Nanopore PCR-cDNA sequencing of the biocontrol isolate Aspergillus flavus AF36 (NRRL 18543) informs gene annotation

ABSTRACT

Toxic molds in the Aspergillus genus synthesize carcinogenic aflatoxins which contaminate crops. The widely applied biocontrol isolate Aspergillus flavus AF36 (NRRL 18543) has a high-quality public genome but lacks corresponding gene annotations. We generated high-quality gene predictions for this isolate by using long-read Nanopore PCR-cDNA sequencing.

ANNOUNCEMENT

Aspergillus flavus AF36 (NRRL 18543), isolated from cottonseed collected in Yuma, Arizona, USA, is the active ingredient in the first non-aflatoxigenic A. flavus-based biocontrol product to be developed for the mitigation of aflatoxins (1–3). It is now sold commercially as AF36 Prevail (Arizona Cotton Research and Protection Council, Phoenix, AZ, USA). A high-quality AF36 genome assembly was previously reported without gene annotations (4). We annotated the AF36 genome incorporating long-read evidence from Nanopore PCR-cDNA sequencing.

Because gene expression might vary between chemical environments and throughout development, we sampled AF36 from high- and low-aflatoxin environments at two time points, resulting in four tissue samples. Fungi were grown in 30 mL modified Czapek’s broth (5) with sucrose replaced by 66 mM glucose, supplemented with 0 µg/mL or 4 µg/mL aflatoxin B1 (Sigma-Aldrich, St. Louis, MO, USA). After inoculation with 10⁵ AF36 spores, flasks were incubated in the dark at 31°C and 150 rpm. After 3 and 7 days of incubation, samples of fungal tissue were collected using vacuum filtration and immediately ground in liquid nitrogen using a sterile mortar and pestle. cDNA was amplified from total RNA (extracted using Plant RNeasy Mini Kit, QIAGEN, Germantown, MD, USA) and sequenced as one library using a PCR-cDNA Sequencing Kit (SQK-PCS109, Oxford Nanopore Technologies, Oxford, UK). Sixteen cycles of PCR were run using a 250-s extension to sequence up to 5 kb. Sequencing was performed using a MinION sequencer with FLO-MIN106D Spot-ON Flow Cell (ONT). MinKNOW and Guppy (MinION Release 19.12.2) were used with the FAST basecalling model and default settings for demultiplexing reads, quality filtering, and adapter sequence trimming. This resulted in 2,450,075 reads with N50 = 1 kb, constituting ~2.1 Gb. Reads were mapped to the AF36 genome (GenBank GCA_012897275.1) (4) using Minimap2 v2.24-r1122 with “--ax splice” (6), resulting in 2,357,784 mapped reads (96.2% of reads). The alignment was converted from SAM to BAM format, sorted by coordinate using samtools v1.9 (7), and input to StringTie2 v2.1.1 with option “-L” for long-read assembly (8). GeneMark v4.59 with options “--ES --fungus” generated initial ab initio gene predictions (9). The genome was cleaned, sorted, soft-masked, and annotated using Funannotate v1.7.4 (10) with options “--protein_evidence” with predicted proteins from A. flavus NRRL 3357, “--genemark_gtf” using the GeneMark GTF output, “--rna_bam” with the sorted BAM alignment of PCR-cDNA reads, and “--transcript_evidence” with the predicted transcripts from StringTie2. For comparison, Funannotate was also run without evidence from PCR-cDNA sequencing.

This pipeline predicted 15,382 transcripts (StringTie2) and 12,894 protein-encoding genes (Funannotate), suggesting ~20% alternative splicing on average (Table 1) (11). BUSCO scores were in the 99th percentile, indicating highly complete gene predictions (Table 1) (12). Without evidence from PCR-cDNA sequencing data, Funannotate generated 12,608 protein-encoding gene predictions, with a corresponding decrease in ortholog recovery (Table 1).

TABLE 1.

Summary of the Aspergillus flavus AF36 gene annotation

	A. flavus strain AF36 gene annotation
	Informed by PCR-cDNA seq	Uninformed by PCR-cDNA seq
Predicted transcripts	15,382	NA
Protein-encoding genes	12,894	12,608
BUSCO db = ascomycota	1,696/1,706 (99.4%)	1,689/1,706 (99%)
BUSCO db = eurotiomycetes	3,513/3,546 (99%)	3,491/3,546 (98.4%)
BUSCO db = eurotiales	4,148/4,191 (99%)	4,108/4,191 (98.1%)

DATA AVAILABILITY

All PCR-cDNA Nanopore reads are available under BioProject ID PRJNA984741 (BioSamples SAMN35777990-SAMN35777993 ). Gene annotation files are available on figshare (11).