We compared a collection of 107 isolates collected from Europe with 194 isolates collected globally (Table 1; Technical Appendix Table 1). All isolates were checked for purity and cultivated on malt extract agar medium (Oxoid, Basingstoke, UK). Cultures were incubated for 2 days at 30°C. A working collection was made by growing C. gattii strains on malt extract agar slants for 2 days at 30°C, after which the strains were stored at 4°C. Strains were stored long term at −80°C by using the Microbank system (Pro-Lab Diagnostics, Richmond Hill, Ontario, Canada).

Genomic DNA extraction, AFLP genotyping, and mating-type determination by amplification of either the STE12a or STE12α locus were performed as described (8). The 7 nuclear consensus MLST loci (CAP59, GPD1, IGS1, LAC1, PLB1, SOD1, and URA5) were amplified by using the preferred primer combinations (9). To compare the current set of C. gattii strains with those from a published C. gattii population biology study, we included the 3 nuclear loci that are not part of the consensus MLST scheme (CAP10, MPD1, and TEF1α) (12). We also included isolates from a global study by Byrnes et al. (11) and Fraser et al. (12) and subjected them to amplification and sequencing of the CAP59, SOD1, and URA5 loci.

Amplifications were conducted in a 25-µL PCR mixture containing 37.5 mmol/L MgCl₂ (Bioline, London, UK), 1× PCR buffer (Bioline), 1.9 mmol/L dNTPs (Bioline), 0.5 U Taq DNA polymerase (Bioline), 5 pmol of both primers (Biolegio, Nijmegen, the Netherlands) (Technical Appendix Table 2), and ≈100 ng of genomic DNA. PCRs were conducted with an initial denaturation step at 94°C for 5 min, followed by 35 cycles of denaturation at 94°C for 30s, annealing for 30s (see Technical Appendix Table 2 for optimal annealing temperatures), extension at 72°C for 1 min, followed by 72°C for 5 min and a final dwell at 21°C.

Sequencing reactions were conducted with the BigDye version 3.1 chemistry kit (Applied Biosystems, Foster City, CA, USA) as described (13). For all amplification products except CAP59, the initial amplification primers were used for sequencing reactions. For CAP59, the newly designed forward primer CAP59L-Fwd and the original reverse primer JOHE15438 (12) were used.

Consensus sequences were assembled and checked for ambiguities by using SeqMan version 8.0.2 (DNASTAR, Madison, WI, USA). Sequence alignments were generated with MEGA version 5 (14) by using the standard settings and manual correction. The genome sequence databases of reference strains H99 (culture collection no. CBS8710; C. neoformans variety grubii; AFLP1/VNI; Broad Institute [www.broadinstitute.org/annotation/genome/cryptococcus_neoformans/]) and JEC21 (culture collection no. CBS10513; C. neoformans variety neoformans; AFLP2/VNIV; Stanford Genome Technology Center [www-sequence.stanford.edu/group/C.neoformans/]) were used to extract the corresponding sequences for all 10 investigated nuclear loci to serve as an outgroup. The best fitting nucleotide substitution model was determined by using MrModeltest version 2 (15) and was conducted for the complete C. gattii 10-loci MLST, for the accepted consensus MLST scheme (CAP59, GPD1, IGS1, LAC1, PLB1, SOD1, and URA5) (9), and a previously launched C. gattii MLST scheme (CAP10, GPD1, IGS1, LAC1, MPD1, PLB1, and TEF1) (12). As a result, the HKY G+I model (Hasegawa-Kishino-Yano plus gamma distributed with invariant sites) was the best model to use for analyzing the phylogeny of the C. gattii isolates for all 3 datasets. The evolutionary history was inferred by using the maximum-likelihood method in MEGA version 5 (14). A bootstrap consensus tree was inferred from 1,000 replicates to show the relevant lineages obtained in this analysis.

