Efficient Homologous and Illegitimate Recombination in the Opportunistic Yeast Pathogen Candida glabrata

Corresponding author: Brendan P. Cormack, Department of Molecular Biology and Genetics, Johns Hopkins Medical School, PCTB 522, 725 N. Wolfe St., Baltimore, MD 21205-2185., bcormack{at}jhmi.edu (E-mail)

Communicating editor: M. JOHNSTON

	ABSTRACT

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

The opportunistic pathogen Candida glabrata causes significant disease in humans. To develop genetic tools to investigate the pathogenicity of this organism, we have constructed ura3 and his3 auxotrophic strains by deleting the relevant coding regions in a C. glabrata clinical isolate. Linearized plasmids carrying a Saccharomyces cerevisiae URA3 gene efficiently transformed the ura3 auxotroph to prototrophy. Homologous recombination events were observed when the linearized plasmid carried short terminal regions homologous with the chromosome. In contrast, in the absence of any chromosomal homology, the plasmid integrated by illegitimate recombination into random sites in the genome. Sequence analysis of the target sites revealed that for the majority of illegitimate transformants there was no microhomology with the integration site. Approximately 0.25% of the insertions resulted in amino acid auxotrophy, suggesting that insertion was random at a gross level. Sequence analysis suggested that illegitimate recombination is nonrandom at the single-gene level and that the integrating plasmid has a preference for inserting into noncoding regions of the genome. Analysis of the relative numbers of homologous and illegitimate recombination events suggests that C. glabrata possesses efficient systems for both homologous and nonhomologous recombination.

CANDIDA species are the causative agents for both mucosal and systemic human infections. Significant progress has been made in our understanding of Candida virulence. Factors thought to contribute to virulence include the ability to grow in different morphological forms and at elevated temperature, the ability to switch between different colony/cellular phenotypes, the ability to adhere to host tissue, and the production and secretion of a variety of hydrolytic enzymes (ODDS 1994 ). The genetic analysis of these characteristics or of other virulence attributes has been complicated in C. albicans because it is diploid and asexual. Thus, in general, two copies of a particular gene of interest must be altered to observe the mutant phenotype. In contrast to C. albicans, C. glabrata, an important opportunistic pathogen responsible for 10–15% of symptomatic vaginal and systemic candidiasis (REDONDO-LOPEZ et al. 1990 ; PFALLER 1996 ), is haploid and therefore more amenable to genetic analysis of yeast pathogenesis. A number of molecular genetic tools exist already in C. glabrata. The URA3 gene has been cloned, and ura3 auxotrophs isolated (ZHOU et al. 1994 ). Disruption of the HIS3 and TRP1 coding regions with URA3 was used to generate his3 and trp1 auxotrophic strains (KITADA et al. 1995 ). C. glabrata autonomously replicating sequences (ARS) and centromere (CEN) sequences have been identified and can function to direct replication and segregation of episomal plasmids (MEHRA et al. 1992 ; KITADA et al. 1996 ).

Here we analyze the fate of transforming DNA in C. glabrata using a virulent clinical isolate (FIDEL et al. 1996 ). We describe unexpectedly high levels of nonhomologous recombination in C. glabrata and characterize the insertion site specificity for these nonhomologous events. We exploit this nonhomologous recombination as a method of insertional mutagenesis.

	MATERIALS AND METHODS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Strains:
The parental C. glabrata isolate (strain B, a gift of Paul Fidel and Jack Sobel) was a clinical isolate from a case of vaginitis that did not respond to fluconazole or boric acid treatment; it is virulent in a murine model of vaginitis (FIDEL et al. 1996 ). All derivative strains are summarized in Table 1. Media were prepared as described (SHERMAN et al. 1986 ). YP media contained glucose (2%); synthetic complete media was supplemented with casamino acids (0.6%), adenine (25 mg · liter^-1), and uracil (25 mg · liter^-1), or with all amino acids (100 mg · liter^-1) except those being tested for specific auxotrophies.

Transformation of C. glabrata:
Transformations were carried out using a variation of the modified LiAc protocol (GIETZ et al. 1992 ). Cells were grown in YPD to early log phase and collected by centrifugation. They were washed twice with water, and resuspended in 0.1 M LiAc, 1 mM EDTA, and 10 mM Tris, pH 7.5. Transforming DNA, was added to the cells with 100 µg of denatured salmon sperm DNA and 0.5 ml of 0.1 M LiAc, 40% PEG (3350), 1 mM EDTA, and 10 mM Tris, pH 7.5 (1 µg). The mix was incubated at 30° for 30 min and heat-shocked at 45° for 15 min.

