Elm yellows phytoplasmas (EY) belonging to the 16SrV-A subgroup were recently proposed as a new candidate species ‘Candidatus Phytoplasma ulmi’. These pathogens infect elm trees, causing leaf yellowing and premature drying. In this study, 25 isolates originating from localities in northeast, east and southwest Serbia were characterized by means of RFLP analysis and DNA sequencing of four genomic loci: 16S rRNA, ribosomal protein rpl22-rps3, secY and map. In total, five different genotypes were identified based on collective sequencing of all four genes. Four of these genotypes showed significant nucleotide changes compared with the EY1T reference strain. Phylogeny based on parsimony analyses of ribosomal protein, secY and map genetic loci indicated a single monophyletic origin of EY1T and the new ‘Ca. Phytoplasma ulmi’ strains. Unlike phylogenetic clustering, DNA sequence comparison of EY1T and the novel strains revealed mutations in oligonucleotide signature sequences for all three genes (16S, rpl22-rps3 and secY) used for the characterization and assignment of 16SrV-A phytoplasmas to the ‘Ca. Phytoplasma ulmi’ species in the original description. Based on their high degree of genetic variability, the Serbian strains were assigned to four different subtypes of ‘Ca. Phytoplasma ulmi’ (EY-S1, EY-S2, EY-S3 and EY-S4). New diagnostic enzymes for practical use in ‘Ca. Phytoplasma ulmi’ identification are proposed for the 16S rRNA, ribosomal protein and secY genes. The implications of genetic variability within signature sequences for taxonomy and identification of ‘Ca. Phytoplasma’ species, as well as the importance of geographic variability and number of strains characterized for species description, are discussed.
Phytoplasmas are wall-less, phloem-limited, non-culturable prokaryotes of the class Mollicutes that are associated with diseases in several hundred species of plants. Because they cannot be cultured on artificial medium and lack measurable phenotypic characters, classification of phytoplasmas has been based primarily on molecular analyses of highly conserved 16S rRNA gene sequences (Lee et al., 1993, 1998). A provisional taxonomic system for uncultured bacteria (Murray & Schleifer, 1994) was recently adopted for naming phytoplasma species candidates, with some modifications as a result of the high level of phytoplasmal 16S rRNA gene variability (IRPCM, 2004). According to the suggested rules, a new ‘Candidatus Phytoplasma’ species should share < 97·5% 16S rRNA sequence identity with any previously described species, unless it has clearly distinct biological, ecological, phytopathological and molecular properties.
The 16SrV elm yellows (EY) phytoplasma group consists of diverse phytoplasma strains with a wide range of biological and ecological properties (such as plant hosts, insect vectors and geographical distribution), which cause diseases in a variety of economically important plant species. They infect woody perennial hosts such as grapevine, alder, elm, blackberry, cherry, Spartium spp. and Ziziphus spp., causing diseases known as flavescence dorée (FD), alder yellows (ALY), elm yellows (EY), rubus stunt (RuS), spartium witches’ broom (SpaWB), cherry lethal yellows (CLY) and jujube witches’ broom (JWB), respectively (Lee et al., 2004). Based on RFLP analyses of 16S rRNA gene sequences, 16SrV group phytoplasmas have been classified into five subgroups (Lee et al., 1998, 2004; Davis & Dally, 2001). The reference phytoplasma strain for the 16SrV group is EY1T of the 16SrV-A subgroup (Lee et al., 1993). Although phytoplasma strains in the 16SrV group share a high degree of 16S rRNA gene similarity, two strains were proposed as separate species of a newly established ‘Candidatus Phytoplasma’ taxon based on genetic variability of less well-conserved genes and distinct ecological properties. Phytoplasma strains infecting Ziziphus species in China and India (16SrV-B subgroup) were proposed as ‘Ca. Phytoplasma ziziphi’ (Jung et al., 2003) and strains infecting Ulmus species in North America and Europe (16SrV-A) as ‘Ca. Phytoplasma ulmi’ (Lee et al., 2004).
Elm yellows disease, previously known as elm phloem necrosis, is caused by phytoplasmas of the 16SrV-A subgroup (Lee et al., 1993, 1998) and was first described in the USA (Swingle, 1938). The disease was reported in the early 1980s in Italy (Pisi et al., 1981), and in the early 1990s in France (Mäurer et al., 1993). Several studies have been performed to determine the relatedness, origin and genetic variability of phytoplasma strains affecting both North American and European elms (Lee et al., 1993; Marcone et al., 1997; Griffiths et al., 1999). These studies found homogeneity among strains and clear separate phylogenetic clustering within 16SrV group phytoplasmas, indicating a common monophyletic origin. Extensive research on the diversity and phylogeny of EY phytoplasma strains infecting elms was completed in 2004, with the description of a novel phytoplasma taxon ‘Ca. Phytoplasma ulmi’, comprising all isolates belonging to the 16SrV-A phytoplasma subgroup, with EY1T designated as the species reference strain (Lee et al., 2004). Although this taxon shares 98·2% 16S rRNA sequence identity with the previously described taxon of the EY phytoplasma 16SrV-B subgroup (‘Ca. Phytoplasma ziziphi’) it’s description was possible as a result of unique ecological properties and clear evidence of molecular diversity. The latter was determined on the basis of unique oligonucleotide sequences in three conserved genes, 16S rRNA, ribosomal protein and secY, and clearly distinct phylogenetic clustering within the 16SrV phytoplasma group.
