in bacteria by spontaneous mutatormutations,in which the mutator alleles raise the error rateduring replication [38].The mutator alleles probably are var-ious DNA replication and repair enzymes.Ten or more genesE.colican develop mutator mutations.Assuming that eachgene has about 1000 bases,the overall mutation rate of muta-,basedon the per base mutation rate in given in Drake et al.[38].Some mutations will be nearly neutral;others will causeextremely high mutation rates and will never increase in fre-quency.Typical E.colicultures accumulate mutator mutants ata frequency of less than 10[78],probably because mostmutations are deleterious and therefore selection does notfavor increased mutation rates.However,mutators can bestrongly favored when the competitive conditions and theselective environment provide opportunities for the mutatorsto generate more beneÞcial mutations than the non-mutators[28,78].In this case,mutators increase because they are linkedwith a higher frequency of beneÞcial mutations.Although mutators are typically rare in freshly grown lab-oratory cultures,hospital isolates of E.colientericasometimes have mutator frequencies above 10[51,65,73].Extensive serial passage in the laboratory can alsolead to high frequencies of mutators [124].Thus,it appearsthat rapid change of hosts or culture conditions can increasethe frequency of mutators 1000-fold relative to stable envi-ronmental conditions.As Drake et al.[38] point out,theorysuggests that mutators can speed adaptation in asexualmicrobes [74,89,135].It would be interesting to comparenaturally occurring frequencies of mutators in stable and rap-idly changing selective environments.Targeting mutations to key loci would be more efficientthan raising the genome-wide mutation rate.Various mecha-nisms can increase the mutation rate over short runs ofnucleotides [47,111].For example,Streptococcus pyogenesits surface with a variable M protein,of which 80 antigeni-cally distinct variants are known [43,71].The amino acidsequence of the M6 serotype revealed repeats in three regionsof the protein [58,59].Sequence analysis of variant M proteins suggests thatmutations occur by generating both gains and losses of theduplications.These mutations probably arise by intragenicrecombination between the DNA repeats,but may be creat-ed by slippage during replication.Slippage mutations overrepeated DNA lead to gain or loss in the number of repeatsand occur at frequencies much higher than typical replicationerrors [29].The repeats of the M protein are multiples ofthree bases;thus changes in repeat number do not causeframeshift mutations.Some of the repeats vary slightly in basecomposition,so recombinations can alter sequence composi-tion as well as total length.Fussenegger [47] reviews severalother cases of bacterial cell-wall proteins that have repeatedCHAPTER 15EVOLUTION OF ANTIGENIC VARIATION sequences,most of which occur in multiples of 3 bp.Repeatsare often associated with binding domains for other proteinsor polysaccharides [145],so perhaps the ability to generatevariable-length domains provides an advantage in attachmentto host tissues or in escape from host immunity.Apart from the well-known case of repeats and replicationslippage,no evidence at present associates antigenic sites withhigher replication errors.But this would certainly be aninteresting problem to study further.One could,for example,focus on associations between mutation rate and nucleotidesequence.Comparison would be particularly interestingbetween epitopes that evolve rapidly and conserved regionsof antigenic molecules that evolve slowly.Such comparisonmay help to identify aspects of nucleotide composition thatpromote higher error rates in replication.In summary,microbial mutation rates per nucleotidedecline with increasing genome size,causing a nearly con-stant mutation rate per genome per generation of about0.003.Genome-wide hypermutation can raise the mutationrate at all sites within the genome.Such mutator phenotypesprobably have altered replication enzymes.Low frequenciesof mutator phenotypes have been observed in stable popula-E.coli,whereas ßuctuating populations appear tomaintain higher frequencies of mutators.In some cases,hypermutation may be targeted to certain genes by DNArepeats and other DNA sequence motifs that promote localreplication errors.15.3.2Stochastic Switching Between ArchivalMany pathogens change critical surface molecules by switch-ing expression between alternative genes.At least four typesof switch mechanisms occur:replication errors that turnexpression on or off,invertible promoters that change thedirection of transcription,gene conversion into Þxed expres-sion sites,and transcriptional silencing of alternative genes.15.3.2.1Regulatory switches by replication errorsofshort repeatsShort,repeated nucleotide sequencesoften lead to high error rates during replication.Repeats haverecurring units typically with 1Ð5 bases per unit.Short,repeated DNA sequences probably lead to replication errorsby slipped-strand mispairing [29,75,87].Errors apparentlyarise when a DNA polymerase either skips forward a repeatunit,causing a deletion of one unit,or slips back one unit,producing a one-unit insertion.Gene expression can beturned on or off by insertions or deletions.Inserted or delet-ed repeats within the coding sequence cause frameshift muta-tions that prevent translation and production of a full protein.For example,the 11 opacity genes of Neisseria meningitidisinßuence binding to host cells and tissue tropism.These geneseach have between eight and 28 CTCTT repeats,which candisrupt or restore the proper translational frame as the num-ber of repeats changes [129,130].The limited repertoire of 11genes and the crude onÐoff switching suggest that variableexpression has more to do with altering cell tropism thanwith escape from host immunity [47].OnÐoff switches canalso be created by short repeats in transcriptional controlregions.Bordetella pertussiscontrols expression of two distinctÞmbriae by transcriptional switching [144].Fimbriae are bac-terial surface Þbers that attach to host tissues.Particular cellsproduce both,only one,or neither of the Þmbrial types.Sequences of about 15 C nucleotides in the transcriptionalpromoters of each of the two genes inßuence expression.Theactual length of the poly-C sequence varies,probably byslipped-strand mispairing during replication.The lengthaffects transcription of the attached gene.Thus,by the sto-chastic process of replication errors,the individual loci areturned on and off.Again,this sort of switching may havemore to do with tissue tropism than with escape fromimmune recognition.15.3.2.2Invertible sequencesE.colistores two alterna-tive Þmbriae genes adjacent to each other on its chromosome[1].A promoter region between the two genes controls tran-scription.The promoter triggers transcription in only onedirection,thus expressing only one of the two variants.Occasionally,the promoter ßips orientation,activating thealternative gene.The ends of the promoter have invertedrepeats,which play a role in the recombination event thatmediates the sequence inversion.mechanism to control ßagellum expression [120].Moraxellaspecies use a different method to vary pilinexpression [79,115].The variable part of the pilin gene hasalternate cassettes stored in adjacent locations.Invertedrepeats ßank the pair of alternate cassettes,causing the wholecomplex occasionally to ßip orientation.The gene starts withan initial constant region and continues into one of the cas-settes within the invertible complex.When the complex ßips,the alternate variable cassette completes the gene.Severalbacteriophage use a similar inversion system to switch genesencoding their tail Þbers,which determine host rangemine host range15.3.2.3Gene conversionSome pathogens store manyvariant genes for a surface antigen,but express only one ofthe copies at any time.For example,there may be a singleactive expression site at which transcription occurs.Occasionally,one of the variant loci copies itself to theexpression site by gene conversion Ð a type of intragenomicrecombination that converts the target without altering thedonor sequence.The genome preserves the archival librarywithout change,but alters the expressed allele.The spirochete Borrelia hermsiihas approximately 30 alter-native loci that encode an abundant surface lipoprotein [12].There is a single active expression site when the spirochete isin mammalian hosts [13].The expression site is changed bygene conversion to one of the variant archival copies at a rateto 10per cell division [14,131].A smallnumber of antigenic variants dominate the initial parasitemiaof this blood-borne pathogen.The host then clears these ini-tial variants with antibodies.Some of the bacteria from this Þrst parasitemia will have changed antigenic type.Thoseswitches provide new variants that cause a second para-sitemia,which is eventually recognized by the host andcleared.The cycle repeats several times,causing relapsingfever.The protozoan T.bruceihas hundreds of alternative locithat encode the dominant surface glycoprotein [16,105].Typically,each cell expresses only one of the alternative loci.Switches in expression occur at a rate of up to 10division [138].The switch mechanism is similar to that inBorrelia hermsiÐ gene conversion of archival copies into atranscriptionally active expression site.T.bruceihas approxi-mately 20 alternative transcription sites,of which only one isusually active.Thus,this parasite can also change expressionby switching between transcription sites.It is not fully under-stood how different transcription sites are regulated.15.3.2.4Transcriptional silencingChanging tran-scriptional activation between different sites appears to be themechanism by which P.falciparumregulates expression of itsmajor surface antigen.P.falciparumexpresses the varwithin erythrocytes.The gene product,PfEMP1,moves tothe surface of the host cell,where it inßuences cellular adhe-sion and recognition by host immunity [37].The varare highly diverse antigenically [133].Each parasite exportsonly one PfEMP1 type to the erythrocyte surface,but a cloneof parasites switches between PfEMP1 types [121].Switchingleads to a diverse population of PfEMP1 variants within ahost and even wider diversity among hosts.It appears thatmany varloci are transcribed during the Þrst few hours aftererythrocyte infection,but only a single vargene transcript isactive when PfEMP1 is translated and moved to the erythro-cyte surface [30,118].It may be that some mechanism shutsdown expression of all but one locus without modifying theDNA sequence.Expression may be inßuenced by an interac-tion between an intron and the promoter,but the details needto be worked out [25,35].15.3.3Intragenomic RecombinationNew variants of alternative genes in archival libraries may becreated by recombination.For example,Rich et al.