Multilocus Sequence Analysis of Cercospora spp. from Different Host Plant Families

Identification of the genus Cercospora is still complicated due to the host preferences of ten being used as the main criteria to propose a new name. We determined the r elationship between host plants and multilocus sequ ence variations (ITS rDNA including 5.8S rDNA, elongation factor 1α, and calmodulin) in Cercospora spp. to investigate the host specificity. We used 53 strains of Cercospora spp. infecting 12 plant families for phylogenetic analysis. The sequences of 23 strains of Cercospora spp. infecting the plant families of Asteraceae, Cucurbitaceae, and Solanaceae were determined in this study. The sequences of 30 strai n of Cercospora spp. infecting the plant families of Fabaceae, Amaranthaceae, Apiaceae, Plumbaginaceae, Malvaceae, Cistaceae, Plantaginaceae, Lamiaceae, and Poaceae were obtained from GenBank. The molecular phylogenetic a nalysis revealed that the majority of Cercospora species lack host specificity, and only C. zinniicola, C. zeina, C. zeae-maydis, C. cocciniae, and C. mikaniicola were found to be host-specific. Closely related species of Cercospora could not be distinguished using molecular analyse s of ITS, EF, and CAL gene regions. The topology of the phylogene tic tree based on the CAL gene showed a better topo logy and Cercospora species separation than the trees developed based on the ITS rDNA region or the EF gene.


Introduction
The genus Cercospora Fresen.1863 was introduced to accommodate hyphomycetes fungi with long and tailed conidia.Members of the genus Cercospora, called cercosporoid fungi, are known as plant pathogens, but are occasionally found as secondary invaders and saprobes [1].Many species of cercosporoid fungi are economically important plant pathogens, causing leaf spot and leaf blight worldwide [2][3][4][5][6].Due to the paucity of useful morphological and physiological characteristics for identification, the taxonomy of the genus Cercospora remains confusing and depends heavily on the host, which has resulted in the description of a large number of species [1].In the MycoBank fungal database, approximately 2,721 species have been described and verified (http://www.mycobank.com/MycoTaxo.aspx,2012).The latest revision of Cercospora, published by Crous and Braun [1], recognized only 659 names in the genus Cercospora, with a further 281 being referred to as C. apii s. lat.(sensu lato).
Since DNA sequence analysis has become more reliable, robust, and widely used for analyzing relationships among fungal taxa, identification of species belonging to the genus Cercospora has shifted from a conventional approach to molecular analysis, and most often involves a combination of these approaches.Several new species of Cercospora, C. arecacearum Hidayat and Meeboon and C. christellae Hidayat and Meeboon, were published using a combination of morphological and molecular analyses [7][8].Molecular analyses of the genus Cercospora have previously relied on sequence analyses of the ITS (internal transcribed spacer) regions of rDNA (ribosomal DNA) including 5.8S rDNA [7][8][9], but ITS rDNA was not sufficient to distinguish between species of Cercospora (especially Cercospora s. lat.), and therefore, additional locus analyses are required [10].A multilocus approach, using a combination of the ITS rDNA, elongation factor 1-α (EF), actin (ACT), calmodulin (CAL), and histone (HIS) genes, has also been developed [11][12][13][14][15]. Cercospora agavicola Ayala-Escobar was proposed as a new species of Cercospora using multilocus sequencing [12].Cercospora acaciae-mangii Crous, Pongpanich and M.J. Wingf.and Cercospora zeae-maydis Tehon and E.Y. Daniels was distinguished from C. apii s. lat.and C. zeina Crous and U. Braun, respectively, based on multilocus sequence analyses [11,15].Furthermore, those analyses also distinguished three closely related species of Cercospora: Cercospora apii Fresen., C. beticola Sacc., and C. apiicola M. Groenew., Crous and U. Braun [13][14].
Generally, species of Cercospora are considered to be host-specific at the level of the plant genus or family [16].This assumption, however, has rarely been tested using molecular analysis.Several reports have been recorded, for example, To-Anun et al. 2011 [17] reported that C. apii was not entirely host-specific based on the host range data of C. apii s. lat.obtained in Thailand.To-Anun et al. 2011 [17] used ITS rDNA sequence data representing phylogenetic affinities from 52 species of Cercospora s. str.(sensu stricto) associated with 29 plant families in Thailand.Furthermore, Groenewald et al. 2005 [13] reported that C. apiicola is host-specific to Apium spp., but C. apii and C. beticola are not.To date, there are no publications related to the relationship between Cercospora species and their hosts at the family level using multilocus-based phylogenetic trees.To-Anun et al. 2011 [17] used ITS rDNA for their analysis although ITS rDNA cannot distinguish Cercospora species (especially Cercospora s. lat.), but they used many species from a wide range of families.On the other hand, Groenewald et al. 2005 [13] used multiple loci for the analysis of just three species, C. apii, C. beticola, and C. apiicola.
To elucidate and clarify the relationship between Cercospora species and their hosts, the analysis of multiple loci from many species of Cercospora is required.In the present study, we examine phylogenetic analyses of Cercospora spp.infecting 12 plant families (Asteraceae, Cucurbitaceae, Solanaceae, Fabaceae, Amaranthaceae, Apiaceae, Plumbaginaceae, Malvaceae, Cistaceae, Plantaginaceae, Lamiaceae, and Poaceae) based on a combination of ITS rDNA regions, including 5.8S rDNA, elongation factor 1-α (EF), and calmodulin (CAL) genes, to test the hypothesis that members of the genus Cercospora are not specific to their host plants.Hopefully, with the addition of more Cercospora species, using multiple loci will enable the elucidation of the relationships between Cercospora species and their hosts.