The 16SrV-group phytoplasmas have a high 16S rRNA gene sequence similarity but are widely ecologically divergent. Because of their economic importance, extensive multigene sequence characterization has been carried out over the years to obtain molecular data that would accurately reflect the biological properties of different strains. Recently, one of these studies designated two additional genes, map and degV, as suitable for 16SrV-group phytoplasma typing (Arnaud et al., 2007). Although the primary focus of that paper was resolving phylogenetic relatedness of FD and ALY phytoplasma strains, multigene analyses also included ‘Ca. Phytoplasma ulmi’ isolates originating from both Europe and North America. Phylogeny based on the housekeeping map gene resulted in higher resolution of relatedness between 16SrV-group phytoplasma strains and revealed the existence of two strain clusters within ‘Ca. Phytoplasma ulmi’. These data indicate a higher genetic diversity within this taxon than was previously assumed.
Infection of Ulmus minor and U. laevis in northeast Serbia by phytoplasmas belonging to the rRNA subgroup 16SrV-A was reported in 2008 (Jovićet al., 2008). RFLP and sequence analyses of the ribosomal protein gene of detected phytoplasmas revealed the presence of diverse strains of ‘Ca. Phytoplasma ulmi’ in Serbian elms, with nucleotide changes located inside the range of unique regions reported by Lee et al. (2004) as being species-specific for ‘Ca. Phytoplasma ulmi’. In the present study, these isolates were molecularly characterized and additional samples collected from different geographical origins in Serbia, using RFLP and DNA sequencing of four conserved genes, 16S rRNA, ribosomal protein rpl22-rps3, secY and map. The objectives of this study were as follows: (i) to determine genetic variability among ‘Ca. Phytoplasma ulmi’ isolates in Serbia; (ii) to define phylogenetic clustering of new strain(s) among presently known members of the 16SrV group; (iii) to re-evaluate the practical use of previously determined unique oligonucleotide sequences of 16S rRNA, rpl22-rps3 and secY genes (Lee et al., 2004); and (iv) to redefine RFLP diagnostic procedures for determination of ‘Ca. Phytoplasma ulmi’ species.
Materials and methods
Plant samples and DNA extraction
Samples of European field elm (U. minor) and European white elm (U. laevis) trees showing symptoms of phytoplasma infection were collected in September of 2007 and 2008 from five different sites in Serbia. Symptoms included leaf yellowing, premature leaf drying and, occasionally, desiccation of whole plants. At least six elm plants were sampled from every site, except from the Tamnič site, where only three elm trees with symptoms were found. A total of 15 U. minor and 15 U. laevis leaf samples were collected (Table 1). Samples from four symptomless young elm trees (U. minor) collected near Belgrade served as the negative controls.
Table 1. Geographic origin, presence of ‘Candidatus Phytoplasma ulmi’ and designation of subtypes, genotypes and representative strains of phytoplasmas detected in samples of Ulmus plants with symptoms in Serbia
No. of samples positive/analyseda
‘Ca. Phytoplasma ulmi’ subtype genotypesc
aPositive samples were determined by amplification of the FD9 marker.
bDesignated representative strains for each genotype/locality. Strain designation is based on unique characteristics of obtained nucleotide sequences of 16S rRNA, rpl22-rps3, secY and map genes and RFLP patterns of 16S rRNA, rpl22-rps3 and secY genes.
cSubtype genotypes designated in this study, based on presence and characteristics of nt changes of 16S rRNA, rpl22-rps3 and secY and map genes for ‘Ca. P. ulmi’ according to Lee et al. (2004), detected by sequencing and/or RFLP analyses. Numbers of isolates belonging to the specific ‘Ca. Phytoplasma. ulmi’ subtype genotype present at each locality are listed in parentheses.
dOrigin of isolates additionally sequenced for the map gene.
EY-S2 (4)d EY-S1 (3)d
EY-S3 (6) EY-S4 (1)
Original type (2)
Leaves and petioles of each sample were put into separate plastic bags and placed in a cooler at 7°C for transport to the laboratory. Prior to DNA extraction, leaf veins and petioles of each sample were frozen in liquid nitrogen and stored at −20°C. Total nucleic acids were extracted using a previously reported CTAB protocol (Angelini et al., 2001) or with the DNeasy Plant Mini kit (QIAGEN) according to the manufacturer’s instructions.
Phytoplasma reference isolates
The following phytoplasma isolates were employed as references for PCR/RFLP analysis: American elm yellows EY1, maintained in experimentally infected periwinkle (Catharanthus roseus) (Angelini et al., 2001); flavescence dorée FD-C isolated from naturally infected field-grown grapevine from the Nišavski region (Serbia); FD-D isolated from naturally infected field-grown grapevine from the Veneto region (Italy) (provided by E. Angelini, Conegliano).