[110]found evidence for recombination between the archived lociof the variable short protein (Vsp) of B.hermsii.They studiedloci within a single clone.loci are silent,archival copies that can,by geneconversion,be copied into the single expression site.Thegenes differ by 30Ð40% in amino acid sequence,providingsufficient diversity to reduce or eliminate antigenic cross-reactivity within the host.Rich et al.[110] used statisticalsequences to infer that past recombinationevents have occurred between archival loci.Those analysesfocus on attributes such as runs of similar nucleotidesbetween loci that occur more often than would be likely ifalleles diverged only by accumulating mutations within eachlocus.Shared runs can be introduced into diverged loci byrecombination.The archival antigenic repertoire of T.bruceievolves rapidly [105].This species has a large archival libraryand multiple expression sites,but only one expression site isactive at any time.New genes can be created within an activeexpression site when several donor sequences convert the sitein a mosaic pattern [10,103].When an active expression sitebecomes inactivated,the gene within that site probablybecomes protected from further gene conversion events[102,104].Thus,newly created genes by mosaic conversionbecome stored in the repertoire.Perhaps new genes in silentexpression sites can move into more permanent archival loca-tions by recombination,but this has not yet been observed.Recombination between silent,archived copies may alsooccur,which,although each event may be relatively rare,could strongly affect the evolutionary rate of the archivedrepertoire.15.3.4Mixing Between GenomesNew antigenic variants can be produced by mixing genesbetween distinct lineages.This happens in three ways.brings together chromosomes from differentlineages.Reassortment of inßuenza AÕs neuraminidase andhemagglutinin surface antigens provides the most famousexample [70].The genes for these antigens occur on two sep-of eight segments.When two or more viruses infect a singlecell,the parental segments all replicate separately and then arepackaged together into new viral particles.This process canpackage the segments from different parents into a newvirus.New neuraminidaseÐhemagglutinin combinations pres-ent novel antigenic properties to the host.Rare segregationevents have introduced hemagglutinin from bird inßuenzainto the genome of human inßuenza [143].The novelhemagglutinins cross-reacted very little with those circulatingin humans,allowing the new combinations to sweep throughoccurs when chromosomes fromdifferent lineages exchange pieces of their nucleotidesequence.In protozoan parasites such as Trypanosomaspecies (e.g.,),recombination happensas part of a typical Mendelian cycle of outcrossing sex[33,64].Recombination can occur in viruses when two ormore particles infect a single cell.DNA viruses may recom-bine relatively frequently because they can use the hostÕsrecombination enzymes [132].RNA viruses may recombinereciprocal exchange of RNA segments.However,manydescriptions of RNA virus recombination have been report-ed [72,112].In all cases,even rare recombination can providean important source for new antigenic variants.Horizontaltransferof DNA between bacteria introduces new nucleotidesequences into a lineage [97].Transformation occurs when acell takes up naked DNA from the environment.Somespecies transform at a particularly high rate,suggesting thatthey have speciÞc adaptations for uptake and incorporationof foreign DNA [48].For example,Neisseriaspecies transformfrequently enough to have many apparently mosaic genesCHAPTER 15EVOLUTION OF ANTIGENIC VARIATION from interspecies transfers [48,127,148],and N.gonorrhoeaehas low linkage disequilibrium across its genome [86].Horizontal transfer also occurs when bacteriophage virusescarry DNA from one host cell to another or when two cellsconjugate to transfer DNA from a donor to a recipient [97].15.4INTERACTIONS WITH HOST IMMUNITYSpeciÞc immunity favors parasites that change their antigensand escape recognition.In this section,I summarize examplesof parasite escape and the consequences for antigenic diver-15.4.1Natural Selection of Antigenic VariantsIn several pathogens,a changing proÞle of antigenic variantscharacterizes the course of infection within a single host.Natural selection favors variants that escape immune recog-nition,although escape is often temporary.Selection may alsofavor diversiÞcation of the pathogens for the ability to attackdifferent types of host cells.I brießy summarize a few15.4.1.1Simian immunodeÞciency virus (SIV) andSoudeyns et al.[126] identiÞed the regions of theHIV-1 envelope under strong selective pressure by analyzingthe pattern of nucleotide changes in the population.Theycompared the rate of non-synonymous nucleotidereplacements that cause an amino acid change versus the rateof synonymous nucleotide replacements that do not causean amino acid change.A high ratio suggests positive nat-ural selection favoring amino acid change;a low suggests negative natural selection opposing change in aminoacids [100].Soudeyns et al.[126] found that regions of theenvelope gene under strong positive selection correspondedto epitopes recognized by CTLs.The non-synonymous sub-stitutions in these epitopes typically abolished recognition bya matching CTL clone.The population of viruses accumulat-ed diversity in the dominant epitopes over the course ofinfection within hosts.These results suggest that CTL attackbased on speciÞc recognition drives the rapid rate of aminoacid replacements in these epitopes.Kimata et al.[67] studied properties of SIV isolated fromearly and late stages of infection within individual hosts.Theearly viruses infected macrophages,replicated slowly,and theviral particles were susceptible to antibody-mediated clear-ance.The late viruses infected T cells,replicated more than1000 times faster than early viruses,and were less sensitive toantibody-mediated clearance.Kimata et al.[67] did not deter-mine the viral amino acid changes that altered cell tropism ofSIV.Connor et al.[32] found that changes in the host-cellcoreceptors used by early and late HIV-1 correlated withchanges in cell tropism,but it is not yet clear which changesare essential for the virusÕs tropic speciÞcity.Connor et al.[32]did show that the population of early viruses used a narrowrange of coreceptors,whereas the late viruses were highlypolymorphic for a diverse range of host coreceptors.Clearly,the virus is evolving to use various cell types.The relative insensitivity of late SIV to antibody appar-ently depended on increased glycosylation of the envelopeproteins.The late viruses with increased glycosylation werenot recognized by antibodies that neutralized the early viruses.Viruses that escape antibody recognition gain signiÞcantadvantage during the course of infection [27,116].Kimataet al.[67] showed that,when injected into a naive host,thelate SIV did not stimulate as much neutralizing antibody asdid the early SIV.Additional glycosylation apparently reducesthe ability of antibodies to form against the viral surface.Presumably,the glycosylation also hinders the ability of thevirus to initiate infection;otherwise both early and late viruseswould have enhanced glycosylation.Both the early,macrophage-tropic SIV and the late,T cell-tropic SIV usedthe host coreceptor CCR5 [67].That observation contrastswith a study of early and late HIV-1 isolated from individualhosts,in which Connor et al.[32] found that early,macrophage-tropic viruses depended primarily on theCCR5 coreceptors,whereas the population of late viruseshad expanded coreceptor use to include CCR5,CCR3,CCR2b,and CXCR4.Many other studies focus on HIV diversiÞcation withinhosts (e.g.[5,49,117]).15.4.1.2Hepatitis C virus (HCV)Farci et al.[41]obtained HCV samples at various stages of infection withinindividual hosts.They sequenced the envelope genes fromthese samples to determine the pattern of evolution withinhosts.They then compared the evolutionary pattern with theclinical outcome of infection,which follows one of threecourses:clearance in about 15% of cases;chronic infectionand either slowly or rapidly progressive disease in about 85%of cases;and severe,fulminant hepatitis in rare cases.Farci et al.