Response from Vilgalys et al

C O M M E N T Phosphoglucomutase I Glucose phosphate isomerase I [ - - I [--I f--I [--I f--I I Genotypes: 1/1 2/3 I 3/3 1/2 and scaling the population genetic methods used 2,3 would contribute greatly to the accurate quantification and comparison of different microbial species. Moreover, the additional use of cladistic analysis would be a useful complement to population genetic analysis. References 1 Jenni, L. et al. (1986)Nature 322, 173-175 2 Tibayrenc,M. (1995)Adv. Parasitol. 36, 47-115 3 Tibayrenc,M. (1996)Annu. Rev. Microbiol. 50, 401-429 4 Tibayrenc,M. et al. (1991)Proc. Natl. Acad. Sci. U. S. A. 88, 5129-5133 5 Caugant,D.A. and Sandven,P. (1993) J. Clin. Microbiol. 31,215-220 6 Pujol,C. etal. (1993)Proc.Natl. Acad. Sci. U. S. A. 90, 9456-9459 7 Gr~ser,Y. et al. (1996)Proc. Natl. Acad. Sci. U. S. A. 93, 12473-12477 Fig. 1. Genetic variation in Trypanosoma cruzi, a parasitic protozoan responsible for Chagas' disease. Two isoenzyme loci have been surveyed: phosphoglucomutase (pgm) and glucose phosphate isomerase (gpJ). At each locus, three alleles (numbered 1-3) are scored. Lack of segregation is observed at each locus. Indeed, at the pgm locus, only two genotypes are observed: the homozygote 1/'1 and the heterozygote 2/3. Missing segregant genotypes include 1/2, 1/3, 2/2and 3/3. A comparable situation is observed at the gpi locus. When the two loci are considered together, it is apparent that there has not been any recombination of these alleles. Indeed, the 1/1 pgm genotype is constantly associated with the 3/3gpigenotype, and the same is true for pgm 2/3 and gpi 1/2. This is an example of linkage disequilibrium; in other words, the non-random association of genotypes occurring at different loci. The genotypes pgm 1 / 2 + gpi 1/2 and pgm 2/2 + gpi 3/3 are never observed, but they should be if the species is sexual. The same isoenzyme patterns have been found in more than 600 strains of T. cruzi (see Ref. 2). analysis used by different authors 2,3. This will help in rigorously evaluating the risk of type II error occurring and will allow reliable comparisons between different categories of microorganisms to be made. It is not easy to compare the work of Grfiser et al. 7, which was performed with DNA markers and relies on linkage disequilibrium analysis of pairs of loci, with the earlier work by Pujol et al. 6, which was based on isoenzyme analysis and overall linkage disequilibrium analysis of all loci. It is even harder to compare Grfiser's results with similar work on bacteria or parasitic protozoa to test, for example, if C. albicans is less clonal than Trypa n o s o m a ¢ruzi 2,3, the agent of Chagas' disease. Second, I have proposed z,3 to broaden the clonality/sexuality debate on microbial population structures. I have suggested that the relevant boundary is, on the one hand, between microbial species whose natural populations consist of discrete and durable genetic lines (either by the existence of hidden biological species or by clonal evolution) and, on the other hand, species that show no such stable subdivisions. The distinction between these two situations is difficult using a mere analysis of departures from panmictic expectations without any point of comparison. Standardizing Response from Vilgalys e t al. s Candida albicans clonal or reThe answer is yes. IAll.combining? species are clonal and recombining to varying degrees. Fungi, in particular, possess mysterious lives that often combine elements of both asexual and sexual reproduction. Researchers have only recently been able to address the life histories of C. albicans and other infectious fungi in more detail with the development of appropriate molecular tools. We agree fully with Michel Tibayrenc that the study of population structure in pathogens is important for understanding microbial epidemiology, and we appreciate the opportunity to respond and amplify his comments on our recent paper. The life history of C. albicans consists solely of a diploid vegetative phase; a sexual stage has never been observed. (However, several related yeast species, including Saccharomyces cerevisiae, have an extremely well-understood sexual cycle.) Without a regular sexual cycle or other means of lateral gene transmission in populations, a clonal population structure is expected for C. albicans. Our recent study scored DNA markers from a local population of C. albicans, and we were Copyright © 1997 Elsevier Science Ltd. All rights reserved. 