Polymerase chain reaction (PCR) amplification
Amplification of the following four phytoplasma genomic loci was carried out for detection and molecular characterization of ‘Ca. Phytoplasma ulmi’ isolates: (i) the 16S ribosomal RNA gene; (ii) ribosomal protein gene operon consisting of the rpl22 and rps3 genes encoding ribosomal proteins L22 and S3; (iii) the FD9 genetic locus, which contains the 3′-end of the rplO gene encoding ribosomal protein L15 and the secY gene encoding a translocase protein; and (iv) the secY-map genetic locus, consisting of 3′-end of the secY gene and the housekeeping map gene, which encodes methionine aminopeptidase. Amplification of the 16S rRNA gene was performed by nested PCR with the universal primers P1/P7 (Deng & Hiruki, 1991; Smart et al., 1996) and P1A/P7A (Lee et al., 2004) or F2n/R2 (Gundersen & Lee, 1996) with reaction conditions according to Lee et al. (2004). For analysis of ribosomal protein genes l22 and s3, amplification was performed using rp(V)F1/rpR1 primers for direct PCR, followed by nested PCR with the rp(V)F1A and rp(V)R1A primer pair, as described by Lee et al. (2004). The FD9 genetic locus was amplified by nested PCR with FD9f/r (Daire et al., 1997), followed by FD9f/r2, FD9f2/r or FD9f3/r2 primers, with PCR conditions according to Angelini et al. (2001). Amplification of the secY-map locus was performed by nested PCR using FD9f5/MAPr1 and FD9f6/MAPr2 primer pairs as previously described (Arnaud et al., 2007). Primers used for amplification of rpl22-rps3, rplO-secY and secY-map genes were specific for 16SrV-group phytoplasmas. PCR products were separated on 1% agarose gels in TBE buffer (Tris-Borate 90 mm, EDTA 1 mm), stained with ethidium bromide and visualized with a UV transilluminator.
RFLP analysis and DNA sequencing
For subgroup and strain differentiation among 16SrV phytoplasma group members, RFLP analysis was conducted on 16S rRNA, ribosomal protein and secY genes of all PCR-positive isolates. Digestion of amplified 16S rRNA and ribosomal protein genetic loci was performed with the recommended diagnostic enzymes for ‘Ca. Phytoplasma ulmi’ detection and differentiation from other 16SrV group phytoplasmas (Lee et al., 2004; Firrao et al., 2005). Amplicons of the 16S rRNA gene obtained by nested PCR with F2n/R2 primers were subjected to restriction digestion with RsaI and BfaI endonucleases (Fermentas), and nested PCR products of ribosomal protein genes l22 and s3 were digested with restriction enzymes MseI and Tsp509I (Fermentas). For FD9 genetic locus, RFLP analysis was performed by digestion of FD9f3/r2 nested PCR products with MseI enzyme (Angelini et al., 2001). All digestions were performed according to manufacturers’ instructions. EY1, FD-D and FD-C phytoplasma isolates were used as references to compare the restriction patterns in the three fragments. Restriction products were separated by electrophoresis on 13% polyacrylamide gels in TBE buffer, stained with ethidium bromide and visualized with a UV transilluminator.
For detailed analyses of strain differences among the ‘Ca. Phytoplasma ulmi’ isolates, sequencing of six isolates from different localities (Table 1) was performed on all four amplified phytoplasma genomic loci. Sequencing of four additional samples (two from Ljubičevo and two from Šuvajić, Table 1) was performed on the map gene locus because differences in this marker gene were not detected by means of RFLP. Amplicons of the 16S rRNA gene amplified with P1A/P7A primers, the ribosomal protein gene fragment amplified with rp(V)F1A/rp(V)R1A primers and the secY gene amplified with FD9f/r2 and FD9f2/r1 primers were all sequenced with the primers used for amplification and with additional intermediate primers. Sequences of the map gene were obtained as described by Arnaud et al. (2007). Final sequences of each gene marker were assembled after all nucleotide positions were covered at least two or three times. Sequencing was performed by BMR Service (Padova, Italy), and nucleotide sequence data were deposited in the GenBank database under the accession numbers HM038455-60 for the 16S rRNA gene, HM038461-64 for ribosomal protein genes, HM038465-70 for FD9 genetic locus and HM038471-80 for the map gene. The remaining two sequences of the ribosomal protein gene fragment of the ‘Ca. Phytoplasma ulmi’ strain from Serbia used in this study were taken from Jovićet al. (2008).
Comparative analyses of nucleotide sequences
Sequences obtained were compared with sequences of representative phytoplasma strains of the 16SrV group (Lee et al., 2004; Arnaud et al., 2007). Sequences of all four genes of representative strains were retrieved online from the National Center for Biotechnology Information (NCBI) website (http://www.ncbi.nlm.nih.gov/BLAST/) and aligned with Serbian ‘Ca. Phytoplasma ulmi’ isolates for comparative analyses using the clustal w program (Thompson et al., 1994). The presence of SNPs (single nucleotide polymorphisms) and indels (insertions and/or deletions) was recorded, and the positions of nucleotide (nt) changes were determined according to EY1T reference strain (Lee et al., 2004). The percentage sequence similarity between different ‘Ca. Phytoplasma ulmi’ strains from Serbia and ‘Ca. Phytoplasma ulmi’ reference strains was determined using the ‘pairwise distance calculation’ option of mega4 software (Tamura et al., 2007).
Virtual restriction analysis and gel plotting
In silico restriction analyses were performed on the obtained sequences of 16S rRNA, ribosomal protein, secY and map genes of Serbian ‘Ca. Phytoplasma ulmi’ isolates to identify suitable restriction enzymes for differentiation of new strains. Inside the range of every genomic locus analysed, the presence of SNPs in recognition sites for restriction endonucleases was determined using the pdraw32 program (AcaClone Software, http://www.acaclone.com).
Sequences of the FD9 marker (rplO-secY gene), the most variable of the four genomic loci analysed, were used for virtual electrophoresis gel analysis. The rplO-secY gene sequence of reference strain EY1T (GenBank acc. no. AY197690) was used as the reference for restriction pattern comparison. Sequences of different ‘Ca. Phytoplasma ulmi’ strains detected in this study were aligned with the reference isolate using clustal w and then trimmed to an approximately 1·2-kb fragment (the FD9f3/r2 fragment, delimited by FD9f3 and FD9r2 primer annealing positions) as previously described (Wei et al., 2007). Trimmed sequences were exported to the pdraw32 program, virtually digested with MseI endonuclease and separated on a 1·5% agarose gel. Virtual restriction patterns were compared with actual enzymatic RFLP patterns of representative amplicons.
