Host-Pathogen Interactions

Total RNA was extracted using TRI-reagent (Sigma) according to the manufacturer’s instructions. Purified RNA was quantified using a Nanodrop spectrophotometer (NanoDrop Technologies, Wilmington, USA) and reverse transcribed into cDNA (20 μg per sample) using Bioscript (Bioline, London, UK) in 30 μL reactions. cDNAs were diluted to 500 μL with TE buffer (pH 8.0) and stored at −20 °C. For qPCR analysis, cDNAs representing uninfected fish (n = 7) and fish exhibiting; early (grade 1; n = 6), moderate (grade 1–2; n = 9), and advanced (grade 2; n = 10 and grade 3; n = 10) stages of clinical disease were examined. Sixty primer sets were used encoding putative cellular markers and immune response genes, as well as primers for the reference gene elongation factor-1α (EF-1α). A full list of all primers and associated information is provided in Additional file 1. EF-1α has been repeatedly demonstrated to be highly consistent as a reference gene in the immune gene expression profiling of fish host-pathogen interactions [13 (link)-15 (link),23 (link)]. The relative parasite prevalence in all tissue samples was assessed by qPCR using T. bryosalmonae-specific primers for the detection of the house-keeping genes; 18S rDNA [EMBL: U70623] and 60S ribosomal protein L18: RPL18 [EMBL: FR852769] [24 (link)]. 18S rDNA primers were modified relative to those in previous PKD studies [25 (link)] and were used to detect the parasite in gDNA samples. Although T. bryosalmonae RPL18 is highly homologous to the rainbow trout homologue (53% amino acid identity), T. bryosalmonae-specific RPL18 primers were designed within the open reading frame aided by the considerable difference in codon usage between host and parasite (typically, 45-55% GC and 25-40% GC respectively). Hence, detection of T. bryosalmonae RPL18 transcripts provides a sensitive measure of only viable parasites in each sample, whereas 18S rDNA detects both live and dead parasite material. With the exception of 18S rDNA, all primer pairs were designed and tested with a set of cDNA and DNA samples to ensure that products could only be amplified from cDNA and not from genomic DNA under the conditions used. Genomic DNA was extracted from the same TRI-reagent tissue homogenates used for RNA extraction, as described previously [13 (link)]. SYBR green (Invitrogen, Paisley, UK) based RT-qPCR using Immolase DNA Polymerase (Bioline) was performed using a Light Cycler® 480 SW 1.5 system (Roche, Mannheim, Germany) as described previously [26 (link)]. For parasite DNA detection, primer efficiency was determined using serial dilutions of reference (internal PCR control) DNA and used for quantification of the DNA concentration. To normalize the level of T. bryosalmonae 18S rDNA to the input genomic DNA, additional qPCR was undertaken using primers to the trout macrophage colony stimulating factor (MCSF) gene, as described previously [13 (link)]. For cDNA detection, the primer efficiency and concentration of each gene transcript was quantified using data generated from serially diluted reference DNA amplified in each PCR run, as described previously [13 (link)]. Since RPL18 mRNA detection is more sensitive than DNA detection and should more accurately reflect the viable parasite prevalence in individual fish, this was used for analysis of the immune gene expression data in addition to the kidney swelling grade assessment. To determine the RT-qPCR detection limit in terms of RPL18 transcript number, a pooled T. bryosalmonae positive sample was obtained from grade 2 cDNAs and serially diluted. A diluted RPL18 reference was included to enable relative quantification. The expression of trout immune genes was initially normalized to the expression of EF-1α and subsequently expressed as fold change relative to the expression level in parasite-naïve uninfected fish. Likewise, parasite RPL18 cDNA levels were normalized to that of trout EF-1α for each sample.

PAST software (V 4.07) (Hammer et al., 2001 ) was used to perform multivariate analyses. We performed the NM-MDS according to the Bray–Curtis similarity index using raw RNA-seq data with individual read values for each replicate (Figures 5A,B) transformed as log10. This analysis was also used to analyze (dis-)similarity between the different bacterial samples also in the expression of subsets of genes involved in host-pathogen interaction and toxin-antitoxin systems.

A web-based database called the Pathogen-Host Interaction Database (PHI-base; Winnenburg et al., 2006 (link)), which comprises experimentally verified pathogenicity, virulence, and effector genes from bacterial and fungal pathogens that infect hosts like plants, animals, fungi, and insects. BLASTP was employed with a cut-off E value of ≤1e−06 in the pathogen-host interaction (PHI) database to find the probable pathogenicity-related genes.

Genes related to cellulose and hemicellulose degradation: Carbohydrate-active enzymes (CAZymes) in the three Penicillium species were annotated using BLAST (Johnson M. et al., 2008 (link)). The dbCAN annotation program HMMER 3 (Finn et al., 2011 (link)) was used to search against the CAZy (carbohydrate-active enzyme) database (Lombard et al., 2014 (link)). The results were combined when e values ≤1e–5. The class II peroxidases and dye-decolorizing peroxidases were further confirmed by BLAST searches against PeroxiBase (Fawal et al., 2012 (link)).
Secondary metabolism genes: Candidate transporter genes in the three Penicillium species were identified based on searches of the Transporter Classification Database (TCDB) (Saier et al., 2014 (link)) with e values ≤1e–5 and identity values ≥40%. The secondary metabolism biosynthesis genes and gene clusters in the genomes of the three Penicillium species were predicted with AntiSMASH 6.0 (Blin et al., 2021 (link)). The Comprehensive Antibiotic Research Database (CARD) (McArthur et al., 2013 (link)) was used to compare coding genes (of the three Penicillium species) involved in antimicrobial resistance.
Virulence associated genes: Candidate pathogen-host interactions (PHI) genes within the genome of the three Penicillium species were identified using BLASTp to search against PHI-base v4.3⁷, and protein alignments were performed to identify putative virulence-associated genes in the three Penicillium species with identity ≥40% and query coverage ≥70%.