Three new Curvularia species from clinical and environmental sources

Abstract Curvularia is a Pleosporalean monophyletic genus with a great diversity of species, including relevant phytopathogenic, animal and human pathogenic fungi. However, their microscopic identification is difficult due to overlapping morphological features amongst species. In recent years, multi-locus sequence analysis using the ITS region of the rDNA and fragments of the genes gapdh and tef1 revealed numerous cryptic species, especially in isolates that commonly produced 3-septate conidia. Therefore, based on sequence analysis of the above-mentioned DNA barcodes recommended for species delineation in Curvularia, we propose three novel species, C. paraverruculosa, C. suttoniae and C. vietnamensis, isolated from soil, human clinical specimens and plant material, respectively, collected in different countries. These new species are morphologically characterised and illustrated in the present study. Curvularia paraverruculosa differs from its counterparts, C. americana and C. verruculosa, mainly by its narrower conidia. Curvularia suttoniae and C. vietnamensis are closely related to C. petersonii, but the former two have larger conidia.


Introduction
The genus Curvularia Boedijn (1933), typified by C. lunata (Wakker) Boedijn, belongs in Pleosporaceae, Pleosporales (Wijayawardene et al. 2018). Members of Curvularia show different life modes, i.e. saprophytic, endophytic and also pathogenic on plants and animals (Marin-Felix et al. 2017a). Phytopathogenic species can affect wild grasses and staple crops, such as rice, maize, wheat or sorghum and give rise to serious losses in agricultural production (Gautam et al. 2013, Manamgoda et al. 2015, Marin-Felix et al. 2017a. The endophytic species have garnered interest in recent years for their use in the production of bio-based products that are beneficial to living organisms and the environment (Bengyella et al. 2019). Since the first report of Curvularia as a human pathogen in a patient with mycetoma (Baylet et al. 1959), other clinical presentations have been reported, such as superficial and deep infections that mainly affect the respiratory tract but can even cause cerebral phaeohyphomycosis with an extremely poor prognosis (de Hoog et al. 2000).
The genus is morphologically distinguished mainly by its asexual morph, which shows sympodial conidiophores with mono-to polytretic conidiogenous cells and transversally septate conidia. Typically, the conidia in Curvularia are curved due to the hypertrophy of one of the intermediate cells and they are euseptate (Ellis 1971), although other authors opine that the conidia in Curvularia are distoseptate (Sivanesan 1987, Seifert et al. 2011, Madrid et al. 2014. The species of Bipolaris and Exserohilum have typically straight and distoseptate conidia; however, some of them have been transferred to Curvularia, based on their DNA sequence analyses (Manamgoda et al. 2012, Hernández-Restrepo et al. 2018). Furthermore, due to the overlapping of morphological characters amongst certain species of Curvularia, such as conidial size, shape and septation, an accurate identification at the species level is difficult without a DNA sequence analysis (da Cunha et al. 2013, Madrid et al. 2014, Manamgoda et al. 2015. Several cryptic species have been described recently using only multi-locus sequence analyses of the recommended DNA barcodes for species delimitation, i.e. the internal transcribed spacer (ITS) region of the rDNA and the protein-coding loci glyceraldehyde-3-phosphate dehydrogenase (gapdh) and translation elongation factor 1-a (tef1) (Marin-Felix et al. 2017a. Nearly 130 species have so far been accepted in Curvularia, including the species classified previously in the teleomorphic genera Cochliobolus and Pseudocochliobolus after applying the current criteria for fungal nomenclature (Manamgoda et al. 2012, 2015, Madrid et al. 2014, Marin-Felix et al. 2017a, 2017b, Dehdari et al. 2018, Heidari et al. 2018, Hernández-Restrepo et al. 2018, Liang et al. 2018, Tibpromma et al. 2018, Kiss et al. 2019, Raza et al. 2019, Zhang et al. 2020.
Based on a polyphasic approach, combining morphological and phylogenetic analyses, three novel Curvularia species are proposed here, isolated from human clinical specimens in the USA, soil in Mexico and seed and plant debris in Vietnam and Indonesia, respectively.

Origin of isolates
Five unidentified Curvularia isolates, maintained in the fungal collection of the Medical School of the Rovira i Virgili University (FMR; Reus, Spain), were included in the study. Two of these (FMR 10992, FMR 11690) were isolated from human specimens in the USA by Deana A. Sutton of the Fungus Testing Laboratory at the University of Texas Health Sciences Center (UTHSC; San Antonio, USA) and the other three (FMR 11956, FMR 17656, FMR 17659) were isolated from environmental samples; the first from sorghum seeds collected in Indonesia, the second from soil collected in the Mexican region of Michoacán and the third from unidentified plant material collected in the north-east of Vietnam.

