Genome-Wide Analysis of Cyclophilin Proteins in 21 Oomycetes

Cyclophilins (CYPs), a highly-conserved family of proteins, belong to a subgroup of immunophilins. Ubiquitous in eukaryotes and prokaryotes, CYPs have peptidyl-prolyl cis–trans isomerase (PPIase) activity and have been implicated as virulence factors in plant pathogenesis by oomycetes. We identified 16 CYP orthogroups from 21 diverse oomycetes. Each species was found to encode 15 to 35 CYP genes. Three of these orthogroups contained proteins with signal peptides at the N-terminal end, suggesting a role in secretion. Multidomain analysis revealed five conserved motifs of the CYP domain of oomycetes shared with other eukaryotic PPIases. Expression analysis of CYP proteins in different asexual life stages of the hemibiotrophic Phytophthora infestans and the biotrophic Plasmopara halstedii demonstrated distinct expression profiles between life stages. In addition to providing detailed comparative information on the CYPs in multiple oomycetes, this study identified candidate CYP effectors that could be the foundation for future studies of virulence.

In mammals and plants, CYPs, FK506 binding proteins, and parvulins are the three immunophilin protein superfamilies. Immunophilins have been demonstrated to bind to cyclosporin A (CsA), an immunosuppressant molecule of fungal origin [14][15][16]. This recognition form contains complexes that affect dendritic and T cells [1,17]. In fungi, CYPs have been identified as targets for CsA. Binding of CsA to Cyp1 compromises the immune response by inhibiting calmodulin-dependent phosphoprotein phosphatase calcineurin [18]. In both wound-infecting Cryphonectria parasitica and appressorium-forming Magnaporthe grisea, mutant ∆cyp1 strains were less virulent on their respective hosts [19,20]. Phenotypically, this was observed as inhibiting appressorium development in M. grisea [19]. These data suggest an important role for Cyp1 in pathogenesis. It is yet to be investigated whether CYP homologs have similar roles in oomycetes, phylogenetically distinct organisms with similar pathogenic lifestyles.

Structure Analysis
The number of CYPs identified for each oomycete varied between species (Table 1 and  Supplementary Table S1). These proteins were clustered into 16 orthogroups, 13 of which were ubiquitous across the 9 genera surveyed. Orthogroups oomcCYP14 and oomcCYP15 were absent in every biotrophic species (Table 1). Orthogroup oomcCYP12 was absent from Albugo spp., an oomycete genus that adapted to biotrophy independently from the downy mildews [24]. Manual curation identified 43 of 472 proteins across 17 species that were likely misannotated ( Table 1, Supplementary  Table S2). Conserved domain analysis identified partial/low scoring CYP domains, whereas orthologs had higher confidence CYP domains. Investigating the annotation identified high scoring CYP domains split across multiple reading frames, implying that a splice site may have not been predicted; this intron position is not conserved across all oomycete species.
Cyclophilins in oomycetes were classified into two major categories: single-domain proteins (five orthogroups) and multi-domain proteins (10 orthogroups; Figure 1). Bigrams, defined as pairs of different domains in a protein, have been reported in eukaryotic species to enable coupling between two distinct cellular processes, and proteins enriched for bigrams may be involved in pathogenicity [25]. The overall number of bigrams in oomycetes was significantly higher than fungi but less than other species (e.g., Drosophila melanogaster) [25]. Investigating the CYP containing bigrams may therefore indicate the role these proteins play in oomycetes. Previously, six additional types of domains were reported in CYP proteins from Phytophthora spp.: a FK506-binding protein (FKBP) immunophilin domain, tetracopeptide repeat (TPR), glutaredoxin (GRX), RNA recognition (RRM), modified DNA-binding ring-finger (U-box), and WD40 repeat domains [17]. This study identified an additional 23 domains that formed bigrams with CYP domains. Five of the previously identified domain combinations were ubiquitous to all oomycetes: FKBP (PF00254; oomcCYP03), GRX (PF00462; oomcCYP04), RRM (PF00076; oomcCYP05), WD40 repeat (PF00400; oomcCYP06), and U-box (PF04564; oomcCYP08) domains [26][27][28][29][30]. FKBP-3TPR-CYP bigram has been reported to be present in unicellular eukaryotes, including ciliophora, oomycetes, diatoms, and dinoflagellates, and as inhibiting calcineurin (protein phosphatase 2B) in the presence of the cognate drugs to exhibit family-specific drug sensitivity [31,32]. This bigram was detected in other stramenopiles and alveolates, but not from Rhizaria, Plantae, or opisthokonts (Supplementary Figure S1). Other bigrams indicate a ubiquitous role in the oomycetes in detoxification, RNA recognition, protein-protein/protein-DNA interactions, and ubiquitination [33][34][35][36][37]. Therefore, CYPs may have a wide range of roles in oomycetes.   There was no strong evidence for secretion signals of all proteins in a single orthogroup, though a few proteins were implicated as being secreted (Supplementary Table S3). This does not preclude these proteins from being secreted or transported to the host through other mechanisms [38,39]. Additionally, secretion signals may be lost if the protein is incorrectly annotated with an early or late Pathogens 2020, 9, 24 5 of 14 start codon predicted. Interestingly, 23 of 26 omcCYP04 proteins had a predicted transmembrane domain; oomcCYP02 proteins contained only a CYP domain, and omcCYP04 CYP proteins were bigrams with GRX ( Figure 1; Supplementary Table S3).

