Background

Phyllosticta citricarpa McAlpine (Synonym: Guignardia citricarpa Kiely) is the fungus responsible for causing citrus black spot (CBS), which is one of the most important diseases affecting the citrus industry worldwide [1, 2]. This disease was first described in Australia, where it caused considerable losses for sweet orange growers [3]. Since then, CBS has seriously affected citrus crops in countries in Africa, Asia, South America, and North America, especially in Argentina, United States and Brazil [1, 2, 4,5,6].

CBS affects almost all commercial varieties of citrus with the main symptom associated with this disease being the development of hard spot lesions in the fruit peel [7]. Advanced stages of the disease lead to maturation and early fall of the fruit [5]. In addition, the affected fruits have their appearance depreciated, making them unsuitable for the fresh fruit market, and therefore the costs associated with chemical control of the disease are significant [1, 7,8,9].

Most of the species in the genus Phyllosticta are plant pathogens of a wide range of hosts [10, 11] and although P. citricarpa is pathogenic, other endophytic and saprophytic species have also been reported for citrus [12,13,14,15]. P. capitalensis is among the endophytic species that lives within citrus and other hosts with a wide geographic distribution [10, 16, 17]. P. capitalensis is commonly found to be associated with lesions in plants; from an economic point of view, this association can be very negative since P. capitalensis is normally confused with the pathogen that is actually responsible for causing the disease. This is an even worse problem if the actual pathogen is a quarantine organism like P. citricarpa that has phytosanitary restrictions, being classified as quarantine A1 in the European Union and A2 in the United States [4, 18]. The similarity between P. citricarpa and P. capitalensis is so close that CBS was erroneously reported in New Zealand [2, 19], with the endophytic species being identified as the species responsible for causing the disease [20]. Due to the high similarity between these species, identifications can only be made at the molecular level [21, 22].

In recent years, the number of works using comparative genomic analysis to understand the genetic basis of the lifestyle of pathogenic and endophytic or symbiont organisms has grown considerably [23,24,25,26]. In this study, the genome of the citrus pathogenic species P. citricarpa was compared to the genome of P. capitalensis, the endophytic species, in order to identify genetic differences that could help understand their different lifestyles.

Results and discussion

Overview of genome sequencing and categorization

The Illumina sequencing of P. citricarpa and P. capitalensis genomes generated a total of 179,880,616 and 148,831,020 paired-end reads, respectively, with more than 90% showing Phred quality > 20 (Q20) for both species. The CG content was slightly lower for P. citricarpa (48.72) compared to P. capitalensis (51.43).

A de novo assembly was done for both genomes, generating drafts of 19,143 contigs in 32.6 Mb for P. citricarpa and N50 of 3049. The assembly of P. capitalensis reads resulted in 11,080 contigs and 33.2 Mb with a N50 of 4925. A Genome Shotgun project has been deposited at DDBJ/EMBL/GenBank under the BioProject number PRJNA486917.

The contigs of both fungi were used for the prediction of coding sequences using Botrytis cinerea, an ascomycete necrotrophic plant pathogen with a broad host range [27] as a reference in Augustus [28]. For P. citricarpa, 15,206 proteins were identified, while for P. capitalensis, the total was 14,797. These proteins were annotated using blastp [29] to find similarities with subjects in the Protein database of the National Center for Biotechnology Information (https://www.ncbi.nlm.nih.gov/protein/) (Additional files 1 and 2). Further characterization of the putative proteins was performed in Blast2GO [30], and when they were distributed in different categories, i.e., cellular component, molecular function, and biological process (level 2).

Even though the distribution of the genes of P. citricarpa and P. capitalensis showed the same pattern for most of the subcategories, with a few additional genes in the latter compared to the former, there were some differences that could be representative for understanding the behavior of both organisms (Fig. 1). In the molecular function category, the binding and catalytic subcategories presented the highest percentage of genes in both genomes while the protein tag subcategory was present only in P. capitalensis (Fig. 1).