DNA extraction, PCR, sequencing and phylogenetic analysis
The fungal DNA was extracted from colonies growing on potato dextrose agar (PDA; Pronadisa, Madrid, Spain) for 7 to 10 days at 25 °C in darkness and following the protocol of Müller et al. (1998). The ITS barcode, including the 5.8S gene and the genes gapdh and tef1 were analysed following Marin-Felix et al. (2017a). Amplification was carried out using the primer pairs ITS5/ITS4 for the ITS region (White et al. 1990), gpd1/gpd2 for gapdh (Berbee et al. 1999) and EF983/2218R for tef1 (Schoch et al. 2009). The PCR products were purified and stored at -20 °C until sequencing. The same pairs of primers used for the amplification were also used to obtain the DNA sequences, which were processed at Macrogen Europe (Macrogen Inc., Madrid, Spain). The sequences of each isolate were edited using SeqMan v. 7.0.0 (DNAStar Lasergene, Madison, WI, USA) to obtain the consensus sequences.
We made a preliminary comparison of gapdh sequences generated from our isolates with those of the National Center for Biotechnology Information (NCBI) using the Basic Local Alignment Search Tool (BLASTn) for their molecular identification. To establish the phylogenetic position of unidentified isolates with respect to the most accepted species in Curvularia, we carried out individual (data not shown) and combined alignments of the three loci complemented by all available sequences of the ex-type and reference strains of Curvularia species retrieved from NCBI (Table 1). Based on this first phylogeny of the genus, a more restricted multi-locus analysis was carried out, including only those Curvularia species most related to the isolates under study. The alignments were made in the MEGA (Molecular Evolutionary Genetics Analysis) software v.6.0. (Tamura et al. 2013), using ClustalW algorithm (Thompson et al. 1994), refined with MUSCLE (Edgar 2004) in the same platform and manually adjusted as necessary. Phylogenetic reconstructions were made using Maximum Likelihood (ML) and Bayesian Inference (BI) approaches under RAxML-HPC2 on XSEDE v.8.2.12 (Stamatakis et al. 2014) in CIPRES Science gateway portal (Miller et al. 2010) andMrBayes v. 3.2.6 (Ronquist et al. 2012), respectively.
For the ML analysis, the best nucleotide substitution model for the combined analysis of ITS, gapdh and tef1, determined using the MEGA programme, was Kimura 2-parameters with Gamma distribution (K2+G); the combined analysis of these three phylogenetic markers was tested through Incongruence Length Difference (ILD) implemented in the Winclada programme (Farris et al. 1994). ML bootstrap values (bs) ≥ 70% were considered significant. Sequences newly generated in this study and novel species proposed are indicated in bold.
For the BI phylogenetic analysis, the best nucleotide substitution model was determined using jModelTest (Posada 2008). For the ITS region, we used Kimura 2-parameter with Invariant sites (K80+I), for gapdh General Time Reversible with gamma distribution (GTR+G) and for tef1 General Time Reversible with invariant sites (GTR+I). The parameter settings used were two simultaneous runs of 5M generations, four Markov chains, sampled every 1000 generations. The 50% majority-rule consensus tree and posterior probability values were calculated after discarding the first 25% of the samples. A posterior probability (pp) value of ≥ 0.95 was considered significant.
Sequence data generated in the present study were deposited in GenBank (Table 1) and the alignments in TreeBASE (http://treebase.org).

Phenotypic study
Macroscopic characterisation of the colonies was made on PDA, oatmeal agar (OA; oatmeal 30 g, agar 13 g, distilled water 1 litre) and potato carrot agar (PCA; potato 20 g, carrot 20 g, agar 13 g, distilled water 1 litre), after 7 days at 25 °C in darkness. Colours of the colonies in descriptions were based on Kornerup & Wanscher (1978). Cardinal temperatures for growth were obtained on PDA after 7 days in darkness.
Microscopic features were studied from the specimens mounted in Shear's solution growing on the same media (Madrid et al. 2014). At least 30 measurements were taken for the calculation of conidial and conidiophores length and width ranges, which are also reported as the mean plus or minus standard deviation in the descriptions.
Photomicrographs were taken using a Zeiss Axio-Imager M1 light microscope (Zeiss, Oberkochen, Germany) with a DeltaPix Infinity X digital camera.
Nomenclatural novelties and descriptions were deposited in MycoBank (Crous et al. 2004). Ex-type cultures and holotypes, which were dried cultures, were deposited at the Westerdijk Fungal Biodiversity Institute from Utrecht (CBS, The Netherlands).