URA3 deletion construct:
The neomycin-resistance gene (neo) of Tn903 was amplified by PCR using primers Tn903-1 and Tn903-2 (sequences shown in Table 2). The Tn903-1 primer incorporates 55 nucleotides upstream of the ATG of the CUP1 gene of C. glabrata. This promoter region is sufficient to transcribe the neo gene at levels sufficient to give resistance to 500 µg G418/ml; the parental strain BG2 is sensitive to G418 at 50 µg/ml. We cloned a 2.2-kb PstI fragment containing the C. glabrata URA3 locus into Puc19 to generate pBC34.1. pBC34.1 was digested with XhoI and HindIII and ligated to the SalI-BamHI PCR fragment containing the neo gene and a BamHI-HindIII fragment containing the 3' region of the URA3 locus, which was generated by PCR amplification with the primers URA3-940 and URA3PST (Table 2). This PCR fragment was sequenced to verify that there were no mutations introduced by PCR amplification. The final construct (pBC39.1) consists of the URA3 gene carried on a PstI fragment in which the coding region of the URA3 gene, from the XhoI site at -85, with respect to the ATG, to the SspI site at 932 has been replaced with the neo gene. To generate the ura3 auxotroph, BG14, strain BG2 was transformed with PstI-digested pBC39.1. The PstI fragment of pBC39.1 contained the 5' and 3' UTR of URA3 flanking the neo gene of Tn903. G418-resistant and 5-fluoroorotic acid (5-FOA)-sensitive recombinants were isolated. Southern analysis showed that the URA3 locus had been replaced by the deleted gene containing the neomycin resistance cassette (data not shown).

HIS3 deletion construct:
We cloned the HIS3 locus as a SphI-BglII fragment (2.5 kb) into Puc19 to generate pBC102.1. The sequence of this 2.5-kb region is available under GenBank accession number AF107116. Primer HIS3PST1, homologous to the 3' end of the His3 coding region, was used to amplify a PstI-BglII fragment and primer HIS3PST2, homologous to the 5' end of the HIS3 coding region, was used to amplify a SphI-PstI fragment. The primer sequences are in Table 2. These fragments were combined with SphI-BamHI-digested YIplac211 (GIETZ and SUGINO 1988 ) in a ligation to generate pBC104. Yiplac211 carries the Saccharomyces cerevisiae URA3 gene that functions in C. glabrata. pBC104 plasmid carries the 5'- and 3'-untranslated regions of HIS3, but is deleted for the entire coding region (nucleotide 1 to 631, with respect to the "A" of the first methionine codon). The flanking region was sequenced to verify that there were no mutations introduced by PCR. PBC104 was linearized by digestion with NotI, for which there is a site in the 3' UTR of the HIS3 gene, and used to transform BG14. Ura⁺ prototrophs were colony purified and then plated on SC plates containing 5-fluoroorotic acid (5FOA; 1.1 g · liter^-1). FOA^R segregants were tested for growth in the absence of histidine. Approximately half of the segregants were His^- and half His⁺, as expected. The structure of the deleted locus for two isolates, BG87b and BG88b, was verified by Southern blot analysis (data not shown). Transformation of BG87b and BG88b with the 2.2-kb fragment of the C. glabrata URA3 gene restored the URA3 locus and generated, respectively, the his3 auxotrophs BG98 and BG99.

Sequence of illegitimate integration sites for YIplac211:
The DNA flanking an insertion site was rescued by digestion of genomic DNA from the transformant with EcoRI or HindIII for which there are no recognition sites in the integrating plasmid. Ligation of this digestion mixture generates a plasmid carrying the genomic DNA flanking the insertion site. Plasmid was recovered by transformation into Escherichia coli strain DH12 (GIBCO/BRL, Rockville, MD). For those events in which two plasmids had integrated in tandem, the ligation mix generated a dimeric plasmid in which there were two copies of the integrating vector as well as the genomic sequences flanking the integration site. In these cases, the rescued plasmid was transformed into the Rec⁺ strain MC1061. The dimeric plasmid resolved into the two component plasmids: a vector-sized plasmid and a vector carrying the chromosomal DNA flanking the original genomic integration site. The flanking genomic DNA in all rescued plasmids was sequenced using primers that hybridize to the vector. The flanking DNA was sequenced on one strand only, but each integrant was sequenced twice from independently prepared templates. The overall sequence quality was excellent, making it unlikely that the lack of ORFs was an artifact of frameshifting because of poor sequence quality. The DNA sequence was used in homology searches using the BLASTX and BLASTN algorithms. For all but two sequences, there were no significant homologies in GenBank.

