UBC Faculty Research and Publications

Comparative hybridization reveals extensive genome variation in the AIDS-associated pathogen Cryptococcus… Hu, Guanggan; Liu, Iris; Sham, Anita; Stajich, Jason E; Dietrich, Fred S; Kronstad, James W Feb 22, 2008

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata


52383-13059_2007_Article_1903.pdf [ 4.72MB ]
JSON: 52383-1.0220707.json
JSON-LD: 52383-1.0220707-ld.json
RDF/XML (Pretty): 52383-1.0220707-rdf.xml
RDF/JSON: 52383-1.0220707-rdf.json
Turtle: 52383-1.0220707-turtle.txt
N-Triples: 52383-1.0220707-rdf-ntriples.txt
Original Record: 52383-1.0220707-source.json
Full Text

Full Text

Open Access2008Huet al.Volume 9, Issue 2, Article R41ResearchComparative hybridization reveals extensive genome variation in the AIDS-associated pathogen Cryptococcus neoformansGuanggan Hu¤*, Iris Liu¤*, Anita Sham*, Jason E Stajich†‡, Fred S Dietrich† and James W Kronstad*Addresses: *Michael Smith Laboratories, The University of British Columbia, 2185 East Mall, Vancouver, British Columbia V6T 1Z4, Canada . †Department of Molecular Genetics and Microbiology, Institute for Genome Sciences and Policy, 287 CARL Building, Duke University Medical Center, Durham, NC 27710, USA. ‡Department of Plant and Microbial Biology, 121 Koshland Hall, University of California, Berkeley, CA 94720-3102, USA. ¤ These authors contributed equally to this work.Correspondence: James W Kronstad. Email: kronstad@interchange.ubc.ca© 2008 Hu et al.; licensee BioMed Central Ltd. This is an open access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.Genome variation in C. neoformans<p>Ext nsive ge ome variation i  the AIDS-associated pathogen <it>Cryptococcus neoformans</it> is revealed through comparative geno e hybridizat on between strains of different mating type, molecular subtype and pl idy.</p>AbstractBackground: Genome variability can have a profound influence on the virulence of pathogenicmicrobes. The availability of genome sequences for two strains of the AIDS-associated fungalpathogen Cryptococcus neoformans presented an opportunity to use comparative genomehybridization (CGH) to examine genome variability between strains of different mating type,molecular subtype, and ploidy.Results: Initially, CGH was used to compare the approximately 100 kilobase MATa and MATαmating-type regions in serotype A and D strains to establish the relationship between the Log2ratios of hybridization signals and sequence identity. Subsequently, we compared the genomes ofthe environmental isolate NIH433 (MATa) and the clinical isolate NIH12 (MATα) with a tiling arrayof the genome of the laboratory strain JEC21 derived from these strains. In this case, CGHidentified putative recombination sites and the origins of specific segments of the JEC21 genome.Similarly, CGH analysis revealed marked variability in the genomes of strains representing the VNI,VNII, and VNB molecular subtypes of the A serotype, including disomy for chromosome 13 in twostrains. Additionally, CGH identified differences in chromosome content between three strainswith the hybrid AD serotype and revealed that chromosome 1 from the serotype A genome ispreferentially retained in all three strains.Conclusion: The genomes of serotypes A, D, and AD strains exhibit extensive variation that spansthe range from small differences (such as regions of divergence, deletion, or amplification) to theunexpected disomy for chromosome 13 in haploid strains and preferential retention of specificchromosomes in naturally occurring diploids.Published: 22 February 2008Genome Biology 2008, 9:R41 (doi:10.1186/gb-2008-9-2-r41)Received: 9 November 2007Revised: 23 December 2007Accepted: 22 February 2008The electronic version of this article is the complete one and can be found online at http://genomebiology.com/2008/9/2/R41Genome Biology 2008, 9:R4141.2http://genomebiology.com/2008/9/2/R41 Genome Biology 2008,     Volume 9, Issue 2, Article R41       Hu et al. RBackgroundThe encapsulated, basidiomycetous fungi Cryptococcus neo-formans and C. gattii cause life-threatening meningoen-cephalitis and pose a significant threat to AIDS patients [1-3].Four serotypes (A to D) of these fungi are recognized, basedon antigenic differences in the capsule polysaccharide, whichis one of the major virulence factors. C. neoformans isolateshave A or D capsular serotypes and mainly infect immuno-compromised individuals [4]. Isolates of serotypes B and Cwere recently re-classified as the separate species C. gattii,which infects both immunocompromised and immunocom-petent patients [3,5-7]. Strains with hybrid serotypes (AD andBD) have also been identified from both clinical and environ-mental sources [8-11]. Serotype A strains are the most preva-lent clinical isolates and account for the majority of infectionsin AIDS patients. Serotype D isolates account for fewer casesof cryptococcosis, and some of these infections may actuallyinvolve AD hybrid strains [12]. Serotype D strains are globalin distribution but are more frequently isolated in Europe [2].The sexual cycle of C. neoformans is well defined and involvesa bipolar mating system with two mating types: a and α [13].The mating-type locus (MAT) is more than 100 kilobases (kb)in length and contains more than 20 genes [9,14,15]. Recom-bination is suppressed within MAT but is elevated in areasadjacent to the region, and extensive sequence divergenceand rearrangements between the MATa and MATα alleleshave been described [14-16]. The majority of strains isolatedfrom clinical and environmental sources are of the α matingtype [13], and a serotype D strain of the α mating type is morevirulent in mice than a congenic a strain [17]. In contrast, thecongenic serotype A strains KN99a and KN99α exhibited nodifference in murine virulence, although the latter strainmore efficiently colonizes the central nervous system [12,18].Further studies revealed that genomic regions outside themating-type locus contribute to differences in virulencebetween a and α cells [19]. Large-scale genomic comparisonsthat may reveal sequences contributing to virulence differ-ences between a and α strains have thus far been limited tothe MAT locus [14,15,20].Cells of C. neoformans are generally haploid, but it is possibleto obtain relatively stable diploid strains through laboratorycrosses and to identify naturally occurring AD hybrids thatappear to result from the fusion of serotype A and D strains[9,21-24]. Most of the environmental and clinical AD isolatesare diploid or aneuploid and contain alleles from both theserotype A and D genotypes [8]. However, recent studies indi-cate that some AD strains contain only a single mating-typelocus (MATa or MATα), either because of deletion of oneallele or the complete absence of one of the MAT chromo-somes [9,25,26]. These observations and the documentedaneuploidy indicate that the genomes of the hybrid strains areone instance, the viable progeny from a self-fertile AD hybridwere found to be either diploid or aneuploid [9].A number of molecular approaches have been used to investi-gate the genetic structure and epidemiologic relationships ofCryptococcus strains [4,8,27-34]. In particular, PCR finger-printing and amplified fragment length polymorphism(AFLP) analysis revealed four major molecular types of C.neoformans [25,29,32,35]: VNI (AFLP1; serotype A), VNII(AFLP1A; serotype A), VNIII (AFLP3; AD hybrid), and VNIV(AFLP2; serotype D). Similarly, four molecular subtypes arefound for C. gattii: VGI to VGIV [35], or AFLP groups 4 to 7[8,32]. Multilocus sequence typing (MLST) has also beenused in the phylogenetic analysis of a large number of isolatesof C. neoformans and C. gattii [36-40]. This work providesinsights into the population structure, geographic distribu-tion, and evolutionary history of the species. For example,Litvintseva and coworkers [38,39] identified a unique groupof serotype A isolates from Botswana (molecular subtypeVNB) that included a significant proportion of fertile strainswith the rare MATa mating type. Interestingly, those investi-gators went on to show that AD hybrids possessing the rareMATa allele clustered phylogenetically with serotype A iso-lates of the VNB subtype from Botswana. In contrast, ADhybrids with the more common MATα allele clustered withserotype A isolates of the VNI molecular subtype that is foundglobally [40].The genomes of C. neoformans and C. gattii have been char-acterized in terms of chromosome content and sequence. Bykaryotype analysis, the genome size of different species andvarieties of Cryptococcus is estimated at 15 to 27 megabases(Mb) and chromosome number varies between 12 and 14[31,32,41-45]. Loftus and coworkers [46] described thegenome sequences for two serotype D strains, B3501A andJEC21, and these turned out to be about 19 Mb in size. Thegenome sequences of one serotype A strain (C. neoformansstrain H99) and two serotype B strains (C. gattii strainsWM276 and R265) have also been completed.Comparative genome hybridization (CGH) is a rapid andcost-effective method to assess the presence, absence, ordivergence of sequences in uncharacterized genomes by com-parison with a reference genome. CGH has been applied tocancer cells, pathogenic and nonpathogenic bacteria andfungi, and other organisms [47-52]. In general, CGH circum-vents the need to sequence multiple closely related genomes.For example, comparison of the genomes of two species in thefungal genus Candida (namely C. albicans and C. dublinien-sis) confirmed the relatedness of the two species and identi-fied a group of unique C. albicans genes that may contributeto virulence [53]. More recent CGH studies in C. albicansidentified genome instability and revealed associationsGenome Biology 2008, 9:R41unstable [9]. AD strains can be sterile or self-fertile, and in thelatter case can produce filaments, basidia, and basidiosporesto yield progeny that generally exhibit poor viability [9,25]. Inbetween aneuploidy, isochromosome formation, and azoleresistance [50,51]. In this study, we used CGH to examine thegenomes of selected strains of the A, D, and AD serotypes41.3http://genomebiology.com/2008/9/2/R41 Genome Biology 2008,     Volume 9, Issue 2, Article R41       Hu et al. Rrepresenting all of the molecular subtypes (VNI to VNIV) of C.neoformans. Specifically, we employed high-density tilingarrays to identify the global genome differences within andoutside of the mating locus between MATa and MATαstrains, to map putative recombination sites in a genomeresulting from a well characterized genetic cross, to distin-guish the molecular subtypes of serotype A