Source of fungi.
A total of 23 Cercospora spp.infected plant families of Asteraceae (12 strains), Cucurbitaceae (6 strains), and Solanaceae (5 strains), were used in this study (Table 1).These species have been identified based on their hosts at the genus level.All species were obtained from the LIPI Microbial Collection (LIPIMC) at the Research Center for Biology, Indonesian Institute of Sciences (LIPI) and the Institut Pertanian Bogor Culture Collection (IPBCC).All isolates were grown on potato dextrose agar (PDA, Difco TM , USA) and incubated at room temperature prior to DNA extraction.

DNA extraction, PCR (polymerase chain reaction)
amplification, and sequencing.Cultures of Cercospora spp.were transferred from PDA to 5 ml of potato dextrose broth (PDB) in tubes, and were incubated in a shaking water bath (Personal-11 Taitec, Tokyo, Japan) at 100 rpm for 10 days at 28 ºC.Fungal mycelia were harvested into 1.5 ml tubes containing 500 µl Milli-Q water.After mechanical treatment (grinding) using a pestle, the mycelial suspensions were centrifuged using a Centrifuge MiniSpin (Eppendorf, Hamburg, Germany) at 14,500 rpm for 10 min.DNA was then extracted from the mycelia using a DNA extraction kit (illustra TM DNA Extraction Kit Parts of the EF and CAL genes were amplified by PCR using the primer pairs of EF1-728F (5'-CATCGAGAAGTTCGAGAAGG-3') and EF1-986R (5'-TACTTGAAGGAACCCTTACC-3') and CAL-228F (5'-GAGTTCAAGGAGGCCTTCTCCC-3') and CAL-737R (5'-CATCTTTCTGGCCATCATGG-3'), respectively [19].PCR reactions for the EF and CAL genes were performed in 25 µl reaction volumes, with each reaction containing 8.75 µl of nuclease free water, 12.5 µl of GoTaq Green Master Mix (Promega, Madison, USA), 0.625 µl for each forward and reverse primer, 0.5 µl of DMSO, and 2 µl of the DNA template.PCR reactions for both regions were performed in the TaKaRa thermocycler (TaKaRa, Japan) as follows: 94º C for 5 min, 35 cycles of 94º C for 30 s, 52º C for 30 s, 72º C for 30 s, followed by a final extension of 7 min at 72º C [14].Approximately 320 nucleotide lengths were generated from the EF and CAL genes [11].
The PCR products were visualized using electrophoresis on a 1% (w/v) agarose gel at 100 V for 30 min.The gel was soaked in an ethidium bromide solution (0.1% v/v) for 1 hour and visualized under UV light (Printgraph).The PCR products were sent to 1st Base (Malaysia) for sequencing.
Phylogenetic analysis.The sequences obtained from the respective primer pairs (ITS5-ITS4, EF1-728F-EF1-986R, and CAL-228F-CAL-737R) were assembled using ChromasPro 1.41 software (Technelysium Pty Ltd., South Brisbane, Australia).The assembled sequences were refined by direct examination, and the nucleotides that were ambiguously aligned were excluded from the sequence.Alignment was conducted using MUSCLE (multiple sequence comparison by log-expectation) in MEGA (molecular evolutionary genetics analysis) version 5.05 [20].Gaps were treated as missing data.Phylogenetic analyses were performed using the neighbor-joining (NJ) method in MEGA.In this analysis, Kimura 2-parameter model algorithms were used to generate distance matrices using pre-aligned DNA sequences from all isolates.Sites containing alignment gaps were excluded entirely from the calculations.The phylogenies, based on estimated distance matrices, were then produced by the NJ method [21].Support for the internal branches was obtained by bootstrap analysis [22] with 1,000 replications.GenBank accession numbers, taxon names, and other information regarding the sequences from GenBank which were used in this study are provided in Table 2. Trees were generated from the analysis and were determined using TreeView version 1.6.6 [23].Mycosphaerella thailandica was assigned as an outgroup in all analyses.