[41] sampled three major periods of infection:the incubation period soon after infection;during thebuildup of viremia but before signiÞcant expression of spe-ciÞc antibodies;and after the hostÕs buildup of speciÞc anti-bodies.The sequence diversity within hosts identiÞed twodistinct regions of the envelope genes.The hypervariableregion evolved quickly and appeared to be under positiveselection from the host immune system,whereas otherregions of the envelope genes had relatively little geneticvariation and did not evolve rapidly under any circumstances.Thus,the following comparisons focus only on the hyper-variable region.Those hosts that eventually cleared the virus had similar orhigher rates of viral diversiÞcation before antibodies appearedthan did those patients that developed chronic infection.Bycontrast,after antibodies appeared,chronic infection was cor-related with signiÞcantly higher viral diversity and rates ofevolution than occurred when the infection was eventuallycleared.It appears that hosts who cleared the infection couldcontain viral diversity and eventually eliminate all variants,whereas those that progressed to chronic infection could not control viral diversiÞcation.The rare and highly virulent ful-minant pattern had low viral diversity and rates of evolution.This lack of diversity suggests either that the fulminant formmay be associated with a single viral lineage that has a strongvirulence determinant or that some hosts failed to mount aneffective immune response.15.4.1.3Generality of within-host evolution of antigensHIV and HCV share several characters that make them par-ticularly likely to evolve within hosts.They are RNA virus-es,which have relatively high mutation rates,relatively sim-ple genomes,simple life cycles,potentially high replicationrates,and potentially high population sizes within hosts.HIVand HCV also typically develop persistent infections withlong residence times in each host.If the mutation rate pernucleotide per replication is 10viruses is of the order of 10within a host,then there arepoint mutations at every site in every generation.Forevery pair of sites,there will usually be at least one virus thatcarries mutations at both sites.Thus,there is a tremendousinßux of mutational variation.Other RNA viruses such asinßuenza also have high mutation rates and potentially largepopulations within hosts,but the hosts typically clear infec-tions within 2 weeks.Some within-host evolution very like-ly occurs,but it does not play a signiÞcant role in the infec-tion dynamics within hosts.DNA-based pathogens producemuch less mutational variation per replication.But largepopulation sizes,long infection times,and hypermutation ofepitopes could still lead to signiÞcant evolution within hosts.At present,the persistent RNA infections have been studiedmost intensively because of their obvious potential for rapidevolutionary change.As more data accumulate,it will beinteresting to compare the extent and the rate of within-hostevolutionary change in various pathogens.15.4.2Pathogen Manipulation of Host Pathogens use several strategies to interfere with hostimmunity.A parasiteÕs exposed surface antigens or candidateCTL epitopes may lack variation because the parasite canrepel immune attack.I do not know of any evidence to sup-port this idea,but it should be considered when studyingcandidate epitopes and their observed level of antigenicvariation.Several reviews summarize viral methods for reducinghost immunity (e.g.[4,128]).Some bacteria also interferewith immune regulation [114].I list just a few viral examples,taken from the outline given by Tortorella et al.[137].Some viruses interfere with MHC presentation of anti-gens.Cases occur in which viruses reduce MHC function atthe level of transcription,protein synthesis,degradation,transport to the cell surface,and maintenance at the cell sur-face.The hostÕs natural killer (NK) cells attack other host cellsthat fail to present MHC class I molecules on their surface.Viruses that interfere with normal class I expression usevarious methods to prevent NK attack,for example,viralexpression of an MHC class I homolog that interferes withNK activation.Host cells often use programed suicide (apoptosis) to con-trol infection.Various viruses interfere with different steps inthe apoptosis control pathway.The host uses cytokines to regulate many immune func-tions.Some viruses alter expression of host cytokines orexpress their own copies of cytokines.Other viruses expressreceptors for cytokines or for the constant (Fc) portion ofantibodies.These viral receptors reduce concentrations offreely circulating host molecules or transmit signals that alterthe regulation of host defense.15.4.3Sequence of Variants in ActiveSome parasites store alternative genes for antigenic surfacemolecules.Each individual parasite usually expresses only oneof the alternatives [36,47].Parasite lineages change expressionfrom one stored gene to another at a low rate.In T.brucei,theswitch rate is about 10per cell division[138].Antigenic switches affect the dynamics of the parasitepopulation within the host.For example,the blood-bornebacterial spirochete B.hermsiicauses a sequence of relapsingfevers [11,12].Each relapse and recovery follows from a spikein bacterial density.The bacteria rise in abundance when newantigenic variants escape immune recognition and fall inabundance when the host generates a speciÞc antibodyresponse to clear the dominant variants.Switches betweentypes within a cellular lineage occur stochastically.But thesequence of variants that dominate sequential waves of para-sitemia tends to follow a repeatable order in T.bruceibruceiand probably in Borrelia[14].Temporal separation in the riseof different antigenic variants allows trypanosomes to contin-ue an infection for a longer period of time [141].If all vari-ants rose in abundance early in the infection,they would allstimulate speciÞc immune responses and be cleared,endingthe infection.If the rise in different variants can be spreadover time,then the infection can be prolonged.The puzzle ishow stochastic changes in the surface antigens of individualparasites can lead to an ordered temporal pattern at the level[3,17,44,139,140].Five hypotheses have been developed,none of which has strong empirical support at present.Ibrießy describe each idea.First,the antigenic variants may differ in growth rate.Those that divide more quickly could dominate the earlyphases of infection,and those that divide more slowly couldincrease and be cleared later in the infection [119].Computerstudies and mathematical models show that variable growthof appearance of different variants [3,69].Only with a verylarge spread in growth rates would the slowest variant be ableto avoid an immune response long enough to develop anextended duration of total infection.Aslam and Turner [8]measured the growth rates of different variants of T.bruceifound little difference between the variants.CHAPTER 15EVOLUTION OF ANTIGENIC VARIATION Second,parasite cells may temporarily express both the oldand new antigens in the transition period after a molecularswitch in antigenic type [3].The double expressers could expe-rience varying immune pressure depending on the time forcomplete antigenic replacement or aspects of cross-reactivity.This would favor some transitions to occur more easily thanothers,leading to temporal separation in the order of appear-ance for different antigenic variants.This model is rathercomplex and has gained little empirical or popular support,as discussed in several papers [2,18,19,21,90].Third,the switch probabilities between antigenic variantsmay be structured in a way to provide sequential dominanceand extended infection [44].If the transition probabilitiesfrom each variant to the other variants are chosen randomly,then an extended sequence of expression does not developbecause the transition pathways are too highly connected.The Þrst antigenic types would generate several variants thatdevelop a second parasitemia.Those second-order variantswould generate nearly all other variants in a random switchmatrix.The variants may arise in an extended sequence if theparasite structures the transition probabilities into separatesets of variants,with only rare transitions between sets.TheÞrst set of variants switches to a limited second set of vari-ants,the second set connects to a limited third set,and so on.Longer infections enhance the probability of transmission toother hosts.Thus,natural selection favors the parasites tostructure their switch probabilities in a hierarchical way inorder to extend the length of infection.Paget-McNicol et al.[101] also developed a model in which switch rates vary,butdid not consider how natural selection might modulateswitch rates.Fourth,Recker et al.[108] noted that hosts with strongercross-reactive immune responses against P.falciparumvariantsare more likely to sustain chronic infections.Presumably,chronic infections mean that the parasiteÕs repertoire of anti-genic surface molecules can be structured into a pattern ofsequential dominance.Based on these points,Recker et al.developed a model in which host immunity develops againsttwo distinct components of the variable surface antigens.One part of the immune response develops lasting immuni-ty against a unique component of each antigen.Another partof the host response develops short-lived immunity against acomponent of the antigenic molecule that is shared by otherantigenic types.With these points in mind,imagine how amalarial infection would play out.One or a few antigenicvariants dominate the initial parasitemia.The host developsspeciÞc immunity against each variant.One part of theimmune response is speciÞc for each variant and long-lived,clearing each variant and preventing another dominant waveof parasitemia by that variant.Another part of the immuneresponse against a particular variant cross-reacts with manyother variants Ð this cross-reactive component lasts only for ashort while.As the initial parasitemia develops,some cells willhave switched expression to other antigenic surface variants.As the Þrst parasitemia clears,the next wave of parasitemiawill develop from those rare variants that are least affected bythe cross-reactive part of the host immune response.As thosefavored types develop into strong parasitemia,the processrepeats,favoring in the subsequent wave those variants thatcross-react least with the previous wave.Molineaux et al.