strains, and toidentify the origins of chromosomes in selected AD hybridstrains.Results and discussionCGH detection of sequence divergence at the MAT locusCGH signal ratios reflect the similarities or differencesbetween reference and test genomes, and can detect theamplification, absence, or divergence of sequences. The CGHapproach was applied to C. neoformans by first designinghigh-density tiling arrays of oligonucleotide probes with anaverage length and spacing of 50 base pairs (bp) and 44 bp,respectively, based on the genomes of strains JEC21 (serotypeD) and H99 (serotype A; see Materials and methods, below).To establish a framework for identifying regions of differencein Cryptococcus genomes, CGH data collected with the tilingarrays was initially calibrated by comparing the Log2 ratios ofthe fluorescence intensity with the corresponding sequenceidentity for previously sequenced mating-type (MAT) regionsof the test and reference genomes (Figure 1). The sequences ofthe approximately 100 kb MAT loci of representative serotypeA (H99 and 125.91) and D (JEC21 and JEC20) strains wereavailable for this analysis [15,46]. The sequences of the MATaand MATα alleles were obtained from GenBank and thesequence identities for the coding regions of 20 genes in eachof these loci were plotted against the corresponding Log2ratios for the hybridization signals of the probes in the genes.That is, the average of the normalized Log2 ratios from everyeight probes (spanning about 400 bp) covering each of the 20MAT genes were compared with the corresponding genomicsequences for the MATa versus MATα alleles. A correlationwas found between sequence identity in the MAT regions andthe Log2 ratios from hybridization signals for each of thecomparisons (r2 = 0.78 for the serotype A strains H99 and125.91, and r2 = 0.81 for the serotype D strains JEC21 andJEC20; Figure 1). The Log2 ratios for the hybridization sig-nals ranged from 0.585 to -4.374, and from 0.971 to -4.382 inMAT regions of the serotype A and D strains, respectively.Based on these comparisons, Log2 ratios between -3.77 and0.49 corresponded to sequence identities in the range ofabout 75% to 100% (Figure 1). We confirmed that these cali-bration values held true for randomly selected regions outsidethe MAT locus by comparing the identity between sevensequenced regions and hybridization probes (198 probes)against the Log2 ratio for those individual probes with theIn general, our observed correlations between the Log2 ratioand sequence identity are similar to CGH results for otherorganisms [53-56]. For example, Log2 ratios between -4.0and 0.5 corresponded to sequence identities (probe/targetidentity) between 81% and 100% in Campylobacter jejuni[56]. Furthermore, CGH data for Chlamydia trachomatisshowed a linear relationship between Log2 ratio andsequence identity between 75% and 99% [57]. In general, aLog2 ratio of ±1.0 is used in many CGH experiments to iden-tify divergent (or deleted) and duplicated genes, representinga conservative threshold for divergent gene detection[55,56,58]. For our analysis, the observed correlationbetween sequence identity and Log2 ratio allowed us to pre-dict whether specific regions in the genomes were divergent,Comparisons of Log2 ratios and sequence identity for the MATa and MATα loci of serotype A and D strainsFigure 1Comparisons of Log2 ratios and sequence identity for the MATa and MATα loci of serotype A and D strains. Each data point represents the average Log2 ratio and sequence similarity of eight probes from a set of 400 base pair windows within 20 genes at the MAT locus. (a) Comparison of the MAT loci of the serotype D strains JEC20 (MATa) and JEC21 (MATα). (b) Comparison of the MAT loci of the serotype A strains 125.91 (MATa) and H99 (MATα). Note that the MATa region of JEC20 originated in strain NIH433 and genomic DNA of NIH433 was used for the hybridization experiment.(a)(b)Log2 ratioLog2 ratioSequence identity (%)Sequence identity (%)JEC20 (MATa) and JEC21 (MATα)125.91 (MATa) and H99 (MATα)Genome Biology 2008, 9:R41related strains NIH433 and NIH12 (described below, anddata not shown).deleted, and/or amplified. The use of these data for wholegenome comparisons for serotype A, D, and AD strains aredescribed below.41.4http://genomebiology.com/2008/9/2/R41 Genome Biology 2008,     Volume 9, Issue 2, Article R41       Hu et al. RComparison of the MATa and MATα genomes of serotype D progenitor strainsInitially, we used CGH to compare the genomes of two sero-type D progenitor strains, NIH433 (MATa) and NIH12(MATα), with the genome of the derived strain JEC21(MATα). JEC21 is commonly used in laboratory experimentsand the genome of this strain has the best annotation of thesequenced cryptococcal genomes [46]. The strain wasobtained from a series of back crosses (starting with NIH433and NIH12) and was the MATα representative of a congenicstrain pair (with JEC20) used to examine the role of matingtype in virulence [17,59]. NIH433 is an environmental isolatefrom pigeon droppings in Denmark, and NIH12 is a clinicalisolate from a patient with osteomyelitis. The cross of NIH12and NIH433 yielded two F1 strains B3501 (MATα) and B3502(MATa), and a cross of these strains yielded JEC20 (B-4476)[17,59]. JEC20 was subsequently used as the parent in aseries of backcrosses to generate JEC21. Thus, approximately50% of the overall genetic background of JEC21 should bederived from each of the NIH12 and NIH433 genomes.Initially, we hybridized the JEC21 array with DNA from theMATa strain NIH433 to compare the genomes of strains ofopposite mating type. As expected, CGH showed that theMATa locus was divergent from the MATα locus of JEC21with Log2 ratios ranging from -5.27 to 1.24 for probes in theregion. However, the analysis revealed an unexpected patternof regions with either similar or divergent hybridization sig-nals along 10 of the 14 chromosomes (Figure 2), in addition tothe divergence observed at the MAT locus (Figures 1 and 3).Interestingly, each of these regions accounted for approxi-mately 50% of the JEC21 genome, with regions of similarityaccounting for about 8.73 Mb (50.6%) and divergent regionsrepresenting about 8.53 Mb (49.4%). Note that the JEC21array did not contain the centromere sequences (or the rDNAcluster) and therefore covered about 17.3 Mb of the approxi-mately 19 Mb genome [46] (Additional data file 1). Theseregions may represent the segments of the JEC21 genomeoriginating from either NIH433 or NIH12, and the borders ofthe regions are likely to be sites of recombination events thatoccurred in the original cross. This idea was tested by hybrid-izing DNA from the MATα parent NIH12 to the JEC21 array,and this analysis revealed a reciprocal pattern of similar anddivergent regions as compared with the results with NIH433(Figure 2). The averages and standard deviations (SDs) ofLog2 ratios of the hybridization intensity of these regionsalong all chromosomes were consistent with the visual obser-vations, in which regions of divergence generally have ahigher SD than regions of similarity (Additional data file 1).This measure of divergence is illustrated quite clearly by sig-nals from the MAT locus. That is, the SD of the MATa locus ofNIH433 is 1.778 (average Log2 ratio of -1.632) upon compar-ison with the MATα locus of JEC21. In contrast, the SD for theThe putative recombination sites were observed in 10 of the14 chromosomes and the number of these sites ranged fromone on chromosome 7 to seven on chromosome 2 (Figure 2).Based on the current annotation, these sites occur in bothintragenic and intergenic regions (Additional data file 2).Chromosomes 10, 11, 13, and 14 did not exhibit variability inLog2 ratios and different segments could not be distin-guished. The SD of the hybridization ratios indicated that allof chromosome 10 may have originated from strain NIH12,with little or no contribution from NIH433. On the otherhand, chromosomes 11, 13, and 14 are remarkably similar toNIH433 but divergent from NIH12, although one of the tel-omeres on chromosome 14 exhibited variability (Figure 2).Overall, these findings indicated that CGH with the JEC21 til-ing array could detect sequence polymorphisms betweenstrains of the same serotype and that these could be used totrack recombination history. For C. neoformans and otherfungal pathogens, the detection of recombination sites mayhave utility for mapping quantitative traits that contribute tovirulence. Similar approaches have been used to examinebreakpoints in the chromosomes of tumors and to character-ize genetic diversity in Saccharomyces cerevisiae [47,60].The CGH analysis also allowed an examination of the MATregions of the serotype D strains NIH12 and NIH433 in com-parison with that of JEC21. Specifically, we observed that theflanking regions of the MAT locus on chromosome 4 con-tained putative sites of recombination presumably from thecross of NIH12 and NIH433 (Figure 3) [59]. The observedsites are consistent with the previous identification of regionsflanking the mating locus that are apparent hotspots ofrecombination [16]. In agreement, CGH revealed two poten-tial recombination sites in a region of about 6 kb (chromo-some 4: 1,525,471 to 1,531,510 and 1,514,930 to 1,520,952), aswell as a third more distant site on the left side, and one sitein a region approximately 27 kb to the right of the MAT locus(chromosome 4: 1,639,910-1,667,015; Figure 3). Within themating-type locus, two divergently transcribed genes, RPO41and BSP2 (chromosome 4: 1,585,174 to 1,591,759), were pre-sumed to be involved in gene conversion because they are99% identical between the two mating-type alleles in phyloge-netic analyses [15,16]. Consistent with this finding, CGHyielded a Log2 ratio near zero, indicating high sequence sim-ilarity (-0.089 and 0.069 for NIH433 and NIH12, respec-tively; Figure 3). Three neighboring genes, LPD1, CID1, andGEF1 (chromosome 4: 1,592,529 to 1,602,117), share similarproperties with RPO41 and BSP2, and Fraser and coworkers[15] described this area as a species-specific, syntenic regionin the MAT locus.Examination of the CGH data outside the MAT locus identi-fied several candidate regions of difference between JEC21and the progenitor strains NIH12 and NIH433. For example,Genome Biology 2008, 9:R41homologous MATα locus of NIH12 is 0.229 (average Log2ratio 0.090).two regions appeared to be present in single copy in NIH433but duplicated in NIH12. One of these regions of approxi-mately 20 kb (chromosome coordinates 1,218,400 to41.5http://genomebiology.com/2008/9/2/R41 Genome Biology 2008,     Volume 9, Issue 2, Article R41       