Genomic islands in the pathogenic filamentous fungus Aspergillus fumigatus

Abstract

We present the genome sequences of a new clinical isolate of the important human pathogen, Aspergillus fumigatus, A1163, and two closely related but rarely pathogenic species, Neosartorya fischeri NRRL181 and Aspergillus clavatus NRRL1. Comparative genomic analysis of A1163 with the recently sequenced A. fumigatus isolate Af293 has identified core, variable and up to 2% unique genes in each genome. While the core genes are 99.8% identical at the nucleotide level, identity for variable genes can be as low 40%. The most divergent loci appear to contain heterokaryon incompatibility (het) genes associated with fungal programmed cell death such as developmental regulator rosA. Cross-species comparison has revealed that 8.5%, 13.5% and 12.6%, respectively, of A. fumigatus, N. fischeri and A. clavatus genes are species-specific. These genes are significantly smaller in size than core genes, contain fewer exons and exhibit a subtelomeric bias. Most of them cluster together in 13 chromosomal islands, which are enriched for pseudogenes, transposons and other repetitive elements. At least 20% of A. fumigatus-specific genes appear to be functional and involved in carbohydrate and chitin catabolism, transport, detoxification, secondary metabolism and other functions that may facilitate the adaptation to heterogeneous environments such as soil or a mammalian host. Contrary to what was suggested previously, their origin cannot be attributed to horizontal gene transfer (HGT), but instead is likely to involve duplication, diversification and differential gene loss (DDL). The role of duplication in the origin of lineage-specific genes is further underlined by the discovery of genomic islands that seem to function as designated "gene dumps" and, perhaps, simultaneously, as "gene factories".

Figure 1. Molecular Divergence in Molds and Yeasts.

A. fumigatus proteins are compared to their orthologs in N. fischeri, A. clavatus, A. terreus, A. oryzae, A. nidulans, and A. niger (mean values: 95%, 84%, 71%, 71%, 68%, and 69%, respectively). Saccharomyces paradoxus, Saccharomyces uvarum, Candida glabrata, and Kluyveromyces lactis are compared to Saccharomyces cerevisiae (adapted from [74,75]). Mean values for these species are 90%, 82%, 64%, and 60%, respectively. Median percent identity between pairs of orthologs from A. fumigatus and each successive genome in the tree is shown. Relative divergence of humans, mice, birds and fish are shown for reference.

Figure 2. Three Closely Related Aspergilli.

The three most closely related aspergilli, which constitute the Affc-core group (A. fumigatus, N. fischeri, and A. clavatus), are in bold black. The maximum-likelihood tree was constructed from an alignment of 90 proteins chosen on the basis of similar lengths and identical number of intron/exon structures in order to minimize the number of inconvenient or incongruent gene models (see Materials and Methods).

Figure 3. Alignment of the A1163, N. fischeri, and A. clavatus Assemblies against the Eight Af293 Chromosomes.

The first three tracks from the top for each reference chromosome show syntenic blocks (horizontal bars) identified in the target genomes, A. fumigatus A1163, N. fischeri, and A. clavatus. Each assembly from the target genomes is represented by a single color. Syntenic blocks are numbered based on the target genome assembly ID and the position of the block in the target genome assembly. Tracks 4 and 5 show Asp-core gene density and blocks (horizontal bars), respectively, in the Af293 genome. Tracks 6 and 7 show Afum-specific gene density and blocks (horizontal bars), respectively. Tracks 8 and 9 show the density of clustered secondary metabolite biosynthesis genes and transposable elements, respectively, found in Af293. Pink vertical bars represent putative centromeres, the purple vertical bar in chromosome 4 represents a region of ribosomal DNA, and horizontal black bars beneath each chromosome designate sequencing gaps.

Figure 4. The Af293 RosA and NosA Proteins.

Shown in bold red are RosA, NosA and Pro1 proteins that have been experimentally characterized are shown in bold black. Branches with a bootstrap of 75% or more are indicated in bold black. The trees are maximum-likelihood trees (see Materials and Methods).

Figure 5. Proteins with Orthologs in the Three Most Closely Related Aspergilli (A. fumigatus, N. fischeri and A. clavatus).

These proteins constitute the Affc-core group, and proteins with no orthologs in N. fischeri and A. clavatus constitute the A. fumigatus-specific group (Afum). The proteins in the Affc-core can be further divided into two groups, Aspergillus-core (Asp-core), which has orthologs in all of the other aspergilli, and the Affc-specific group, which is comprised of the rest of the Affc-core.

Figure 6. A. fumigatus-Species Specific Genes Supported by Homology and Expression Data.

Genes with no orthologs in N. fischeri and A. clavatus constitute the A. fumigatus-specific group (Afum). Genes that have homologs in other fungal genomes constitute the Homology group. Genes differentially expressed in microarray studies represent the Expressed group.

Figure 7. Four Common Topologies Detected by Phylogenetic Analysis of N. fischeri-Specific Proteins.

The N. fischeri proteins under consideration are in bold red. The bootstrap supporting the clade containing the N. fischeri is also in bold red. Other N. fischeri proteins are shown in bold black. Blue species names correspond to the recipient genome when different from N. fischeri. Systematic gene names are indicated. Branches with a bootstrap of 75% or more are indicated in bold black. The trees are maximum-likelihood trees (see Materials and Methods). A. Set1 protein evolved by probable duplication, differentiation and differential loss in other Aspergillus species (DDL). B. Set 2 protein evolved by probable HGT from Sordaryomycetes into the N. fischeri lineage. C. Set 2 protein evolved by probable DDL and a Fusarium solani protein (in blue) evolved by probable HGT from the N. fischeri lineage into Sordaryomycetes. D. Set 2 protein showing similarity to a protein from the Sordaryomyce Chaetomium globosum.