Morphological, phylogenetic, and genomic evidence reveals the causal agent of thread blight disease of cacao in Peru is a new species of Marasmius in the section Neosessiles, Marasmius infestans sp. nov.

Sampling and morphological analysis

Specimens were obtained during an expedition into the Imaza District (4°47′09.4″S 78°16′51.6″W) in the department of Amazonas, Peru (Figure 1), during August–September 2022 (Table 1). Collections were performed under the authorization N° AUT-IFL-2021-052 granted by the Peruvian National Forestry and Wildlife Service Agency - SERFOR. Morphological studies were conducted in the Plant Health Laboratory of the National University Toribio Rodriguez de Mendoza de Amazonas, Peru (UNTRM-A). The macroscopic characterization was made from fresh pilei. We described the color, shape, and size using a stereomicroscope SMZ18 (NIKON, Tokyo, Japan). The microscopic description was made from fresh and dried material. First, a tiny piece of dried material was carefully sectioned and placed onto a microscope slide. We applied 95% ethanol to the sample for 30 seconds, and then a drop of distilled water was added. The slides were dried out and 5% KOH was applied. Additionally, ethanol-washed sections were mounted on Melzer’s reagent as in.³ We placed a cover slip onto the sample and observed it under an OLYMPUS DP74 (Tokyo, Japan) microscope. Dimensions of microstructures were reported with the ranges between the 5^th and 95^th percentile, with extreme values in parentheses, as in.¹⁶ Additionally, for basidiospores, the arithmetic mean ± standard deviation (x_m) and the quotient of the length by the width, expressed as range (Q) and arithmetic mean (Q_m), were calculated. The dimensions of basidia and basidiospores were measured with at least thirty individual structures, i.e., n ≥ 30. Finally, pure cultures were obtained from rhizomorph tissues by the hyphal tip technique. The holotype specimen (TAIM04) was deposited in the KUELAP herbarium of UNTRM-A under voucher number KUELAP-2940.

Figure 1. Collection points of specimens in Imaza district, Bagua province, Amazonas department, Peru.

Phylogenetic analysis

DNA extractions were performed from pure cultures using the Wizard® genomic DNA purification kit (Catalogue number A1120; Promega, Wisconsin, USA). DNA was quantified with the BioSpectrometer® Basic (Eppendorf, New Jersey, USA), and diluted to 0.5 ng/μl for PCR reactions. The internal transcribed spacer region (ITS1-5.8S-ITS2, or simply ITS), the large subunit (LSU) rDNA gene, and partial fragments of the genes Elongation Factor 1-α (EF1α) and the largest subunit of RNA polymerase II gene (RPB1) were amplified following.¹⁷ The amplified PCR products were Sanger sequenced at MACROGEN (Seoul, South Korea). LSU, EF1α, and RPB1 sequences from the West African TBD-causing specimen GHA37 were retrieved from the recently published genome.¹⁵ In addition, we included the mitochondrial cytochrome oxidase I gene (COX1) from isolate INDES-AFHP31, which was retrieved from the mitogenome generated in this study, and from other phylogenetic-related species.¹² The introns of COX1 sequences were removed as in.¹⁸ Most other sequences used in phylogenetic analyses were obtained from other relevant literature.^{3,4,11,13,14,19–25} We also included unpublished sequences available in NCBI from M. tenuissimus-phylogenetically related isolates C2/06 and C2/33 (accession numbers KM246261 and KM246277 for ITS, and KM246066 and KM246082 for LSU, respectively).

Generated sequences were edited and assembled with Sequencher v.5.4 (Gene Codes, USA). Once all sequences were gathered, they were aligned with MUSCLE²⁶ implemented in MEGA-X,²⁷ and concatenated with SeaView 4.7.²⁸ We used jModelTest v2²⁹ to identify the most appropriate nucleotide evolution models under the Akaike information criterion. Phylogenetic analysis was performed with IQ-TREE v2³⁰ which implements the Maximum Likelihood algorithm,³¹ in the CIPRES Science Gateway v3.3 portal.³² The phylogenetic trees were visualized and edited with FigTree.³³

Genomic analysis

DNA extraction, sequencing, and assembly

Total DNA extraction was performed using the Wizard Purification Kit (Promega Corp., Madison, Wisconsin) following the manufacturer’s instructions. For nuclear and mitogenome sequencing, we used the strain INDES-AFHP31 from Northern Peru, previously reported.¹⁴ A paired-end sequencing library was constructed using the TruSeq Nano DNA Kit, according to the manufacturer’s instructions. The library was sequenced on an Illumina NovaSeq 6000 platform in paired-end, 2 × 150 format.