Results and Discussion
For each of the three sequenced loci, approximately 500, 300, and 300 bases were determined for the ITS rDNA, EF, and CAL regions, respectively.A partition homogeneity test showed that the ITS rDNA, EF, and CAL datasets were combinable into a single phylogenetic tree (p = 0.50) [24].Each of the ITS, EF, CAL, and multilocus-based phylogenetic trees consisted of 53 sequences of Cercospora (including the outgroups), of which 23 sequences were obtained from this study and 30 sequences were retrieved from GenBank (http://www.ncbi.nlm.nih.gov).The strains of M. thailandica (CPC 10547, CPC 10548, CPC 10549, and CPC 10621) were used as outgroups.The phylogenetic trees of ITS, EF, CAL, and their combinations (multilocus) are shown in Figures 1, 2, 3,  and 4, respectively.
The phylogenetic tree generated from the ITS sequences divided the Cercospora into two major clades (Figure 1).The first clade contained a majority of Cercospora species (41 Operational Taxonomic Units/OTU) with 92% bootstrap support (BS) and small independent clades included in the clade.The second clade consisted of C. zeae-maydis and C. zeina on Zea mays (Poaceae), and C. mikaniicola on Mikania micrantha (Asteraceae) with 62% BS.The sequences of Cercospora from different plant families were mixed together in both clades.A lack of small, independent clades in the ITS tree indicated that the resolution of the ITS region in determining Cercospora species from different hosts was limited.The same results were found by other researchers [11,13,25].
Similar to the phylogenetic tree based on the ITS rDNA sequences, Cercospora species were divided into two clades in the phylogenetic tree based on the EF gene sequence analysis (Figure 2 Momordica charantia (Cucurbitaceae) were nested in the same clade with 97% BS, and two sequences of C. zinniicola on Zinnia elegans (Asteraceae) also formed an independent clade with 61% BS.In general, the phylogenetic tree based on the EF gene produced a higher resolution than the ITS sequence region.
The CAL gene was found to be more informative than the ITS rDNA and the EF gene for separating Cercospora species (Figure 3).This was displayed by a higher number of clades and greater branch lengths being formed in the phylogenetic tree, based on the CAL sequence (Figure 3).Due to the high variability within the CAL sequence, The topology of the multilocus tree was generally congruent with the CAL tree, but the multilocus tree has higher bootstrap support and greater branch length than the CAL tree.In the multilocus tree, the first clade had a 97% BS (Figure 4) and the CAL tree had a 96% BS (Figure 3).Both of the phylogenetic trees consisted of the most Cercospora species used in this study (19 OTU).The differentiation between the multilocus tree and the CAL tree is reflected in the position of C. 2001 [25] and Crous et al. 2006 [15].