We obtained the sequence of flanking DNA for 37 integrants. For 17 of these, we obtained the sequence of the undisrupted genomic target site. Of these 17, 14 were insertions in a single 1.4-kb locus. For these, the sequence of the flanking DNAs for each of the insertions was assembled to give the sequence of the entire locus, including the integration sites for each insertion. For inserts 15, 16, and 17 (Figure 4) the sequence of the flanking DNA was used to synthesize primers, which were subsequently used to amplify (via PCR) the integration site from genomic DNA of the parental wild-type strain. The primers used for amplification of the insertion sites are shown in Table 2. Thus, for each insertion, we obtained the sequence of both junctions for the integrant into genomic DNA and for the undisrupted integration site.

View larger version (15K): In this window In a new window Download PPT slide   Figure 1. Structure of disrupted loci. (a) Structure of disrupted URA3 locus. The 2.2-kb PstI fragment is shown. The URA3 coding region is shown as a crosshatched box. The neo cassette used to disrupt the URA3 gene is shown. The region of the URA3 locus replaced by the neo cassette is indicated by connecting lines. The coding region for the neo gene is preceded by 54 bp of promoter from the C. glabrata CUP1 gene (indicated by a filled box). (b) Structure of disrupted HIS3 locus. The 2.8-kb SphI-BglII fragment is shown with the HIS3 and DED1 ORFs indicated by hatched boxes. The region deleted in his3 strains is indicated.

View larger version (68K): In this window In a new window Download PPT slide   Figure 2. Fate of transformed YIplac211 in BG14. (a) Genomic DNAs from 12 transformants were prepared and digested with PstI, which cleaves once in the integrating plasmid. Lane 1, YIplac211 digested with PstI. Lanes 2–13, DNA from transformants digested with PstI. The Southern blot was probed with [32P]dCTP-labeled YIPlac 211. (b) Auxotrophs isolated among illegitimate integrants of YIPlac 211 in BG14. A total of 4000 transformants were tested by replica plating onto media lacking a given amino acid. Ten transformants were unable to grow without supplementation of the indicated amino acids.

View larger version (6K): In this window In a new window Download PPT slide   Figure 3. Physical map of 14 insertions at the EPA1 locus. Insertion sites are marked with vertical hatch marks. The open reading frame is shown as a hatched box. The insertions correspond to sequences 1–14 of Figure 4. The extent and position of the deleted fragment in transformant 10 is shown by a bar. The position and size of the insertion in transformant 12 is shown by the small triangle. A "T" marks a consensus TATA box.

View larger version (41K): In this window In a new window Download PPT slide   Figure 4. Flanking sequences and target sites for BamHI- and XbaI-digested YIplac211 transformed into BY14. Each target sequence is shown as the center double-stranded DNA. Above and below the target sequence is the sequence of the digested ends of YIplac-211. The flanking sequences are aligned so that the single strand of the integrating end is immediately adjacent to the first base unambiguously derived from the target sequence. Underlined are nucleotides in the overhang of the integrating end that are homologous to the target sequence. The SED (single end directed), NED (neither end directed), or BED (both ends directed) classification is based how many ends of the integrating plasmid have microhomology with the target site. For the sake of consistency and comparison with the work in S. cerevisiae of SCHIESTL et al. 1994 , we score as microhomologies those events involving at least 2 bp immediately adjacent to the recombination point. In bold are sequences immediately adjacent to the site of insertion that match the consensus for topoisomerase recognition [(G/C)(A/T)T]. In capitals are nucleotides present in the junction sequence that are not templated by target or plasmid, and which were presumably added during the recombination reaction. For recombinant 10, two target sites are shown because in this recombinant, the intervening 98 nucleotides between the two target sites were deleted. For recombinant 12, only one of the ends shown is plasmid derived, because there was a 125-bp fragment of unknown origin inserted in between the target and the second plasmid end.