The raw reads were checked by FastQC v.0.11.9. Also, quality trimming (Phred Q > 25) and remotion of adapters were conducted with Trimmomatic v0.36³⁴ and TrimGalore software,³⁵ respectively. Jellyfish v.2.³⁶ was used for k-mer counting, and Genome Scope v1.0.0³⁷ for assessing genome size, repeat content, and heterozygosity rate. This analysis involved utilizing the output of Jellyfish and the count of 17-mers for k-mer analysis. Additionally, k-depth estimation was performed to identify a predominant single-peak pattern in the frequency distribution analysis of k-mers.

De novo assembly was performed with two assembly algorithms: SOAPdenovo2 v.2.04,³⁸ and Masurca v.4.0.6.³⁹ Next, we used QUAST v.5.2.0⁴⁰ to evaluate the statistics of assemblies. The assembly validation process employed two distinct methods. First, the filtered paired-end Illumina reads were realigned to identify any errors in the assembly. This was accomplished using Bowtie2 v.2.4.2⁴¹ and SamTools v.1.7⁴² software. Second, the completeness of the assembly was evaluated using the Agaricales-specific profile of the BUSCO strategy.⁴³ To identify vector contamination, we employed VecScreen JCVI (https://github.com/tanghaibao/jcvi), which utilizes the Univecdatabase (https://ftp.ncbi.nlm.nih.gov/pub/UniVec/). We also performed a BLAST v.2.2.26⁴⁴ analysis, mapping the scaffolds against the nt/nr NCBI database (found at https://www.ncbi.nlm.nih.gov/). Following this, any contaminated scaffolds and vectors were eliminated before submitting the remaining data to the NCBI database. This assembly has been deposited at DDBJ/ENA/GenBank under the accession: JANHQD01.

Nuclear genome annotation

Genome annotation was performed using Funannotate v1.8.12.⁴⁵ Repetitive elements were soft-masked by TANTAN.⁴⁶ Ab initio gene prediction was carried out using AUGUSTUS v.3.3.3,⁴⁷ GlimmerHMM v.3.0.4,⁴⁸ and SNAP⁴⁹ trained with alignments of the BUSCO agaricales_odb10 dataset,⁴⁶ and with GeneMark-ES v.4.69⁵⁰ self-trained on the repeat-masked genome sequence. Protein evidence from the UniProt/SwissProt database⁵¹ was mapped to the genome using DIAMOND v.2.0.15⁵² and Exonerate v.2.4.0.⁵³ Finally, EVidenceModeler v.1.1.1⁵⁴ was used to generate consensus gene models based on all the above ab initio and evidence-based gene models. The tRNA genes were identified with tRNAscan-SE v.2.0.9.⁵⁵ Functional annotations were assigned by similarity to UniProtKB (2021_02),⁵¹ InterPro v89.0,⁵⁶ Pfam v.34.0,⁵⁷ EggNOG v.5.0,⁵⁸ BUSCO (agaricales_odb10 dataset),⁴³ dbCAN v.10.0,⁵⁹ and MEROPS v.12.0.⁶⁰ Phobius⁶¹ and SignalP v.5b⁶² were used to predict transmembrane topology and signal peptides, respectively. DeepLoc 2.0⁶³ was used to determine protein localization. Effectors were predicted by EffectorP 3.0⁶⁴ based on the subset of extracellular proteins. AntiSMASH v.6.0⁶⁵ was used to predict secondary metabolite gene clusters (SMGCs) and secondary metabolite Clusters of Orthologous Groups (smCOGs).

