Multilocus Sequence Typing of the Pathogenic Fungus Aspergillus fumigatus

The 100 isolates of A. fumigatus used in this study were sourced from both clinical and environmental settings (Table 1). They were taken from our own stock collection and provided by the kindness of collaborators. Each isolate originated from a separate source except for IHEM20373, which was obtained from a culture collection but which was found, after it had been typed, to be from the same source as J960330. The isolate originally received as TX2684 was found to consist of colonies of opposite mating types, which were typed separately by MLST as TX2684A and TX2684B. For four isolates, no information concerning the origins was known. Among the remaining 96, 46 came from the United Kingdom, 19 from the United States, 17 from Australia, 8 from France, and 2 each from Belgium, Canada, and The Netherlands (Table 1). Fifty-six isolates had been originally cultured since 2000, and 36 were cultured in the 1990s, 1 in the 1980s, and 2 in the 1970s. Eighteen isolates had previously been typed by PCR of genes encoding the intergenic spacer regions of the large and small ribosomal subunits (41). The fungi were stored at −80°C as spore suspensions in YPG-glycerol (1:1) (1% yeast extract [Difco, Sparks, MD], 2% Bacto peptone [Difco], 2% glucose [Fisher, Loughborough, United Kingdom], 50% glycerol [Fisher]) and then streaked onto Sabouraud agar and grown at 35°C until conidia formed.

The identity of the isolates as A. fumigatus and not a closely related species, such as Aspergillus lentulus or Aspergillus udagawae, was confirmed by PCR and StyI restriction digestion as described elsewhere (3).

Mycelia were grown from conidial inocula in 20 ml of liquid YPG with shaking at 200 rpm at 30°C for 3 days. The mycelia were strained to remove the supernatant, and then 0.3-g glass beads (0.45- to 0.52-mm diameters; Sigma, St. Louis, MO), 300 μl extraction buffer (100 mM Tris-HCl, pH 8, 100 mM NaCl, 2% Triton X-100, 1% sodium dodecyl sulfate, 1 mM EDTA), and 300 μl phenol-chloroform (1:1) were added. The mycelia were disrupted by vortexing for 30 min. To the lysate, 200 μl TE (10 mM Tris-HCl, pH 8, 1 mM EDTA) was added, and insoluble debris was removed in a bench-top centrifuge run at full speed for 5 min. The DNA in the aqueous phase was extracted a second time by adding an equal volume of phenol-chloroform (1:1), vortexing it for 30 s, and repeating the centrifugation step. Nucleic acid from the aqueous phase was precipitated with 1 ml ethanol and pelleted as before. The pellet was resuspended in TE containing 250 μg ml⁻¹ RNase (Sigma) and incubated for 30 min at 37°C. DNA was precipitated with 1 ml isopropanol and 10 μl 3 M sodium acetate. Following centrifugation for 10 min, DNA pellets were dried and resuspended in 50 μl water.

MAT1-1 and MAT1-2 sexual idiomorphs were differentiated by PCR as described previously (37). Briefly, a multiplex PCR was performed with reaction volumes of 25 μl containing 1 μl of genomic DNA as the template, 5 μl 5× GoTaq buffer (Promega, Southampton, United Kingdom), 2 mM MgCl₂ (Promega), 200 μM deoxynucleoside triphosphate mix (Invitrogen, Paisley, United Kingdom), and 2.5 units GoTaq DNA polymerase (Promega). The reaction mixture contained 0.64 μM of reverse primer AFM3 (5′-CGGAAATCTGATGTCGCCACG-3′), which is common to both idiomorphs, and 0.32 μM each of forward primer AFM1, which is specific to MAT-1 (5′-CCTTGACGCGATGGGGTGG-3′), and forward primer AFM2, which is specific to MAT-2 (5′-CGCTCCTCATCAGAACAACTCG-3′). PCR was performed with a thermocycler (TC-412; Techne, Cambridge, United Kingdom) programmed as follows: 95°C for 5 min; 35 cycles of 95°C for 30 s, 60°C for 30 s, and 72°C for 1 min; and a final elongation step at 72°C for 5 min.