Construction of Tn903 directed integration vectors:
Fragments of the neo gene of Tn903 were amplified by PCR and cloned as EcoRI-BamHI fragments into a version of YIPLAC-211 (GIETZ and SUGINO 1988 ) from which the HindIII site had previously been deleted by incubation with dNTPs and the Klenow fragment of E. coli polymerase I and religation. The primers include EcoRI sites and BamHI sites and were chosen so that the natural SmaI or HindIII sites of the neo gene would be in the exact middle of the amplified fragment. The primers (A-GTn903 xxx) are shown in Table 2. The numbers refer to the nucleotide position of the neomycin resistance gene with respect to the ATG. This nucleotide is at the 5' end of the PCR oligo and, after PCR amplification, is at the end of the generated fragment. Primers with the same letter designation were used in one PCR amplification. The size of the PCR fragments for a given primer pair can be obtained by subtracting the two nucleotide number designations. Of the seven resulting plasmids, then, three contain inserts that flank the HindIII site, and four contain inserts flanking the SmaI site.

	RESULTS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Construction of auxotrophic strains of C. glabrata:
To generate a ura3 auxotroph of C. glabrata, a wild-type clinical isolate of C. glabrata [strain BG2, isolate B (FIDEL et al. 1996 )] was transformed with a linear DNA fragment consisting of the neo gene of Tn903 flanked by 5' and 3' URA3 sequences. In four isolates (strains BG12, BG13, BG14, BG15), the URA3 locus had been replaced by the the neomycin resistance cassette (data not shown). To generate a his3 auxotroph of C. glabrata, the HIS3 coding region was precisely deleted in a standard two-step disruption (STRUHL 1983 ; details of strain construction in MATERIALS AND METHODS). The structures of the HIS3 and URA3 loci in these strains (Figure 1A) were verified by Southern analysis (data not shown).

Nonhomologous recombination in C. glabrata:
Consistent with what has been previously reported (ZHOU et al. 1994 ), we found that a S. cerevisiae CEN URA3 vector (pYCplac33, GIETZ and SUGINO 1988 ) transforms a C. glabrata ura3 mutant strain (BG14) at high efficiency (2–5 x 10⁵/µg) and replicates as a circular plasmid in these transformants (data not shown).

To examine efficiencies of illegitimate integative transformation, the ura3 strain BG14 was transformed with the S. cerevisiae URA3 integrating vector YIplac211, which has no homology with the C. glabrata genome. Before transformation, the vector was linearized with BamHI or XbaI, enzymes that recognize polylinker sites in the plasmid and do not cleave within the URA3 gene. Transformation by illegitimate recombination occurred with a frequency of 5 x 10²/µg, compared to 1–5 x 10⁰/µg seen in S. cerevisiae (SCHIESTL et al. 1993 ). Genomic DNA from transformants was analyzed by Southern blot analysis probing with the S. cerevisiae URA3 gene (Figure 2a). The sites of integration showed no obvious bias for a particular locus, and for most transformants only a single copy of the transforming vector had integrated. Of 59 total transformants analyzed, 45 (76%) were simple insertions, 9 (15%) were tandem insertions of two copies of the plasmid, and 4 (7%) were nontandem integrations of two copies of the plasmid. Ten of 4000 transformants (0.25%) were unable to grow on minimal media supplemented with only adenine and uracil. This frequency of amino acid auxotrophy is close to the predicted frequency (1–2%) for random integration if we assume 100 genes capable of mutating to auxotrophy. Several different auxotrophies were found (Figure 2b), again suggesting a fairly random distribution of the integrated plasmid in the genome.

To examine the target site specificity, we rescued the integrated plasmid from 37 integrants and determined the flanking DNA sequence. The 37 integrants sequenced included 2 colony morphology mutants, 6 auxotrophs, and 13 mutants with no apparent phenotype; in addition, we had identified 16 insertions that had lost the ability to adhere to eukaryotic cells (B. P. CORMACK and S. FALKOW, unpublished results). Only 2 of the insertions (both with no apparent phenotype) were in coding regions; the other 35 mutants apparently inserted in noncoding regions because only small ORFs of less than 250 nucleotides were present at the insertion site. None of these small ORFs had homology to genes in the S. cerevisiae database, on the basis of BLASTN or BLASTX homology searches (see MATERIALS AND METHODS). The loci targeted in the mutants with adherence phenotypes were characterized in detail. Of the 16 mutants, 14 were insertions in the same 1.4-kb locus [the Epithelial Adhesin 1 (EPA1) gene] but were all independent insertion events because the precise insertion points were not the same. A physical map of the insertions is shown in Figure 3. The majority (13/14) of insertions form one clustered group ~1.2 kb upstream of the start site of translation. To assess whether there was any microhomology between the ends of the integrating plasmids and the genomic target site, we determined the DNA sequence of the undisrupted genomic target site in a total of 17 of the insertions. Of this total, 14 (1–14; Figure 4) were insertions at the EPA1 locus; 3 others were insertions with no phenotype (15–17; Figure 4). For nine (53%) of the integration events, there was no homology between either end of the integrating plasmid and the site of integration. For eight (47%) of the integrations, there was microhomology between one end of the integrating plasmid and the target site.