Fig. 1
figure 1

Categorization of all genes of P. citricarpa and P. capitalensis. The genes were automatically classified based on GO (Gene Ontology) and were distributed according to their function performed in fungus cells. The genes were distributed into three categories: cellular component, molecular function and biological process annotation. P. capitalensis (red bars) and P. citricarpa (blue bars)

Full size image

For biological processes, the most representative sub categories were metabolic processes, cellular processes, establishment of localization, and localization. The growth, multicellular organismal processes and viral reproduction subcategories were unique for the endophytic fungus and in addition, it also showed a higher percentage of genes in the death and reproduction subcategories (Fig. 1).

BUSCO analysis confirms that the genomes are not complete with 502 and 765 proteins missing for P. capitalensis and P. citricarpa, respectively (Additional file 3) in a total of 3156 proteins. Other genomes of Phyllosticta spp. are much more complete and therefore were used for validating the results obtained with our genomes.

Phylogenetic analysis of Phyllosticta species

To verify the evolutionary relationship of different Phyllosticta species, we used the two genomes sequenced in this work together with seven other genomes available in public databases (P. capitalensis strains: CBS128856, Gm33, and LGMF01; P. citricarpa strains: CBS141350, Gc12, and LGMF06, P. citribraziliensis strain CBS100098, and P. citrichinaensis strain CBS130529). The phylogenetic tree obtained using the 3185 single copy-ortholog genes (Fig. 2) confirms that P. citricarpa and P. citriasiana are very closely related and that P. capitalensis is distant from this group, confirming the result obtained with ITS, LSU, TEF1, ACT and GPDH sequences alignment [10, 31]. P. capitalensis and P. citribraziliensis are endophytes in Citrus while P. citricarpa and P. citriasiana are pathogenic. P. citrichinaensis, on the other hand, causes minor disease symptoms [32] and therefore, pathogenicity in Citrus may be related to the phylogeny of Phyllosticta species at least in part since the two pathogenic species cluster together.

Fig. 2
figure 2

Maximum likelihood tree generated from genomic sequence analysis of the Phyllosticta species. The maximum likelihood phylogenetic analysis was performed using 3.185 single copy-ortholog genes with support values obtained for 1000 bootstrap replicates using the JTT + F + R4 model. Bootstrap support values for maximum likelihood (ML) are shown above the branches. The root of ML tree was inferred the midpoint rooting method

Full size image

Major functional categories enriched in P. citricarpa compared with P. capitalensis

To better evaluate the differences at the genetic level between one pathogenic and one endophytic species, the sequences from P. citricarpa and P. capitalensis were used in a functional enrichment analysis to verify potential molecular mechanisms associated with their interaction with citrus (Fig. 3).

Fig. 3
figure 3

PHANTHER GO-Slim biological process subcategories enriched in P. capitalensis LGMF6 and P. citricarpa LGMF01. List of subcategories obtained for Phyllosticta species in enrichment analysis using the gene set enrichment analysis. Each category is associated with the number of genes from each fungi in comparison to the total number of genes of the reference organism (Saccharomyces cerevisiae) with the same GO term. Note that the same gene may have more than one associated ontology

Full size image

It was observed that enrichment analysis resulted in a difference between P. citricarpa and P. capitalensis. For the Biological Process category, P. citricarpa showed a higher number of enriched subcategories, GO:0044848, GO:0051704, GO:0000003, and GO:0032502, which are all related with growth and development. This result may indicate that P. citricarpa presents a more complex living compared to P. capitalensis, which could result from the distant relationship in evolution. However, even though it is an appealing hypothesis for testing, it would be necessary to evaluate other sequenced genomes that show closer phylogenetic relationship to P. capitalensis. Other works reporting the colonization of P. capitalensis and P. citricarpa in Citrus are also missing, in particular those of global gene expression.