0966 842)(/97/$17.00 TRENDS 254 voL 5 NO. surprised to find several straightforward and easily detected examples of intergenic (and even intragenic) recombination 1. Tibayrenc has contrasted our results with a previous study by Pujol et al. 2, which concluded that C. albicans is primarily clonal. While Pujol's study used different markers and a different population sample, the results from both studies yield remarkably similar results that suggest a combination of clonality and recombination occurs in the population structure in C. albicans. Which genetic markers should be used for C. albicans? One possible source of controversy in the clonality debate can be traced to the types of genetic markers used by different laboratories to characterize populations of C. albicans. A broad variety of molecular typing methods has been applied to study the genetics of C. albicans 3. Although all of these approaches are useful for genetic analysis, the resuits are not always directly comparable when different methods are used or even among laboratories using similar techniques. For example, two commonly employed DNA typing methods, DNA fingerprinting and AP-PCR (RAPD) [arbitrarily primed-PCR (random amplified polymorphic DNA)], can both PIE S0966-842X(97)01076-7 7 .ltJt.Y 1 9 9 7 C O M M E N T yield complex electrophoretic patterns that are often able to show genetic differences among strains but which provide less information about the exact nature of the genetic differences. As markers detected using DNA fingerprinting and APPCR are dominant, it is difficult, if not impossible, to distinguish between strains that are homozygous or heterozygous. In addition, some complex fingerprint patterns present problems with scoring and reproducibility, especially when results are compared between different laboratories. These features of DNA fingerprinting methods diminish their value for studying population structure. In contrast, both Pujol's study2, which was based on isozyme evidence, and o u r s 1, which used DNA sequence data, analyzed codominant, single-locus markers that permit direct inference of genotypes at each locus. As different DNA typing methods have their own advantages and limitations, it may be desirable to use a combination of different methods for some purposes. For example, single-locus markers are ideal for analyzing population structure because the information they provide can be analyzed using population genetic methods. As many population genetic analyses require information about possible linkage relationships among marker loci, linkage among markers can be established by hybridization against blots of electrophoretically separated chromosomes. For identifying genetically identical strains (clones) in a population, which can be a useful adjunct to standard genetic markers, DNA fingerprinting methods may provide a higher level of discrimination, despite the limitations noted above. Sequence data from individual loci also make it feasible to examine the molecular basis of genetic variation and confirm the identity of alleles. Criteria for clonality On the subject of clonality in microorganisms, Tibayrenc et al. 4 illustrated how the analysis of population genetics can be used to test hypotheses about genetic structure and clonality in natural populations of microorganisms. Three criteria TRENDS for the assessment of clonal population structure in microorganisms, including C. albicans, were outlined" (1) apparent overrepresentation of one or more genotypes within a population (especially where many genotypes are possible), (2) evidence for non-random segregation at individual loci and (3) reduced segregation between loci, which is detected by testing for linkage disequilibrium. We tested all three criteria for our population sample and concluded, as Pujol et al. 2 did before us, that C. albicans does not exhibit a random population structure. C. albicans, therefore, must be clonal to some degree. So how clonal is C. albicans? The primary null hypothesis for the clonal criteria proposed by Tibayrenc et al. 4, namely random mating (panmixia), is the backbone against which all