Phylogenetic relationships among members of 16SrV group and Serbian ‘Ca. Phytoplasma ulmi’ isolates were assessed on the basis of sequences of all four genomic loci analysed. Sequences of 16S rRNA, rpl22-rps3, secY and map genes of representative phytoplasma strains in the 16SrV group (Lee et al., 2004; Arnaud et al., 2007) were retrieved from the NCBI website, as described above, and used to construct phylogenetic trees. Phylogenetic analysis using maximum parsimony was conducted with mega4 software. The reliability of the analyses was subjected to a bootstrap test with 1000 replicates.
Detection and RFLP characterization of ‘Ca. Phytoplasma ulmi’ isolates
‘Ca. Phytoplasma ulmi’ was detected in 25 out of 30 elm trees with symptoms in the sampled areas of Serbia (Table 1). None of the symptomless plants were positive for the presence of phytoplasmas. All four phytoplasma genomic loci were successfully amplified from every positive sample and 16S, rpl22-rps3 and secY genes were characterized by means of RFLP.
Digestion of F2nR2 amplicons of the 16S rRNA gene was performed with RsaI and BfaI diagnostic enzymes. Restriction profiles obtained with RsaI enzyme were identical among all 25 isolates and with the reference strains EY1, FD-C and FD-D, as expected (Fig. 1). By contrast, RFLP analysis of the same amplicons with BfaI endonuclease showed the presence of phytoplasmas belonging to the 16SrV-A subgroup in only two samples, both from the Tamnič site in East Serbia. The other 23 isolates had restriction patterns identical to the FD-C and FD-D reference strains (16SrV-C and 16SrV-D subgroups) (Fig. 1). For the ribosomal protein gene operon, MseI digestion of rp(V)F1A/rp(V)R1A amplicons showed the presence of 16SrV-A subgroup ‘Ca. Phytoplasma ulmi’ in all samples, but profiles obtained with Tsp509I were identical to the EY1T reference strain only in the two samples from Tamnič (data not shown). Tsp509I profiles of other phytoplasma isolates present in elm plants with symptoms were different from EY1 and more similar to FD-C and FD-D reference strains, as previously reported by Jovićet al. (2008).
RFLP analysis of the FD9 marker (rplO-secY gene) amplified with FD9f3r2 primers and digested with MseI endonuclease showed the presence of five different profiles among the phytoplasma isolates from Serbian elms (Fig. 2). None of the profiles obtained were identical to the EY1T reference strain. Samples from east Serbia (Tamnič site, as represented by strain EY24_SRB) had profiles similar to the EY phytoplasma strain ULW, as described by Angelini et al. (2001), and to strain EY626 belonging to the secYV-M subgroup (Lee et al., 2004). A second profile (as represented by strains EY1_SRB and EY6_SRB) was observed in all five isolates from Srednjevo, all four isolates from Ljubičevo and in three out of seven isolates from Šuvajić, all in northeast Serbia (Table 1). Another four isolates from Šuvajić had a third distinct restriction profile, which was detected only in samples from this locality (as represented by strain EY10_SRB). In southwest Serbia, six isolates of elm yellows phytoplasma had a fourth MseI profile type of the FD9 marker (as represented by strain EY18_SRB) and one sample had a fifth distinct profile type, which was detected only in this sample (as represented by strain EY20_SRB).
Genetic variation assessed by comparative analysis of nucleotide sequences
Genetic variation in 16SrV phytoplasma strains was determined on the basis of 16S, rpl22-rps3, secY and map genetic loci. Comparison between sequences of six ‘Ca. Phytoplasma ulmi’ strains from Serbia (designated in Table 1) and reference ‘Ca. Phytoplasma ulmi’ strains (EY1T, EY125, EY626 and EY627) showed different degrees of genetic variability across these four genes. In total, five different genotypes were recognized based on the collective sequences of all analysed marker genes because EY1_SRB and EY6_SRB strains had identical nucleotide composition. The sequence similarity in the 16S rRNA gene ranged from 99·8% to 100% among analysed members of the 16SrV-A phytoplasma subgroup, and only strain EY24_SRB showed complete identity with the EY627 reference strain. All other sequenced Serbian ‘Ca. Phytoplasma ulmi’ strains showed identical 16S gene nucleotide compositions with one another, but not with any of the reference strains. More variability was detected in sequences of the other three less well-conserved genes. Four analysed strains (EY1_SRB, EY10_SRB, EY18_SRB and EY20_SRB) had identical sequences of ribosomal protein genes, but shared only 99·3–99·5% sequence homology with reference ‘Ca. Phytoplasma ulmi’ strains. For the FD9 genetic marker, 99·3–99·6% sequence identity was detected among these strains compared to 98·5–98·8% similarity with reference strains. The map gene locus had 99·9% sequence identity within the group of Serbian strains, and 98·8–99·1% identity with the reference strains. Strain EY24_SRB proved to be most similar to the reference strain EY1T, with 100% sequence identity in the ribosomal protein genetic locus and 99·8% in the secY gene. This strain showed 100% identity with the E04-D714 strain of the MAP-EY1 phylogenetic cluster which is identical to the EY1 strain for the map locus (Arnaud et al., 2007).