Gene families expansion in Phyllosticta

We also did an analysis of gene families expansion in the genomes of Phyllosticta species. This analysis was done based on the phylogeny with the five different species of Phyllosticta used in this work (Fig. 4). The analysis of gene family expansion shows a high number of families being expanded from ancestral 5 in relation to 17, in the P. capitalensis clade (Additional file 4). This seems to be much more related to the species evolution than the lifestyle of the organisms since P. citribraziliensis, a Citrus endophyte, is not in the same clade of P. capitalensis. In addition, a family of HET-domain containing proteins is expanded in all the P. capitalensis analyzed, being a more recent event in evolution of the species. HET proteins were identified as important for vegetative incompatibility, in the formation of heterokaryon [33]. These proteins induce cell death avoiding the formation of heterokaryon [34]. Therefore this genomic data make us believe that P. capitalensis may be more effective in avoiding heterokaryon formation than the other Phyllosticta species.

Fig. 4
figure 4

Phylogenetic inference and gains/losses of gene for Phyllosticta species. Number of gene gains and losses are shown on tree branches and indicated with + and −, respectively

Full size image

In the other clade, fewer families were expanded and besides, most of them encodes uncharacterized proteins (Fig. 4). However it is worth mentioning the expansion of a family of putative chitin recognition proteins that happens just in the P. citricarpa genome reported in this work (LGMF06). LysM domain proteins bind chitin and are important for fungus defense against the host response [35]. Because the expansion occurs just in the isolate sequenced in our lab, we suggest that the environment may have an importance in modulating the genome plasticity in P. citricarpa, as known for other filamentous fungi [36].

Comparative genomics

The phylogenetic analysis with all the genes confirmed that indeed P. citricarpa and P. capitalensis are distant species as previously reported. Therefore, the changes observed in their genomes may result from the evolution of the species, including gain and loss of genes. Even distant, a comparative genomics analysis is appealing because of the different lifestyle and it was carried out to identify genes exclusively found in the genomes of P. citricarpa LGMF06 and P. capitalensis LGMF01. A reciprocal blastp analysis was carried out using the protein sequences encoded in both genomes. The exclusive proteins were considered those that either had a positive hit with an E-value above 1 e− 30 or resulted in a “no hit”. These proteins were reciprocally re-evaluated in blastp using the proteomes from P. citricarpa CBS141350 and P. capitalensis CBS128856 to make sure that the differences were not a problem resulting from the incompleteness of the genomes (Table 1; Additional files 5 and 6). These genes were further evaluated in order to better understand the pathogenicity of P. citricarpa.

Table 1 Number of exclusive proteins for P. citricarpa and P. capitalensis after analyses by blastp
Full size table

Genes related to pathogenicity

To identify putative genes involved in pathogenicity, we analyzed the P. citricarpa and P. capitalensis-exclusive genes for similarity with subjects in the pathogen-host interaction gene database (PHI-base).

Of the 2896 gene sequences of P. citricarpa (Additional file 5), 123 were identified as having similarity to putative PHI proteins (Additional file 7). For P. capitalensis this total was 210 out of 4164 exclusive genes (Additional files 6 and 8).

Among the 123 genes of P. citricarpa and 210 of P. capitalensis, 18 genes are common to both fungi (Fig. 5; Additional file 9). Therefore, 70 genes were unique in the pathogenic species, while 95 were exclusive to the citrus endophytic species (Fig. 5; Additional file 9).

Fig. 5
figure 5

Pathogenicity gene orthologs in P. citricarpa and P. capitalensis in comparison to other fungal species using PHI-base data. Distribution of putative P. citricarpa, P. capitalensis and common genes involved in pathogenicity

Full size image

Among the PHI exclusive genes of P. citricarpa, genes involved in fungus infection, toxin synthesis, effector protein secretion system (important for the pathogenicity of pathogenic organisms), signaling, detoxification systems, virulence factors and transcription factors were identified (Additional file 5). Because the role of these genes in pathogenicity and virulence is already known for other interactions, we assume that they are important in the development of CBS in citrus [37].