standard measures of conventional population structure are derived (e.g. Wright's F statistics, the principle of Hardy and Weinberg, linkage disequilibrium). Although the data on C. albicans obtained from our studyI and Pujol's 2 clearly vary from random expectations, we believe it is also useful to examine clonality itself as the alternative null hypothesis against which data can be tested. Several lines of evidence from both studies suggest that recombination occurs in populations of C. albicans. One indirect line of evidence supporting the recombination hypothesis is our inability (in many cases) to reject the null hypothesis that loci may in fact be segregating at random. Three out of 12 loci (25%) in our study did not differ significantly from Hardy-Weinberg expectations, and the majority of pairwise tests (73-80%) failed to detect significant linkage disequilibrium between loci 1. Tibayrenc suggests that our inability to detect significance for some tests may be attributed to a lack of statistical power (type II error) associated with small sample sizes. The sample size from our study (52 strains) is comparable with Pujol's previous study (55 strains) 1,2. However, much larger sample sizes would certainly be needed to detect significant values of disequilibrium (D), especially if these values are low for the whole IN MICROBIOLOGY 255 voL 5 NO. population. It is less clear what the expected value of D should be for strictly clonal populations. For some pairs of loci, D would be expected to approach 1.0, as clonal lineages diverge in genotype. (Values of D estimated from our data ranged from 0 to 0.21, with an average of 0.034.) In contrast to the conventional use of disequilibrium testing of random association of alleles among loci, we are unaware of any current test that could be used to predict and test values of D for populations under the hypothesis of strict clonality. From our study, the strongest evidence favoring recombination in C. albicans comes from direct examination of the actual data 1. For several pairs of loci, all possible combinations of genotypes were detected at both loci. The simplest explanation for this observed pattern of recombination is sexual reproduction. If sex does not occur in C. albicans, then other explanations for recombination still need to be found. Testing clonality in C. albicans The absence of a sexual stage in the diploid C. albicans has predictable consequences for partitioning genetic variation in populations. For purely clonal populations, genealogical relationships among strains should also be reflected in the genealogy of their gene sequences. This principle was elegantly developed and tested for Escherichia coli by Dykhuizen and Greens, who compared different gene trees for the same set of strains and found evidence for intergenic recombination. If populations were completely clonal, transmission of genetic information at one or more loci would always be vertical, as depicted in Fig. 1. Conversely, even moderate amounts of recombination are likely to result in reticulate patterns of gene transmission across generations. A major effect of recombination is to erase any clonal pattern of genealogical descent that might exist. If C. albicans was strictly clonal (with no recombination), then different strains would be expected to show an unambiguous hierarchical pattern (phylogenetic structure) reflecting their descent within different lineages via clonal reproduction 7 JuLY 1997 C O M M E N T (a) Ancestor (AA) 112 3 41 gg gg gg AA cc gg gg gg ct gg dd gg gg tt cc gg dd ~ aa gg tt cc aa aa aa gg tt gg aa aa aa Clonal reproduction (b) ag AATaa Am a I/I Aa Aa Aa aa gg aa gg aa m aa gg aa gg aa Aa Aa aa aa Random mating Fig. 1. Clonal vs random-mating models for transmission of genetic markers under different life histories. (a) With clonal reproduction, relatedness among genotypes arises from a vertical pattern of genealogical transmission. (b) By contrast, sexual populations with random mating among individuals are characterized by a reticulate pattern of marker transmission with a less predictable relationship between genotypes. (Fig. 1). Such strictly clonal relationships should also be inferable using appropriate phylogenetic approaches. One such approach was recently developed by Burt et al. 6, who proposed