Multidomain Analysis
The CYP domains of oomycete species ranged from 121 to 259 residues in length. The consensus sequence for nine of the orthogroups contained five motif blocks ( Figure 2) that were conserved in other eukaryotic CYP domains. The amino acid motifs QGGD and KHVVFG are associated with protein folding and stabilization in humans [40], and were present in the consensus sequence of 11 of the 16 orthogroups. The consensus sequence for all CYP orthogroups, except oomcCYP04, showed conservation of 65 to 130 residues dispersed across the CYP protein, including a CsA binding site and three conserved residues required for PPIase catalysis [41] (Supplementary Figure S2, Figure 3). When each ortholog of oomcCYP04 was aligned against PPIase, 55 residues were positionally conserved with the PPIase sequence. The annotated oomcCYP04 proteins were conserved across the oomycetes, although highly diverged from other orthogroups of oomycete cyclophilins ( Figure 4).

Phylogenetics of CYPs in Oomycetes
The phylogenetic tree based on the CYP domain showed clustering correlated with orthology based on all-by-all protein alignments, though orthogroup oomcCYP00 was split into five clades (oomcCYP00-i, oomcCYP00-ii, oomcCYP00-iii, oomcCYP00-iv, and oomcCYP00-v) ( Figure 5, Supplementary Table S4). OomcCYP00 was highly similar in all orthogroups (using P. infestans and P. sojae CYP sequence as the sample; Supplementary Table S4). As this was the largest orthogroup with proteins often only containing a single CYP domain (Figure 1), it is possible that multiple paralogs were assigned to a common orthogroup. Signal peptides or trans-membrane domains were often found encoded in proteins belonging to clade oomcCYP00-i. Although the majority of CYP domains clustered phylogenetically, there were some instances where clades containing CYP domains were assigned to different orthogroups (i.e., oomcCYP-v, Figure 5 inset). The phylogenetic analysis and annotations supported that downy mildew and Albugo species assemblies do not contain oomcCYP14 and oomcCYP15 cyclophilins. Additionally, two oomcCYP00 clades (oomcCYP00-iv and oomcCYP00-v) were not detected from these species ( Figure 5). Like downy mildews, Albugo spp. are thought to have adapted to biotrophy from a non-biotrophic ancestor [42], meaning that these CYP proteins may have been lost from at least two lineages that independently adapted to biotrophy. If these proteins are not required for biotrophy, then a lack of selection and drift may have resulted in their loss. A similar conclusion was made for Pathogens 2020, 9, 24 6 of 14 biotrophic downy mildews, which exhibited a depletion of pathogenicity as well as transporter and carbohydrate-associated domains, when compared to hemibiotrophs [43].   To study the relationship between oomycete CYPs and plant or fungal CYPs, the top 10 plant and fungal CYPs from National Center for Biotechnology Information (NCBI), ranked by percent identity, were added to the alignments, using Phytophthora proteins as queries. Phylogenetic trees were constructed from diverse oomycete, plant, and fungal CYP sequences. In most cases, oomycete orthogroups clustered together, away from plant and fungal CYP sequences (Supplementary Figure  S3). This was not observed for oomcCYP01; proteins annotated in Saprolegnia and Aphanomyces species appeared closer to plant CPYs than other oomycete CYPs. Reciprocal BLAST [44] of P. infestans annotations supported orthology of oomycete proteins with plant and fungal proteins for CYPs belonging to oomcCYP00-iii, oomcCYP01, oomcCYP05, oomcCYP06, oomcCYP07, oomcCYP08, oomcCYP09, and oomcCYP13 (Supplementary Table S5). In addition, oomcCYP00-i, oomcCYP00-ii, oomcCYP03, and oomcCYP04 had the best reciprocal BLAST hits with one of either fungi or plants, but not both, supporting shared ancestry (Supplementary Table S5). Additional domains fused to CYP proteins may have resulted in the top hit identified being non-orthologous, such as for oomcCYP04, a CYP-GRX bigram (Figure 1). For the other eight P. infestans CYP proteins, the reciprocal BLAST hit for fungal and plant results was to other P. infestans CYP proteins (Supplementary Table S5). Only two of these eight had reciprocal BLAST hits when non-oomycete stramenopiles were surveyed (Supplementary Table S5). Therefore, these six protein lineages may be unique to the oomycetes. Interestingly, one of these lineages, oomcCYP14, was not detected in biotrophic oomycete species (Table 1).