Hu et al. R1,239,200; average Log2 ratio of -0.062 in NIH433 and 0.997in NIH12) is present on chromosome 3 and contains genesencoding several putative functions (PM-scl autoantigen,nicotinate-nucleotide diphosphorylase [carboxylating], phos-phate transporter, and kynureninase). The other region ofapproximately 13 kb on chromosome 13 (chromosome coor-dinates 508,400-521,200) encodes putative functions includ-ing a metal transporter, a carboxylesterase, a (R, R)-butanediol dehydrogenase, and a formaldehyde dehydroge-nase (glutathione; average Log2 ratio of 0.016 in NIH433 and01.284 in NIH12). Additionally, an approximately 29 kb seg-ment near the right telomere of chromosome 5 (encodingmostly hypothetical proteins) was highly divergent inNIH433 (average Log2 ratio of -3.232, SD of 1.496) and hadstrains H99 and JEC21 [61]. The sequences of the regionshare 98.5% identity between the two genomes - a level thatis approximately 10% higher than the average identity acrossthe entire genomes. Kavanaugh and coworkers [61] foundthat the approximately 40 kb identity island was the apparentresult of a nonreciprocal transfer event from a serotype Agenome to a serotype D genome about 2 million years ago.They also surveyed 12 serotype D strains and found that ten(including NIH12) had the identity island that originatedfrom the serotype A sequence and that two strains, NIH430and NIH433, retained the original serotype D version of thesequence. Therefore, the analysis reported here for NIH12and NIH433 matches the findings of Kavanaugh and cowork-ers [61], and thus demonstrates the utility of the CGHGenome hybridization to compare the progenitor strains NIH12 and NIH433 to the reference strain JEC21Figure 2Genome hybridization to compare the progenitor strains NIH12 and NIH433 to the reference strain JEC21. Regions with higher variability in Log2 ratios in the test genomes are more divergent from the JEC21 sequence; regions with Log2 ratios close to zero have greater similarity (Additional data file 1). A reciprocal pattern of similar and divergent segments is found upon hybridization of genomes of NIH12 and NIH433 to the JEC21 array. The scale of chromosome coordinates for the JEC21 genome is indicated at the top of the figure, and gaps in the chromosomes represent putative centromeric regions [46]. The borders of segments are probable sites of recombination events that occurred during the mating of NIH12 and NIH433, and the subsequent backcrossing to obtain JEC21 (Additional data file 2) [59].Genome Biology 2008, 9:R41an average Log2 ratio of 0.829 (SD of 0.382) in NIH12 (Fig-ure 2). This region is part of an approximately 40 kb so-called'identity island' in the genomes of the serotype A and Dapproach for detecting these types of sequence anomalies inC. neoformans genomes.41.6http://genomebiology.com/2008/9/2/R41 Genome Biology 2008,     Volume 9, Issue 2, Article R41       Hu et al. RNIH433 and NIH12 are both virulent, but NIH433 requiresmore time than NIH12 to cause equal mortality when injectedintravenously into mice [59]. In addition to the MAT locus,differences in genetic background were proposed to beimportant contributors to the virulence of these strains [59].In this context, the differences observed for chromosomes 3,5, and 13 or undetected polymorphisms (single nucleotidechanges) could potentially influence virulence. For example,one region on chromosome 3 (2,022,000 to 2,024,000; Log2ratio -3.161) contains a gene for a predicted O-acetyltrans-ferase (CNC06920) that is deleted in NIH433, as confirmedby PCR (data not shown). O-acetyl substituents are found onthe polysaccharide capsule that is the major virulence deter-minant of C. neoformans, and a mutant defective in O-acetylation was found to be hypervirulent [62,63].Genomic differences between serotype A strains representing three molecular subtypesSerotype A strains of C. neoformans are responsible for themajority of clinical cases of cryptococcosis, particularly inAIDS patients, and three molecular subtypes (VNI, VNII, andVNB) have been identified by MLST and AFLP analyses[2,29,38,39]. Given that CGH detected variation within theserotype D strains, we next considered whether the sameapproach would detect genomic differences in serotype Astrains of opposite mating type and different molecular sub-types. The majority of serotype A isolates have the VNI molec-ular subtype and the MATα mating type. MATa strains arerare among clinical and environmental isolates comparedidentified: 125.91 from Tanzania [64] and IUM 96-2828 fromItaly [65]. Strain 125.91 (VNI) was found to mate with a sub-set of MATα serotype A strains, but was unable to mate withthe reference strain H99 [12,64]. Recently, detailed MLSTand AFLP analyses of a large population of serotype A strainsfrom a global collection identified MATa isolates among theVNB molecular subtype that is uniquely present in Botswana;these isolates included strain Bt63, which can mate with H99[38,39]. With the emerging view of the serotype A populationin mind, we examined the genomes of two MATa strains(125.91 and Bt63), a VNII strain (WM626 [MATα]), and aVNI strain (CNB7779 [MATα]) reported to have a smallgenome [32]. For this analysis, we employed the tiling arraybased on the sequence of the VNI strain H99 and the corre-sponding chromosome numbers from the genome sequenceassembly [61].Our analysis of strains 125.91 and Bt63 revealed that, asexpected, the Log2 ratios of hybridization signals in the MATlocus region were highly variable and ranged from -4.345 to0.445, and from -4.24 to 0.488, respectively. This indicatesextensive sequence divergence between these MATa regionsand the MATα locus of H99 (Figures 1 and 4, and Additionaldata file 3). Note that the MAT locus is on chromosome 5 inthe serotype A strain H99 used for the analysis of the MATregions of the serotype A strains 125.91 and Bt63; MAT is onchromosome 4 in strain JEC21 (serotype D). The averageLog2 ratios were -2.200 (SD 1.319) and -2.204 (SD 1.401) for125.91 and Bt63, respectively. These results mirror theSequence divergence and putative recombination sites at the MAT locus in serotype D strainsFig re 3Sequence divergence and putative recombination sites at the MAT locus in serotype D strains. The chromosomal coordinates are shown at the top and the annotated genes in the region are indicated as boxes. Putative recombination sites are marked with arrows. Specific genes that are known to have high sequence similarity between the MATa and MATα alleles [15], and that exhibit a corresponding Log2 ratio close to zero are labelled with letters: a, RPO41 (XM_570485.1; The Institute for Genomic Research [TIGR] locus tag CND05820); b, BSP2 (XM_570482.1; TIGR locus tag CND05830); c, LPD1 (XM_570114.1; TIGR locus tag CND05840); d, CID1 (XM_570548.1; TIGR locus tag CND05850); and e, GEF1 (XM_570546.1; TIGR locus tag CND05860). The gap in the hybridization signal for NIH12 and NIH433 centered on position 1,610,000 resulted from the presence of repeated sequences in this region that reduced probe density [15].Chromosome 4Genome Biology 2008, 9:R41with the prevalence of the MATα mating type. In fact, MATastrains of serotype A were thought to be extinct or to exist asa vestigial, nonfunctional form until two clinical strains werecomparisons of the MATa and MATα alleles for the serotypeD strains presented above (Figures 1 to 3). Outside the MATlocus, the relative extent of divergence in hybridization sig-41.7http://genomebiology.com/2008/9/2/R41 Genome Biology 2008,     Volume 9, Issue 2, Article R41       Hu et al. Rnals between Bt63 and H99 (molecular subtypes VNB andVNI, respectively) versus 125.91 and H99 (both VNI) sug-gested that a higher level of genome variability existsbetween, versus within, molecular subtypes (Table 1). A largenumber of regions of difference were detected for both 125.91and Bt63 relative to H99 (Additional data files 3 and 4). Sev-eral regions were also different between the two MATastrains. These include the region from 339,600 to 350,000 onchromosome 1 that appears to be absent from or highly diver-gent in 125.91 but potentially amplified in Bt63 (Additionaldata file 4) and an approximately 1.2 kb deletion (1,443,600to 1,445,200) on chromosome 3 in Bt63 that was confirmedby PCR (Log2 ratio of -2.868 in the deleted region; data notshown). In addition, Bt63 appears to have a duplication of aregion (1,776,000 to 1,786,400) on chromosome 5 that con-tains genes encoding putative myoinositol transporters. Thisobservation is interesting because Xue and coworkers [66]recently showed that inositol stimulates mating in C. neofor-mans, and it is known that Bt63 mates more robustly withH99 than does 125.91 [64].The genome variability between VNI and VNB strainsrevealed by CGH prompted us to compare the genome of theVNII strain WM626 with the H99 genome. VNII strains mayrepresent up to 20% of the serotype A population worldwide,and WM626 is a clinical isolate from Sydney, Australia [29].The results indicated that the WM626 genome is quite diver-gent from the H99 genome (Figure 4, Table 1, and Additionaldata file 3), suggesting considerable variation between thedeleted regions on 12 of the 14 chromosomes relative to thecorresponding locations in Bt63 (Additional data file 4).We also used CGH to examine the genome content of the clin-ical strain CBS7779 (VNI) from Argentina that was reportedto have a small genome size as estimated by electrophoretickaryotype analysis [32]. Specifically, the genome size forCBS7779 was estimated to be 15 Mb, which is considerablysmaller than the approximately 19 Mb genomes of thesequenced strains. We initially confirmed the published kary-otype of CBS7779 (data not shown), and subsequent CGHanalysis of the genome of this strain with H99 revealed Log2ratios near zero for all of the chromosomes, suggesting a closerelationship between the two strains (Figure 4 and Table 1).The similarities in the genomes were particularly evident incontrast to the results for the hybridizations with Bt63 andWM626 (Figure 4). These results suggested that CBS7779shared most, if not all, of its genome with H99, and missingsequences that would account for the smaller estimatedgenome size were not found. This outcome may reflect thechallenge in using electrophoretic karyotyping to measuregenome size, particularly for strains that exhibit chromosomelength polymorphisms that may result in co-migrating chro-mosomes. It is also possible that the genomes of these strainsmay contain different amounts of repetitive DNA. The com-parison of the CBS7779 and H99 genomes did reveal a