P. capitalensis also presents genes involved in fungal infection, cell wall degradation, growth, cell death regulation and signaling (Additional file 6). Although this fungus is an endophyte in citrus and a weak plant pathogen with a worldwide distribution presently known from 70 plant families [17], it is the causal agent of the leaf spot in the orchid Bifrenaria harrisoniae [14], which explains the presence of pathogenicity genes in the genome of this fungus.

The molecular components of Phyllosticta spp. pathogenicity in citrus are poorly known. In fact to our knowledge there is one report of identification of putative proteins involved in pathogenicity of P. citricarpa in citrus [38], in which the authors suggest a possible major role of pectinases for this organism. Therefore, this work opens new perspectives in understanding the pathogenicity of these Phyllosticta species.

Analysis of Carbohydrate-Active Enzymes (CAZymes) in P. citricarpa and P. capitalensis.

The function of CAZymes is breakdown, biosynthesis or modification of glycoconjugates, oligo- and polysaccharides. These enzymes are produced by phytopathogens and play a central role in the breakdown and synthesis of plant cell walls during host-pathogen interactions [39]. Due to the importance of CAZymes in fungal pathogenicity, we examined the presence of these enzymes in both fungi, comparing and evaluating the exclusive genes of each species.

The search for CAZymes revealed 23 in P. citricarpa, which were distributed across CBMs (carbohydrate-binding modules), CEs (carbohydrate esterases), GHs (glycoside hydrolases), GTs (glycosyl transferases) and AAs (auxiliary activities) families (Fig. 6, Table 2). It is also noted there was diversity within each family (and subfamily), mainly GTs and GHs (Table 2).

Fig. 6
figure 6

Comparative analysis of CAZymes in P. citricarpa and P. capitalensis. Number of CAZymes in P. citricarpa compared to P. capitalensis and their distributions among different families. CBM, carbohydrate-binding module; CE, carbohydrate esterases; GHs, glycoside hydrolases; GTs, glycosyl transferases; AAs, auxiliary activities

Full size image
Table 2 Overview of CAZyme and number of genes in each CAZyme category. Color tones differentiate families from enzymes
Full size table

The diversity within the families in the pathogenic species was also higher in relation to endophytic species (Fig. 6, Table 2). P. capitalensis presents 44 genes coding for CAZymes distributed among CBM, CE, GH, GT and AA families, the latter being the only one to present more genes in this species (Fig. 6, Table 2).

We analyzed which of these CAZymes may be secreted by P. citricarpa and P. capitalensis. Of the 128 enzymes of the pathogenic species, 4 are possibly secreted: 1 of the GT family (34), and 3 of the GH family (16, 17, and 37) (Additional file 10). The endophytic species presented 1 GBM (63) secreted, 3 enzymes of the CE family (4, 5, and 6) and 5 proteins of the family GH (16, 18, 3, 43, and 55), totaling 9 CAZymes secreted (Additional file 10). The results show that P. capitalensis presents a greater amount of these enzymes in relation to P. citricarpa, besides presenting a higher diversity of representatives of the different CAZymes families. Since these two species present high similarity and identification can only be made at the molecular level [21, 22], the divergency between the two species regarding CAZymes may be an important distinguishing characteristic, allowing the development of new markers for differencing the two species, like antibodies for instance.

Most of the enzymes secreted by P. citricarpa are from the GH family. These are involved in hydrolysis of glycosidic bonds between or within carbohydrate molecules [39].