For the 16S rRNA gene, 1457-bp sequences were obtained. In this region of the gene only one SNP mutation was detected among the Serbian strains, which was present in five out of six samples analysed by sequencing. However, comparative analysis showed that this mutation was located in one of the supposed unique oligonucleotide sequences that represent the signature sequence for ‘Ca. Phytoplasma ulmi’ species. The mutation was present at the 1100-bp position of the EY1T reference strain in the unique sequence encompassing positions from 1098 to 1108 bp (Lee et al., 2004). This SNP generates an additional recognition site for BfaI endonuclease and, in consequence, an RFLP pattern identical to the FD-D and FD-C strains (Fig. 1).
Analysis of ribosomal protein gene variability was performed on 1153-bp sequences comprising the complete rpl22 and rps3 genes. Collectively, six nucleotide changes were observed in five out of six sequenced strains, of which three were located in the range of species-specific oligonucleotide sequences for ‘Ca. Phytoplasma ulmi’ (Table 2). Two of these point mutations, at the 912- and 930-bp positions, were within the range of species signature sequences, causing different RFLP patterns in the new strains. As a consequence of these SNPs, Serbian ‘Ca. Phytoplasma ulmi’ strains (except for those from Tamnič) lack two recognition sites for Tsp509I at the 911- and 928-bp positions.
Table 2. Positions of SNPs in rpl22-rps3 genes of new strains of ‘Candidatus Phytoplasma ulmi’ (rpV marker)
In the most variable analysed genetic locus, FD9, five diverse genotypes were detected among phytoplasmas infecting Serbian elms, congruent with observed RFLP profiles. Within these genotypes, SNP mutations as well as insertions and deletions were detected (Table 3). Length variation in obtained sequences was from 1299 bp in strains with indels to 1326 bp in strains without indels. Only the EY24_SRB strain had a sequence with unique sites designated in the ‘Ca. Phytoplasma ulmi’ species description (Lee et al., 2004). Four other strains showed significant nt changes in the whole range of analysed marker genes, including species-specific sites. In total, 32 variable nt sites were detected, of which five were located within the range of three out of four secY gene signature sequences (Table 3). The insertion detected in three different genotypes was located at the 386-bp position of the EY1T reference strain in the secY gene and resulted in the addition of two amino acids in the encoded translocase protein. The deletion of 33 nt, causing loss of 11 amino acids in the encoded translocase protein, was observed in two strains from different localities (EY1_SRB and EY6_SRB; genotype EY-S1). The position of the deletion encompassed almost the complete range of two unique regions specific for ‘Ca. Phytoplasma ulmi’ (Table 3).
Table 3. Single nucleotide polymorphism (SNP) and indel (deletion/insertion) positions on secY gene in new strains of ‘Candidatus Phytoplasma ulmi’ in Serbia
For the map gene, 675- or 676-bp sequences were obtained from 10 sequenced isolates. Differences in sequence length were caused by a 1-nt insertion located in the intergenic sequence between the protein translocase and the map gene of the EY18_SRB strain. Five different map genotypes were observed, consistent with the secY genotypes. Genetic variability in this gene sequence was much lower than that of the FD9 marker and involved individual SNP mutations, except in the strain EY24_SRB, which differed from the other strains by 6–7 bp substitutions.
Phylogenetic positions and relatedness of the ‘Ca. Phytoplasma ulmi’ strains in Serbian elms with respect to one another and reference isolates of 16SrV-group phytoplasmas were determined by maximum parsimony analyses of 16S rRNA, ribosomal protein, secY and map genes. Phylogeny based on 16S rRNA gene sequences with AshY phytoplasma (ash yellows, ‘Ca. Phytoplasma fraxini’, 16SrVI group) as the outgroup resulted in 186 equally parsimonious trees. One of the most parsimonious trees showed a high degree of similarity and relatedness among 16SrV-group phytoplasma strains (Fig. 3). 16SrV phytoplasmas were grouped into only two phylogenetic clusters, supported by high bootstrap values (Fig. 3). One group of 16SrV-B phytoplasma strains (JWB, CLY-5, PY-In) formed a clear separate cluster with bootstrap support of 98%. The second cluster was comprised of the other 16SrV subgroups with a bootstrap value of 85%. Phylogenetic relationships among strains in this cluster were not well resolved, as indicated by the low bootstrap value (< 50%). With respect to reference ‘Ca. Phytoplasma ulmi’ strains, they formed a separate phylogenetic lineage within this cluster together with Serbian strain EY24_SRB, although with low bootstrap support. New strains of ‘Ca. Phytoplasma ulmi’ identified in this study formed a separate lineage in a poorly supported branch with 16SrV-C and 16SrV-D strains, being most closely related to the FD-C strain.