Results
BLASTn results with gapdh sequences showed that the isolate FMR 17656 was ≤ 97.6%, similar to C. verruculosa CPC 28792; FMR 11956 and FMR 17659 showed a similarity of 93.31% and 93.6%, respectively, with C. spicifera CBS 198.31; and isolates FMR 10992 and FMR 11690 both exhibited a similarity of 94.7% with the ex-type strain of C. petersonii (BRIP 14642). Sequence similarity with this marker between FMR 11956/17659 and FMR 10992/11690 was 97%. These values suggested that the unidentified isolates represented putative new species for the genus, which were then confirmed by multi-locus sequence analysis of ITS, gapdh and tef1 barcodes. The combined analysis included 128 sequences representing 126 taxa in the genus Curvularia and these were rooted with Bipolaris maydis (CBS 136.29) and B. saccharicola (CBS 155.26) (Suppl. material 1: Fig. S1). The alignment comprised a total of 1928 bp (ITS 432, gapdh 573 bp and tef1 923 bp), including 546 variable sites (ITS 119 bp, gapdh 253 bp and tef1 174 bp) and 445 phylogenetically informative (ITS 83 bp, gapdh 233 bp and tef1 129 bp). The unidentified isolates were allocated to three single lineages in the same clade (74/0.99) together with sequences of the ex-type strains of C. americana (UTHSC 08-3414), C. petersonii (BRIP 14642) and C. verruculosa (CBS 150.63), but with enough distance to be considered distinct species. The two clinical isolates (FMR 10992 and FMR 11690) formed a fully-supported clade closely related to isolates FMR 11956 and FMR 17659, which were collected in Indonesia and Vietnam, respectively and to C. petersonii. The fifth isolate (FMR 17656) was related to C. verruculosa and C. americana, but formed an independent and distant branch from the previouslymentioned species.
Cardinal temperature for growth. Notes. See C. suttoniae described above.

Discussion
As in other Pleosporalean genera, Curvularia is currently a well-delineated genus on the basis of molecular data (Manamgoda et al. 2015, Marin-Felix et al. 2017a). However, morphological features and analyses of the ITS barcode are insufficient to accurately identify Curvularia species. Thus, the multi-locus sequence analysis of different gene markers (i.e. LSU, ITS, gapdh, rpb2 and tef1) has been used to study the species diversity in Curvularia and phylogentic relationships with other similar genera (Hernández-Restrepo et al. 2018, Manamgoda et al. 2012, 2015, Madrid et al. 2014, Marin-Felix et al. 2017a, 2017b Zhang et al. 2020). Novel species are found, not only on fresh material collected in various geographical regions, but also in reevaluation of Curvularia isolates deposited in fungal collections and earlier identified by morphological features or ITS sequence analysis. The five isolates, studied here, showed morphological similarity with C. americana or C. lunata (Sivanesan 1987, Madrid et al. 2014), but they also showed subtle variations that did not match with these species. Multi-locus analysis of the recommended barcodes facilitated the delineation of the novel species C. paraverruculosa, C. suttoniae and C. vietnamensis, which were closely related to the known species C. americana, C. petersonii and C. verruculosa (Fig 1).
As in the case of C. suttoniae, other related species, such as C. americana and C. verruculosa, have also been associated with clinical specimens previously (da Cunha et al. 2013, Madrid et al. 2014). However, the role of all these fungi in human diseases has never been proven. Contrary to that, the recently described species C. coimbatorensis and C. tamilnaduensis were shown to be causal agents of fungal keratitis in India (Kiss et al. 2019). These two latter species, as with C. suttoniae and C. vietnamensis in our case, could only be molecularly differentiated by gapdh and tef1 loci; ITS sequence similarity between C. coimbatorensis and C. tamilnaduensis was 99% (Kiss et al. 2019) and between C. suttoniae and C. vietnamensis, it was 100%. Therefore, considering clinical laboratories commonly use ITS barcode for fungal diagnosis, not only will the diversity of Curvularia species remain obscure in the clinical setting, but also, subsequently, the epidemiology of its species associated with human or animal diseases. Our results suggest that gapdh and tef1 loci could be good alternatives as barcodes for Curvularia identification, since both have a high discriminatory power amongst species. However, gapdh would be the recommended locus because there are more sequences available for different species in the genus.
The ITS analysis revealed that C. palmicola, only known for its type specimen found on dead branches of Acoelorrhaphe wrightii in Thailand , is also closely related to the novel species described here. However, this fungus was not included in our concatenate analysis since sequences of gapdh and tef1 were not available for comparison. Nevertheless, C. palmicola can be distinguished morphologically from our species mainly by having conidia with constricted wall at the septum level. Furthermore, C. palmicola has longer conidia (23.9-34.7 μm) than C. suttoniae (8-22 μm) and C. vietnamensis (15-28 μm) and it differs from C. paraverruculosa by its smooth-walled conidia.