[88]developed a more complex model of P.falciparumdynamics and host immunity.Their model includes Þtted val-ues for how the various components of immunity clear par-asites and variation in growth rate of different variants.This isan interesting analysis,but with so many parameters,it is dif-Þcult to determine whether the good Þt with data arisesfrom so many degrees of freedom or from a model that prop-erly highlights the essential features of antigenic variation.Turner [139] proposed a Þfth explanation for high switchrates and ordered expression of variants.The parasite faces atrade-off between two requirements.On the one hand,com-petition between parasite genotypes favors high rates ofswitching and stochastic expression of multiple variants earlyin an infection.On the other hand,lower effective rates ofswitching later in an infection express variants sequentiallyand extend the total length of infection.Many T.bruceitions in the Þeld probably begin with inoculation by multi-ple parasite genotypes transmitted by a single tsetse ßy vector[77].This creates competition between the multiple genotypes.According to Turner [139],competition intensiÞes the selec-tive pressure on parasites to express many variants Ð variationallows escape from speciÞc immunity by prior infections andhelps to avoid cross-reactivity between variants expressed bydifferent genotypes.These factors favor high rates of stochas-tic switching.The effective rate of switching drops as theinfection progresses because the host develops immunity tomany variants.Effective switches occur when they producenovel variants,and the rate at which novel variants arisedeclines over the course of infection.Those novel variants,when they do occur,can produce new waves of parasitemia,promoting parasite transmission.TurnerÕs idea brings out many interesting issues,particu-larly the role of competition between genotypes within ahost.But his verbal model is not fully speciÞed.For example,delayed expression of some variants and extended infectiondepend on the connectivity of transition pathways betweenvariants,an issue he does not discuss.The problem calls formathematical analysis coupled with empirical study.Connectivity of transition pathways between variants playsan important role in most theories.In Agur et al.Õs [3] model,host immunity acting differentially on double expressers dur-ing the switch process favors some transitions over others.InFrankÕs [44] model,the different rates of molecular switchingbetween variants provides structure to transition pathways.InRecker et al.Õs [108] model,short-lived and cross-reactivehost immunity favors particulars sequences of antigenic dom-inance.TurnerÕs [139] model is not fully speciÞed,but towork it must also provide a tendency for some transitions tobe favored over others Ð this may occur by chance with ran-dom and rare switching or perhaps may favor commonswitches early and rare switches later in the sequence,moreor less as in FrankÕs [44] model. Connectivity of transition pathways has not been studiedempirically.Frank and Barbour [46] have recently discussedthis issue based on reanalysis of earlier data from B.hermsii15.5EXPERIMENTAL EVOLUTIONExperimental evolution manipulates the environment of apopulation and observes the resulting pattern of evolutionarychange.This allows one to study the selective forces thatshape antigenic diversity.For example,one could manipulateimmune selection by exposing parasites to different regimesof monoclonal antibodies.The parasitesÕevolutionaryresponse reveals the adaptive potential and the constraintsthat shape patterns of antigenic variation.In this chapter,I describe experimental evolution studiesof foot-and-mouth disease virus (FMDV).I also use this virusas a case study to show how different methods combine toprovide a deeper understanding of antigenic variation.Theseapproaches include structural analysis of the virion,function-al analysis of epitopes with regard to binding cellular recep-tors,sequence analysis of natural isolates,and experimentalanalysis of evolving populations.15.5.1Antigenicity and Structure of FMDVFMDV is an RNA virus that frequently causes disease indomesticated cattle,swine,sheep,and goats [107].FMDVpopulations maintain antigenic diversity in several rapidlyevolving epitopes [42,85].The most important epitopes occur on the GH loop ofthe VP1 surface protein [82,84,125].This loop has about 20amino acids that contribute to several overlapping epitopes.These antibody-binding sites appear to be determined most-ly by the amino acids in the GH peptide (a continuous epi-tope).Antibodies that bind to an isolated GH peptide alsoneutralize intact viruses.Many antibody escape variants occur in the GH loop,leading to extensive genetic variation in this region.However,a conserved amino acid triplet,Arg-Gly-Asp(RGD),also binds to antibodies.This conserved triplet medi-ates binding to integrin host-cell receptors typically used inFMDV attachment and entry [20,92,125].The GH loop ofVP1 contains continuous epitopes that together deÞne thehypervariable antigenic site A common to all serotypes.Discontinuous epitopes occur when amino acid residuesfrom widely separated sequence locations come togetherconformationally to form a binding surface for antibodies.Two antigenic sites of serotypes A,O,and C have discontin-uous epitopes that have received widespread attentionead attention15.5.2Antibody Escape MutantsMany antibody escape mutants have been sequenced (refer-ences in [80]).One can develop a map of natural escape vari-ants by comparing changes in sequence with differences inTwo problems of interpreting selective pressures arise froman escape map based on natural variants.First,Þeld isolates donot control the multitude of evolutionary pressures on varia-tion.Mutants may spread either in direct response to anti-body pressure,in response to other selective pressures,or bystochastic ßuctuations independent of selective forces.Lackof variability may result either from lack of antibody pressureor from constraining selective pressures such as binding tohost receptors.The second problem for interpreting selective pressuresfrom natural isolates concerns lack of control over geneticbackground.Whether a particular amino acid site affectsantibody affinity may depend on conformation-changingvariants at other sites.Site-directed mutagenesis controls amino acid replace-ments in a Þxed genetic background.One can alter sites thatdo not vary naturally to test for effects on antibody binding.Site-directed mutagenesis has provided useful informationfor FMDV [83].But this method can only deÞne changes inantibody binding;it does not show how viral populationsactually respond to immune pressure.Several studies haveapplied monoclonal or polyclonal antibodies to FMDV inlaboratory culture [82,125].This allows direct control ofselective pressure by comparing lines with and without expo-sure to antibodies.In addition,cultures can be started withgenetically monomorphic viruses to control genetic back-ground.Martinez et al.[80] began laboratory evolution studiesfrom a single viral clone of serotype C.These viruses weregrown on baby hamster kidney cells (BHK-21).All host cellswere derived from a single precursor cell.Two separate virallines were established.C-S8c1 developed through three suc-cessive plaque isolations.C-S8c1p100 began with C-S8c1and developed through 100 serial passages on a monolayer ofBHK-21 cells.The host cells were refreshed from independ-ent stock in each passage and therefore did not coevolve withthe virus over the passage history.In natural isolates,extensive sequence variability in theGH loop of VP1 correlates with escape from antibody neu-tralization.However,the Arg-Gly-Asp (RGD) sequence nearthe center of this GH loop is invariant in Þeld isolates [125].Controlled studies of laboratory evolution provide someinsight into the evolution of this region.The monoclonal antibody SD6 binds to an epitope span-ning residues 136Ð147 in the GH loop of VP1.Martinez et al.[80] applied selective pressure by SD6 after establishment ofthe separate viral lines C-S8c1 and C-S8c1p100 by growing acloned (genetically monomorphic) isolate in the presence ofthe antibody and sampling escape mutants.Nucleotidesequences of escape mutants were obtained.Each mutant(except one) escaped antibody neutralization by a single aminoacid change.The different locations of these mutations in theoriginal (C-S8c1) line compared with the serially passaged (C-S8c1p100) line provide the most striking result of this study.The original line conserved the Arg-Gly-Asp (RGD) motif atpositions 141Ð143.By contrast,the serially passaged line hadCHAPTER 15EVOLUTION OF ANTIGENIC VARIATION numerous mutations within the RGD motif.Figure 15.1 con-trasts the location of mutants for the two lines.Variants in the RGD motif had not previously beenobserved in spite of neutralizing antibodiesÕaffinity for thisregion.The RGD motif was thought to be invariant becauseof its essential role in binding to the host cell.Yet,the serial-ly passaged line accumulated variants in this region.Thosevariants replicated with the same kinetics as the parentalviruses of C-S8c1p100,with no loss in fitness.Baranowskiet al.[9] showed that lineages with an altered RGD motif usean alternative pathway of attachment and entry to host cells.Martinez et al.[80] sequenced the capsid genes from theoriginal line,the serially passaged line,and an escape mutantof the serially passaged line.The escape mutant from the seri-ally passaged line differed from the parental virus of this lineonly at a single site in the RGD region.Tolerance to replace-ments in the RGD region must follow from the differencesaccumulated by C-S8c1p100 during serial passage.Six aminoacids differed between the original and serially passaged lines.Apparently,those substitutions changed cell tropism proper-ties of the virus and allowed variation in the previouslyinvariant RGD motif.15.5.3Cell Binding and TropismAttachment and entry to host cells impose strong naturalselection on some regions of the viral surface.Experimentalevolution provides one approach to analyze those selectiveforces,as described in the previous section.In this section,Ibrießy summarize further studies of amino acid variation inthe FMDV capsid and the consequences for attachment andentry to host cells.Jackson et al.