Similar to S. cerevisiae (SCHIESTL et al. 1993 ), a small number of illegitimate events were not simple insertions. In one case (recombinant 10; Figure 4) the integration generated a target site deletion of 98 bp. In another (recombinant 12; Figure 4), there was an insertion of 125 bases of nonhomologous DNA at the locus immediately adjacent to the plasmid insertion. This inserted sequence had no significant homology to sequences in the database. In an additional four events (recombinants 2, 8, 10, and 14; Figure 4) a nonencoded A or T was added to the integrating end of the plasmid as part of the overall process of integration.

To show that the phenotype of a given strain was the result of the inserted plasmid, we chose one of the lysine auxotrophs, one of the mutants with an altered colony morphology, and all the mutants with a strong adherence phenotype and reintroduced the YIplac211 plasmid into the same locus in the parental strain BG14. To do this, we used the rescued plasmid (YIplac211 carrying the flanking genomic DNA for the insertion site) for each strain to transform the parental strain to Ura⁺, which disrupts the gene by homologous recombination between the flanking DNA in the vector and the chromosomal DNA. Of such transformants tested, ~90% had the same phenotype as the original mutants, showing that the phenotype in these strains was indeed linked to the insertion of the plasmid. The remaining 10% are presumably due to the insertion of the plasmid by nonhomologous recombination, which is 10-fold less efficient than homologous recombination (see below).

Homologous and nonhomologous transformation:
To characterize the sequence requirements for homologous and nonhomologous recombination in C. glabrata, a set of seven integrating plasmids was made carrying DNA from the Tn903 neo gene that we had used to disrupt the URA3 locus to generate the strain BG14. Plasmids integrating by homologous recombination disrupted the neo gene rendering the strain sensitive to G418; on the other hand, plasmids integrating by nonhomologous recombination left the strain resistant to G418. The neo fragments ranged in size from 50–400 bp and were chosen so that in the middle of each fragment was either the naturally occurring SmaI site or the naturally occurring HindIII site. Each of these plasmids was linearized with HindIII or SmaI and used to transform the strain BG14. A simple comparison of the ratio of G418-resistant and G418-sensitive integrants for each plasmid gave the frequency of homologous vs. nonhomologous integration. The frequency of total, homologous, and nonhomologous integrants for each of the plasmids is shown in Figure 5. With <50 bases of homology, the frequency of homologous recombination is very low. With >200 bases of homology, total transformation efficiency increases dramatically and >90% of the integrants are homologous recombinants. We confirmed by PCR analysis (data not shown) that 20/20 neomycin-sensitive integrants were indeed integrants at the ura3::neoR locus, while 20/20 neomycin-resistant integrants were not integrants at the ura3::neoR locus.

View larger version (14K): In this window In a new window Download PPT slide   Figure 5. Influence of amount of homology on rates of homologous and illegitimate recombination. (A) Number of homologous recombination events expressed as a percentage of total transformants plotted against the exact size (in base pairs) of the insert carried on the integrating plasmid. For details on insert size, see MATERIALS AND METHODS. HindIII and SmaI refer to the enzyme used to linearize the plasmid in the precise middle of the neo gene fragment (see MATERIALS AND METHODS). (B) Number of transformants (, illegitimate recombinants; , homologous recombinants) per microgram of integrating plasmid plotted against size (in base pairs) of the insert carried on the integrating plasmid.

	DISCUSSION

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

C. glabrata is a significant cause of human disease in spite of the fact that it is phylogenetically closely related to the nonpathogenic S. cerevisiae. In this investigation, we used standard one- and two-step disruptions to generate ura3 and his3 auxotrophs in which the coding regions are deleted. These ura3 and his3 strains of C. glabrata should be useful in molecular-genetic analysis of multiple aspects of C. glabrata virulence, including generating prototrophic hybrids for use in parasexual analysis of C. glabrata after spheroplast fusion (WHELAN and KWON-CHUNG 1987 ).