Phylogenetics of CYPs in Oomycetes
The phylogenetic tree based on the CYP domain showed clustering correlated with orthology based on all-by-all protein alignments, though orthogroup oomcCYP00 was split into five clades (oomcCYP00-i, oomcCYP00-ii, oomcCYP00-iii, oomcCYP00-iv, and oomcCYP00-v) ( Figure 5, Supplementary Table S4). OomcCYP00 was highly similar in all orthogroups (using P. infestans and P. sojae CYP sequence as the sample; Supplementary Table S4). As this was the largest orthogroup with proteins often only containing a single CYP domain (Figure 1), it is possible that multiple paralogs were assigned to a common orthogroup. Signal peptides or trans-membrane domains were often found encoded in proteins belonging to clade oomcCYP00-i. Although the majority of CYP domains clustered phylogenetically, there were some instances where clades containing CYP domains were assigned to different orthogroups (i.e., oomcCYP-v, Figure 5 inset). The phylogenetic analysis and annotations supported that downy mildew and Albugo species assemblies do not contain oomcCYP14 and oomcCYP15 cyclophilins. Additionally, two oomcCYP00 clades (oomcCYP00-iv and

Expression of CYPs in Different Life Stages of P. infestans and P. halstedii
Expression of CYPs was characterized in different asexual life stages of the hemibiotroph P. infestans and the biotrophic P. halstedii ( Figure 6). For P. infestans, life stage replicates clustered together, inferring a robust expression profile within biological replicates. The majority of CYPs were expressed in most life stages, except in zoospores, where the most variation between-replicates was observed ( Figure 6). Generally, the highest CYP expression was detected in the mycelia time-point, where plant infection, including appressorium and haustoria formation, takes place [45]. Expression of oomcCYP14 and oomcCYP15 was highest in sporangia and slightly reduced in cleaving sporangia, zoospores, and germ tube forming time-points. Transcription of these genes was greatly reduced in the mycelia (Figure 6). OomcCYP01 and oomcCYP06 were upregulated in the cleaving sporangia stage. OomcCYP01 was identified as closely related to fungal PPIase-1 (Pin1) (e.g., XP_003177293.1 and KZZ96398.1; Supplementary Figure S3). Pin1 participates in the phosphorylation-dependent prolyl isomerization that changes the conformation of its substrates, thus controlling cell cycle progression in fungi [46]. of pathogenicity as well as transporter and carbohydrate-associated domains, when compared to hemibiotrophs [43]. To study the relationship between oomycete CYPs and plant or fungal CYPs, the top 10 plant and fungal CYPs from National Center for Biotechnology Information (NCBI), ranked by percent identity, were added to the alignments, using Phytophthora proteins as queries. Phylogenetic trees were constructed from diverse oomycete, plant, and fungal CYP sequences. In most cases, oomycete orthogroups clustered together, away from plant and fungal CYP sequences (Supplementary Figure  S3). This was not observed for oomcCYP01; proteins annotated in Saprolegnia and Aphanomyces species appeared closer to plant CPYs than other oomycete CYPs. Reciprocal BLAST [44] of P. infestans annotations supported orthology of oomycete proteins with plant and fungal proteins for CYPs belonging to oomcCYP00-iii, oomcCYP01, oomcCYP05, oomcCYP06, oomcCYP07, oomcCYP08, oomcCYP09, and oomcCYP13 (Supplementary Table S5). In addition, oomcCYP00-i, oomcCYP00-ii, oomcCYP03, and oomcCYP04 had the best reciprocal BLAST hits with one of either fungi or plants, but not both, supporting shared ancestry (Supplementary Table S5). Additional domains fused to CYP proteins may have resulted in the top hit identified being non-orthologous, such as for oomcCYP04, a CYP-GRX bigram (Figure 1). For the other eight P. infestans CYP proteins, the reciprocal BLAST hit for fungal and plant results was to other P. infestans CYP proteins (Supplementary Table S5). Only two of these eight had reciprocal BLAST hits when non-oomycete stramenopiles were surveyed (Supplementary Table S5). Therefore, these six protein lineages may be unique to the oomycetes. Interestingly, one of these lineages, oomcCYP14, was not detected in biotrophic oomycete species (Table 1). The analysis of CYP expression during P. halstedii infection revealed that the expression profiles of CYPs between the infection time point (early stage of infection) and the spores time point were almost inverts of each other, except for oomcCYP00-i and oomcCYP07 (Figure 6b), indicating distinct, life stage-dependent expression of each orthologous group in P. halstedii. The sporulation and spore profiles were more similar to one another; the expression of oomcCYP01, oomcCYP04, oomcCYP05, and oomcCYP06 were very similar (Figure 6b). These proteins include CYP bigrams with GRX, RRM, and WD40 (Supplementary Table S3), indicating that these CYP proteins may be less important to establishing an infection. During infection, only oomcCYP02, oomcCYP10, oomcCYP12, and oomcCYP13 were highly expressed, consistent with a role in establishing infection. These proteins were not annotated as encoding additional domains, signal peptides, or transmembrane domains. In P. halstedii, cyclophilins phylogenetically linked to fungal Pin1 (oomcCYP01) had low expression levels during infection, but higher expression in spores, the opposite of what was observed in P. infestans. The difference between P. halstedii and P. infestans suggests that many cyclophilins may have opposite roles in the life-cycle for these two oomycetes.
zoospores, and germ tube forming time-points. Transcription of these genes was greatly reduced in the mycelia (Figure 6). OomcCYP01 and oomcCYP06 were upregulated in the cleaving sporangia stage. OomcCYP01 was identified as closely related to fungal PPIase-1 (Pin1) (e.g., XP_003177293.1 and KZZ96398.1; Supplementary Figure S3). Pin1 participates in the phosphorylation-dependent prolyl isomerization that changes the conformation of its substrates, thus controlling cell cycle progression in fungi [46]. The analysis of CYP expression during P. halstedii infection revealed that the expression profiles of CYPs between the infection time point (early stage of infection) and the spores time point were almost inverts of each other, except for oomcCYP00-i and oomcCYP07 (Figure 6b), indicating distinct, life stage-dependent expression of each orthologous group in P. halstedii. The sporulation and spore profiles were more similar to one another; the expression of oomcCYP01, oomcCYP04, oomcCYP05, and oomcCYP06 were very similar (Figure 6b). These proteins include CYP bigrams with GRX, RRM, and WD40 (Supplementary Table S3), indicating that these CYP proteins may be less important to establishing an infection. During infection, only oomcCYP02, oomcCYP10, oomcCYP12, and oomcCYP13 were highly expressed, consistent with a role in establishing infection. These proteins were not annotated as encoding additional domains, signal peptides, or transmembrane domains. In P. halstedii, cyclophilins phylogenetically linked to fungal Pin1 (oomcCYP01) had low expression levels during infection, but higher expression in spores, the opposite of what was observed in P. infestans. The difference between P. halstedii and P. infestans suggests that many cyclophilins may have opposite roles in the life-cycle for these two oomycetes. The proteins absent in the biotrophic P. halstedii were poorly expressed in the mycelia of P. infestans, consistent with a non-critical role in infection. These proteins were commonly expressed in later stages of infection, including sporangia formation and cleavage. This expression pattern coincided with the necrotic stage of P. infestans infection, which was absent in downy mildews. The expression profile of these genes in P. infestans may be part of a transcription-level molecular signature for the onset of the hemibiotrophic phase. The absence of these genes in the genome of P. halstedii possibly reflects the lack of a selective pressure to maintain them during the evolution of its biotrophic life style.