Mitochondrial genome annotation

The mitochondrial genome was confirmed using the default Geneious Prime 2023.1 setup (Biomatters Ltd., Auckland, New Zealand) and GetOrganelle v1.7.6.1.⁶⁶ Genes were annotated with MITOS,⁶⁷ MFannot⁶⁸ and manually confirmed with ORFfinder available in NCBI, and tRNAscan-SE 1.21,⁶⁹ finally adjusted in Geneious. A physical map of the mitogenome was created with OGDRAW v 1.2.⁷⁰ Our results were compared to the ones reported for another published M. aff. tenuissimus mitogenomes.¹²

Sampling and morphological analysis

We collected new cacao tissues infected by TBD-bearing fruiting bodies of the pathogen in Imaza province, department of Amazonas, in Northern Peru. Besides AFHP-31 from a previous study,¹⁴ and for which no basidiocarps and only pure agar culture were preserved, two additional specimens were included in this study: TAIM-04 and AFHP-101. The presence of white rhizomorphs, and white pilei no larger than 5 mm in diameter, smooth and non-intervenose are macro morphology hallmarks of the TBD- causing specimens in Amazonas, Peru (Figure 2). In terms of micromorphological features, these specimens produce ellipsoid and smooth basidiospores with dimensions of (6.2-)7.0-8.6(-8.8) × (3.7-)3.8-5.0(-5.2) μm; they also produce Siccus-type cheilocystidia and pileipellis broom cells; cheilocystidia are cylindrical slightly narrower at the base, and pileipellis broom cells are ovoid to globose shape (Figure 3). The examination of tissues in Melzer’ reagent revealed that most micro structures are inamyloid, except for caulocystidia which reacted in a strongly dextrinoid manner (Figure 4). All these morphological features provided the first point of evidence that the cacao TBD-causal agent in Peru is a new species.

Figure 2. Macroscopic features Marasmius infestans sp. nov. (TAIM 04, holotype).

a) Small basidiocarps growing on decomposing cacao leaves, b) Mature basidiomata, c) Young basidiomata, d) White-rhizomorphs colonizing cacao stem. Photographed by A. F, Huaman-Pilco. Scale bar in b) and c) = 1 mm.

Figure 3. Microscopic features of Marasmius infestans sp. nov. (TAIM 04, holotype), a) Basidia, b) Basidiospores, c) Siccus-type broom cells of Pileipellis, d) Cheilocystidia of Siccus-type broom cells, e) Caulocystidia.

Scale bar = 20 μm. Drawings by T. A. Ramos-Carrasco.

Figure 4. Melzer’s reagent examination of Marasmius infestans sp. nov. (TAIM 04, holotype), a) Inamyloid basidiospores, b) Inamyloid pileipellis broom cell, c) Inamyloid pileus trama hyphae, d) Strongly dextrinoid caulocystidia, e-f) Inamyloid stipitipellis hyphae and caulocystidia.

Phylogenetic analysis

We used fifty taxa (type and other reference specimens) of Maramius spp. in the Neosessiles and other closely related sections. As expected for paraphyletic groups, the phylogenetic analysis grouped sect. Neosessiles taxa in two different parts of the trees built with the ITS, LSU, ITS-LSU concatenated dataset, and six-gene-multilocus concatenated dataset (Figure 5).^13,24 The Peruvian specimens were grouped together in a highly supported clades in all phylogenetic analyses (85–100% bootstrap support; Figure 5). The closest related clade was composed of the unpublished Marasmius isolates (C2/33, C2/06), putatively soybean endophytes according to their NCBI passport information. These two clades were phylogenetically distinct to the M. tenuissimus sensu stricto clade conformed by the reference strains NW192, and NW199, and other reference specimens (Figure 5). Therefore, these results, combined with the unique morphological characteristics of the Peruvian specimens, support they belong to a new species within the sect. Neosessiles, herein after called Marasmius infestans sp. nov.