The current phylogenetic study has shown that most of the plant pathogenic Cercospora spp.are not restricted to their hosts at both the genus and family levels of their plant hosts.Cercospora sequences from the same host family were found to be polyphyletic and intermixed with other sequences from different host families.Based on the multilocus tree (Figure 4), polyphyletic families are showed in Asteraceae, Cucurbitaceae, Solanaceae, Apiaceae, and Amaranthaceae.[15] based on multilocus sequences rDNA, elongation factor 1-α, calmodulin, histone H3, and actin genes).Cercospora cocciniae were found to be host specific at the family level but the other species were found to be host specific at the species, genus, and family levels.Those species formed an independent lineage in the multilocus tree (Figure 4).
The current study showed that the identification of Cercospora species using ITS rDNA is clearly insufficient.Many species belonging to C. apii sensu lato (e.g. C. apii and C. beticola) are nested in the same clade in all of the phylogenetic trees (Figures 1-4).
Pairwise sequence analyses of all Cercospora included in this study clearly show that few informative characteristics are available in the ITS region for the determination of Cercospora species (Iman Hidayat, unpublished data).Although ITS regions of rDNA are usually used to identify species in the majority of fungal groups, this region could not resolve the problematic taxonomy of the genus Cercospora [10].The elongation factor 1-α gene, which was used in the identification of the genus Fusarium [27], could not distinguish between C. apii and C. beticola.Our analyses showed that the calmodulin (CAL) gene is more informative than the ITS rDNA and EF genes (Figures 1-3).Cercospora apii and C. beticola were successfully distinguished based on phylogenetic analyses of CAL gene sequences (Figure 3).Groenewald et al. 2005 [13] reported that the CAL gene was found to be very effective for separating species, and further developed species-specific primers using the CAL gene.Multilocus sequence analyses have been shown to be more reliable in determining species of Cercospora than single loci sequence analyses (ITS rDNA, EF, and CAL) (Figure 4).However, as several species of Cercospora were difficult to differentiate using a combination of ITS regions of rDNA, EF gene, and CAL gene sequences, additional gene loci, such as actin (ACT) and histone (HIS), and additional taxa of Cercospora should be included in analyses in order to fully resolve the taxonomy of this pathogenic fungus.Pathogenicity tests should be conducted to support the phylogenetic tree.

Conclusions
In this study, the phylogenetic trees showed that most Cercospora species are not restricted to their hosts at the genus and family levels; one species of host plant can be infected by two species of Cercospora.Conversely, one species of Cercospora can infect two or more host species.The species of Cercospora have a wide host range and only C. zinniicola, C. zeina, C. zeae-maydis, C. cocciniae, and C. mikaniicola were found to be host-specific.Closely related species of Cercospora could not be distinguished using molecular analyses of ITS, EF, and CAL gene regions.However, among those regions, the CAL gene produced a better topology of phylogenetic tree than that of the ITS rDNA region and the EF gene.The topologies of the CAL gene and the multilocus based trees were congruent, but the multilocus based tree has higher bootstrap support than that of the CAL gene tree.