a phylogenetic test for strict clonality that employs multilocus genotypes as 'taxa' and individual loci as characters for parsimony analysis. By using PAUP (phylogenetic analysis using parsimony) 7, we analyzed phylogenetic relationships among the 27 12-locus genotypes from our study 1. In accordance with a hypothesis of strict clonality, each diploid locus was treated as an ordered character (with heterozygotes as intermediates). Parsimony analysis using a simple heuristic search option of PAUP failed to identify a unique phylogenetic history for the 27 genotypes and instead yielded 3569 equally parsimonious trees whose consensus is an unresolved 'rake' (see Fig. 2). Thus, phylogenetic analysis rejects TRENDS cc ga ga aa aa cc tg dd cc gg aa cc ag gg aa cc ag gg dd cc ag tt c c g g d d cc gg aa tt a a cc cc gg dd cc aa gg tt ag tt c c g g d d cc aa gg tt aa cg ag ag cc gg ag aa gg tt c c gg dd cc gg aa tt aa gg dd cc ga ga tt a a tt aa cc gg dd ag gg gg cc ct gg gg gg cc cc gg dd cc ga ga cc g g d d cc gg aa gg tt aa gg tt aa gg aa gg tt cc gg dd cc gg ga ct gg gg aa aa tt cc gg dd cc aa gg tt a a ag gg gg cc ct gg dd cc gg tt cc gg dd cg ga gg gg gg aa tt a a tt a g cg ga ga tt cc gg cc ga ga tt aa aa gg gg cc cc tt d d c c g g g g tt a a gg aa gg tt c c tt d d c c a a g g cc gg gg gg gg tt gg aa gg tt c c g g d d cc gg ga tt a a aa aa gg tt c t g g d d cc gg gg tt aa gg aa gg cc ag gg aa aa gg tt gg tt dd dd cc ga ga (2 strains) tt a a ga cc ct gg dd (5strains) (14strains) tt a a gg cc cc gg Aa Aa cc tt a a ga ga tt aa gg gg aa gg gg AA Aa Aa A A A a cc cc ga ga gg dd cc - - a a l m m- Aa Aa A a a a a a dd gg aa gg tt mmm A A ~ A a 7 [8 9 lo]rJ3 J tt cc gg tt a a cc gg dd cc ga ga t t a a (2 cc g g b b cc aa gg tt aa strains) Fig. 2. Phylogenetic relationships inferred for 27 Candida albicans genotypes using parsimony analysisL Genotypes are shown for 12 diploid loci (1-12) scored for either DNA base polymorphism (aa, ag, etc.) or the presence/absence of length mutations (dd, bb). Boxed loci are genetically linked. Several genotypes were detected more than once (indicated to the right of each genotype). The consensus tree shows unresolved relationships resulting from the presence of recombined genotypes among some loci. the hypothesis of strict clonality. Both the tree length (53 steps) and low consistency index (0.358) from this analysis suggest that the considerable homoplasy in these data could only result from recombination among loci. Cryptic sex in fungi? At least two important questions remain: how much recombination occurs (or has occurred) in C. albicans and how might recombination take place in the absence of a sexual cycle? Clonality, and its absence resuiting from recombination, is a preoccupying theme of many current studies on fungal population biology8-1°. Our demonstration of both recombination and clonality in C. albicans is typical of a growing number of studies that are employing molecular approaches to assess ,N MICROBIOLOGY 256 voi. 5 No. population structure in fungi. Indeed, in a similar report, Burt et al. 6 present evidence for recombination in another pathogenic fungus, Coccidioides immitis, which also lacks a known sexual cycle. Do other species of fungi harbor secret sexual cycles in nature? The possibility of cryptic sex in many species is a real possibility, and additional studies using population genetics and phylogenetic analysis are needed to address this question. Rytas Vilgalys Dept of Botany, Duke University, Durham, NC 27708, USA, Yvonne Gr~iser, Gabriele Sch6nian and Wolfgang Presber 7 .JULY 1997 C O M M E N T Dept of Microbiology and Hygiene (CharitY), Humboldt University, Dorotheenstrasse 96, D-10098 Berlin, Germany, and Thomas G. Mitchell Dept of Microbiology, Duke University Medical Center, Durham, NC 27710, USA References 1 Gr~iser,Y. et al. (1996)Pro& Natl. Acad. Sci. U. S. A. 93, 12473-12477 2 Pujol,C. et al. (1993) Proc. Natl. Acad. Sci. U. S. A. 90, 9456-9459 3 Magee,P.T., Bowdin,L. and Staudinger, J. (1992)J. Clin. Microbiol. 30, 2674-2679 4 Tibayrenc,M. et al. (1991) Proc. Natl. Acad. Sci. U. S. A. 88, 5129-5133 5 Dykhuizen,D.E. and Green, L. (1991) J. Bacteriol. 173, 7257-7268 6 Burt, A. etal. (1996)Proc. Natl. Acad. Sci. U. S. A. 93, 770-773 7 Swofford,D.L. (1991) PAUP: Phylogenetic Analysis Using Parsimony, Illinois Natural History Survey 8 McDonald, B.A. and McDermott,J.M. (1993) Biosci. Biotechnol. Biochem. 