A selection of 27 A. fumigatus genes from 0.7 to 3.3 kb was sequenced across a set of 12 isolates chosen to maximize the diversities of geographical sources, anatomical properties, and dates of isolation and therefore increase the likelihood of representation of diverse genotypes. Primers for PCR amplification and sequencing of the seven fragments were designed from sequences taken from GenBank (http://www.ncbi.nlm.nih.gov/ ) and A. fumigatus GeneDB (http://www.genedb.org/genedb/asp/ ). Results from these pilot sequences were used to determine a final panel of seven gene fragments that gave the highest discrimination for MLST analysis ( Table 2). In accordance with feasible automated-sequencing run length, fragments of up to 590 bp containing the most discriminatory single-nucleotide polymorphisms (SNPs) were identified.

Gene fragments were amplified by PCRs in 96-well microdilution plates to amplify gene fragments of <1 kb with the primers listed in Table 2. Reaction volumes of 50 μl contained 2 μl of genomic DNA as a template, 10 μl 5× Green GoTaq buffer (Promega), 1.5 mM MgCl₂ (Promega), 100 μM deoxynucleoside triphosphate mix (Invitrogen), 0.2 μM each of forward and reverse primers (Invitrogen), and 5 U GoTaq DNA polymerase (Promega). PCR was performed with a TC-412 thermocycler (Techne, Cambridge, United Kingdom), with a cycle program of 94°C for 5 min; 35 cycles of 94°C for 1 min, 50°C for 1 min, and 68°C for 1 min; and a final elongation step at 68°C for 10 min. Amplified DNA fragments were purified by combining PCR mixtures with 60 μl of 20% polyethylene glycol 8000 (Sigma) and 2.5 M NaCl solution for 30 min at room temperature and then centrifuging 96-well plates at 2,254 × g for 60 min. The plates were inverted onto blotting paper to remove the supernatant. The DNA pellets were washed with 150 μl 70% ethanol and centrifuged for 10 min, and the supernatant was removed by blotting. Plates inverted on tissue paper were centrifuged for 1 min at 500 × g to remove residual ethanol. DNA was resuspended in 50 μl distilled H₂O for use as template DNA for subsequent sequencing reactions. Reaction volumes of 5 μl were set up in 96-well plates and contained 1 μl of purified gene fragment, 2 μl of 0.67 μM forward primer, 0.25 μl Big Dye (Applied Biosystems, Warrington, United Kingdom), 0.875 μl 5× buffer (Applied Biosystems), and 0.875 μl distilled H₂O. Reaction mixtures were prepared as described above with the reverse primer. Sequencing PCR was performed with 25 cycles of 96°C for 10 s, 52°C for 5 s, and 60°C for 2 min. DNA was purified by addition of 50 μl of a 250:1 solution of ethanol-sodium acetate and incubation at room temperature for 45 min. DNA was pelleted and washed in 96-well plates as described above. Automated DNA sequencing was performed at the Department of Zoology, University of Oxford, United Kingdom. Forward and reverse DNA sequence chromatograms were analyzed with DNASTAR software to identify interstrain SNPs.

MLST sequence data were analyzed to determine relationships between the 100 strains. Concatenated SNPs for all MLST alleles of each strain were input with MEGA 3.0 software (26) to generate an unweighted-pair group method with arithmetic averages (UPGMA) dendrogram based on p-distance with 1,000 bootstrap replications. eBURST version 3 (http://eburst.mlst.net/ ) was used to ascertain clonal cluster relationships between isolates on the basis of single allele differences, regardless of numbers of SNP differences between alleles.

A total of 27 gene fragments was sequenced for a set of 12 isolates, expected to be genetically diverse, to determine an optimal panel of gene fragments for MLST. Initially, housekeeping genes were considered potential MLST fragment candidates. These are defined as genes for which the ratio of nonsynonymous/synonymous amino acid changes resulting from SNPs is <1. Traditionally, only such genes were used in bacterial schemes when MLST was first devised (32). A. fumigatus housekeeping genes showed such low interstrain sequence variability that we eventually included nonhousekeeping genes to obtain a higher frequency of interisolate SNPs. The 20 genes that were considered but not chosen for the final selection were AREA, CALN, CSN, CPCA, CRNA, FACC, HIS3, HSP70, LEU2, MAN70, MANA, NIAD, NIIA, PEP, PREA, PREB, SOK, STEA, STUA, and TRPC.