[62] compared the affinityof different viral genotypes for two integrin receptors,.The standard RGD motif was required for bothreceptors.The following amino acid at the RGDtion influenced relative affinity for the two integrin types.For ,several different amino acids at RGD1 allowedbinding,consistent with this receptorÕs multifunctional rolein binding several ligands.By contrast,has narrowerspecificity,favoring a leucine at RGD1.Jackson et al.[62]compared two viruses that differed only at RGD1,thefirst with an arginine and the second with a leucine.Thefirst virus had relatively higher affinity for compared.By contrast,the second virus had relatively highercompared with .For at least someantibodies that recognize RGDL,loss of leucine at1 abolishes recognition (see Fig.15.1) [80].Thirtytype O and eight type A field isolates had leucine at1.By contrast,five SAT-2 isolates had arginine,twoAsia-1 isolates had methionine,and one Asia-1 isolate hadleucine [62].These and other data suggest that mostserotypes have leucine at RGD.SAT-2 may either have greater affinity foror its binding may be conditioned by amino acid vari-In another study,Jackson et al.[63] analyzed FMDV bind-ing to a different integrin,.This integrin binds relative-ly few host ligands and depends on an RGDLXXL motif4.Most FMDV isolateshave leucines at those two positions.does not havestringent requirements at those sites,suggesting that may be an important natural receptor.Overall,RGDLXXLbinds to the widest array of integrins,at least over those stud-ied so far,although relative affinities for different integrinsmay be modulated by substitutions at RGDother sites.It would be interesting to sample isolates fromvarious host tissues that differ in the densities of the variousintegrin receptors and analyze whether any substitutionsappear relative to isolates in other body compartments of theViral success in different cell types or in different hostsmay depend on variations in nonstructural genes that donot mediate binding and entry to host cells.For example,Nunez et al.[95] serially passaged FMDV in guinea pigs.FMDV does not normally cause lesions in guinea pigs,butafter serial passage,viral variants arose that caused disease.Among the several amino acid substitutions that arose dur-ing passage,a single change from glutamine to arginine atposition 44 of gene 3A provided virulence.The function of3A in FMDV is not known.In poliovirus,a distantly relat-ed picornavirus,3A plays a role in virus-specific RNA syn-thesis.These studies show the potential power of experi-mental evolution in studying evolutionary forces,particularly when combined with analysis of naturallyoccurring variation. TTTYTASARGDLAHLTTTHARHLP INEVSVGPDGNIRR Fig.15.1.Amino acid sequence in the central region of the VP1GH loop of FMDV.The start and stop numbers label amino acidpositions.The box shows the RGD motif at positions 141Ð143.Themonoclonal antibody SD6 recognizes the continuous epitopedeÞned by the underlined positions.Black triangles show positionsat which most replacement amino acids greatly reduce binding bySD6;in other words,a single replacement at any of these sites cre-ates an escape mutant.The white triangles denote positions that cantolerate certain amino acid replacements without greatly affectingantibody binding.Unmarked positions in the epitope can varywithout much change in binding.The letters above the sequencesummarize the escape mutants of C-S8c1 (original line);lettersbelow the sequence summarize escape mutants of C-S8c1p100(passaged line).Letters denote amino acids according to the standardsingle-letter code.Redrawn from [80]. 15.5.4Fitness Consequences of SubstitutionsAntibody escape mutants are typically isolated in one of twoways.First,pathogens may be grown This creates selective pressure for substitutions that escapeantibody recognition.Second,naturally occurring variantsfrom Þeld isolates may be tested against a panel of antibodies.Certain sets of antibodies may bind most isolates,allowingidentiÞcation of those variants that differ at commonly rec-ognized epitopes.Escape variants gain a Þtness advantage by avoiding anti-body recognition targeted to important epitopes.However,those pathogen epitopes may also play a role in binding tohost cells,in release from infected cells,or in some otheraspect of the pathogenÕs life cycle.Functional and structuralstudies of amino acid substitutions provide one method ofanalysis.That approach has the advantage of directly assessingthe mechanisms by which amino acid variants affect multiplecomponents of parasite Þtness,such as escape from antibodyrecognition and altered host attachment characteristics.Although functional and structural approaches can directlymeasure binding differences caused by amino acid substitu-tions in different genetic backgrounds,they cannot provide agood measure of all the Þtness consequences associated withchanges in genotype.Carrillo et al.[26] used an alternative approach to analyzethe consequences of amino acid substitutions.They studiedthe relative Þtnesses in vivoof a parental FMDV genotype andthree mutant genotypes derived from the parental type.Theymeasured relative Þtness by competing pairs of strains withinlive pigs.The parental type,C-S8c1,came from a C serotypeisolated from a pig.The Þrst monoclonal antibody-resistantmutant,MARM21,arose in a pig infected with C-S8c1.MARM21 differs from C-S8c1 by a single change from ser-ine to arginine at VP1 139 (Fig.15.1),providing escape fromthe monoclonal antibody SD6.The second mutant,S-3Tcame from a blood sample of a pig 1 day after experimentalinoculation with C-S8c1.That isolate had a single changefrom threonine to alanine at VP1 135 (Fig.15.1).Only oneof 58 monoclonal antibodies differentiated between theparental type and S-3T,and the difference in affinity wassmall.This supports the claim in Figure 15.1 that position 135is not strongly antigenic.The third mutant,C-S15c1,derivedfrom a Þeld variant of type C1 isolated from a pig.Thismutant type had eight amino acid differences in VP1 com-pared with C-S8c1.C-S15c1 did not react with monoclonalOne of the three mutants was coinoculated with theparental type into each experimental pig.Two replicate pigswere used for each of the three pairs of mutant and parentaltypes.Fever rose 1 day after infection and peaked 2 or 3 dayspost infection.All animals developed vesicular lesions 2Ð4days post infection.For each animal,between two and sevensamples were taken from lesions,and the relative proportionsof the competing viruses were assayed by reactivity to mon-oclonal antibodies.The small sample sizes do not allow strongconclusions to be drawn.Rather,the following two resultshint at what might be learned from more extensive studies ofthis sort.First,the parental type strongly dominatedMARM21 in all seven lesions sampled from the two experi-mental animals,comprising between 80% and 94% of theviruses in each lesion.The MARM21 mutation appears toconfer lower Þtness in vivo,at least in the two animals tested.The lower Þtness may arise because the mutant was clearedmore effectively by antibodies,bound less efficiently to hostcells,or had reduced performance in some other Þtness com-ponent.Second,S-3Tabundance relative to the parental typevaried strongly between lesions.In the two lesions analyzedfrom one animal,the parental type comprised 67 standard deviation).In the other ani-mal,the three lesions analyzed had parental-type percentages4.1%%,25 2.8%,and 5.9 1.2%.Differences indominance between lesions also occurred between C-S15c1and the parental type.Variations in dominance may arise fromstochastic sampling of viruses that form lesions,from differ-ences in tissue tropism,or from some other cause.Furtherstudies of this sort may provide a more reÞned understandingof the multiple Þtness consequences that follow from partic-ular amino acid changes,their interactions with the geneticbackground of the virus,the role of different host genotypes,and the effect of prior exposure of hosts to different antigenicvariants.15.6MEASURING SELECTION WITHPOPULATION SAMPLESExperimental evolution provides insight into kinetic andmechanistic aspects of parasite escape from host immunity.Such experimental studies clarify selective forces that inßu-ence change at certain amino acid sites.But experimentalstudies provide only a hint of what actually occurs in naturalpopulations,in which selective pressures and evolutionarydynamics differ signiÞcantly from those in controlled labora-tory studies.It is important to combine experimental insightswith analyses of variation in natural populations.In this sec-tion,I discuss how population samples of nucleotidesequences provide information about natural selection ofantigenic variation.I focus on themes directly related to thegoal of this chapter Ð the synthesis between different kinds ofbiological analyses.In particular,I show how analysis of pop-ulation samples complements studies of molecular structureand experimental evolution.Several books and articles reviewthe methods to analyze population samples and the many dif-ferent types of applications [23,34,56,60,68,76,91,93,94,9815.6.1Positive and Negative SelectionThe genetic code maps three sequential nucleotides (acodon) to a single amino acid or to a stop signal.The four dif-ferent nucleotides combine to make 464 differentCHAPTER 15EVOLUTION OF ANTIGENIC VARIATION codons.The 64 codons specify 