Phylogeny based on the ribosomal protein, secY and map genes revealed more phylogenetic divergence and clearer lineage sorting in the 16SrV phytoplasma group. One of the 97 most parsimonious trees for the ribosomal protein gene fragment, the most parsimonious tree of the FD9 genetic locus and one of the 49 most parsimonious trees of the map gene are shown in Figure 4. The highest bootstrap support and number of lineage appropriations was for the FD9 genetic locus. Topologies of the trees obtained for all three genomic loci analysed were similar, with clearly separate and highly supported grouping of reference and new ‘Ca. Phytoplasma ulmi’ strains in one of the three previously determined major phylogenetic groups of the 16SrV phytoplasma group (Lee et al., 2004). Two distinct lineages were resolved within this phylogenetic cluster: (i) a lineage comprised of reference ‘Ca. Phytoplasma ulmi’ strains and strain EY24_SRB from elms in east Serbia (here designated as rpV-EY1, SecY-EY1 and Map-EY1 lineages; Fig. 4) and (ii) a lineage comprised of new, more divergent strains of ‘Ca. Phytoplasma ulmi’ detected in Serbian elms EY1_SRB, EY10_SRB, EY18_SRB and EY20_SRB (here designated as rpV-EY2, SecY-EY2 and Map-EY3 lineages; Fig. 4). Because of the high genetic variability and resolving power of the FD9 genetic locus, the SecY-EY2 phylogenetic lineage split into two homogenous strain clusters comprised of isolates from northeast and from southwest Serbia. In conclusion, parsimony analyses of ribosomal protein, FD9 and map genetic loci congruently indicated a single monophyletic origin of reference and new ‘Ca. Phytoplasma ulmi’ strains, as they clustered on a branch supported by high bootstrap values of 99% (rpV), 100% (FD9) and 81% (map).
Putative restriction maps, diagnostic enzymes and new RFLP subgroup designation
Based on sequence analysis and putative restriction maps, reference and new strains of ‘Ca. Phytoplasma ulmi’ could be differentiated from other strains belonging to the 16SrV group by restriction analysis of 16S rRNA or the rpl22-rps3 ribosomal protein gene fragments. The 16S gene diagnostic enzyme is RsaI, as previously suggested (Lee et al., 2004; Firrao et al., 2005) which distinguishes between 16SrV-A and 16SrV-B subgroups. However, the BfaI enzyme was not applicable, as the new ‘Ca. Phytoplasma ulmi’ strains detected in this study have BfaI restriction profiles identical to the 16SrV-C, 16SrV-D and 16SrV-E subgroups (Fig. 1). MaeIII (‘GTnAC_) and BamHI (G_GTAC’C) are proposed as additional enzymes for practical use in diagnosis of ‘Ca. Phytoplasma ulmi’, as none of the 16SrV-A phytoplasma subgroup members have these recognition sites. In contrast, all members of the 16SrV-C and 16SrV-D subgroups have at least one MaeIII restriction site, as do most of the 16SrV-E subgroup members, with the exception of strain RUS (AY197648). Of all the presently known members of the 16SrV group, only strain RUS has the specific BamHI recognition sequence. The recognition sites for both proposed diagnostic enzymes are within a region of oligonucleotide sequence complementary to the unique region of the ‘Ca. Phytoplasma ulmi’ 16S rRNA, from the 827- to 831-bp position of the reference strain. The signature sequence is GGAAA, resulting in the absence of both MaeIII and BamHI recognition sequences in ‘Ca. Phytoplasma ulmi’ species members. In summary, the absence of MaeIII and BamHI recognition sites, together with RsaI restriction profile affiliation, can be used for diagnostic purposes for the analysis of ‘Ca. Phytoplasma ulmi’ species. Because of the characteristic BfaI restriction profile of strains EY1_SRB, EY10_SRB, EY18_SRB and EY20_SRB, and their low genetic variability (only one SNP mutation), they are assigned to subgroup 16SrV-A1 as a restriction subprofile group of the 16SrV-A subgroup (Table 4).
Table 4. Designations of ‘Candidatus Phytoplasma ulmi’ strains from Serbia among members of the EY phytoplasma group (16SrV)
Based on the RFLP patterns obtained and putative restriction maps of ribosomal protein gene fragments, only MseI endonuclease is confirmed as a proper diagnostic enzyme for identification of reference and new ‘Ca. Phytoplasma ulmi’ strains. Differentiation from other 16SrV group members can be made only for the reference 16SrV-A subgroup strains, but not for new ones, using the previously proposed diagnostic enzyme Tsp509I (Lee et al., 2004; Firrao et al., 2005) (Table 2). The new strains showed similar ribosomal protein gene fragment Tsp509I restriction profiles to the FD-D and FD-C strains of the 16SrV-C and 16SrV-D subgroups, confirming their nucleotide sequence similarity with members of these subgroups, as previously shown for the 16S gene. Because MseI differentiates all 16SrV-A subgroup members (reference and new strains) from other 16SrV-group phytoplasmas, it is proposed that this enzyme remain diagnostic for ‘Ca. Phytoplasma ulmi’ species and that enzyme Tsp509I should be used for specific identification of reference strains. Additionally, a new diagnostic enzyme, AccI (GT’mk_AC), is proposed for specific identification of new ‘Ca. Phytoplasma ulmi’ strains because only these strains have the AccI recognition sequence. Recognition sites for both Tsp509I and AccI enzymes occur inside the region of species-specific oligonucleotide sequences for ‘Ca. Phytoplasma ulmi’ (Table 2). The proposed three-enzyme diagnostic procedure ensures differentiation of 16SrV-A subgroup members from other elm yellows group phytoplasmas, and at the same time ensures identification of new and reference strains. Based on these restriction characteristics the new strains have been assigned to the rpV-N RFLP subgroup (Table 4).