rela-tively small number of short regions of divergence. Two largerdifferences for CBS7779 relative to H99 included an apparentdeleted or divergent region of 17 kb on chromosome 10 and anTable 1Comparison of Log2 ratios and standard deviations for all 14 chromosomes of four serotype A strainsChr BT63 (VNB) 125.91 (VNI) CBS7779 (VNI)a WM626 (VNII)aAverage Log2 ratioSD Average Log2 ratioSD Average Log2 ratioSD Average Log2 ratioSD1 0.007 0.674 -0.019 0.438 -0.011 0.212 -0.014 0.6342 0.013 0.634 0.007 0.312 -0.013 0.159 -0.019 0.6373 0.034 0.600 0.020 0.272 -0.018 0.190 -0.011 0.6604 -0.051 0.733 -0.002 0.329 -0.018 0.170 -0.105 0.7785 -0.093 0.900 -0.097 0.623 -0.005 0.212 0.006 0.6226 0.019 0.632 0.023 0.290 -0.035 0.263 -0.042 0.6877 0.013 0.686 0.016 0.338 -0.029 0.258 -0.022 0.7158 -0.048 0.773 0.020 0.323 -0.014 0.183 -0.032 0.7249 0.031 0.625 0.018 0.292 -0.036 0.246 -0.026 0.64310 0.050 0.612 0.033 0.310 -0.081 0.522 0.034 0.61111 0.013 0.653 0.012 0.289 -0.019 0.212 -0.047 0.72612 0.019 0.668 0.075 0.441 0.005 0.232 -0.085 0.82713 0.025 0.627 -0.026 0.345 0.562 0.214 0.721 0.88414 0.026 0.622 -0.006 0.328 -0.027 0.232 -0.060 0.685aThe log2 ratios for chromosome (Chr) 13 in these strains indicates a copy number greater than one. SD, standard deviation.Genome Biology 2008, 9:R41VNI and VNII subtypes (although a larger survey is needed).Compared with the results for 125.91 and Bt63, the CGH datafor WM626 revealed a large number of highly divergent oramplified region of 20 kb on chromosome 5 (Additional datafile 4).41.8http://genomebiology.com/2008/9/2/R41 Genome Biology 2008,     Volume 9, Issue 2, Article R41       Hu et al. RChromosome 1Chromosome 5Chromosome 13Genome Biology 2008, 9:R41Figure 4 (see legend on next page)41.9http://genomebiology.com/2008/9/2/R41 Genome Biology 2008,     Volume 9, Issue 2, Article R41       Hu et al. RThe evaluation of the serotype A strains also allowed a moredetailed look at the 'identity island' discussed above. Thisregion is on the left end of chromosome 5, based the assem-bled H99 genome sequence, and Kavanaugh and coworkers[61] found that several genes (The Institute for GenomicResearch [TIGR] IDs: CNE05310, CNE05340, andCNE05350) in this region were present in H99 and 125.91,but not Bt63. CGH confirmed these findings and furtherrevealed that other genes within the region (CNE05250,CNE05290, CNE05320, CNE05330, CNE05300, andCNE00020) were present in 125.91 (VNI) but absent in Bt63(VNB; Additional data file 4). Interestingly, several geneswithin the region (CNE05310, CNE05340, CNE05350,CNE05320, and CNE05330) appeared to be duplicated inCBS7779 (VNI; Additional data file 4). These genes were alsopresent in the VNII strain WM626, although a region of about6 kb on the telomere proximal side appeared to be lost. Foranother region of high sequence identity between H99 (chro-mosome 10) and JEC21 (chromosome 13) [61], CGH revealedthat three consecutive genes, namely CNM02580,CNM02590 and CNM02600, were conserved in all of thestrains (H99, Bt63, 125.91, and CBS7779). However,CNM02570, which encodes a putative receptor protein nearto the telomere, was present in H99, Bt63, CBS7779, andWM626, but not in 125.91. Kavanaugh and coworkers [61]found that repetitive elements were associated with the iden-tity island on chromosome 5 and suggested that one of these(Cnl1) may have participated in translocation of the island inthe serotype D strain. It is possible that these elements alsocontribute to the variability in these and other regionsobserved by CGH.The CGH analysis also revealed an anomaly in the hybridiza-tion signals for chromosome 13 in strains CBS7779 andWM626, such that the Log2 ratio was above zero (0.562 and0.721, respectively) across the entire chromosome (Table 1,Figure 4, and Additional data file 3). The simplest explanationfor this is that chromosome 13 in these strains is present inmore than one copy in some or all of the cells. Alternatively,duplicated chromosome 13 segments could reside elsewherein the genome, but this is less likely, given that the entirechromosome exhibited an elevated Log2 ratio. Quantitativereal-time PCR with three loci on chromosome 13 in strainsH99, WM626, and CBS7779 supported the conclusion of anincreased copy number in the latter two strains (Additionalwith the single copy found upon transformation of strain H99(Additional data file 7). However, integration at the APT1gene in strain CBS7779 yielded transformants with only theneomycin marker replacement of the gene. We hypothesizethat instability at chromosome 13 in this strain may haveresulted in the loss of one copy during the transformationprocess (see Additional data file 8). This idea is supported byquantitative real-time PCR data that confirmed a differencein copy number for chromosome 13 before and after transfor-mation of the strain (Additional data files 5 and 6). StrainsH99, WM626, and CBS7779 also exhibit phenotypic differ-ences (for instance, in melanin formation) that could poten-tially result from the difference in chromosome content or theregions of divergence detected by CGH (Additional data file9).Other studies have documented genome anomalies for C.neoformans, including segmental duplications for serotype Dstrains [67] and chromosome length polymorphisms [32,68].The possibility of elevated copy number for specific chromo-somes in C. neoformans suggests that there is potentiallygreater genome variability among isolates than was previ-ously appreciated. Chromosome copy number variation mayhave implications for differences in phenotypic propertiesbetween isolates. For example, variability in virulence hasbeen observed for clinical isolates of serotype A and among A,D, and AD strains [69,70]. Additionally, C. neoformansexhibits phenotypic switching that influences the expressionof virulence traits, interactions with the host immune system,the outcome of chronic infection and symptom development(specifically, intracranial pressure) [71]. It is possible thatswitching could result from changes in chromosome copynumber, perhaps through a positive or negative influence ongene expression. In this regard, Torres and coworkers [72]recently showed that the gain of extra chromosomes in S. cer-evisiae had a major influence on cellular processes, includinggene expression, proliferation, metabolism, and proteinturnover.Taken together, the CGH data with selected serotype A strainsrevealed extensive variability, in agreement with the classifi-cation of the strains into different molecular subtypes. Specif-ically, variation in the Log2 ratios and SDs for eachchromosome supports the grouping of H99 with the otherVNI strains 125.91 and CBS7779 (lower SDs) and indicatesVariation in chromosomes 1, 5, and 13 for four serotype A strainsFigure 4 (see previ us page)Variation in chromosomes 1, 5, and 13 for four serotype A strains. The DNA from the strains was hybridized to the array from the genome of strain H99, and gaps in the chromosomes represent the positions of repetitive sequences that represent putative centromeres or repeated elements in the MAT locus. The spikes in the Log2 ratios (for instance, for chromosome 1 of CBS7779) represent individual probes with high Log2 ratios (3 to 4); the sequences of these probes are present in single copy in the H99 genome but may be part of repetitive sequences in the other strains. For all of the chromosomes in all four strains, the hybridization data are shown in Additional data file 10, the Log2 ratios are listed in Table 1 and the regions of difference are listed in Additional data file 3.Genome Biology 2008, 9:R41data files 5 and 6). Furthermore, replacement of the APT1gene on chromosome 13 with a neomycin marker confirmedthe presence of two copies of the gene in WM626, compareddivergence from the VNII strain WM626 and the VNB strainBt63 (higher SDs; Table 1). Although we examined a smallnumber of strains that may not be completely representative,41.10http://genomebiology.com/2008/9/2/R41 Genome Biology 2008,     Volume 9, Issue 2, Article R41       Hu et al. Rthe variable regions may define signature differences thatcould facilitate further characterization of molecular subtypesand epidemiologic studies. A wide variety of hypotheticalgenes and genes with predicted functions were present in theregions of difference, but no clear pattern of variation wasobserved, with the exception that variability was often associ-ated with repetitive elements and/or present at subtelomericregions (Additional data file 4). As described above, repeatedsequences such as those associated with mobile genetic ele-ments may contribute to genome instability and exampleshave been noted for C. neoformans [61]. For the telomericand subtelomeric regions, the observed variability probablyreflects rapid structural evolution of these regions in C. neo-formans, similar to that observed at telomeric or subtelom-eric regions of S. cerevisiae, A. fumigatus, Magnaporthegrisea, and many other organisms [47,73-77]. For example,CGH experiments with the genome of Aspergillus fumigatusstrain Af293 as a reference revealed 2,557 genes that areabsent or diverged in two additional strains of A. fumigatusand three closely related species, namely A. clavatus, Neosa-rtorya fischeri, and N. fennelliae [75]. These absent or diver-gent genes exhibited a bias toward subtelomeric locations[74]. The identification of similar variable segments in C. neo-formans may guide future analyses to assess whether genomevariation contributes to the differences in virulence betweenserotype A strains [70]. Of course, the approach reported herewould not detect regions that are present in the test genomesbut not in H99.The chromosome complements of serotype AD hybrid strainsClinical and environmental isolates of C. neoformans are nor-mally haploid, and the diploid phase is a transient part of thesexual phase of the life cycle. Some clinical and environmen-tal isolates possess a hybrid AD serotype and are presumed toresult from natural fusions between A and D parental strains[78]. Fluorescence-activated cell sorting and PCR analysesalso revealed that AD strains are diploid or aneuploid (>1nbut <2n) [9,26]. We used the tiling arrays for the referencegenomes of H99 (serotype A) and JEC21 (serotype D) todetermine the utility of the CGH approach for characterizinghybrid strains. Three hybrid strains (KW5, CDC228, andCDC304) were chosen because they had previously beencharacterized with respect to virulence and they exhibitedserotype-specific differences at some loci, including genes atthe MAT locus [9].Hybridization of DNA from the AD