20 different amino acids plusa stop signal,leading to an average of $64/21 3 differentcodons for each amino acid or stop signal.This degenerateaspect of the code means that some nucleotide substitutionschange are called synonymous;those that do change theencoded amino acid are called non-synonymous.Synonymoustherefore should not be affected by natural selection of phe-notype.By contrast,non-synonymous substitutions can beaffected by selection because they do change the encodedprotein.If there is no selection on proteins,then the sameforces of mutation and random sampling inßuence allnucleotide changes,causing the rate of non-synonymous,to equal the rate of synonymous substitu-ymous substitu-When natural selection favors change in amino acids,thenon-synonymous substitution rate rises.Thus,measured in a sample of sequences implies that natural selec-tion has favored evolutionary change.This contribution ofselection to the rate of amino acid change above the back-ground measured by positive selection.Parasite epi-topes often show signs of positive selection as they change toescape recognition by host immunity [147].By contrast,negative selection removes amino acidchanges,preserving the amino acid sequence against thespread of mutations.Negative selection reduces the non-syn-onymous substitution rate,causing .The great major-ity of sequences show negative selection,suggesting that mostamino acid replacements are deleterious and are removed bynatural selection.In cases where positive selection does occur,the non-synonymous replacements often cluster on proteinsurfaces involved in some sort of speciÞc recognition.In thesepositively selected proteins,amino acid sites structurally hid-den from external recognition often show the typical signs ofnegative selection.15.6.2Positive Selection to Avoid HostMany examples of positive selection come from genesinvolved in hostÐparasite recognition [40,60,147].Thesesequence analyses provide information about how selectionhas shaped the structure and function of proteins.For exam-ple,one may combine analysis of positive selection withstructural data to determine which sites are exposed to anti-body pressure.In the absence of structural data,sequences canbe used to predict which sites are structurally exposed andcan change and which sites are either not exposed or func-tionally constrained.I brießy summarize one example.The tick-borne protozoan Theileria annulatain cattle [52].The surface antigen Tams1 induces a strongantibody response and has been considered a candidate fordeveloping a vaccine.However,Tams1 varies antigenically;thus studies have focused on the molecular nature of the vari-ability to gain further insight.The structure and function ofTams1 have not been determined.Recently,Gubbels et al.[52] analyzed a population sample of nucleotide sequences topredict which domains of Tams1 change in response to hostimmunity and which domains do not vary because of struc-tural or functional constraints.They found seven domainswith elevated rates of non-synonymous substitutions com-pared with synonymous substitutions (Fig.15.2),suggestingthat these regions may be exposed to antibody pressure.Somedomains had relatively little non-synonymous change,indi-cating that structural or functional constraints preserve aminoacid sequence.These inferences provide guidance in vaccinedesign and point to testable hypotheses about antigenicityand structure.15.6.3Phylogenetic Analysis of NucleotideInitial studies of selection often used small numbers ofsequences,typically fewer than 100.Small sample sizesrequired aggregating observations across all nucleotide sites togain sufficient statistical power.Conclusions focused onwhether selection was positive,negative,or neutral whenaveraged over all sites.With slightly larger samples,one coulddo a sliding window analysis as in Figure 15.2 to infer thekind of selection averaged over sets of amino acids that occurcontiguously in the two-dimensional sequence [40].Majorchanges in binding and antigenicity often require only one ora few amino acid changes [45].The analytical methods that 300400500600700800Nucleotide position at start of sliding windowPercent of maximum Fig.15.2.The seven peaks identify the major regions of positiveselection in the Tams1 protein.The 18 sequences analyzed in this Þg-ure have about 870 nucleotides.The analysis focused on a slidingwindow [40] of 60 nucleotides (20 amino acids).For each windowshown on the -axis,the numbers of non-synonymous and synony-mous nucleotide substitutions were calculated by comparing the 18sequences.The -axis shows the strength of positive selection meas-ured as follows.For each window of 60 nucleotides,each pair ofsequences was compared.Each paired comparison was scored for thestatistical signiÞcance of positive selection based on the numbers ofnon-synonymous and synonymous changes between the pair,with ascore of zero for nonsigniÞcant,a score of 1 for signiÞcant,and ascore of 2 for highly signiÞcant.The maximum score is twice thenumber of comparisons;the actual score is the sum of signiÞcancevalues for each comparison;and the percentage of the maximum isthe actual divided by the maximum multiplied by 100.From [52]. aggregate over whole sequences or sliding windows often failwhich appears to be the proper scale for understanding anti-genic evolution.Recently,larger samples of sequences haveprovided the opportunity to study the rates of synonymousand non-synonymous substitutions at individual nucleotidesites.Each individual substitution occurs within a lineal his-tory of descent,that is,a change occurs between parent andoffspring.To study each substitution directly,one must Þrstarrange a sample of sequences into lineal relationships bybuilding a phylogenetic tree.From the tree,one can infer thenucleotide sequence of ancestors,and therefore trace the his-tory of each nucleotide change through time.Eachnucleotide change can be classiÞed as synonymous or non-synonymous.For each amino acid site,one can sum up thenumbers of synonymous and non-synonymous nucleotidechanges across the entire phylogeny and derive the associatedrates of change.With appropriate statistics,one determinesfor each amino acid site whether non-synonymous changesoccur signiÞcantly more or less often than synonymouschanges [24,53,87,134,142,147].The concepts of measuringpositive and negative selection remain the same.However,forthe Þrst time,the statistical power has been raised to the pointwhere analysis of population samples provides signiÞcantinsight into the evolution of antigens.The power derivesfrom studying the relative success of alternate amino acids ata single site.Important selective forces include the aminoacids at other sites as well as binding properties to hostimmune molecules and other host receptors.Haydon et al.[54] analyzed selection on individual aminoacid sites of FMDV.Most sites showed mild to strong nega-tive selection,as usually occurs.At 17 sites,they found evi-dence of signiÞcant positive selection.Twelve of these posi-tively selected sites occurred at positions that had previouslybeen observed to develop escape mutants in experimentalevolution studies that imposed pressure by monoclonal anti-bodies.The other Þve sites indicate candidates for furtherexperimental analysis.Haydon et al.Õs [54] study of natural isolates gives furtherevidence that a small number of amino acid sites determines alarge fraction of antigenic evolution to escape antibody recog-nition.The combination of analyses on structure,experimen-tal evolution,and natural variation provide an opportunity tostudy how complex evolutionary forces together determinethe evolutionary dynamics of particular amino acids.15.6.4Predicting EvolutionStudies on positive selection in FMDV [54] and HIV [146]could not correlate amino acid substitutions at particular siteswith the actual success of the viruses.In each case,selectionwas inferred strictly from the patterns of nucleotide substitu-Bush et al.Õs [24] study of inßuenza takes the next step byassociating particular amino acid substitutions with the suc-cess or failure of descendants that carry the substitutions.Inßuenza allows such studies because sequences have beencollected each year over the past several decades,providing ahistory of which substitutions have led to success overtime.The inßuenza data can be used to predict future evolu-tion by two steps.First,previous patterns of substitutions andsites contain variants that enhance Þtness.Second,new vari-ants arising at those key sites are predicted to be the progen-itors of future lineages.Bush (this volume) discusses thesemethods applied to the inßuenza data.15.7SHAPE, CHARGE, BINDING KINETICS, cules inßuence binding of those molecules,which deÞnes thenature of hostÐparasite recognition.Binding reactions deter-mine the course of infection within each host,and the advan-tages and disadvantages of different antigenic variants of theparasite.Those advantages and disadvantages set the course ofsuccess for the different variants,changing the frequency ofvariants over time and space and determining the evolutionof antigenic variation.ABBREVIATIONSAIDS:Acquired immunodeÞciency syndromeCTL:Cytotoxic T lymphocyteFMDV:Foot-and-mouth disease virusHCV:Hepatitis C virusHIV:Human immunodeÞciency virusMabs:Monoclonal antibodiesMHC:Major histocompatibility complexNK:Natural killerPlasmodium falciparumerythrocyte membraneproteinSIV:Simian immunodeÞciency virusGLOSSARY:A molecule that induces an immune response.Antigenic variation:Molecular variation between individualparasites in a particular antigenic molecule,usually a speciÞchost immune response directed against one variant does notrecognize other variants as well.Archival copies:Different genetic loci that store and do notexpress variant genes for an antigenic molecule.:The presence or amount of bacteria in the blood.:The