Putative restriction maps of the secY gene and actual RFLP analysis of FD9f3r2 amplicons (Fig. 2) indicated that MseI restriction analysis of this genetic locus can be used as a diagnostic for differentiation all five ‘Ca. Phytoplasma ulmi’ genotypes present in Serbian elms, as well as the reference strain EY1T (Table 3). MseI restriction enables recognition of not only point mutations, but also of insertions and/or deletions, which are important features of each strain (Fig. 2). MseI restriction patterns indicate affiliation to the specific strain by the number and size of fragments up to 71 bp in length. All other fragments of the RFLP profiles are too short, resulting in poor pattern resolution, and thus are not useful for routine RFLP diagnostic procedures (Fig. 2). In particular, the presence of a 168-bp fragment in the MseI profile directly indicates the insertion of six nt at the 386 bp position of the reference strain, whilst a fragment of 162 bp is characteristic of a new strain without this insertion, EY10_SRB (Fig. 2). The size of a second fragment in the profile of the new strains (110 or 143 bp) directly points to the presence or absence of 33 nucleotides at the 596-bp position of the reference strain. Additionally, all reference and new ‘Ca. Phytoplasma ulmi’ strains have an 84-bp fragment, but only strain EY18_SRB also has a fragment of 96 bp in length (as a consequence of a point mutation at the 329-bp position), whilst all other strains show a fragment of 71 bp. Because reference ‘Ca. Phytoplasma ulmi’ strains EY1T and EY626 lack a MseI recognition sequence at the 529-bp position (Table 3), and EY626 also lacks a recognition site at the 628-bp position, the longest fragments in the profile of these strains are 261 and 305 bp, respectively. Only the new strain EY24_SRB and the reference strain EY626 cannot be differentiated by MseI restriction profiles. For differentiation of these strains, MboI is proposed as an applicable diagnostic enzyme based on putative restriction map analysis (Table 3). Using the unique nucleotide sequences properties and RFLP profiles, the new strains are assigned to the secYV-O, -P, -Q, -R and -S subgroups (Table 4).
Although five different genotypes were detected among ‘Ca. Phytoplasma ulmi’ strains infecting Serbian elms based on analysis of the map genetic locus, no congruent restriction enzymes could be designated as appropriate for RFLP diagnosis method on this gene.
The specific ecological, epidemiological and molecular properties of individual ‘Candidatus Phytoplasma’ species, as well as their economic importance and geographic distribution, have driven the effort to achieve proper species description by molecular characterization and phylogenetic reconstruction. The same is true for ‘Ca. Phytoplasma ulmi’, an economically important species in North America (NA) but not Europe, and whose distribution has to date been determined for midwestern and eastern NA states, but for only a few Western European countries. This phytoplasma species shares high 16S rRNA sequence similarity with other members of 16SrV-group phytoplasmas, but has specific biological properties, including specific host-plant associations and two confirmed specific vector species, Scaphoideus luteolus in NA (Baker, 1949) and Macropsis mendax in Italy (Carraro et al., 2004). A description of species with such diverse bioecological characteristics could be expected to be based on many strains of different geographical origins. However, the rules recommended by the Phytoplasma Taxonomy Group of the International Research Program on Comparative Mycoplasmology (IRPCM, 2004) for new ‘Ca. Phytoplasma’ species descriptions do not define the number of strains that should be characterized, so it is therefore left for authors to decide on a number. Consequently, the number of characterized genomic markers and/or different strains required for the description of each phytoplasma taxon differs significantly among different taxa; especially among those that are clearly differentiated on the basis of 16S rRNA gene sequences and those that are not. The incompatibility of available data is particularly problematic for pathogens like phytoplasmas which are not cultivable and are identified by means of nucleic acid sequence analysis.
As shown here, the limited number of strains analysed for species description can cause a problem in subsequent species identification because of the possibility of much higher genetic variability in species of interest than was assumed in the initial characterization. This could be a consequence of not only the number of strains analysed, but also the fact that strains within described species have not been previously elaborated with respect to geographic variability. The historical description of ‘Ca. Phytoplasma ulmi’ is a good example of such a case. This description has been based on multigene characterization of only four isolates, which were homogeneous in 16S rRNA and ribosomal protein genes, with only minor diversity in the secY gene. Strains characterized in the present study, with different geographical origins than those used for previous species description, suggest the presence of high genetic variability in elm yellows phytoplasma belonging to the 16SrV-A subgroup and point to different, more diverse multigene characteristics of this candidate species.
Considering the data included in the description of ‘Ca. Phytoplasma’ species that form the baseline for subsequent species identification, Firrao et al. (2005) proposed that the complete 16S rRNA gene sequence distinctive for the described species (upon which species phylogenetic relatedness was determined) need not be questioned, but that the phytoplasma species habitat (plant host) and characteristic oligonucleotide signature sequences should be verified. The results of the present research are in agreement with this proposal and highlight the need for continuous re-evaluation and verification of the latter species-unique characteristics. Verification, correction and/or update of species-specific oligonucleotide sequences, as shown in the case of ‘Ca. Phytoplasma ulmi’ strains infecting elm trees in Serbia, also indicates the possibility that widely used RFLP diagnostic procedures for phytoplasma identification may need to be modified as well.
Signature sequences of the 16S rRNA gene are important features for designating taxa for prokaryotic species, especially putative ones. These oligonucleotide sequences, which are complementary to the unique region of the 16S rRNA, enable clear recognition of the taxon of interest and differentiation from related taxa (e.g. Murray & Schleifer, 1994; IRPCM, 2004; Firrao et al., 2005; Brown et al., 2007). Prior to extensive availability of PCR for DNA amplification, short oligonucleotide sequences comprising variable parts of otherwise conserved 16S genes were used for phylogenetic analysis and species recognition and differentiation (e.g. Woese et al., 1980). Considering the tools from molecular biology now available for DNA analysis and data processing (especially comparative analysis of nucleotide sequences and phylogenetic analysis), these short oligonucleotide signature sequences could be more problematic than useful for species identification when applied to uncultured, highly variable, putative taxa like ‘Ca. Phytoplasma ulmi’. As shown here, misleading putative identification of taxa can further arise from more detailed description of signature sequences, including designation of species-specific sequences for highly variable genes.
