Aspergillus Section Flavi, Need for a Robust Taxonomy

REPLY

In a recent letter to the editor, Houbraken et al. (23) provide a series of recommendations to the microbiological community to prevent the taxonomic misidentification of genome-sequenced fungal strains. In the era of genomics and bioinformatics, postulating that 1 nucleotide (nt) within a gene can “correctly” identify a species does not seem plausible. However, the authors of the letter call this the “calmodulin barcode,” meaning nucleotide substitutions within a 506-nt region of the calmodulin gene (1). After the evolutionarily conserved rRNA (18S rRNA, internal transcribed spacer [ITS], 28S rRNA) and RNA polymerase II (2–4) showed no differences between Aspergillus flavus S- and L-morphotypes, attention shifted toward the calmodulin gene. Thus, without sequencing 18S rRNA, 28S rRNA, or the largest RNA polymerase II subunit, at least 34 new species of Aspergillus were named by Houbraken, Frisvad, Visagie, and coworkers (1, 5, 6). However, in a phylogenetic tree of 152 Aspergillus section Flavi isolates using the calmodulin 506-nt region, 40 Aspergillus minisclerotigenes isolates had only two nucleotide substitutions in common, namely, 100C>A and 269A>G, both of which are silent mutations (Fig. 1). However, only 269A>G discriminates A. minisclerotigenes from A. flavus, since 100C>A is present in three A. flavus isolates (GenBank accession numbers MK451387, MK451365, and MG517986) identified by the authors of the letter. We all agree that species identification is important; paradoxically, the calmodulin barcode assigns species based on a single-nucleotide polymorphism (SNP), while there are between 133,000 and 179,000 SNPs within A. flavus S- and L-morphotypes, respectively (7).

FIG 1.

Maximum likelihood phylogenetic tree using 506 nt of the calmodulin gene. (Top) Partial sequence of the calmodulin gene of Aspergillus, highlighting the substitutions at positions 100 and 269. (Bottom) Phylogram of 152 isolates of Aspergillus species section Flavi, including 87 A. flavus isolates (14 SNPs), 40 A. minisclerotigenes isolates (10 SNPs), 17 Aspergillus agricola isolates (4 SNPs), 3 Aspergillus pipericola isolates (2 SNPs), and 5 A. flavus S-morphotype isolates (3 SNPs) from Ethiopia; numbers in parentheses correspond to SNPs found within each group. Scale bar, number of substitutions per 100 nt. Blue arrows indicate the 3 A. flavus strains with 100C>A substitutions; the yellow area highlights isolates with 269A>G substitutions. After sequence alignment, the neighbor-joining method was used as the construction method with the Jukes-Cantor substitution model including rate variation, processed in CLC Genomics Workbench v.21.0. (Qiagen).

Another limitation of Aspergillus taxonomy is the chemotypes resulting from 30 genes in the aflatoxin biosynthesis gene cluster (ABC) (8), e.g., A. flavus produces B-aflatoxins and Aspergillus parasiticus produces B and G types (9). Despite that a single nucleotide change in one ABC gene can prevent aflatoxin production (10), the inheritance of the ABC is favored by environmental pressure (11), and Aspergillus spp. are not physically or reproductively isolated; intraspecies and interspecies crosses can result in gain of function, e.g., G-type aflatoxin production (9, 12, 13). Hence, a new species named by one author of the letter was later reversed to its initial name by the same author because of the chemotype, i.e., A. flavus S-morphotype to Aspergillus parvisclerotigenus (14) and back to A. flavus (6). Other groups utilized the calmodulin gene and a single deletion in the ABC to name three new Aspergillus species (15, 16).

Analysis of 25 insertion/deletions (indels) within the ABC revealed almost as many Aspergillus genotypes as the number of isolates (17, 18). That work was performed not to identify new species but to detect predominant genotypes, sequence their genomes (19, 20), and find targets for gene silencing. Isolates in those publications had been identified by Dr. Bruce Horn comparing morphology to type strains in the culture collection at the National Peanut Research Laboratory, as described (17, 21). Furthermore, SNP analysis of five Ethiopian S-morphotype isolates versus the model genome A. flavus NRRL3357 was performed by combining 20,213 nt of evolutionarily conserved genes: rRNA cistron (7,796 nt) and DNA-dependent RNA polymerase largest subunit (RPB1) (5,857 nt) and second largest subunit (RPB2) (6,560 nt). Four S-morphotype isolates had 99.2% identity to A. flavus NRRL3357, with SNP detection processed in CLC Genomics Workbench v.21.0 (Qiagen, Denmark). Genome-wide SNP analysis of 24 Aspergillus strains, including 13 A. flavus isolates (10 L-morphotype and 3 S-morphotype), 10 Aspergillus oryzae isolates, and 1 A. parasiticus, concluded that, despite their morphological differences, L-type and S-type isolates have 99.2% identity; therefore, they belong to the same A. flavus species (7). The fifth S-morphotype isolate from Ethiopia presented a 189-nt insertion in its rRNA cistron. Unfortunately, the full-length rRNA cistron of A. minisclerotigenes is not available for comparison, since it was not reported with its genome (GenBank accession number SWDZ00000000) (22). The current Aspergillus taxonomy is based on PCR amplification of short DNA fragments, does not examine the most evolutionarily conserved genes, adopts single indels within the ABC as a classification tool, and observes chemotypes whose detection can vary by orders of magnitude, depending on the analysis methods. New technologies are available, and we think there is a need for an upgrade to a more robust system of classification.

Data availability.

The data used to generate the figure are openly available in Harvard Dataverse (https://doi.org/10.7910/DVN/LVJ7ZN).