Figure 5. Phylogenetic analyses of Marasmius taxa in the Neosessiles and other closely related sections conducted in this study. a) Phylogenetic analysis with the internal transcribed spacer (ITS1-5.8S-ITS2, ITS) rDNA region using the Hasegawa-Kishino-Yano nucleotide substitution model with a discrete Gamma model for the rate of heterogeneity (HKY + G). b) Phylogenetic analysis using the large subunit (LSU) rDNA region using the transition model 3 with equal frequency allowing for proportion of invariable sites (TIM3+I). c) Phylogenetic analysis with the ITS-LSU concatenated dataset using the transversion model with unequal base frequency with invariable site plus discrete Gamma model for the rate of heterogeneity (TVM + I +G). d) Phylogenetic analysis with the multi-locus dataset composed by ITS, LSU, and partial sequences of the Translation Elongation Factor 1-α (EF1α), the largest subunit of RNA polymerase II gene (RPB1), and the mitochondrial cytochrome oxidase I (COX1) genes, using the transversion model with unequal base frequency, with a discrete Gamma model for the rate of heterogeneity (TVM + G).

All phylogenetic trees were midpoint-rooted and built under the maximum likelihood framework. Models of evolution were determined by jModelTest2.²⁹ Values above branches = Maximum likelihood bootstrap values. Taxa in blue are the specimens causing thread-blight disease of cacao in Amazonas, Peru, and conform a new species clade, Marasmius infestans sp. nov. Type specimens are marked with an asterisk.

West African specimens were phylogenetically distinct to both, M. tenuissimus sensu stricto, and the Peruvian-specimen clade (Figure 5). They were grouped in at least four phylogenetic clades. The informally described M. neosessiliformis nom. prov. was grouped together with isolate GHA64.

Nuclear and mitochondrial genomic analyses

The nuclear genome assembly of Marasmius infestans AFH-31 reveals a size of 84.7 Mb, organized in 3,213 contigs (≥1,000 bp) with a GC content of 49.32% and a N50 value of 42,194 kb. We predicted 21,762 protein-coding genes and 656 tRNA genes in the nuclear genome of M. infestans strain AFHP31. Functional annotation resulted in the identification of 15,414 Pfam domains; 29,349 InterPro protein families; 31,346 Clusters of Orthologous Groups of proteins (COGs) and EggNog proteins; 606 proteases/protease inhibitors, and 679 carbohydrate-active enzymes (CAZymes). Moreover, we predicted 1,823 signal peptides and 4,010 transmembrane regions. About 8.9% of the proteome (2,152) are extracellular proteins, of which 25.6% (550) and 16.5% (356) were predicted to be apoplastic and cytoplasmic effectors, respectively. The genome contained 11 SMGCs, 18 biosynthetic enzymes, and 27 smCOGs. The draft nuclear genome of M. infestans AFHP-31 is larger in size than isolate GHA-37 genome size and, despite having fewer proteins, has more effectors (Table 2).

The mitochondrial genome of Marasmius infestans is circular, 47,389 bp long, and contains 42 genes. It is A + T rich (72.83%) and includes 25 tRNA (trnR, trnL and trnS occur in duplicate, while trnM in triplicate), 14 ribosomal proteins, two rRNA (rnl, rns), and one orf (orf868) (Figure 6). A comparison with the mitochondrial genome of TBD-causal agents from West Africa reveals M. infestans differs to them in size, and gene number, and has lost all their introns in COX1 (Table 3).

Figure 6. Circular map of Marasmius infestans sp. nov. mitochondrial genome.

The genes inside and outside the circle are transcribed in clockwise and counterclockwise directions, respectively. Genes belonging to different functional groups are shown in different colors. The dark gray area in the inner circle denotes GC content.

Taxonomy:

Marasmius infestans Huamán, Ramos C. & Díaz Val., sp. nov. IF 901138 (Figures 1-2).

Etymology: ‘infestans’, in reference to the capacity to infect branches and leaves of cacao trees.

Diagnosis: Similar to Marasmius tenuissimus and M. neosessilis but M. infestans has white to white-cream pilei at young and mature stages, lamella non-intervenose, and presence of ovoid to globose pileipellis broom cells. It is also phenetically similar to M. griseoroseus but M. infestans lacks Rotalis- and Amyloflagellula-type pileipellis broom cells and pileus-trama is inamyloid.

Type: PERU: Amazonas department, Bagua province, Imaza district, Pumpu native community; -4.785944, -78.281000; leg. Tito Ramos-Carrasco; Sept. 2022. Holotype specimen TAIM-04 (voucher KUELAP-2940).