2 Figure 1 .Figure 2 .
Figure 1.Phylogenetic Tree of Cercospora spp.Based on Neighbor-joining Method of the Internal Transcribed Spacer Region of rDNA

Figure 3 .
Figure 3. Phylogenetic Tree of Cercospora spp.Based on Neighbor-joining Method of Calmodulin Gene

Table 2 . List of Fungal Strains of Cercospora Retrieved from the GenBank (NCBI) Database Used in Phylogenetic Analyses
Groenewald et al. 2005 [13]utilized the CAL gene to develop a species-specific diagnostic assay for Cercospora species.In the current tree, the sequences of C. physalidis on Solanum verbascifolium (LIPIMC 0776) and S. nigrum (IPBCC 13.1011) and two sequences of C. zinniicola on Z. elegans (LIPIMC 0752 and LIPIMC 0771) form independent clades with 97% BS.Cercospora zeaemaydis CBS 117757 and C. zeina CPC 11995 on Z. mays form independent clades with 84% BS.Cercospora cocciniae on C. grandis (Cucurbitaceae) and on M. charantia (Cucurbitaceae) were nested in the same clade with 100% BS.Based on this result, the phylogenetic tree based on the CAL sequence analyses could not resolve the host specificity among Cercospora species from different host families.
The separation of C. zeae-maydis, C. zeina, and C. mikaniicola from the majority of Cercospora species (Figures1, 2, and 4) indicates some degree of evolutionary divergence from the other Cercospora species.The evolutionary divergence of C. zeae-maydis and C. zeina was previously noted byGoodwin et al.
multilocus phylogenetic tree.The third clade in the CAL tree consisted of five C. apiicola and C. mikaniicola with no bootstrap support.The fourth clade contains a sister group of C. zeina and C. zeae-maydis with 100% BS in the multilocus tree and 84% BS in the CAL tree.
Tanacetum parthenium, while C. citrullina has been found on Benincasa spp., Bryonopsis laciniata, Citrullus spp., Coccinia spp., Cucumis spp., Cucurbita spp., Lagenaria spp., Luffa spp., Melothria pendula, Momordica spp., Sechium edule, Telfairia pedata, Trichosanthes spp., and [1] topology of the phylogenetic tree based on multilocus sequences did not support host specificity.One species of host plant can be infected by two species of Cercospora (e.g.Z.mays by C. zeina and C. zeae-maydis; B. vulgaris by C. apii and C. beticola).Moreover, three species of Cercospora, C. beticola, C. apii, and C. apiicola, were found to be infecting Apium graveolens.Crous and Braun[1]noted that several species of Cercospora have the ability to infect multiple hosts.In the current study, Cercospora physalidis (LIPIMC 0776 and IPBCC 13.1011) was found on S. verbascifolium and S. nigrum (Solanaceae), and C. citrullina (IPBCC 13.1006 and IPBCC 13.1007) was found on Trichosanthes anguina and Citrullus vulgaris (Cucurbitaceae).Although C. physalidis and C. citrullina could infect plants at a same genera and/or family levels, those species were not host specific, since C. physalidis was found to be polyphyletic and C. citrullina was found to be paraphyletic in clade 1 on the multilocus tree (Figure4).Two morphologically similar species, C. apii and C. beticola, were recognized as Cercospora species with broad host ranges.Cercospora apii (CBS 152.

Phylogenetic Tree of Cercospora spp. Based on Neighbor-joining Method of Multilocus ITS Region of rDNA, EF, and CAL Genes Thladiantha
sp. (http://nt.ars-grin.gov/fungaldatabases/fungushost/fungushost.cfm,2013).Based on this information, it was clear that the identification of Cercospora species based on host species must be reconsidered, because analysis based on molecular data does not support this.Cercospora zeina and C. zeae-maydis were found on Z. mays (Poaceae) (http://www.mycobank.com/MycoTaxo.aspx,2013).Wang et al. 1998 [26] considered C. zeae-maydis and C. zeina as sibling species based on AFLP profiles and ITS nucleotide sequences.Both species were further distinguished by Crous et al. 2006