strains to the JEC21 andH99 arrays suggested that most chromosomes (chromo-somes 2 to 4 and 9 to 14 [numbers based on the JEC21genome]) were represented by copies from both the A and Dgenomes because the average Log2 ratios were close to zero(Figure 5; Additional data file 10). This observation was sup-Log2 ratios for chromosome 1 in the AD strains ranged from-1 to -4 upon hybridization to the JEC21 array, suggesting theabsence of sequences of chromosome 1 originating from a Dgenome, whereas the ratios for hybridization to the H99 arraywere 0.5 to 1, suggesting the presence of one or two copies ofchromosome 1 from the parental A genome. Note that the VNmolecular subtype of the parental strains for the hybrids isnot known, and the variation in the Log2 ratios may reflectsequence divergence between H99 and the serotype A parent.The average Log2 ratios for each chromosome of the threestrains are listed in Additional data file 10. Overall, the resultsindicated that chromosome 1 of KW5, CDC208, and CDC304originated from an A genome, suggesting that the D genomeversion of chromosome 1 may have been lost. Similarly, basedon the Log2 ratios, chromosomes 6 and 7 of KW5 may haveonly a serotype A version, chromosome 8 of KW5 appears tobe from a D genome, and chromosome 5 of CDC304 may haveonly a serotype D version.The predictions about the presence of specific chromosomeswere tested by PCR-RFLP analysis of selected regions.Specifically, tests were performed with chromosome 5 as arepresentative chromosome only from serotype D in CDC304and with chromosome 1 as representative of a chromosomeonly from serotype A in all three strains. For comparison,chromosomes 2 and 3 were included as examples ofchromosomes that were present from both serotype A and Dparents (Figure 5). Initially, PCR-RFLP analysis of a region inthe JEC21 gene CNE04380 on chromosome 5 that was con-served between the A and D genomes confirmed that KW5and CDC228 have both a serotype A and serotype D copy ofchromosome 5, whereas CD304 only has the serotype Dsequence of this region (Figure 6). Digestion with a differentenzyme confirmed these results and revealed that the sero-type A copy in KW5 has a different RFLP pattern from that inCDC228, most likely due to restriction site polymorphisms.Parallel PCR-RFLP analyses for chromosomes 2 and 3 con-firmed the CGH prediction that the AD hybrids containedthese chromosomes from both the A and D genomes.Chromosome 1 in the AD hybrids was also examined by PCR-RFLP analysis of a segment of the conserved JEC21 geneCNA01230 (Figure 6). This analysis initially indicated that allthree strains had only a serotype A version of chromosome 1,although closer examination revealed faint additional bandsfrom strain CDC304 that could result from incomplete diges-tion, sequence polymorphisms at the site, or the presence ofthe serotype D version of chromosome 1. Restriction digestswith additional enzymes suggested that CDC304 might havea serotype D version of chromosome 1 present at a low level(Figure 6). One possibility was that strain CDC304 containeda mixed population of cells in which the majority had aserotype A copy of chromosome 1 and a minority also con-Genome Biology 2008, 9:R41ported by the finding that other chromosomes of the A and Dgenomes were not equally represented in the three AD hybridstrains (Figure 5 and Additional data file 10). Specifically,tained a copy from serotype D. To test this hypothesis, we iso-lated and analyzed four different colonies of CDC304 andfound that the cells in three colonies had only a serotype A41.11http://genomebiology.com/2008/9/2/R41 Genome Biology 2008,     Volume 9, Issue 2, Article R41       Hu et al. Rversion of chromosome 1. The other colony contained cells inwhich the serotype A and D versions of chromosome 1 werepresent in approximately equal abundance (data not shown).We therefore hypothesize that most cells in strain CDC304have lost the serotype D copy of chromosome 1, but that aminor population retains this chromosome. Thus, CDC304may still be in the process of losing chromosomes from theoriginal A and D parents, and this observation is consistentwith the genome instability observed for AD hybrid strains[9].The AD hybrid strains of Cryptococcus examined in thisstudy showed that chromosomes of both A and D serotypesare not equally represented in each AD hybrid strain. That is,specific chromosomes were represented by sequences frommosomes are present in two copies because the average Log2ratios were about 0.4. It is notable that all three AD strainspreferentially contained chromosome 1 sequences from theserotype A parental strain but not from the D strain. Tostrengthen the conclusion that chromosome 1 from serotypeA is preferentially retained in hybrids, we obtained 16 addi-tional AD hybrid strains and examined the origin of chromo-some 1 using the RFLP-PCR method (Additional data file 11).We found that 11 possessed chromosome 1 sequences onlyfrom serotype A and the other five had copies of chromosome1 from both A and D parents. These results agree with those ofNakamura and coworkers [79] and Okabayashi andcolleagues [80], who used CAP59 gene sequences from chro-mosome 1 to examine phylogenetic relationships betweenserotypes. They found that the three AD hybrid strains thatHybridization analysis of three AD hybrid strainsFigure 5Hybridization analysis of three AD hybrid strains. The chromosome (chr) numbers listed at the bottom of each panel follow those of the reference genomes on the tiling arrays; for example, the chromosome numbers from JEC21 are used for the hybridization of DNA from each strain to the JEC21 array. Note that the JEC21 and H99 genomes are largely co-linear, but some of the homologous chromosomes have been assigned different numbers in the current genome assemblies [61]. Most of the chromosomes of the AD hybrid strains are represented by copies from both A and D genomes. However, the hybridization signals indicate that chromosome 1 is only represented by sequences from a serotype A genome in all three strains. Similarly, chromosomes 5, 6, and 7 (and chromosome 14 in KW5) are represented by sequences from only one of the serotypes (either A or D). The average Log2 ratios and standard deviations for all of the chromosomes are listed in Additional data file 6, and PCR-RFLP confirmation for selected chromosomes is shown in Figure 6.Log2 ratioLog2 ratioLog2 ratioLog2 ratioLog2 ratioLog2 ratioGenome Biology 2008, 9:R41only one serotype. Although, we have not determined thecopy number for chromosomes that are represented by a sin-gle serotype sequence, the possibility exists that these chro-they tested grouped with serotype A strains, suggesting thatthese strains also had only the serotype A allele of CAP59.Similarly, Xu and coworkers [24] found evidence for loss of41.12http://genomebiology.com/2008/9/2/R41 Genome Biology 2008,     Volume 9, Issue 2, Article R41       Hu et al. Rheterozygosity in AD hybrid strains using MSLT analysis withfour genes and identified strains that cluster with serotype Astrains as well as others that lacked consistent grouping withone serotype.The emerging picture of frequent loss of heterozygosity inhybrid strains and the retention of specific chromosomesraises questions about the mechanisms of loss and retention.It is possible that chromosome 1 of serotype A carries a geneor genes that confer a selective advantage relative to the sero-type D version of the chromosome, and/or that the latterchromosome has a disadvantageous combination of genes. Inthis case, hybrids that spontaneously lose chromosome 1 fromserotype A might be at a selective disadvantage. It is also pos-sible that chromosome 1 from the serotype D parent has aninherent defect in replication or transmission, such that it ispreferentially lost or that incompatibility exists for somechromosomes. The mechanisms underlying these patterns ofchromosome content, and possible relevance for virulence,remain to be investigated. It is interesting, however, that theutes a selective advantage in the mammalian hostenvironment.A set of genes or regions in the AD hybrids have been ampli-fied and compared previously [9]. We examined the CGHdata for these sequences, and the Log2 ratios are consistentwith the interpretations of chromosome content (data notshown). The three AD strains have also been tested previouslyfor virulence in comparison with strain H99 [9]. These assaysindicated that the AD hybrids are less virulent than H99 anddiffer from each other. Specifically, infected mice succumbedto H99 by day 25, to KW5 by day 100, and to CDC228 orCDC304 by day 150. It is known that serotype A strains aregenerally more virulent that serotype D strains, althoughstrain variation exists in both groups [69,70]. Differences inchromosomal content could potentially influence virulencethrough the contributions of specific alleles, such that agreater content of the A genome (as found in KW5) may resultin enhanced virulence. Of course, other explanations arepossible, including the accumulation of mutations in key vir-PCR-RFLP confirmation of the presence of serotype-specific chromosomes in three AD hybrid strainsFigure 6PCR-RFLP confirmation of the presence of serotype-specific chromosomes in three AD hybrid strains. Agarose gels are shown in which lane 1 for each contains size markers (1 kilobase [kb] ladder); lanes 2, 4, 6, 8, and 10 contain undigested PCR fragments; and the remaining lanes contain the same fragments after restriction enzyme digestion. (a) Digestion of PCR fragments (primers CNA01230 F/R) from chromosome (chr) 1 with AvaI (left panel) or StuI (right panel). (b) Digestion of PCR fragments (primers CNE04380 F/R) from chromosome 5 with TaqI (left panel) or HindIII (right panel). (c) Digestion of a PCR fragment (primers CNB01970 F/R) from chromosome 2 with NdeI (left panel) and a fragment (primers acidphos F/R) from chromosome 3 with TaqI (right panel).