We calculated the haplotype diversity (H_R), equal to the Simpson diversity index (D), by using the Microsoft Excel (Microsoft, Redmond, WA, USA) add-in called Haplotype Analysis (16). For this purpose, sequences were collapsed into sequence type numbers (Technical Appendix Table 1).

The 301 C. gattii isolates collected from Europe and other areas around the world could be divided into the following genotypes: 146 AFLP4/VGI (50.2%; 72 from Europe), 22 AFLP5/VGIII (7.6%; 1 from Europe), 108 AFLP6/VGII (37.1%; 23 from Europe), 13 AFLP7/VGIV (4.5%; 2 from Europe), and 2 AFLP10/VGIV (0.7%; both from Europe). From 10 isolates (7 AFLP8 [5 from Europe] and 3 AFLP9 [2 from Europe]), genotypes represented interspecies C. neoformans × C. gattii hybrids. These 10 isolates were excluded from further analysis because amplified fragments of hybrid isolates will result in mixtures of different alleles. The 57 human patient isolates from Europe were obtained from 40 patients (Technical Appendix Table 1). The genotypic diversity of the remaining 291 C. gattii isolates (100 from Europe, 191 from other areas) with a haploid genotype is shown in the Table.

Sequence type diversity was calculated for each of the MLST loci, all 10 loci, and the combined datasets according to Fraser et al. (12) and Meyer et al. (9) (Technical Appendix Table 3). The CAP10 locus showed the lowest overall diversity (n_ST = 19; D_ST = 0.765) and the IGS1 locus showed the highest diversity (n_ST = 52; D_ST = 0.930). The 10-loci MLST dataset showed the highest diversity (n_ST = 150; D_ST = 0.975), followed by the MLST scheme of Meyer et al. (D_ST = 0.971) (9) and Fraser et al. (D_ST of 0.959) (12). The latter MLST scheme differentiated more sequence types than the consensus MLST scheme (n_ST = 136 vs. 127).

Maximum-likelihood analysis of the MLST data showed that the C. gattii isolates clustered in 5 monophyletic clusters, which were highly supported and agreed with the AFLP genotypes (Figure, Technical Appendix Figure). When each genotypic cluster was separately analyzed, support values of the branches were low, <75 (Technical Appendix Figure). For genotype AFLP4/VGI isolates, bootstrap support values for nearly all branches were low; for the second largest group formed by AFLP6/VGII isolates, branches were better supported.

Phylogenetic analysis demonstrated that several human patient genotype AFLP4/VGI isolates from Europe had an autochthonous origin because they formed a separate cluster with environmental and animal isolates from the same area. A set of isolates from human patients on the Iberian Peninsula (CCA232, CCA242L, CCA242T, CCA311, CCA312, CL6148) were found to be genetically indistinguishable from isolates obtained from animals or the environment in the same area (indicated in Technical Appendix Figure as the European Mediterranean cluster). The human patient isolates from Europe with genotype AFLP4/VGI, which were probably acquired within Europe because patients did not have histories of travel outside Europe, are IP2005/215 and IP2006/958 (France), RKI85/888 (Germany), 5UM and 75UM (Italy), RKI97/482 (Portugal), and CBS2502 (the Netherlands). Some of these human patient isolates were closely related (e.g., IP2005/215 and RKI97/482) or even genetically indistinguishable (e.g., CBS2502 and RKI85/888).

A large proportion of the human patient and environmental isolates of C. gattii AFLP4/VGI from Italy and Spain formed a novel autochthonous Mediterranean MLST cluster that was genetically homogeneous, irrespective of their origin or mating type (Figure, Technical Appendix Figure). C. gattii AFLP4/VGI was involved with numerous small outbreaks among goats in Spain, and the isolates were genetically indistinguishable from recently obtained human patient, animal, and environmental isolates from different provinces in Spain as well as from AFLP4/VGI mating-type a isolates from Italy (Technical Appendix Figure).