C. glabrata can be transformed with circular plasmids carrying ARS elements from S. cerevisiae (ZHOU et al. 1994 ) or from C. glabrata itself (MEHRA et al. 1992 ); these plasmids replicate as unstable, high-copy extrachromosomal elements. Transformation of C. glabrata is extremely efficient: we found here that linearized plasmids not carrying an ARS element transformed C. glabrata at high efficiency, integrating by homologous recombination when the plasmid carried as little as 100 bp of DNA homologous with the genome. Plasmids carrying no C. glabrata genomic DNA integrated by nonhomologous recombination at a frequency ~10% of that seen for homologous integration. Thus, C. glabrata possesses efficient systems for homologous and nonhomologous recombination.

In some respects, illegitimate recombination in C. glabrata is similar to that in S. cerevisiae. Integration is relatively random in overall distribution in the genome, as suggested by Southern analysis, by the auxotroph frequencies among illegitimate recombinants, and by the sequence of target sites. No consensus site for integration was apparent with the possible exception of a slight preference for C (18/34) and T (18/34) as the first and second nucleotides 3' to the integration site, respectively (Figure 4).

Three lines of evidence suggest that illegitimate recombination in C. glabrata differs significantly from that in S. cerevisiae. First, our sequence of the target sites showed illegitimate recombination in C. glabrata has only a slight dependence on microhomology between plasmid ends and target site. For example, none of the illegitimate recombination events in C. glabrata had homology between both ends of the plasmid and the target DNA, compared with 43% (10/23) in S. cerevisiae; 47% (8/17) of C. glabrata events had microhomology between one end and the target site compared to 74% (17/23) in S. cerevisiae (SCHIESTL and PETES 1991 ; SCHIESTL et al. 1993 , SCHIESTL et al. 1994 ; ZHU and SCHIESTL 1996 ; Table 3). Indeed, in C. glabrata, the average number of homologous bases between the overhanging four bases of the integrating end and the target site was only 1.3, close to what is expected (1.0) on a purely random basis.

The second major difference with illegitimate recombination in S. cerevisiae is the role of topoisomerase I. In S. cerevisiae, toposiomerase I has been implicated in illegitimate recombination by several independent lines of evidence (ZHU and SCHIESTL 1996 ), including a higher-than-expected proportion of integration events immediately adjacent to consensus sites for topoisomerase I cleavage [consensus (C/G) (A/T) (T)]. By contrast, in C. glabrata, this bias for integration at Top1p sites was not observed. We do not know the consensus sequence for Top1p recognition in C. glabrata. However, (C/G) (A/T) (T) is the consensus site for the eukaryotic topoisomerase I in species ranging from S. cerevisiae (ZHU and SCHIESTL 1996 ) and Tetrahymena pyriformis (ANDERSEN et al. 1985 ) to rats (BEEN et al. 1984 ). It is therefore likely that the Top1p site in C. glabrata falls within this consensus. For the 17 events analyzed, only 5 (15% of target sites) had adjacent (C/G) (A/T) (T) sites (Table 3). This is only marginally higher than the number expected on a random basis (~2/17 target sites). In S. cerevisiae CTT is the primary site for Top1p action in mediating illegitimate recombination, because, in cells overexpressing Top1p, 12/16 consensus sequences adjacent to integration sites are CTT (ZHU and SCHIESTL 1996 ). In C. glabrata, by contrast, the 5 Top1p sites adjacent to insertion sites are distributed between all 4 possible sites (GTT, CTT, GAT, and CAT; Figure 4). While we cannot rule out that the presence of these 5 Top1p consensus sites reflects a modest role for TOP1 in illegitimate recombination in C. glabrata, the representation of all 4 possible sites suggests that they are present fortuitously.