reaction of an antibody with an antigenother than the one that gave rise to it.:Strength of binding by a parasite to the surfaceCHAPTER 15EVOLUTION OF ANTIGENIC VARIATION :Molecules secreted by certain cells of the immunesystem that have an effect on other cells.:The part of an antigenmolecule to which an antibody attaches itself;continuous ifin the protein chain;discontinuous if,during protein folding,the epitope forms from disparate parts of the amino acidsequence.Escape mutant:A genetic variant of a parasite epitope in whichthe original type was recognized by a particular host immuneresponse and the mutant is not.Glycosylation:The addition of molecular sugar components toa protein,sometimes protects an antigen from being recog-nized by the host immune response.:An immune cell that devours invading pathogens;stimulates other immune cells by presenting them with smallpieces of the invader.MHC class I:Molecules of the major histocompatibility com-plex that bind small peptides within cells and then present theMHCÐpeptide complex on the surface of cells for interactionwith T cells;class I can stimulate cytotoxic T lymphocytes.Monoclonal antibodies:An antibody produced by a single cloneParasitemia:The presence or amount of parasites in the blood.Polyclonal antibodies:An antibody produced by a multiple dis-tinct clones of cells and consisting of diverse,distinct anti-:A particular molecule on a host cell surface to whichSite-directed mutagenesis:Experimentally controlled mutationto a particular part of a gene,causing a mutational change inTropism:Tendency of a particular parasite variant to bindto a particular kind of cell.:The presence or amount of viruses in the blood.1.Abraham JM,Freitag CS,Clements JR,Eisenstein BI.Aninvertible element of DNA controls phase variation of type 1fimbriae of Escherichia coliProc Natl Acad Sci USA2.Agur Z.Mathematical models for African trypanosomiasis.Parasitol Today3.Agur Z,Abiri D,van der Ploeg,LHT.Ordered appearance ofantigenic variants of African trypanosomes explained in amathematical model based on a stochastic switch process andimmune-selection against putative switch intermediates.Natl Acad Sci USA4.Alcami A,Koszinowski UH.Viral mechanisms of immune eva-Trends Microbiol5.Allen TM,OÕConnor DH,Jing P,et al.Tat-speciÞc cytotoxicT lymphocytes select for SIV escape variants during resolutionof primary viraemia.6.Almarri A,Batchelor JR.HLA and hepatitis B infection.7.Andreasen V,Lin J,Levin SA.The dynamics of cocirculatinginßuenza strains conferring partial cross-immunity.8.Aslam N,Turner CMR.The relationship of variable antigenexpression and population growth rates in Trypanosoma bruceiParasitol Res9.Baranowski E,Ruiz-Jarabo CM,Sevilla N,Andreu D,Beck E,Domingo E.Cell recognition by foot-and-mouth disease virusthat lacks the RGD integrin-binding motif:ßexibility in aph-thovirus receptor usage.10.Barbet AF,Kamper SM.The importance of mosaic genes totrypanosome survival.Parasitol Today11.Barbour AG.Immunobiology of relapsing fever.ContribMicrobiol Immunol12.Barbour AG.Linear DNA of Borreliavariation.Trends Microbiol13.Barbour AG,Burman N,Carter CJ,Kitten T,Bergstrom S.Variable antigen genes of the relapsing fever agent Borreliahermsiare activated by promoter addition.14.Barbour AG,Stoenner HG.Antigenic variation of Borreliahermsii.In Genome Rearrangement(eds I.Herskowitz and M.I.Simon),UCLA Symposium on Molecular and CellularBiology,New Series,Vol.20.Alan Liss,Inc.,New York,1985.15.Barry JD.Antigenic variation during Trypanosoma vivaxtions of different host species.Parasitology16.Barry JD.The relative signiÞcance of mechanisms of antigenicvariation in African trypanosomes.Parasitol Today17.Barry JD,McCulloch R.Antigenic variation in trypanosomes:enhanced phenotypic variation in a eukaryotic parasite.AdvParasitol18.Barry JD,Turner CMR.The dynamics of antigenic variationand growth of African trypanosomes.Parasitol Today19.Barry JD,Turner CMR.Mathematical models for African try-panosomiasis Ð reply.Parasitol Today20.Berinstein A,Roivainen M,Hovi T,Mason PW,Baxt B.Antibodies to the vitronectin receptor (integrin binding and infection of foot-and-mouth disease virus to cul-tured cells.21.Borst P,Rudenko G,Blundell PA,et al.Mechanisms of anti-genic variation in African trypanosomes.Behring Institute22.Brannan LR,Turner CMR,Phillips RS.Malaria parasitesundergo antigenic variation at high rates in vivo.23.Bush RM.Predicting adaptive evolution.24.Bush RM,Bender CA,Subbarao K,Cox NJ,Fitch WM.Predicting the evolution of inßuenza A. 25.Calderwood MS,Gannoun-Zaki L,Wellems TE,Deitsch KW.Plasmodium falciparum vargenes are regulated by two regionswith separate promoters,one upstream of the coding regionand a second within the intron.J Biol Chem26.Carrillo C,Borca M,Moore DM,Morgan DO,Sobrino F.Invivo analysis of the stability and Þtness of variants recoveredfrom foot-and-mouth disease virus quasispecies.J Gen Virol27.Chackerian B,Rudensey LM,Overbaugh J.SpeciÞc N-linkedand O-linked glycosylation modiÞcations in the envelope V1domain of simian immunodeÞciency virus variants that evolvein the host after recognition by neutralizing antibodies.28.Chao L,Cox EC.Competition between high and low mutat-Escherichia coliEvolution29.Charlesworth B,Sniegowski P,Stephan W.The evolutionarydynamics of repetitive DNA in eukaryotes.30.Chen Q,Fernandez V,Sundstrom A,et al.Developmentalvargene expression in Plasmodium falciparum31.Coffin JM.Retroviridae:the viruses and their replication.InFields Virology(eds B.N.Fields,D.M.Knipe,and P.M.Howley),3rd edn.Lippincott-Raven,Philadelphia,1996,pp.763Ð843.32.Connor RI,Sheridan KE,Ceradini D,Choe S,Landau NR.Change in coreceptor use correlates with disease progressionin HIV-1-infected individuals.33.Conway DJ,Roper C,Oduola AMJ,et al.High recombinationPlasmodium falciparum34.Crandall KA (ed.).The Evolution of HIV.Johns HopkinsUniversity Press,Baltimore,MD,1999.35.Deitsch KW,Calderwood MS,Wellems TE.Cooperativevar36.Deitsch KW,Moxon ER,Wellems TE.Shared themes of anti-genic variation and virulence in bacterial,protozoal,and fun-37.Deitsch KW,Wellems TE.Membrane modiÞcations in ery-throcytes parasitized by Plasmodium falciparumMol BiochemParasitol38.Drake JW,Charlesworth B,Charlesworth D,Crow JF.Rates ofspontaneous mutation.39.Drake JW,Holland JJ.Mutation rates among RNA viruses.Proc Natl Acad Sci USA40.Endo T,Ikeo K,Gojobori T.Large-scale search for genes onwhich positive selection may operate.Mol Biol Evol41.Farci P,Shimoda A,Coiana A,et al.The outcome of acute hep-atitis C predicted by the evolution of viral quasispecies.42.Feigelstock DA,Mateu MG,Valero ML,Andreu D,DomingoE,Palma EL.Emerging foot-and-mouth disease virus vari-ants with antigenically critical amino acid substitutions pre-dicted by model studies using reference viruses.Vaccine43.Fischetti VA.Streptococcal M protein.44.Frank SA.A model for the sequential dominance of antigenicvariants in African trypanosome infections.45.Frank SA.Immunology and Evolution of Infectious DiseasePrinceton University Press,Princeton,NJ,2002.46.Frank SA,Barbour AG.Within-host dynamics of antigenicvariation.Infect Genet Evol47.Fussenegger M.Different lifestyles of human pathogenic pro-caryotes and their strategies for phase and antigenic variation.48.Fussenegger M,Rudel T,Barten R,Ryll R,Meyer TF.Transformation competence and type-1 pilus biogenesis inNeisseria gonorrhoeae:a review.49.Goulder PJR,Brander C,Tang Y,et al.Evolution and trans-mission of stable CTL escape mutations in HIV infection.50.Gray AR.Antigenic variation in a strain of Trypanosoma bruceitransmitted by Glossina morsitansG.palpalis51.Gross MD,Siegel EC.Incidence of mutator strains in Escherichiaand coliforms in nature.52.Gubbels M-J,Katzer F,Hide G,Jongejan F,Shiels BR.Generation of a mosaic pattern of diversity in the major mero-zoite-piroplasm surface antigen of Theileria annulataBiochem Parasitol53.Hasegawa M,Rienzo AD,Kocher TD,Wilson AC.Toward amore accurate time scale for the human mitochondrial DNAtree.J Mol Evol54.Haydon DT,Bastos AD,Knowles NJ,Samuel AR.Evidencefor positive selection in foot-and-mouth disease virus capsidgenes from Þeld isolates.55.Hill AVS.The immunogenetics of human infectious diseases.Annu Rev Immunol56.Hillis DM,Moritz C,Mable BK (eds).Sinauer Associates,Sunderland,MA,1996.57.Hohler T,Gerken G,Notghi A,et al.HLA-DRB11302 protect against chronic hepatitis B.58.Hollingshead SK,Fischetti VA,Scott JR.Complete nucleotidesequence of type 6 M protein of the group A streptococcus:repetitive structure and membrane anchor.59.Hollingshead SK,Fischetti VA,Scott JR.Size variation ingroup A streptococcal protein is generated by homologousrecombination between intragenic repeats.60.Hughes AL.Adaptive Evolution of Genes and Genomes.OxfordUniversity Press,Oxford,1999.61.Iida S,Meyer J,Kennedy KE,Arber W.A site-specific,con-servative recombination system carried by bacteriophageP1:mapping the recombinase gene and the crossoverfor the inversion of the C-segment.62.Jackson T,Blakemore W,Newman JWI,et al.Foot-and-mouthdisease virus is a ligand for the high-affinity binding confor-mation of integrin :inßuence of the leucine residue with-in the RGDL motif on selectivity of integrin binding.CHAPTER 15EVOLUTION OF ANTIGENIC VARIATION 63.Jackson T,Sheppard D,Denyer M,Blakemore WE,KingAMQ.The epithelial integrin is a receptor for foot-and-mouth disease virus.64.Jenni L,Marti S,Schweizer J,et al.