(a)(b)(c)Chromosome 1Chromosome 5Chromosome 2 Chromosome 3Genome Biology 2008, 9:R41AD strains are clinical isolates, which raises the possibilitythat retention of chromosome 1 from the serotype A genome(serotype A strains are known to be more virulent) contrib-ulence traits in the less virulent AD strains and epigeneticphenomena. Barchiesi and coworkers [69] suggested that thepresence of the serotype A, MATα mating type allele in either41.13http://genomebiology.com/2008/9/2/R41 Genome Biology 2008,     Volume 9, Issue 2, Article R41       Hu et al. Rhaploid or diploid (aneuploid) strains is correlated withvirulence, whereas the MATa in serotype A or MATα allele inserotype D is associated with moderate or no virulence. Inthis regard, Lengeler and colleagues [9] showed that KW5 isheterozygous for the mating-type locus with the serotype DMATa and the serotype A MATα loci present. In contrast, theCDC228 and CDC304 strains have the MATα locus from aserotype D parent and appear to have the MATa locus from aserotype A parent. Therefore, the contributions of matingtype to virulence are unclear, with contributions likely bothfrom specific alleles and variable chromosome complementsin the AD hybrids.ConclusionThe CGH data presented here corroborated the aneuploidy ofAD hybrids and identified preferential chromosome retentionin some strains. Coupled with the discovery of copy numberdifferences for chromosome 13 in two serotype A strains,these results identified an unexpected level of genomevariation in C. neoformans. This extensive genome variationis in contrast to previous reports of extensive clonality in alarge number of isolates representing the environmental pop-ulation [81]. This may partly reflect one of the advantages ofCGH over AFLP analysis, in that the latter method is typicallyunable to identify chromosome or segmental duplication.Many fungi have a remarkable tolerance to variability in chro-mosome content, and this may be advantageous for adapta-tion to different environmental niches. For example,aneuploidy is common in C. albicans, and aneuploidy andisochromosome formation can contribute to drug resistance[50,51]. Additionally, changes in chromosome copy numbercan influence virulence [82]. Given these findings, one canenvisage more detailed CGH experiments to determinewhether genome variation contributes to the phenotypic andvirulence differences between strains, mutants, and switchvariants. Variation may also occur during passage through ananimal or during antifungal drug treatment. In combinationwith physical mapping [83] and sequencing, CGH will alsoallow detailed characterization of emerging strains of clinicalsignificance and novel populations such as the unusual VNBisolates that appear to be restricted to Botswana [39]. In thislight, we are also using the sequenced genomes of C. gattiistrains and CGH to investigate genome variability in strainsfrom the outbreak that is ongoing on Vancouver Island [7,84].The combined view of genome variability in C. neoformansand C. gattii may thus provide insights into the mechanismsof genome microevolution in these pathogens.Materials and methodsStrains, genomic DNA extraction and array hybridizationDNA for hybridization experiments was isolated as previouslydescribed [85]. The genome sequences of strain JEC21 [46]and strain H99 were obtained from public databases [86,87].The assembled sequences were used by NimbleGen Systems,Inc. (Madison, WI, USA) [88] to design and manufacturehigh-density oligonucleotide genomic arrays. The design forthe arrays was the same for each genome, with oligonucle-otide probes that cover all 14 chromosomes for each genometiled at an average interval spacing of 44 bp on one strand.The average length of the probes was 50 bp (range 45 to 85bp) and the average melting temperature (Tm) of the probeson each array was 76°C. The number of oligonucleotideprobes for each genome were as follows: 386,279 for H99 and380,236 for JEC21. The probes for each array were designedto uniquely match a single sequence in the genome, andhighly repetitive centromeric regions and the rDNA repeatcluster were not included. In H99, a region of the matinglocus (chromosome 5: 203,813 to 220,976) was consideredrepetitive and not included on the array. The comparativestandard operation procedures of NimbleGen Systems, Inc.were followed for hybridization (42°C), array scanning, anddata acquisition, as described previously [60]. Genomic DNAfrom the test strains was labelled with Cy3 and that of refer-ence strains with Cy5. The data were expressed as Log2 ratiosof Cy3/Cy5 fluorescence intensity. The arrays are availablefrom NimbleGen Systems, Inc.Data analysisThe data were extracted from scanned array images usingNimbleScan 2.0 software (NimbleGen Systems, Inc.) andwere provided to us as an initial DNA segmentation analysisof the normalized data. This analysis included a window aver-aging step, in which the probes that fell into a defined basepair window size were averaged using the Tukey biweightmean. The adjacent windows were averaged to reduce the sizeof the dataset and the noise in the data. In the present study,the data were analyzed using 400, 800, and 2,000 bp win-dows for averaging; these sizes represent 10, 20, and 50 timesthe length of an average probe. In most cases, a 400 bp win-dow was used for the analysis. The segmentation of the aver-aged log2 ratio data was determined based on a circularbinary segmentation algorithm [89]. We also examined thedata for each individual probe in the analyses shown in Figure1 and Additional data file 4. The sample key for the rawhybridization data is provided in Additional data file 13 andthe actual data files are available on our website [90].The CGH results were viewed and analyzed as GFF files withSignalMap (NimbleGen Systems, Inc.) using 400, 800, and2,000 bp windows for averaging. The data for the MAT loci(Figure 1) were analyzed using Log2 ratios from segmenta-tions based on a 400 bp window and the nucleotide sequenceidentity for the corresponding region of available homologousGenome Biology 2008, 9:R41Additional data file 12 lists the strains used in this study. Thestrains were maintained on yeast extract, peptone, dextrosemedium (YPD; Difco, Sparks, MD, USA) and the genomicsequences from 20 genes in the loci. The sequences of theMAT loci were obtained from GenBank, and sequence iden-tity was determined from alignments with the BioEdit41.14http://genomebiology.com/2008/9/2/R41 Genome Biology 2008,     Volume 9, Issue 2, Article R41       Hu et al. RSequence Alignment Editor [91]. The accession numbers forthe sequences are as follows: serotype A strains H99 (MATα,AF542529) and 125.91 (MATa, AF542528), and serotype Dstrains JEC21 (MATα, AF542531) and JEC20 (MATa,AF542530). The relationship between Log2 ratio values andsequence identity was examined by linear regression analysisin Excel.Verification of CGH results by PCR amplification and RFLP analysisSelected regions identified by CGH as candidates for deleted,duplicated, or divergent sequences were examined by PCRamplification and sequence analysis. PCR amplification fromgenomic DNA was performed as described by Hu and Krons-tad [85], with primers designed from the sequences flankingeach predicted region of difference, so that the amplified frag-ment spanned the region. The primers were obtained fromInvitrogen (Burlington, Ontario, Canada) and theirsequences are listed in Additional data file 14. PCR-amplifiedDNA fragments were either sequenced using the dideoxychain-termination method or digested with selected restric-tion enzymes in the case of AD hybrids. Primers for the anal-ysis of AD hybrid strains were designed to target highlyconserved regions in the genome to ensure that the primerswould bind to both the serotype A and D alleles in thesestrains. These conserved regions were identified by aligningthe sequences from H99 (representing serotype A) and JEC21(representing serotype D) with BioEdit [91].AbbreviationsAFLP, amplified fragment length polymorphism; bp, basepairs; CGH, comparative genome hybridization; kb, kilo-bases; Mb, megabases; MLST, multilocus sequence typing;PCR, polymerase chain reaction; RT, reverse transcription;SD, standard deviation; TIGR, The Institute for GenomicResearch.Authors' contributionsGH, IL and JWK conceived and designed the study. FSD, GH,JWK, IL, and JES analyzed the data and wrote the paper. ASanalyzed the CGH data for the MAT loci.Additional data filesThe following additional data are available. Additional datafile 1 is a table of Log2 ratios of divergent and conserved seg-ments of the JEC21 genome relative to the genomes of theprogenitor strains NIH12 and NIH433. Additional data file 2is a table of the regions of putative recombination sites acrossall of the chromosomes in the JEC21 genome. Additional datafile 3 is a figure of the variation in the genomes of strains rep-pared with the sequenced genome of strain H99. Additionaldata file 5 is a table of the quantitative RT-PCR analysis ofgene copy number relative to the H99 genome. Additionaldata file 6 is a table of the quantitative RT-PCR analysis ofgene copy number relative to the JEC21 genome. Additionaldata file 7 is a figure showing the replacement of a gene onchromosome 13 to examine copy number. Additional data file8 describes the genomic hybridization analysis of transform-ants carrying a replacement at the APT1 gene on chromosome13 in strains H99, CBS7779, and WM626; this file alsopresents an estimation of relative copy number by quantita-tive real-time PCR of selected markers on chromosome 13 instrains H99, JEC21, CBS7779, and WM626. Additional datafile 9 is a figure showing the phenotypic differences betweenstrains H99, CBS7779, and WM626. Additional data file 10 isa table of the average Log2 ratios of the chromosomes in threeAD hybrid strains upon hybridization to the tiling arrays ofthe JEC21 and H99 genomes. Additional data file 11 is a figureshowing an RFLP-PCR analysis of the origin of chromosome1 in 16 AD hybrid strains. Additional data file 12 is a table listof Cryptococcus neoformans and C. gattii strains. Additionaldata file 13 is a sample key for the CGH data. Additional datafile 14 is a table list of primer sequences. Additional data file15 contains the figure legends for the figures presented inAdditional data files 3, 7, 9 and 11.Additional data file 1Log2 ratios of divergent and conserved segments of the JEC21 genomePresented is tab of Log2 ratios of divergent and conserved seg-m nts of the JEC21 ge me relative to the genomes f the prog n-it r str in  NIH1  and NIH433.Click h re for file 2Re io s of pu tiv r co bination sites across ll of the chromo-s m i  nthe regi put tiv  recombi ation sit s acro ll of the ch om somes  the JEC21 g ome.3Va iati n in g n e  of str ins r pr senti g th r molecular ub ype within the A s rotype f C. n ofo mansfigu e of the vari on in    t ain  r pr sen ng the hre  mol cula  subtyp s withi  the A se oty e of C. n of m n . 4i f ce i  the gen m s f fou  ser type A veru  the s qu nc d g nom of s rain H99r gion  of dif re c  i  t e g nom s of f uryp A ra ns p r  w th th  eq enced gen m  f straiH99. 