Other examined specimens: AFHP-101 and dried culture of AFHP-31 (voucher KUELAP-2251).

Pileus: (0.7-)1.1-4.0(-4.9) mm diam., convex, smooth, glabrous, dull to shiny, white to white-cream (n = 84). Context white, thin. Lamellae adnate, distant (1-5) with 0 series of lamellulae, narrow, smooth, non-intervenose, white, non-marginate. Stipe: absent or extremely rudimentary, eccentric, cylindrical, insititious not arising from rhizomorphs; white rhizomorphs frequently present. Odor not distinctive.

Basidiospores: (6.2-)7.0-8.6(-8.8) × (3.7-)3.8-5.0(-5.2) μm [x_m = 7.8± 0.5 × 4.6 ± 0.4 μm, Q = 1.4-2.0, Q_m = 1.7, n = 45], hyaline, inamyloid, ellipsoid, smooth, thin-walled (n = 44). Basidia: length (17.3-)17.7-21.5(-22.2) μm, thicker part (7.0-)7.3-10.1(-10.2) μm, thinner part (2.6-)3.1-6.1(-6.3) μm, hyaline, inamyloid, cylindrical to clavate (n = 30). Cheilocystidia: Siccus-type broom cells, hyaline, inamyloid; main body 17.8-20.6 × 7.3-9.9, cylindrical, slightly narrower at the base, thin-walled; apical setulae length 2.3-3.8(-4.1) μm, obtuse, thin-walled. Caulocystidia: setoid, hyaline, strongly dextrinoid, thin- to thick-walled, main body length (41.7-)46.5-85.1(-87.5) μm, thicker part (7.4-)7.6-10.5(-10.9) μm, thinner part 3.0-5.0(-5.1) μm (n = 5). Pileipellis: hymeniderm, mottled, composed of Siccus-type broom cells; main body (9.6-)9.8-12.1(-12.2) × (8.5-)8.9-16.8(-17.9) μm, ovoid to globose, light-brown to hyaline, thin- to thick-walled (n = 7). Pileus trama: interwoven, inamyloid. Lamellar trama: hyphae (2.4-)2.6-3.9(-4.0) μm diam., hyaline, inamyloid, non-gelatinous. Stipitipellis: hyphae diam 3.4-4.6 μm., hyaline, inamyloid, non-gelatinous. Clamp connections: present in all tissues.

Habit and habitat: Rhizomorphic colonizing leaf and twigs of cacao, which subsequently causes thread blight disease. Rhizoids are white; fruiting bodies develop on dead tissue. The fungus occurs on poorly managed cacao farms with excessive shade and high relative humidity, typical characteristics of the tropical rainforest.

Geographic distribution Imaza province, Amazonas department, Peru.

Notes: Marasmius infestans differs from M. tenuissimus in the color and size of the pilei, and lamellae appearance. Marasmius infestans pilei are white to white-cream, while M. tenuissimus is between light grayish fuscous to rusty brown when fresh,⁹ or greyish or pale orange to golden brown, light brown or orangish white.^3,4,9 Pilei in M. infestans can reach up to 5 mm broad, while pilei in M. tenuissimus can have diameters typically of 7–40 mm.^3,4 Lamellae in M. infestans lacks reticulation while in M. tenuissimus lamellae is heavily reticulate. Additionally, West African species within the M. tenuissimus species complex can produce pure-white to brown-colored rhizomorphs,¹¹ while M. infestans produces pure-white rhizomorphs. At the genomic level, M. infestans is 13,636,061 pb larger than M. aff. tenuissimus GHA37. Marasmius infestans also has 3,229 fewer genes, and 231 more effectors (Table 2). Moreover, M. infestans differs from M. neosessilis and its varieties (M. neosessilis var. neosessilis and M. neosessilis var. montepiensis) in that these taxa can have various pilei colors such as apricot, salmon-flesh, and gray.^9,71 Also, M. neosessilis and the other phenetically similar species M. griseoroseus have pseudoamyloid (dextrinoid) pileus-trama, while pileus-trama of M. infestans is inamyloid.