In the description of ‘Ca. Phytoplasma ulmi’, Lee et al. (2004) included signature sequences not only for the 16S gene (which is the most studied gene not only for phytoplasmas but for all prokaryotes), but also for two additional gene regions, rpl22-rps3 and secY. At the time, this strategy may have seemed the most promising for a more precise definition of ‘Ca. Phytoplasma ulmi’ species, based on nucleotide sequence analyses of four strains with low genetic variability, and considering that this proposed putative taxon shares more than 97·5% 16S rRNA sequence identity with the previously described taxon ‘Ca Phytoplasma ziziphi’. Based on clearly separate and high bootstrap value-supported phylogenetic clustering of proposed ‘Ca. Phytoplasma ulmi’ reference strains for ribosomal protein and secY gene fragments, designation of signature sequences in these genes seemed to have valuable diagnostic importance. The results of the present study have confirmed that ‘Ca. Phytoplasma ulmi’ is a separate taxon, as defined by Lee et al. (2004), but demonstrate that the taxon has a much higher strain diversity than previously assumed, presumably because of the genetic variability of strains from geographically distant localities. Phylogenetic relatedness and congruent lineage clustering of three gene fragments, rpl22-rps3, secY and map, confirm the common origin of both reference and newly detected ‘Ca. Phytoplasma ulmi’ strains infecting elm trees in Serbia. However, unlike phylogenetic clustering, nucleotide sequence comparison revealed the presence of mutations inside signature sequences of all three genes, 16S, rpl22-rps3 and secY, used for characterization and assignment of 16SrV-A phytoplasmas to the ‘Ca. Phytoplasma ulmi’ species. In addition, the presence of insertions and/or deletions was recorded in secY, with deletions located in the range of two unique oligonucleotide sites. This gene was recently evaluated for usefulness in phytoplasma taxonomy and was found to be more variable than other genes; thus, it is useful for the differentiation of closely related strains within the same 16Sr group, and for inferred phylogenetic clustering congruent to 16S gene inferred phylogeny (Lee et al., 2010).
In the present study, comparative analysis of reference ‘Ca. Phytoplasma ulmi’ strains and newly-detected strains infecting elm trees in Serbia was performed using the highly conserved 16S rRNA gene as a primary parameter for characterization and differentiation, and three more variable phylogenetic markers (rpl22-rps3, secY and map) with higher resolving power for finer discrimination. Based on these results, four subtypes are proposed, representing four genotypes of ‘Ca. Phytoplasma ulmi’: EY-S1, EY-S2, EY-S3 and EY-S4 (Table 1). Designation of subtypes is established by clear genetic differences and divergence of strains detected in Serbian elms compared with reference strains, as well as separate phylogenetic lineage sorting. As shown by RFLP and a nucleotide sequence analyses of 16S and rpV gene markers, these ‘Ca. Phytoplasma ulmi’ subtypes diverge from the reference strain EY1T by mutations in unique oligonucleotide sequences, but are identical to one other. Analysis of map and FD9 genetic loci, which are more variable and thus more efficient marker genes for differentiation of closely related strains, revealed the presence of four different strains (subtypes). In particular, the identity of new ‘Ca. Phytoplasma ulmi’ subtypes is 98·5–98·8% divergent within the FD9 marker sequence compared to the reference strains. These subtypes can be precisely and reliably identified by RFLP analysis of the FD9 locus using the diagnostic enzyme MseI, as shown in Figure 2.
It seems that the high diversity of EY phytoplasmas in elm trees in Serbia is not unique. Recent findings and preliminary characterization of ‘Ca. Phytoplasma ulmi’-related strains from France and the Czech Republic also indicates the presence of genetically divergent isolates (Malembic-Maher et al., 2009; Navrátil et al., 2009), although more comprehensive research is required to determine the level of diversity and distribution patterns of ‘Ca. Phytoplasma ulmi’ strains across Europe. Considering the obvious diversification of ‘Ca. Phytoplasma ulmi’ strains and the wide geographic distribution of this species, it is likely that more different strains will be detected and characterized in the coming years.
The high genetic diversity of ‘Ca. Phytoplasma ulmi’ strains in Serbia, as well as in France and the Czech Republic, indicates that this pathogen could be native to Europe, although available data are limited. In support of this proposition, other 16SrV-group phytoplasmas, such as flavescence dorée strains (16SrV-C and 16SrV-D subgroups), which were long assumed to be of NA origin, have been found to be autochthonous to Europe (Arnaud et al., 2007; Filippin et al., 2009). More strains, especially those from NA, should be analysed to verify this hypothesis. Extensive genotyping of FD phytoplasma strains associated with Alnus glutinosa in France (Arnaud et al., 2007) and Clematis vitalba in Italy and the Balkans (Filippin et al., 2009) strongly suggest that they are original host plants of the disease, which has significant implications for the ecological and epidemiological cycle of these pathogens. Thus, further studies of the diversity and virulence of different ‘Ca. Phytoplasma ulmi’ strains associated with elm trees in particular geographic regions is of great importance to predict and prevent potential new shifts from native phytoplasma reservoirs to plants of economic importance.
We are very grateful to Elisa Angelini for a critical review of the manuscript and helpful suggestions, as well as for providing phytoplasma reference isolates. This research was supported by grant III43001 from the Ministry of Science and Technological Development of the Republic of Serbia.