Several C. gattii genotype AFLP6/VGII infections were found to have originated in Europe (e.g., the IP1998/1037–1 and −2, IP2003/125, and CCA242 isolates), and all fell within a cluster that could not be linked to an environmental source. The same holds true for the human patient isolates from Greece (AV54S, –W, and IUM01–4731), which came from the same patient who had no history of travel outside Greece.

In the phylogenetic analysis, isolates from human patients in Europe were also observed next to those originating from other geographic areas. The most striking example was a set of 4 human patient isolates from citizens of Denmark, the Netherlands, Germany, and Switzerland, in whom cryptococcosis developed after they had visited Vancouver Island, British Columbia, Canada. The isolates obtained from these tourists (CBS10485, RKI06/496, RKI01/774, and CBS10866, respectively) had MLST profiles identical to that of C. gattii AFLP6A/VGIIa, the genotype that caused the Vancouver Island outbreak (Figure, Technical Appendix Figure) (17–19). The outbreak-related sets of AFLP6A/VGIIa and AFLP6C/VGIIc isolates from the Vancouver Island and Pacific Northwest outbreaks, respectively, are within outbreak-specific clusters (Figure, Technical Appendix Figure). A similar finding was observed for a set of 5 human patient C. gattii AFLP6B/VGIIb isolates (IP1996/1120–1 and −2, IP1999/901–1 and −2, and IP2000/87) in France, which had been obtained from patients who had emigrated from Africa to France and which were genetically indistinguishable from an isolate (IP2001/935–1) from a resident of Senegal (Technical Appendix Figure). This set of 5 isolates from France and 1 from Senegal were closely related to isolates from the Vancouver Island and Pacific Northwest outbreaks; however, it seems unlikely that the patient from Senegal had traveled to these outbreak areas, and none of the patients in France reported having traveled to Canada or the United States.

A set of C. gattii AFLP4/VGI mating-type α isolates from Portugal (IP1997/18) and Belgium (IHEM19725B and –S) was found to be indistinguishable from human patient isolates from the Democratic Republic of the Congo and Rwanda (B3939 and CBS6289, which were both mating-type a, and IHEM10602S, IHEM10769S, and IHEM10769W, which were mating-type α) (Figure, Technical Appendix Figure). Both phenotypically different isolates of IHEM19725 were obtained from an HIV-infected patient from Rwanda who had emigrated to Belgium. The CBS1622 isolate from Europe, obtained from a patient with the oldest documented case of a C. gattii infection (20), was found to be genetically indistinguishable from a set of isolates from North America.

The single C. gattii AFLP5/VGIII isolate from a 24-year-old immunocompetent patient from Germany (isolate RKI97/310) (21) clustered with human patient and environmental isolates from Mexico (Technical Appendix Figure). The 2 C. gattii AFLP7/VGIV isolates CBS7952D and CBS7952S, obtained from an HIV-infected patient in Sweden, were genetically indistinguishable from each other and had a unique MLST profile that clustered with human patient isolates from Africa. According to the patient’s history, years before the onset of cryptococcal infection, she had emigrated from Zambia to Sweden (Technical Appendix Figure). Another example of a reactivated dormant C. gattii infection is that of the human patient isolate from Italy, IUM92–6682 (AFLP4/VGI), obtained from an immunocompetent immigrant from Brazil, which had an identical MLST profile to 6 human patient mating-type α isolates (IHEM14934, IHEM14956, IHEM14965, IHEM14968, IHEM14976, and IHEM14984) from Brazil (Figure, Technical Appendix Figure).

Infection was acquired outside Europe for 24 of these patients (31 isolates) and within Europe for 16 patients (26 isolates) (Table, Technical Appendix Figure). Among these 57 human patient isolates, most (47 [82.5%]) were obtained since 1995, the remaining 10 (17.5%) were isolated during 1895–1994. One of these isolates originated from 1985, another 2 isolates (CBS1622 and CBS2502), from 1895 and 1957, were retrospectively found to represent C. gattii isolates from Europe (20,22).