The fact that CAZymes are responsible for the breakdown of cell wall components suggests they are strictly related to the successful infection process of the fungus. The number of CAZymes is higher in pathogenic fungi [26]. These results are in agreement with those found by [38] where it was verified that P. citricarpa produces more endoglucanases in relation to P. capitalensis, in addition to producing more amylases and pectinases. The authors concluded that these differences could be associated with differences in pathogenicity for citrus plants. Similar to classical necrotrophic fungi, such as Botrytis cinerea and Sclerotinia sclerotiorum, the P. citricarpa genome is predicted to encode a large number of CAZymes involved in plant cell wall degradation [40, 41]. However, because P. capitalensis has more CAZymes in its genome compared to P. citricarpa, we could not do any association of the number of these enzymes with the lifestyle of the fungi. But the possibly secreted CAZymes of P. citricarpa is different from the ones from P. capitalensis, which could have an influence in the pathogenicity of the fungus.

All of our analyses could not give enough support to state that the different lifestyle could be explained by the genome differences observed between P. citricarpa and P. capitalensis. Indeed there is a debate on whether endophytes are pathogenic or not, or may turn pathogenic under certain conditions [42].

Conclusion

There are many different genomes sequenced for Phyllosticta species. In this work we presented the partial genomes of two isolates, one from P. citricarpa and another one from P. capitalensis, which present different lifestyles and are also distantly related in evolution. Different features were identified in each genome and could be used for understanding a little more about the biology on the different fungi. Nevertheless some of these features seem to be related to the phylogeny of the group and it would be very interesting to have more genomes available from other closely related species in special to P. capitalensis to confirm these findings. However, our analyses showed interesting trends like the HET-domain containing proteins expansion in all the P. capitalensis strains or the expansion of a family of putative chitin recognition proteins in the P. citricarpa genome reported in this work (LGMF06). These findings seem to be related to the evolution of the species detected through the genome analyses but drivers are acting in the genome of the different species that could leads to the formation of new species in the future.

Methods

Fungal isolates, culture condition and DNA isolation

The fungi isolates used for genome sequencing are from the biological bank of the Laboratory of Microorganisms (LabGeM) at the Federal University of Paraná, Curitiba, Paraná, Brazil. The pathogenic species P. citricarpa (LGMF06) was isolated from CBS lesions in fruits from a plant grown in the State of São Paulo, Brazil, and tested for pathogenicity according to [43]. The endophytic species P. capitalensis (LGMF01) was isolated from asymptomatic fruits from a plant also grown in the State of São Paulo, Brazil. These fungi were grown in liquid Citrus Medium Fabris-Nishimura (CFN) [44] and incubated under agitation (65 rpm) at 25 °C for 7 days in the dark. The mycelium was removed from the medium, washed with distilled water, macerated under liquid nitrogen and stored at − 80 °C for later use in the DNA extraction procedure [12].

Genome sequencing, assembly and annotation

A total of 10 μg of DNA from each fungus (P. citricarpa and P. capitalensis) was sent to Macrogen Inc. (South Korea) for sequencing using the HiSeq 2000 platform (Illumina Inc.). All procedures were performed according to Illumina’s protocols. Paired-end fragments were generated with 101 base pairs (bp). One lane was used per library. The results were sent in fastq format. De novo assembly was carried out using CLC Genomics Workbench software (https://www.qiagenbioinformatics.com/products/clc-genomics-workbench/) with default parameters.

The prediction of open read frames (ORFs) was conducted using Augustus software v3.2 [32] with Botrytis cinerea (Botryotinia fuckeliana) as the model species. The annotation and categorization of ORFs were automatically carried out using Blast2GO with an E-value cut-off ≤10− 6 [34]. The tool called WEGO (Web Gene Ontology Annotation Plot) was used for plotting GO annotation GO [45]. The completeness of the gene prediction was assessed using BUSCO v3 [46] software package and pezizomycotina_odb9 gene set.