(1986).Hybrid formationbetween trypanosomes during cyclical transmission.65.Jyssum K.Observations on two types of genetic instability inEscherichia coliActa Pathol Microbiol Scand66.Kamp D,Kahmann R,Zipser D,Broker TR,Chow LT.Inversion of the G DNA segment of phage Mu controls phageinfectivity.67.Kimata JT,Kuller L,Anderson DB,Dailey P,Overbaugh J.Emerging cytopathic and antigenic simian immunodeÞciencyvirus variants influence AIDS progression.68.Kimura M.The Neutral Theory of Molecular EvolutionCambridge University Press,Cambridge,1983.69.Kosinski RJ.Antigenic variation in trypanosomes:a computeranalysis of variant order.Parasitology70.Lamb RA,Krug RM.Orthomyxoviridae:the viruses and theirreplication.In Fields Virology(eds D.M.Knipe and P.M.Howley),4th edn.Lippincott-Raven,Philadelphia,2001,pp.1487Ð531.71.LanceÞeld RC.Current knowledge of type-speciÞc M anti-gens of group A streptococci.J Immunol72.Laukkanen T,Carr JK,Janssens W,et al.Virtually full-lengthsubtype F and F/D recombinant HIV-1 from Africa and SouthAmerica.Virology73.LeClerc JE,Li B,Payne WL,Cebula TA.High mutation fre-Escherichia colipathogens.74.Leigh EG.Natural selection and mutability.75.Levinson G,Gutman GA.Slipped-strand mispairing:a majormechanism for DNA sequence evolution.Mol Biol Evol76.Li W-H.Molecular Evolution.Sinauer Associates,Sunderland,MA,1997.77.MacLeod A,Turner CMR,Tait A.A high level of mixed T.bru-infections in tsetse ßies detected by three hypervariableMol Biochem Parasitol78.Mao EF,Lane L,Lee J,Miller JH.Proliferation of mutators inJ Bacteriol79.Marrs CF,Reuhl WW,Schoolnik GK,Falkow S.Pilin genephase variation of Moraxella bovisis caused by an inversion ofthe pilin gene.J Bacteriol80.Martinez MA,Verdaguer N,Mateu MG,Domingo E.Evolutionsubverting essentiality:dispensability of the cell attachment Arg-Gly-Asp motif in multiply passaged foot-and-mouth diseasevirus.Proc Natl Acad Sci USA81.Martinson JJ,Chapman NH,Rees DC,Liu Y-T,Clegg JB.Global distribution of the CCR5 gene 32-basepair deletion.82.Mateu MG.Antibody recognition of picornaviruses andescape from neutralization:a structural view.83.Mateu MG,Escarmis C,Domingo E.Mutational analysis of dis-continuous epitopes of foot-and-mouth disease virus using anunprocessed capsid promoter precursor.84.Mateu MG,Hernandez J,Martinez MA,et al.Antigenic het-erogeneity of a foot-and-mouth disease virus serotype in theÞeld is mediated by very limited sequence variation at several85.Mateu MG,Silva DA,Rocha E,et al.Extensive antigenic het-erogeneity of foot-and-mouth disease virus of serotype C.Virology86.McMichael AJ,Phillips RE.Escape of human immunodeÞ-ciency virus from immune control.Annu Rev Immunol87.Meyer TF.Molecular basis of surface antigen variation inNeisseriaTrends Genet88.Molineaux L,Diebner HH,Eichner M,Collins WE,JefferyGM,Dietz K.Plasmodium falciparumparasitaemia described bya new mathematical model.Parasitology89.Moxon ER,Rainey PB,Nowak MA,Lenski RE.Adaptiveevolution of highly mutable loci in pathogenic bacteria.Curr90.Munoz-Jordan JL,Davies KP,Cross GA.Stable expression ofmosaic coats of variant surface glycoproteins in Trypanosoma91.Nee S,Holmes EC,Rambaut A,Harvey PH.Inferring popu-lation history from molecular phylogenies.Philos Trans R Soc92.Neff S,Sa-Carvalho D,Rieder E,et al.Foot-and-mouth dis-ease virus virulent for cattle utilizes the integrin receptor.93.Nei M.Molecular Evolutionary Genetics.Columbia UniversityPress,New York,1987.94.Nei M,Kumar S.Molecular Evolution and Phylogenetics.OxfordUniversity Press,Oxford,2000.95.Nunez JI,Baranowski E,Molina N,et al.A single amino acidsubstitution in nonstructural protein 3A can mediate adapta-tion of foot-and-mouth disease virus to the guinea pig.96.OÕBrien SJ,Dean M.In search of AIDS-resistance genes.97.Ochman H,Lawrence JG,Groisman EA.Lateral gene transferand the nature of bacterial innovation.98.Otto SP.Detecting the form of selection from DNA sequenceTrends Genet99.Overbaugh J,Bangham CRM.Selection forces and constraintson retroviral sequence variation.100.Page RDM,Holmes EC.Molecular Evolution:A PhylogeneticApproach.Blackwell ScientiÞc,Oxford,1998.101.Paget-McNicol S,Gatton M,Hastings I,Saul A.ThePlasmodium falciparum vargene switching rate,switching mech-anism and patterns of parasite recrudescence described byParasitology102.Pays E.Gene conversion in trypanosome antigenic variation.Prog Nucleic Acid Res Mol Biol103.Pays E.Pseudogenes,chimeric genes and the timing of anti-gen variation in African trypanosomes.Trends Genet104.Pays E,Lheureux M,Steinert M.The expression-linked copyof the surface antigen gene in Trypanosomais probably the onetranscribed. 105.Pays E,Nolan D.Expression and function of surface proteinsTrypanosoma bruceiMol Biochem Parasitol106.Preston BD,Dougherty JP.Mechanisms of retroviral mutation.Trends Microbiol107.Racaniello VR.Picornaviridae:the viruses and their replication.Fields Virology(eds D.M.Knipe and P.M.Howley),4th edn.Lippincott-Raven,Philadelphia,2001,pp.685Ð722.108.Recker M,Nee S,Bull PC,Kinyanjui S,Marsh K,NewboldC,Gupta S.Transient cross-reactive immune responses canorchestrate antigenic variation in malaria.109.Reeder JC,Brown GV.Antigenic variation and immune eva-Plasmodium falciparummalaria.Immunol Cell Biol110.Rich SM,Sawyer SA,Barbour AG.Antien polymorphism inBorrelia hermsii,a clonal pathogenic bacterium.Proc Natl Acad111.Ripley LS.Predictability of mutant sequences:relationshipsbetween mutational mechanisms and mutant speciÞcity.New York Acad Sci112.Robertson DL,Hahn BH,Sharp PM.Recombination inAIDS viruses.J Mol EvolRodrigo AG,Learn GH,Jr (eds).Evolutionary Analysis of HIV Molecular Sequences.KluwerAcademic Publishers,Boston,2000.114.Rottem S,Naot Y.Subversion and exploitation of host cells bymycoplasmas.Trends Microbiol115.Rozsa WF,Marrs CF.Interesting sequence differences betweenthe pilin gene inversion regions of Moraxella lacunataATCCMoraxella bovisJ BacteriolRudensey LM,Kimata JT,Long EM,Chackerian B,Overbaugh J.Changes in the extracellular envelope glycopro-tein of variants that evolve during the course of simianimmunodeÞciency virus SIVMne infection affect neutralizingantibody recognition,syncytium formation,and macrophagetropism but not replication,cytopathicity,or CCR-5 corecep-tor recognition.117.Saha K,Zhang J,Gupta A,Dave R,Yimen M,Zerhouni B.Isolation of primary HIV-1 that target CD8using CD8 as a receptor.118.Scherf A,Hernandez-Rivas R,Buffet P,et al.Antigenic varia-tion in malaria:in situ switching,relaxed and mutually exclu-sive transcription of vargenes during intra-erythrocytic devel-Plasmodium falciparum119.Seed JR.Competition among serologically different clones ofTrypanosoma brucei gambiensein vivo.J Protozool120.Silverman M,Zieg J,Hilmen M,Simon M.Phase variation in:genetic analysis of a recombinational switch.Natl Acad Sci USA121.Smith JD,Chitnis CE,Craig AG,et al.Switches in expressionPlasmodium falciparum vargenes correlate with changes inantigenic and cytoadherent phenotypes of infected erythro-122.Smith MW,Dean M,Carrington M,et al.Contrasting genet-ic inßuence of CCR2 and CCR5 variants on HIV-1 infectionand disease progression.123.Smith MJ,Smith NH,OÕRourke M,Spratt BG.How clonalare bacteria? Proc Natl Acad Sci USA124.Sniegowski PD,Gerrish PJ,Lenski RE.Evolution of highmutation rates in experimental populations of E.coli125.Sobrino F,Saiz M,Jimenez-Clavero MA,et al.Foot-and-mouth disease virus:a long known virus,but a current threat.Vet Res126.Soudeyns H,Paolucci S,Chappey C,et al.Selective pressureexerted by immunodominant HIV-1-speciÞc cytotoxic Tlymophocyte responses during primary infection drives genet-ic variation restricted to the cognate epitope.Eur J Immunol127.Spratt BG,Bowler LD,Zhang Q-Y,Zhou J,Smith JM.Roleof interspecies transfer of chromosomal genes in the evolutionof penicillin resistance in pathogenic and commensal NeisseriaJ Mol Evol128.Spriggs MK.One step ahead of the game:viral immunomod-ulatory molecules.Annu Rev Immunol129.Stern A,Brown M,Nickel P,Meyer TF.Opacity genes inNeisseria gonorrhoeae:control of phase and antigenic variation.130.Stern A,Meyer TF.Common mechanism controlling phaseand antigenic variation in pathogenic neisseriae.131.Stoenner HG,Dodd T,Larsen C.Antigenic variation ofBorrelia hermsiiStrauss EG,Straus JH,Levine AJ.Virus evolution.InFundamental Virology(eds B.N.Fields,D.M.Knipe,and P.M.Howley),3rd edn.Lippincott-Raven,Philadelphia,1996,pp.133.Su X,Heatwole VM,Wertheimer SP,et al.A large and diversegene family (var) encodes 200Ð350 kD proteins implicated inthe antigenic variation and cytoadherence of Plasmodium falci-Ðinfected erythrocytes.134.Suzuki Y,Gojobori T.A method for detecting positive selec-Mol Biol Evol:1315Ð28.135.Taddei F,Radman M,Smith JM,Toupance B,Gouyon PH,Godelle B.Role of mutator alleles in adaptive evolution.136.Thursz M,Kwiatkowski D,Allsopp CEM,Greenwood BM,Thomas HC,Hill AVS.Association of an HLA class II allelewith clearance of hepatitis B virus infection in The Gambia.137.Tortorella D,Gewurz BE,Furman MH,Schust DJ,Ploegh HL.Viral subversion of the immune system.Annu Rev Immunol138.Turner CMR.The rate of antigenic variation in ßy-transmit-ted and syringe-passaged infections of Trypanosoma brucei139.Turner CMR.Antigenic variation in Trypanosoma bruceitions:an holistic view.140.Turner CMR.A perspective on clonal phenotypic (anti-genic) variation in protozoan parasites.Parasitology:S17Ð23.141.Vickerman K.Trypanosome sociology and antigenic variation.Parasitology142.Wakely J.Substitution rate variation among sites in hypervari-able region I of human mitochondrial DNA.J Mol EvolCHAPTER 15EVOLUTION OF ANTIGENIC VARIATION 143.Webster RG,Shortridge KF,Kawaoka Y.Inßuenza:interspeciestransmission and emergence of new pandemics.Immunol Med Microbiol144.Willems R,Paul A,van der Heide HGJ,ter Avest AR,Mooi FR.Fimbrial phase variation in Bordetella pertussis:a novel mecha-nism for transcriptional regulation.145.Wren BW.A family of clostridial and streptococcal ligand-binding proteins with conserved C-terminal repeat sequences.146.Yamaguchi-Kabata Y,Gojobori T.Reevaluation of amino acidvariability of the human immunodeÞciency virus type 1gp120 envelope glycoprotein and prediction of new discon-tinuous epitopes.147.Yang ZH,Bielawski JP.Statistical methods for detectingTrends Ecol Evol148.Zhou J,Spratt BG.Sequence diversity within the Neisseria meningitidisÑinterspecies reco-gene.