The third and most important difference with illegitimate recombination in S. cerevisiae is that illegitimate events in C. glabrata were targeted to noncoding regions. The ratio of coding and noncoding sequences in C. glabrata is probably similar to that found in S. cerevisiae because the genome size of C. glabrata is very similar to that of S. cerevisiae (MALESZKA and CLARK-WALKER 1993 ). Moreover, for at least two loci, HIS3–DED1 (Figure 1) and CUP2-YGL164c, the intergenic regions are approximately the same size and gene organization is conserved. Strikingly, for 37 integration events analyzed, all but 2 (which were in open reading frames homologous to S. cerevisiae genes YIL049 and YPL224C) were targeted to noncoding sequence. The 35 insertions in noncoding regions are likely in promoter regions. We have detailed information about three loci in which insertions afffected adherence to epithelial cells. The 14 insertions at the EPA1 locus are between 0.3 and 1.75 kb upstream of the translational start (Figure 3). Strikingly, in screening for such insertion mutants, we did not isolate a single insertion in the coding region of the EPA1 gene, even though such an insertion would result in a nonadherent phenotype (B. P. CORMACK and S. FALKOW, unpublished results). This emphasizes the bias of illegitimate recombination to noncoding regions. We also sequenced 2.5 kb of two other loci in which insertions had strong adherence phenotypes. In one case, the insertion was 250 bases upstream of a long (>1 kb) ORF. In the other case, the insertion was 0.5 kb upstream of a 500-bp ORF.

The bias of illegitimate recombination to noncoding regions in C. glabrata could result from several different mechanisms: (1) the recombination complex might be recruited to the site of assembly of interaction with specific host factors; (2) the targeted intergenic regions might be generally more accessible than coding regions of the genome. These same mechanisms have been proposed to explain targeting of members of the Ty retrotransposon family to regions upstream of RNA polymerase III genes or to regions of silenced chromatin (CHALKER and SANDMEYER 1992 ; JI et al. 1993 ; ZOU and VOYTAS 1997 ; KIM et al. 1998 ). An analysis of illegitimate recombination in C. glabrata might reveal aspects of chromosome structure important in targeting the recombination machinery. Moreover, because insertions are exclusively or nearly exclusively targeted to intergenic regions, mutagenesis by illegitimate recombination may be a useful method of altering transcriptional regulation of endogenous genes.

Insertion by illegitimate recombination appears to be a useful method of general mutagenesis. The sequence of the integration sites and the variety of auxotrophic mutants that we recovered suggest a fairly random genomic distribution of insertions. Moreover, we were able to isolate insertions upstream of EPA1, a gene important in adherence to epithelial cells (B. P. CORMACK and S. FALKOW, unpublished results), on the basis of phenotype caused by the insertion. We demonstrated that, for strains showing a phenotype, the rescued plasmid/flanking DNA vector could be used to reintroduce the insertion at the original locus and regenerate a strain with the same phenotype. Because C. glabrata is asexual, putative mutants cannot be analyzed to see whether the insertion and the phenotype cosegregate in the spores resulting from a sexual cross. However, reintroduction of the plasmid into the same locus of the parental strain gives the same information and should be useful as a rapid second screen in C. glabrata mutant hunts.

	ACKNOWLEDGMENTS

We thank Dennis Thiele for the C. glabrata URA3 gene and Paul Fidel and Jack Sobel for providing the original clinical C. glabrata isolate. We thank Jeff Corden, Carol Greider, Jef Boeke, and Lalita Ramakrishnan for comments on the manuscript. This work was supported by a grant to S.F. from SmithKline Beecham. B.P.C. was supported by a postdoctoral fellowship from the Helen Hay Whitney Foundation.

Manuscript received September 21, 1998; Accepted for publication December 7, 1998.

	LITERATURE CITED

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

ANDERSEN, A. H., E. GOCKE, B. J. BONVEN, O. F. NIELSEN, and O. WESTERGAARD, 1985  Topoisomerase I has a strong binding preference for a conserved hexadecameric sequence in the promoter region of the rRNA gene from Tetrahymena pyriformis. Nucleic Acids Res. 13:1543-1557[Abstract/Free Full Text].

BEEN, M. D., R. R. BURGESS, and J. J. CHAMPOUX, 1984  Nucleotide sequence preference at rat liver and wheat germ type 1 DNA topoisomerase breakage sites in duplex SV40 DNA. Nucleic Acids Res. 12:3097-3114[Abstract/Free Full Text].

CHALKER, D. L. and S. B. SANDMEYER, 1992  Ty3 integrates within the region of RNA polymerase III transcription initiation. Genes Dev. 6:117-128[Abstract/Free Full Text].

FIDEL, P. L., JR., J. L. CUTRIGHT, L. TAIT, and J. D. SOBEL, 1996  A murine model of Candida glabrata vaginitis. J. Infect. Dis. 173:425-431[Medline].

GIETZ, R. D. and A. SUGINO, 1988  New yeast-Escherichia coli shuttle vectors constructed with in vitro mutagenized yeast genes lacking six-base pair restriction sites. Gene 74:527-534[Medline].