Identification of CYPs from Oomycete Species
Oomycete genomes and annotations were downloaded from their respective sources (Table 2). InterProScan v5.33 [47] was run on the entire dataset and queried for proteins encoding a CYP domain (PF00160). Additional domains encoded in the same proteins were also identified. Signal peptide and transmembrane predictions were performed with SignalP4.0 [48] and TMHMM Server v2.0 [49], respectively. All annotations were run through OrthoFinder v2.2.1 [50] and queried for orthogroups containing CYPs. Consensus domain architecture for each orthogroup was defined and proteins that deviated from this consensus were subject to further manual inspection.

Multiple Sequence Alignment and Phylogenetic Analysis
Phylogenetics was used to investigate the evolutionary relationships among the oomycete CYPs. Coordinates for the CYP domains were obtained from searches against the NCBI conserved domain database [17] and InterProScan [47] using PF00160 to filter the latter [52]. The sequences were manually extracted. The protein sequences of the CYP domains were aligned using MAFFT v7.245 [53]. Consensus protein sequences were obtained from alignments using the CLC Genomics Workbench v 8.0.1 (https: //www.qiagenbioinformatics.com; https://secure.clcbio.com/helpspot/index.php?pg=kb.page&id=78). Conserved amino acid motifs were identified using the MEME v5.0.5 suite (http://meme-suite.org/) [54] with default parameters (zero-ordered model of sequences, minimum width equal to 6 and maximum width equal to 50). P. infestans sequences were queried against the NCBI nucleotide (nt) database to independently identify cyclophilins of plants (taxid: 3193), animals (taxid: 33208), fungi (taxid: 4751), Rhizaria (taxid: 543769), Alveolata (taxid: 33630), and stramenopiles (taxid: 33634), using taxid numbers to reduce the database size. An additional search of stramenopiles was conducted excluding oomycetes (taxid: 4762). The top 10 non-redundant plant and fungal hits were aligned with each oomycete orthogroup. A maximum likelihood protein tree was produced using RAxML v8.2.9, with 1000 bootstraps and a GAMMA substitution model [55]. Alignments and trees were visualized using Geneious version R10 [56]. Reciprocal BLAST of the top fungal and plant hit was carried out against the P. infestans assembly to infer support for orthology.

Expression Analysis of Phytophthora infestans and Plasmopara halstedii
Previously published transcriptome data of P. infestans and P. halstedii (SRR5179148 to SRR5179157 and ERR583683 to ERR583685) were used to investigate the transcription of CYPs at distinct asexual life stages. Reads were mapped to their respective assembly using STAR v2.6.0c (-quantMode GeneCounts) [57], trimmed means of M normalization was applied to the mapped reads [58,59], and they were analyzed in RStudio [60]. Heatmaps for life stage specificity of the expression of CYP proteins of P. infestans and P. halstedii were generated in RStudio using tidyverse, ggplot2 [61], gplots [62], and edgeR [63].

Conclusions
We conducted a comprehensive sequence analysis of the CYPs encoded in the genome assemblies of 23 oomycetes, from 21 species. The oomycete CYPs were clustered into 16 orthogroups, largely supported by phylogenetic analysis of the CYP domains. Six CYP orthogroups included proteins that formed bigrams with a diverse range of domains indicative of a wide diversity of functions, which may include virulence. Significantly, the CYP-FBKP bigram (oomcCYP03) was found to be unique to stramenopiles and alveolates, and was not detected in Rhizaria, Plantae, or Opisthokonta. The function of these proteins is yet to be elucidated. Variable transcription of every CYP encoded by the hemibiotroph P. infestans and the biotroph P. halstedii was detected at different times throughout the course of infection. The differential expression of CYPs during an infection cycle in these oomycetes is consistent with CYPs playing diverse functions including, but not exclusively, pathogenicity.  Author Contributions: Y.Z. led the data curation, data analysis, investigation, and writing-original drafting of the manuscript; K.F. made significant contributions to the conceptualization, supervision, and writing-reviewing and editing; R.H. contributed to the data analysis and writing-reviewing and editing. R.M. contributed to writing-reviewing and editing. R.Y. contributed to writing-reviewing and editing. All authors have read and approved the final draft. All authors have read and agreed to the published version of the manuscript.