43, 311-319 9 Anderson,J.B. and Kohn, L.M. (1995) Annu. Rev. Phytopathol. 33, 369-391 10 Milgroom, M.G. (1996) Annu. Rev. Phytopathol. 34, 457-477 Virus entry: two receptors are better than one Frank Tufaro iruses face a serious problem when their host cell is unable to harbor them any longer; they must seek out and gain entry to a new host. Enveloped viruses, which are surrounded by a lipid coating, achieve this feat by attaching to specific cell-surface components (virus receptors) in order to trigger the events that draw the virus inexorably into the cytoplasm of the cell. Once inside the cell, the viral genome can avail itself of any machinery required to ensure its persistence or propagation. It is not surprising, therefore, that viruses have evolved extraordinary mechanisms to ensure that they can engage the components of the host cell surface reliably to ensure their survival as successful pathogens. Two recent papers, one by Montgomery et al. 1 on herpes simplex virus (HSV) and one by Endres et al. 2 on HIV, provide striking new evidence that the infection strategies used by these pathogens are more elaborate and perhaps more insidious than once envisioned. HSV, an enveloped virus containing at least a dozen proteins in its viral membrane, encodes at least two viral glycoproteins (gB and gC) that are responsible for virus attachment to the cell surface and at least four glycoproteins (gB, gD, gH and gL) that are required for penetration 3. It is known from cell cul- V ture experiments that HSV infects a broad range of cells, and it is reasonable to ask what all of these proteins are doing in the envelope. One possibility is that HSV recognizes multiple receptors via different viral glycoproteins. The identity of one such receptor was revealed a few years ago when cell surface glycosaminoglycans, which are the polyanionic sugar moieties decorating cell surface proteoglycans, were shown to be responsible for the initial attachment of HSV to the cell surface 3. However, the situation was complicated when it was demonstrated that certain cell types, such as Chinese hamster ovary (CHO) cells4 and swine testis cells s, are not susceptible to HSV infection, even though they express glycosaminoglycans on their surface. The converse experiment was also informative; mutant cells completely lacking in glycosaminoglycans could still be infected by HSV, albeit at low efficiency 6. Given that glycosaminoglycans do not appear to be absolutely essential for infection, it is likely that HSV requires a second receptor to mediate infection. F. Tufaro is in the Dept of Microbiology and Immunology, University of British Columbia, #100-2386 East Mall, Vancouver, BC, Canada V6T 1Z3. tel: +I 604 822 0580, fax: +1 604 822 0607, e-mail: [email protected] Copyright © 1997 Elsevier Science Ltd. All rights reserved. 0966 842X/97/$17.00 I RENI)S IN MICR()BIOL()(Iy 257 VO'. 5 NO. The TNF/NGF receptor family An exciting breakthrough in the search for a second HSV receptor was recently reported by Montgomery and colleagues 1, implicating a novel member of the TNF/ NGF (tumor necrosis factor/nerve growth factor) receptor family as a mediator of HSV infection. These workers devised an elegant approach for identifying the molecules that would render C H O cells susceptible to HSV infection. Pools of cDNA clones generated from HeLa cell RNA were transferred into naturally resistant C H O cells, which were then exposed to a recombinant HSV-1 strain that expresses [3-galactosidase following infection. After exposure to the virus, infected cells were identified by incubation with the lactose analogue X-gal to detect [3-galactosidase activity. The amount of virus used for selection was capable of infecting all the control HeLa cells but none of the CHO cells; therefore, the appearance of infected cells in a dish would signal that a receptor was being expressed from a cDNA. Ultimately, this process led to the isolation of a 1698-bp cDNA that encoded a previously unrecognized member of the TNF/NGF receptor family. This protein, designated H V E M (herpes virus entry mediator), was shown to mediate the entry of several wild-type HSV-1 and PII: S0966-842X{97)01057-3 7 JuLY 1997