Comparative genomic analysis and secretome profiling of Penicillium species isolated from cured meat products

Motivation. Although the knowledge of the contribution of molds in fermented foods is still very limited, compared to bacteria and yeasts, several fungal species are commonly used which contribute to the improvement of food safety, nutritional value, organoleptic quality, and food preservation. However, filamentous fungi can also be considered food contaminants leading to food spoilage and the production of toxic metabolites. Penicillium is among the predominant fungal genera capable of growing on the surface of dry fermented meats, and several species, such as P. nalgiovense, have been domesticated by continuos selection for beneficial traits. Here we investigate the genome of six Penicillium species isolated from Italian salami, which have been shown to be adapted to the curing conditions occurring during the production of fermented sausage, with the aim of characterizing the differences in secretomic potential, which may have been related to the maturation of meat products. Methods. The genome sequencing and annotation of Penicillium brevicompactum ITEM 18316, P. biforme ITEM 15300, P. chrysogenum ITEM 4680, P. cvjetkovicii ITEM 18317, P. nalgiovense ITEM 18323, P. salamii ITEM 15291, and P. solitum ITEM 18327 isolated from dry cured meat was part of a collaborative sequencing project involving JGI, PNNL and CNR-ISPA. Draft genomes and transcriptomes were generated using Illumina technology. Genome assemblies were masked for repeats using RepeatMasker, with the RepBase library and RepeatScout, and the measures of genome assembly completeness were performed with BUSCO. Nuclear genomes were annotated using the JGI Annotation pipeline, and further improved. Protein-coding gene models were generated using a combination of ab initio, homology-based, and transcriptome-based gene predictors. Predicted proteins were functionally annotated using SignalP, TMHMM for transmembrane domains, InterProScan for protein domains, and Blastp alignments against the NCBI NR, SwissProt, KEGG, and KOG. Transcription factors were assigned based on Pfam. GO terms were assigned based on InterPro and SwissProt hits. TCDB and MEROPS were used for transporter and peptidase classifications, respectively. CAZymes, secondary metabolites, and cytochrome 450 subfamilies were also annotated and antiSMASH fungi was used to improve cluster detection. Subcellular localization of proteins was predicted with DeepLoc and proteins with extracellular localization were realigned with Diamond and reannotated with eggNOG-mapper. Proteins and peptide domains with interesting catalytic or antimicrobial activities that can describe the contribution of molds in meat curing have been interrogated and retrieved. Additioanlly, the mycotoxin production potential of each fungal species was also evaluated. Orthofinder analysis was also implemented to provide comprehensive statistics for the comparative genomics of the six Penicillium isolates, and genome of other isolates of the same species deposited on NCBI, if available. Results. The draft genome assembly resulted in 165 to 750 contigs and 156 to 692 scaffolds for Penicillium brevicompactum ITEM 18316 and P. salamii ITEM 15291, respectively. The gene prediction identified 9657 to 12781 protein-coding genes for P. cvjetkovicii ITEM 18317 and P. biforme ITEM 15300, respectively. Using whole genome sequence information and bioinformatics tools, various genomic features were explored to characterize the secretory potential of the studied fungal species, focusing on several interesting properties. Complete clusters for mycotoxin biosynthesis were identified in P. brevicompactum ITEM 18316 (mycophenolic acid), and P. biforme ITEM 15300 (PR-toxin). Protease and peptidase enzymes that complete the hydrolysis of proteins and peptides into