Seven MLST fragments that represented the minimum set needed to differentiate the panel of 12 isolates were selected (Table 2). This MLST scheme was used to type 100 A. fumigatus isolates, including both clinical and environmental specimens, from our culture collection (Table 1). The polymorphic sites in the MLST fragments for A. fumigatus are shown in Table 3. Details of the MLST system can be found at http://pubmlst.org/afumigatus/ . Among the 100 strains analyzed, we distinguished 30 STs (Table 1), giving our scheme a discriminatory index of 0.93 (21). ST 5 was the most common (Table 1), with 18 isolates, followed by ST 7 (12 isolates), ST 14 (8 isolates), and ST 20 (7 isolates). For 19 STs, only one or two examples each were found among the isolates tested (Table 1).

Polymorphic sites were found predominantly in coding sequences; however, some were found in introns that were discovered by comparison of our direct sequence data with cDNA sequences in GenBank (Table 3). The BGT1 fragment used for MLST contained a 71-bp intron without polymorphic sites. The CAT1 fragment contained two introns of 66 bp and 105 bp, which included, respectively, one and two of the eight SNPs for this gene. One SNP was upstream of the start codon in the BGT1 fragment (Table 3). MLST can be based on coding and noncoding sequences because the technology relies upon the digital information of genomic DNA sequences and not expressional information. The amino acid substitutions for polymorphisms within the coding sequence were determined. The ratio of nonsynonymous to synonymous amino acid substitutions was determined for each MLST fragment (Table 3) and found to be >1 for five of the seven genes. Therefore, to determine the functionally relevant substitutions, the ratio of substitutions that maintained amino acid side chain polarity and acidity/basicity to those that did not was also calculated. These ratios were <1 for five gene fragments, which we consider acceptably low (Table 3).

In addition to applying the MLST scheme to genotype our A. fumigatus collection, we determined the mating type for each strain, using a multiplex PCR with primers AFM1, AFM2, and AFM3 (Fig. 1), as described previously (37). The locus-specific amplicon of the sexual idiomorph MAT1-1 is 834 bp, and that of MAT1-2 is 438 bp. It should be noted that the MAT1-2 fragment sequenced in the MLST scheme can be amplified from all A. fumigatus isolates because both MAT loci carry the 3′ MAT1-2 coding region, within which lies the 135-bp MLST fragment (Fig. 1). The position of the MLST MAT1-2 fragment is from nucleotides +1106 to +1240 of MAT1-2. Of the 100 A. fumigatus strains that we tested, 29% were sexual idiomorph MAT1-1 and 71% were MAT1-2, a ratio of approximately 1:2. There was no association between mating type and ST, since both mating types were represented among the most populous STs.

An eBURST analysis of the MLST data, which constructs a tree of STs joined to indicate pairs that differ in only one of the seven alleles sequenced, is shown in Fig. 2. All but 5 of the 30 STs formed a single clonal cluster of related isolates, with ST 20 as the putative founding isolate, according to eBURST. Figure 2 shows the genotype assignments of the 18 isolates that had been typed in a previous study on the basis of intergenic spacer regions in ribosome-encoding DNA (41). It is obvious that no relation could be found between the positions of STs on the branches of the eBURST clonal cluster and previously determined genotypes.

A UPGMA dendrogram of the MLST data, based on p-distance (Fig. 3), separated four of the five isolates that were singletons in the eBURST snapshot (Fig. 2) as distant relatives of the rest. The UPGMA dendrogram divided the isolates along lines similar to those for eBURST, but with ST 23, the fifth eBURST singleton, included with the main cluster in this analysis (Fig. 3). Isolates with STs 8, 17, 18, and 22 remained well distanced from other isolates. The main set of isolates could be differentiated into two subgroups by UPGMA, with strong bootstrap support (Fig. 3). The subgroup assignments of the 100 isolates are listed in Table 1, together with other genotype details. No significant association was found between the subgroups and the geographic origins or mating types of the isolates.