Phylogenomic reconstruction

To reconstruct a phylogenetic hypothesis for nine Phyllosticta species (P. capitalensis strains: CBS128856, Gm33, LGMF01; P. citricarpa strains: CBS141350, Gc12, LGMF06, P. citribraziliensis strain CBS100098, P. citrichinaensis strain CBS130529), in total, we used 3.185 single copy-ortholog genes identified with OrthoFinder [47]. Proteins sequences were aligned using MAFFT v7.271 [48] and trimmed using TrimAL v1.4 [49]. The Maximum Likelihood (ML) phylogenetic analysis was performed using IQtree v1.6.5 [50], with support was assessed via 1000 bootstrap replicates using the JTT + F + R4 model. The model of nucleotide sequence evolution was inferred using software ModelFinder [51] with both the Akaike information criterion (AIC) [52]. The root of ML tree was inferred the midpoint rooting method [53].

Gene set enrichment analysis

To realize a functional enrichment analysis of P. citricarpa and P. capitalensis, protein sequences was submitted to a search of orthologous groups based on reciprocal best hits between each fungus and the yeast reference organism Saccharomyces cerevisiae. The search of orthologous groups was performed using the Orthofinder v2.2.7 software. Subsequently, we have selected groups present only in the genomes of P. capitalensis/Saccharomyces and P. citricarpa/Saccharomyces. The set of genes retrieved was submitted for GO enrichment analysis in the PANTHER classification system [54] and Saccharomyces Genome database (SGD) [55].

Expansion and retraction of gene families

We used the maximum-likelihood approach implemented in BadiRate v1.35 [56] to estimate the gene birth, death, and innovation rates in the Phyllosticta gene families during the evolution. In brief, we first inferred orthologous groups based on reciprocal best hits within and between gene families of nine Phyllosticta genomes using the Orthofinder v2.2.7 software [47]. Single-copy orthologists and lineage-specific gene families were extracted using using in-house Perl script. The ML tree previously inferred from the 3.185 single copy orthologs and orthologous gene groups were utilizeded by the program as input and with the following parameters: -anc 1 -bmodel FR -model BDI -ep ML. For each orthologous group, gene gain and loss events were counted from the number of members at internal nodes inferred by maximum likelihood under the BDI stochastic model [57], assuming that each branch has its own specific turnover rates.

Comparative genomics

Comparative genomics analysis between P. citricarpa and P. capitalensis were conducted using only exclusive sequences for each species. These sequences were obtained after different rounds of BLAST [29]. Initially, a blastp analysis was carried out with unique coding sequences extracted from the genome of each organism using Augustus against the proteins from the genomes of the other species. The results were filtered in order to just get hits presenting E values >1e-50. These subsets of coding sequences were then evaluated reciprocally against proteomes available at the JGI (Phyllosticta citricarpa, CBS 127454 v 1.0; Phyllosticata capitalensis, CBS 128856, v 1.0) using blastp. Once again an E value filter (1e-50) was used and only the sequences showing results below the cutoff were kept for further analyses, being considered as the exclusive sets for each species.

PHI-base analyzes of the exclusive genes of P. citricarpa and P. capitalensis

The P. citricarpa and P. capitalensis, exclusive gene sequences for each, were used to investigate whether they had been verified to be pathogenic genes using annotations from the PHI-base using blastp (e-value < 0.001) [58].

Analysis of carbohydrate-active (CAZy) enzymes in P. citricarpa and P. capitalensis

The CAZymes are involved in plant cell wall degradation. For this reason, the genes of both fungi, pathogenic and endophytic, were subjected to CAZy annotation using dbCAN [59], which is based on the CAZy (Carbohydrate-Active Enzyme) database classification [39].

Secretome

To verify which proteins are secreted the following pipeline was used: the secreted proteins were identified using SignalP 3.0 [60], which detects the presence and location of signal peptide cleavage sites in proteins. In addition, we also used SecretomeP 1.0f [61] to predict protein secretion by nonclassical pathways. Afterwards, TMHMM [62] was used to identify the sequences that have transmembrane domains that were excluded from further analyses. Therefore, the sequences presenting signal peptide and no transmembrane domains were selected. These sequences were then submitted to Cello 2.5 [63] analysis to predict cellular localization of the sequences.