5Qua ativ RT-PCR an lysi gen  copy u b r r l ive t he g m qu tative RT-PCR analysi of g n  c py umb r lativ t t H99 om .6JEC 1 genome JEC21 g om .7pl c of a ge  n ch om s m 13 o x in  cop u b rr  sh w n  he repla of n chr mo 13 xam n  c y umb r.8G o ic hybr dizatio analy is f a f r a c rying a l c t at h  APT1 n c m 13 tr ins H99, BS7779, and WM6 6ocu d bi g th g n mi hy idizati  naly tr ns o a s a ry a r pl ent at  APT1 g no h s me 13 i s r H 9, CBS7779, WM626. his f ll  p s n st io   el e c py umb r b  quantita iva - PC  o  l ct d k on hro  13 n st a s, JEC21, CBS77 9, d WM626.9h otyp c di r c b tw n r n H99, CBS7779, d WM626 sh in he p typic d ff s b twet s H99 a0v rag L g2 t os o s in thr e AD h br s ra sup br z i  til ng a ys  h  JEC21 an H99 s age Log2 r i c mon thre  AD hybr rai up hyb z  til g ray fth JEC21 and H99 .1FLP-P R an ly i g f ch s 1 i  16 D ybr dRFLP-PCR ysis f the if ro i 1 D br d in .2yp coc us e fo d C. a t i s a slat d li t f r pt c us f ma s C. t i . 3S le k y fo CGH ts mpl k y f CG d a.4me e p m r s.5Figu  l g s f r Ad i l a 3, 7, 9 ande h  i ur g s f r  pr t d i  Addii l d a ile 3, , a 11AcknowledgementsWe thank Drs Teun Boekhout, Joseph Heitman, June Kwon-Chung, KirstenNielsen, Wieland Meyer, and J-P Xu for providing strains, and the BroadInstitute for access to the H99 sequence. This work was supported by theNational Institute of Allergy and Infectious Disease (R01 AI053721) and bythe Canadian Institutes of Health Research. JWK is a Burroughs WellcomeFund Scholar in Molecular Pathogenic Mycology.References1. Casadevall A, Cassone A, Bistoni F, Cutler JE, Magliani W, MurphyJW, Polonelli L, Romani L: Antibody and/or cell-mediatedimmunity, protective mechanisms in fungal disease: anongoing dilemma or an unnecessary dispute?  Med Mycol 1998,36(Suppl 1):95-105.2. Casadevall A, Perfect JR: Cryptococcus neoformans Washington, DC:American Society for Microbiology Press; 1998. 3. Sorrell TC: Cryptococcus neoformans variety gattii.  Med Mycol2001, 39:155-168.4. Franzot SP, Fries BC, Cleare W, Casadevall A: Genetic relation-ship between Cryptococcus neoformans var. neoformansstrains of serotypes A and D.  J Clin Microbiol 1998, 36:2200-2204.5. Kwon-Chung KJ, Wickes BL, Stockman L, Roberts GD, Ellis D,Howard DH: Virulence, serotype, and molecular characteris-tics of environmental strains of Cryptococcus neoformans var.gattii.  Infect Immun 1992, 60:1869-1874.6. Kwon-Chung KJ, Varma A: Do major species concepts supportone, two or more species within Cryptococcus neoformans?FEMS Yeast Res 2006, 6:574-587.7. Bartlett KH, Kidd SE, Kronstad JW: The emergence of Cryptococ-cus gattii in British Columbia and the Pacific Northwest.  CurrInfect Dis Rep 2008, 10:58-65.8. Boekhout T, Theelen B, Diaz M, Fell JW, Hop WC, Abeln EC,Dromer F, Meyer W: Hybrid genotypes in the pathogenic yeastCryptococcus neoformans.  Microbiology 2001, 147:891-907.Genome Biology 2008, 9:R41resenting the three molecular subtypes within the A serotypeof C. neoformans. Additional data file 4 is a table of regions ofdifference in the genomes of four serotype A strains com-9. Lengeler KB, Cox GM, Heitman J: Serotype AD strains of Crypto-coccus neoformans are diploid or aneuploid and are hetero-zygous at the mating-type locus.  Infect Immun 2001, 69:115-122.10. Bovers M, Hagen F, Kuramae EE, Diaz MR, Spanjaard L, Dromer F,41.15http://genomebiology.com/2008/9/2/R41 Genome Biology 2008,     Volume 9, Issue 2, Article R41       Hu et al. RHoogveld HL, Boekhout T: Unique hybrids between the fungalpathogens Cryptococcus neoformans and Cryptococcus gattii.FEMS Yeast Res 2006, 6:599-607.11. Cogliati M, Esposto MC, Tortorano AM, Viviani MA: Cryptococcusneoformans population includes hybrid strains homozygousat mating-type locus.  FEMS Yeast Res 2006, 6:608-613.12. Nielsen K, Cox GM, Wang P, Toffaletti DL, Perfect JR, Heitman J:Sexual cycle of Cryptococcus neoformans var. grubii and viru-lence of congenic a and alpha isolates.  Infect Immun 2003,71:4831-4841.13. Hull CM, Heitman J: Genetics of Cryptococcus neoformans.  AnnuRev Genet 2002, 36:557-615.14. Lengeler KB, Fox DS, Fraser JA, Allen A, Forrester K, Dietrich FS,Heitman J: Mating-type locus of Cryptococcus neoformans: astep in the evolution of sex chromosomes.  Eukaryot Cell 2002,1:704-718.15. Fraser JA, Diezmann S, Subaran RL, Allen A, Lengeler KB, Dietrich FS,Heitman J: Convergent evolution of chromosomal sex-deter-mining regions in the animal and fungal kingdoms.  PLoS Biol2004, 2:e384.16. Hsueh YP, Idnurm A, Heitman J: Recombination hotspots flankthe Cryptococcus mating-type locus: implications for theevolution of a fungal sex chromosome.  PLoS Genet 2006,2:e184.17. Kwon-Chung KJ, Edman JC, Wickes BL: Genetic association ofmating types and virulence in Cryptococcus neoformans.  InfectImmun 1992, 60:602-605.18. Nielsen K, Cox GM, Litvintseva AP, Mylonakis E, Malliaris SD, Ben-jamin DK Jr, Giles SS, Mitchell TG, Casadevall A, Perfect JR, HeitmanJ: Cryptococcus neoformans α strains preferentially dissemi-nate to the central nervous system during coinfection.  InfectImmun 2005, 73:4922-4933.19. Nielsen K, Marra RE, Hagen F, Boekhout T, Mitchell TG, Cox GM,Heitman J: Interaction between genetic background and themating-type locus in Cryptococcus neoformans virulencepotential.  Genetics 2005, 171:975-983.20. Ren P, Roncaglia P, Springer DJ, Fan J, Chaturvedi V: Genomicorganization and expression of 23 new genes from MATal-pha locus of Cryptococcus neoformans var. gattii.  Biochem Bio-phys Res Commun 2005, 326:233-241.21. White CW, Jacobson ES: Occurrence of diploid strains of Cryp-tococcus neoformans.  J Bacteriol 1985, 161:1231-1232.22. Sia RA, Lengeler KB, Heitman J: Diploid strains of the pathogenicbasidiomycete Cryptococcus neoformans are thermallydimorphic.  Fungal Genet Biol 2000, 29:153-163.23. Xu J, Luo G, Vilgalys RJ, Brandt ME, Mitchell TG: Multiple origins ofhybrid strains of Cryptococcus neoformans with serotype AD.Microbiology 2002, 148:203-212.24. Xu J, Mitchell TG: Comparative gene genealogical analyses ofstrains of serotype AD identify recombination in populationsof serotypes A and D in the human pathogenic yeast Crypto-coccus neoformans.  Microbiology 2003, 149:2147-2154.25. Cogliati M, Allaria M, Tortorano AM, Viviani MA: Genotyping Cryp-tococcus neoformans var. neoformans with specific primersdesigned from PCR-fingerprinting bands sequenced using amodified PCR-based strategy.  Med Mycol 2000, 38:97-103.26. Cogliati M, Esposto MC, Clarke DL, Wickes BL, Viviani MA: Originof Cryptococcus neoformans var. neoformans diploid strains.  JClin Microbiol 2001, 39:3889-3894.27. Varma A, Kwon-Chung KJ: Restriction fragment polymorphismin mitochondrial DNA of Cryptococcus neoformans.  J GenMicrobiol 1989, 135:3353-3362.28. Meyer W, Mitchell TG: Polymerase chain reaction fingerprint-ing in fungi using single primers specific to minisatellites andsimple repetitive DNA sequences: strain variation in Crypto-coccus neoformans.  Electrophoresis 1995, 16:1648-1656.29. Meyer W, Marszewska K, Amirmostofian M, Igreja RP, Hardtke C,Methling K, Viviani MA, Chindamporn A, Sukroongreung S, John MA,Ellis DH, Sorrell TC: Molecular typing of global isolates of Cryp-tococcus neoformans var. neoformans by polymerase chainreaction fingerprinting and randomly amplified polymorphicDNA-a pilot study to standardize techniques on which tobase a detailed epidemiological survey.  Electrophoresis 1999,20:1790-1799.sive conditions.  J Infect Dis 1996, 173:754-758.31. Boekhout T, van Belkum A: Variability of karyotypes and RAPDtypes in genetically related strains of Cryptococcusneoformans.  Curr Genet 1997, 32:203-208.32. Boekhout T, van Belkum A, Leenders AC, Verbrugh HA, Mukamu-rangwa P, Swinne D, Scheffers WA: Molecular typing of Crypto-coccus neoformans: taxonomic and epidemiological aspects.Int J Syst Bacteriol 1997, 47:432-442.33. Viviani MA, Wen H, Roverselli A, Caldarelli-Stefano R, Cogliati M,Ferrante P, Tortorano AM: Identification by polymerase chainreaction fingerprinting of Cryptococcus neoformans serotypeAD.  J Med Vet Mycol 1997, 35:355-360.34. Diaz MR, Boekhout T, Theelen B, Fell JW: Molecular sequenceanalyses of the intergenic spacer (IGS) associated withrDNA of the two varieties of the pathogenic yeast, Crypto-coccus neoformans.  Syst Appl Microbiol 2000, 23:535-545.35. Ellis D, Marriott D, Hajjeh RA, Warnock D, Meyer W, Barton R: Epi-demiology: surveillance of fungal infections.  Med Mycol 2000,38(Suppl 1):173-182.36. Kidd SE, Guo H, Bartlett KH, Xu J, Kronstad JW: Comparativegene genealogies indicate that two clonal lineages of Crypto-coccus gattii in British Columbia resemble strains from othergeographical areas.  Eukaryot Cell 2005, 4:1629-1638.37. Fraser JA, Giles SS, Wenink EC, Geunes-Boyer SG, Wright JR, Diez-mann S, Allen A, Stajich JE, Dietrich FS, Perfect JR, Heitman J: Same-sex mating and the origin of the Vancouver Island Crypto-coccus gattii outbreak.  Nature 2005, 437:1360-1364.38. Litvintseva AP, Marra RE, Nielsen K, Heitman J, Vilgalys R, MitchellTG: Evidence of sexual recombination among Cryptococcusneoformans serotype A isolates in sub-Saharan Africa.Eukaryot Cell 2003, 2:1162-1168.39. Litvintseva AP, Thakur R, Vilgalys R, Mitchell TG: Multilocussequence typing reveals three genetic subpopulations ofCryptococcus neoformans var. grubii (serotype A), including aunique population in Botswana.  Genetics 2006, 172:2223-2238.40. Litvintseva AP, Lin X, Templeton I, Heitman J, Mitchell TG: Manyglobally isolated AD hybrid strains of Cryptococcusneoformans originated in Africa.  PLoS Pathog 2007, 3:e114.41. Perfect JR, Magee BB, Magee PT: Separation of chromosomes ofCryptococcus neoformans by pulsed field gel electrophoresis.Infect Immun 1989, 57:2624-2627.42. Perfect JR, Ketabchi N, Cox GM, Ingram CW, Beiser CL: Karyotyp-ing of Cryptococcus neoformans as an epidemiological tool.  JClin Microbiol 1993, 31:3305-3309.43. Polacheck I, Lebens GA: Electrophoretic karyotype of the path-ogenic yeast Cryptococcus neoformans.  J Gen Microbiol 1989,135:65-71.44. Wickes BL, Moore TD, Kwon-Chung KJ: Comparison of the elec-trophoretic karyotypes and chromosomal location of tengenes in the two varieties of Cryptococcus neoformans.  Micro-biology 1994, 140:543-550.45. Forche A, Xu J, Vilgalys R, Mitchell TG: Development and charac-terization of a genetic linkage map of Cryptococcus neoform-ans var. neoformans using amplified fragment lengthpolymorphisms and other markers.  Fungal Genet Biol 2000,31:189-203.46. Loftus BJ, Fung E, Roncaglia P, Rowley D, Amedeo P, Bruno D,Vamathevan J, Miranda M, Anderson IJ, Fraser JA, Allen JE, Bosdet IE,Brent MR, Chiu R, Doering TL, Donlin MJ, D'Souza CA, Fox DS, Grin-berg V, Fu J, Fukushima M, Haas BJ, Huang JC, Janbon G, Jones SJ, KooHL, Krzywinski MI, Kwon-Chung JK, Lengeler KB, Maiti R, et al.: Thegenome of the basidiomycetous yeast and human pathogenCryptococcus neoformans.  