GIETZ, D., A. ST, R. A. JEAN, R. H. WOODS AND, and R. H. WOODS ANDSCHIESTL, 1992  Improved method for high efficiency transformation of intact yeast cells. Nucleic Acids Res. 20:1425[Free Full Text].

JI, H., D. P. MOORE, M. A. BLOMBERG, L. T. BRAITERMAN, and D. F. VOYTAS et al., 1993  Hotspots for unselected Ty1 transposition events on yeast chromosome III are near tRNA genes and LTR sequences. Cell 73:1007-1018[Medline].

KIM, J. M., S. VANGURI, J. D. BOEKE, A. GABRIEL, and D. F. VOYTAS, 1998  Transposable elements and genome organization: a comprehensive survey of retrotransposons revealed by the complete Saccharomyces cerevisiae genome sequence. Genome Res. 8:464-478[Abstract/Free Full Text].

KITADA, K., E. YAMAGUCHI, and M. ARISAWA, 1995  Cloning of the Candida glabrata TRP1 and HIS3 genes, and construction of their disruptant strains by sequential integrative transformation. Gene 165:203-206[Medline].

KITADA, K., E. YAMAGUCHI, and M. ARISAWA, 1996  Isolation of a Candida glabrata centromere and its use in construction of plasmid vectors. Gene 175:105-108[Medline].

MALESZKA, R. and G. D. CLARK-WALKER, 1993  Yeasts have a four-fold variation in ribosomal DNA copy number. Yeast 9:53-58[Medline].

MEHRA, R. K., J. L. THORVALDSEN, I. G. MACREADIE, and D. R. WINGE, 1992  Cloning system for Candida glabrata using elements from the metallothionein-IIa-encoding gene that confer autonomous replication. Gene 113:119-124[Medline].

ODDS, F. C., 1994  Pathogenesis of Candida infections. J. Am. Acad. Dermatol. 31:S2-5[Medline].

PFALLER, M. A., 1996  Nosocomial candidiasis: emerging species, reservoirs, and modes of transmission. Clin. Infect. Dis. 22(Suppl. 2):S89-94.

REDONDO-LOPEZ, V., M. LYNCH, C. SCHMITT, R. COOK, and J. D. SOBEL, 1990  Torulopsis glabrata vaginitis: clinical aspects and susceptibility to antifungal agents. Obstet. Gynecol. 76:651-655[Abstract/Free Full Text].

SCHIESTL, R. H. and T. D. PETES, 1991  Integration of DNA fragments by illegitimate recombination in Saccharomyces cerevisiae. Proc. Natl. Acad. Sci. USA 88:7585-7589[Abstract/Free Full Text].

SCHIESTL, R. H., M. DOMINSKA, and T. D. PETES, 1993  Transformation of Saccharomyces cerevisiae with nonhomologous DNA: illegitimate integration of transforming DNA into yeast chromosomes and in vivo ligation of transforming DNA to mitochondrial DNA sequences. Mol. Cell. Biol. 13:2697-2705[Abstract/Free Full Text].

SCHIESTL, R. H., J. ZHU, and T. D. PETES, 1994  Effect of mutations in genes affecting homologous recombination on restriction enzyme-mediated and illegitimate recombination in Saccharomyces cerevisiae. Mol. Cell. Biol. 14:4493-4500[Abstract/Free Full Text].

SHERMAN, F., G. R. FINK and J. B. HICKS, 1986 Methods in Yeast Genetics. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY.

STRUHL, K., 1983  The new yeast genetics. Nature 305:391-397[Medline].

WHELAN, W. L. and K. J. KWON-CHUNG, 1987  Parasexual genetics of Torulopsis glabrata.. J. Bacteriol. 169:4991-4994[Abstract/Free Full Text].

ZHOU, P., M. S. SZCZYPKA, R. YOUNG, and D. J. THIELE, 1994  A system for gene cloning and manipulation in the yeast Candida glabrata. Gene 142:135-140[Medline].

ZHU, J. and R. H. SCHIESTL, 1996  Topoisomerase I involvement in illegitimate recombination in Saccharomyces cerevisiae. Mol. Cell. Biol. 16:1805-1812[Abstract].

ZOU, S. and D. F. VOYTAS, 1997  Silent chromatin determines target preference of the Saccharomyces retrotransposon Ty5. Proc. Natl. Acad. Sci. USA 94:7412-7416[Abstract/Free Full Text].