Science 2005, 307:1321-1324.47. Winzeler EA, Castillo-Davis CI, Oshiro G, Liang D, Richards DR,Zhou Y, Hartl DL: Genetic diversity in yeast assessed withwhole-genome oligonucleotide arrays.  Genetics 2003,163:79-89.48. Watanabe T, Murata Y, Oka S, Iwahashi H: A new approach to spe-cies determination for yeast strains: DNA microarray-basedcomparative genomic hybridization using a yeast DNAmicroarray with 6000 genes.  Yeast 2004, 21:351-365.49. Pinkel D, Albertson DG: Comparative genomic hybridization.Annu Rev Genomics Hum Genet 2005, 6:331-354.50. Selmecki A, Bergmann S, Berman J: Comparative genome hybrid-Genome Biology 2008, 9:R4130. Chen SC, Brownlee AG, Sorrell TC, Ruma P, Nimmo G: Identifica-tion by random amplification of polymorphic DNA of a com-mon molecular type of Cryptococcus neoformans var.neoformans in patients with AIDS or other immunosuppres-ization reveals widespread aneuploidy in Candida albicanslaboratory strains.  Mol Microbiol 2005, 55:1553-1565.51. Selmecki A, Forche A, Berman J: Aneuploidy and isochromo-some formation in drug-resistant Candida albicans.  Science41.16http://genomebiology.com/2008/9/2/R41 Genome Biology 2008,     Volume 9, Issue 2, Article R41       Hu et al. R2006, 313:367-370.52. Parker CT, Quiñones B, Miller WG, Horn ST, Mandrell RE: Com-parative genomic analysis of Campylobacter jejuni strainsreveals diversity due to genomic elements similar to thosepresent in C. jejuni strain RM1221.  J Clin Microbiol 2006,44:4125-4135.53. Moran G, Stokes C, Thewes S, Hube B, Coleman DC, Sullivan D:Comparative genomics using Candida albicans DNA micro-arrays reveals absence and divergence of virulence-associ-ated genes in Candida dubliniensis.  Microbiology 2004,150:3363-3382.54. Carret CK, Horrocks P, Konfortov B, Winzeler E, Qureshi M, New-bold C, Ivens A: Microarray-based comparative genomic anal-yses of the human malaria parasite Plasmodium falciparumusing Affymetrix arrays.  Mol Biochem Parasitol 2005, 144:177-186.55. Dunn B, Levine RP, Sherlock G: Microarray karyotyping of com-mercial wine yeast strains reveals shared, as well as unique,genomic signatures.  BMC Genomics 2005, 6:53.56. Taboada EN, Acedillo RR, Luebbert CC, Findlay WA, Nash JH: Anew approach for the analysis of bacterial microarray-basedComparative Genomic Hybridization: insights from anempirical study.  BMC Genomics 2005, 6:78.57. Brunelle BW, Nicholson TL, Stephens RS: Microarray-basedgenomic surveying of gene polymorphisms in Chlamydiatrachomatis.  Genome Biol 2004, 5:R42.58. Fukiya S, Mizoguchi H, Tobe T, Mori H: Extensive genomic diver-sity in pathogenic Escherichia coli and Shigella Strainsrevealed by comparative genomic hybridization microarray.J Bacteriol 2004, 186:3911-3921.59. Heitman J, Allen B, Alspaugh JA, Kwon-Chung KJ: On the origins ofcongenic MATalpha and MATa strains of the pathogenicyeast Cryptococcus neoformans.  Fungal Genet Biol 1999, 28:1-5.60. Selzer RR, Richmond TA, Pofahl NJ, Green RD, Eis PS, Nair P, Broth-man AR, Stallings RL: Analysis of chromosome breakpoints inneuroblastoma at sub-kilobase resolution using fine-tilingoligonucleotide array CGH.  Genes Chromosomes Cancer 2005,44:305-319.61. Kavanaugh LA, Fraser JA, Dietrich FS: Recent evolution of thehuman pathogen Cryptococcus neoformans by intervarietaltransfer of a 14-gene fragment.  Mol Biol Evol 2006,23:1879-1890.62. Cherniak R, Valafar H, Morris LC, Valafar F: Cryptococcus neofor-mans chemotyping by quantitative analysis of 1H nuclearmagnetic resonance spectra of glucuronoxylomannans witha computer-simulated artificial neural network.  Clin Diagn LabImmunol 1998, 5:146-159.63. Janbon G, Himmelreich U, Moyrand F, Improvisi L, Dromer F: Cas1pis a membrane protein necessary for the O-acetylation ofthe Cryptococcus neoformans capsular polysaccharide.  MolMicrobiol 2001, 42:453-467.64. Lengeler KB, Wang P, Cox GM, Perfect JR, Heitman J: Identificationof the MATa mating-type locus of Cryptococcus neoformansreveals a serotype A MATa strain thought to have beenextinct.  Proc Natl Acad Sci USA 2000, 97:14455-14460.65. Viviani MA, Esposto MC, Cogliati M, Montagna MT, Wickes BL: Iso-lation of a Cryptococcus neoformans serotype A MATa strainfrom the Italian environment.  Med Mycol 2001, 39:383-386.66. Xue C, Tada Y, Dong X, Heitman J: The human fungal pathogenCryptococcus neoformans can complete its sexual cycle dur-ing a pathogenic association with plants.  Cell Host Microbe 2007,1:263-273.67. Fraser JA, Huang JC, Pukkila-Worley R, Alspaugh JA, Mitchell TG,Heitman J: Chromosomal translocation and segmental dupli-cation in Cryptococcus neoformans.  Eukaryot Cell 2005,4:401-406.68. Fries BC, Chen F, Currie BP, Casadevall A: Karyotype instability inCryptococcus neoformans infection.  J Clin Microbiol 1996,34:1531-1534.69. Barchiesi F, Cogliati M, Esposto MC, Spreghini E, Schimizzi AM,Wickes BL, Scalise G, Viviani MA: Comparative analysis of path-ogenicity of Cryptococcus neoformans serotypes A, D and ADin murine cryptococcosis.  J Infect 2005, 51:10-16.70. Clancy CJ, Nguyen MH, Alandoerffer R, Cheng S, Iczkowski K, Rich-ardson M, Graybill JR: Cryptococcus neoformans var. grubii iso-71. Jain N, Guerrero A, Fries BC: Phenotypic switching and its impli-cations for the pathogenesis of Cryptococcus neoformans.FEMS Yeast Res 2006, 6:480-488.72. Torres EM, Sokolsky T, Tucker CM, Chan LY, Boselli M, Dunham MJ,Amon A: Effects of aneuploidy on cellular physiology and celldivision in haploid yeast.  Science 2007, 317:916-924.73. Bond U, Neal C, Donnelly D, James TC: Aneuploidy and copynumber breakpoints in the genome of lager yeasts mappedby microarray hybridisation.  Curr Genet 2004, 45:360-370.74. Galagan JE, Calvo SE, Cuomo C, Ma LJ, Wortman JR, Batzoglou S, LeeSI, Bas¸türkmen M, Spevak CC, Clutterbuck J, Kapitonov V, Jurka J,Scazzocchio C, Farman M, Butler J, Purcell S, Harris S, Braus GH,Draht O, Busch S, D'Enfert C, Bouchier C, Goldman GH, Bell-Peder-sen D, Griffiths-Jones S, Doonan JH, Yu J, Vienken K, Pain A, FreitagM, et al.: Sequencing of Aspergillus nidulans and comparativeanalysis with A. fumigatus and A. oryzae.  Nature 2005,438:1105-1115.75. Nierman WC, Pain A, Anderson MJ, Wortman JR, Kim HS, Arroyo J,Berriman M, Abe K, Archer DB, Bermejo C, Bennett J, Bowyer P,Chen D, Collins M, Coulsen R, Davies R, Dyer PS, Farman M, Fedor-ova N, Fedorova N, Feldblyum TV, Fischer R, Fosker N, Fraser A,García JL, García MJ, Goble A, Goldman GH, Gomi K, Griffith-JonesS, et al.: Genomic sequence of the pathogenic and allergenicfilamentous fungus Aspergillus fumigatus.  Nature 2005,438:1151-1156.76. Bhattacharyya MK, Lustig AJ: Telomere dynamics in genomestability.  Trends Biochem Sci 2006, 31:114-122.77. Rehmeyer C, Li W, Kusaba M, Kim YS, Brown D, Staben C, Dean R,Farman M: Organization of chromosome ends in the rice blastfungus, Magnaporthe oryzae.  Nucleic Acids Res 2006,34:4685-4701.78. Tanaka R, Taguchi H, Takeo K, Miyaji M, Nishimura K: Determina-tion of ploidy in Cryptococcus neoformans by flow cytometry.J Med Vet Mycol 1996, 34:299-301.79. Nakamura Y, Kano R, Watanabe S, Hasegawa A: Molecular analysisof CAP59 gene sequences from five serotypes of Cryptococcusneoformans.  J Clin Microbiol 2000, 38:992-995.80. Okabayashi K, Kano R, Nakamura Y, Watanabe S, Hasegawa A: Cap-sule-associated genes of serotypes of Cryptococcusneoformans, especially serotype AD.  Med Mycol 2006,44:127-132.81. Litvintseva AP, Kestenbaum L, Vilgalys R, Mitchell TG: Comparativeanalysis of environmental and clinical populations of Crypto-coccus neoformans.  J Clin Microbiol 2005, 43:556-564.82. Chen X, Magee BB, Dawson D, Magee PT, Kumamoto CA: Chromo-some 1 trisomy compromises the virulence of Candidaalbicans.  Mol Microbiol 2004, 51:551-565.83. Schein JE, Tangen KL, Chiu R, Shin H, Lengeler KB, MacDonald WK,Bosdet I, Heitman J, Jones SJ, Marra MA, Kronstad JW: Physicalmaps for genome analysis of serotype A and D strains of thefungal pathogen Cryptococcus neoformans.  Genome Res 2002,12:1445-1453.84. Kidd SE, Chow Y, Mak S, Bach PJ, Chen H, Hingston AO, KronstadJW, Bartlett KH: Characterization of environmental sources ofthe human and animal pathogen Cryptococcus gattii in BritishColumbia, Canada, and the Pacific Northwest of the UnitedStates.  Appl Environ Microbiol 2007, 73:1433-1443.85. Hu G, Kronstad JW: Gene disruption in Cryptococcus neoform-ans and Cryptococcus gattii by in vitro transposition.  Curr Genet2006, 49:341-350.86. Resources for fungal comparative genomics   [http://fungal.genome.duke.edu/]87. The Broad Institute   [http://www.broad.mit.edu/annotation/fgi/]88. NimbleGen Systems, Inc.   [http://www.nimblegen.com/products/cgh/index.html]89. Olshen AB, Venkatraman ES, Lucito R, Wigler M: Circular binarysegmentation for the analysis of array-based DNA copynumber data.  Biostatistics 2004, 5:557-572.90. Kronstad laboratory web site   [http://www.kronstadlab.msl.ubc.ca/filesharing/CGHdatafiles.zip]91. BioEdit   [http://www.mbio.ncsu.edu/BioEdit/bioedit.html]92. Hu G, Steen BR, Lian TS, Sham AP, Tam N, Tangen KL, Kronstad JW:Transcriptional regulation by protein kinase A in Cryptococ-cus neoformans.  PLoS Pathog 2007, 3:e42.Genome Biology 2008, 9:R41lates recovered from persons with AIDS demonstrate a widerange of virulence during murine meningoencephalitis thatcorrelates with the expression of certain virulence factors.Microbiology 2006, 152:2247-2255.93. Zhu X, Williamson PR: A CLC-type chloride channel gene isrequired for laccase activity and virulence in Cryptococcusneoformans.  Mol Microbiol 2003, 50:1271-1281.94. Livak KJ, Schmittgen TD: Analysis of relative gene expression41.17http://genomebiology.com/2008/9/2/R41 Genome Biology 2008,     Volume 9, Issue 2, Article R41       Hu et al. Rdata using real-time quantitative PCR and the 2-ΔΔCTmethod.  Methods 2001, 25:402-408.95. Ferreira ID, do Rosario VE, Cravo PVL: Real-time quantitativePCR with SYBR Green I detection for estimating copy num-bers of nine drug resistance candidate genes in Plasmodiumfalciparum.  Malaria J 2006, 5:1.96. Yan Z, Li X, Xu J: Geographic distribution of mating type alle-les of Cryptococcus neoformans in four areas of the UnitedStates.  J Clin Microbiol 2002, 40:965-972.Genome Biology 2008, 9:R41


Citation Scheme:


Citations by CSL (citeproc-js)

Usage Statistics



Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            async >
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:


Related Items