DNA samples from ‘Ca. P. asteris’ HP, RhY, and PaWB-Japan strains, and from ‘Ca. P. phoenicium’ PEY were extracted from infected Hydrangea spp. (Takinami et al., 2013), Rhus javanica (Takinami et al., 2013), Paulownia tomentosa (Kakizawa et al., 2006), and periwinkle (Maejima et al., 2014), respectively.

The phytoplasma species and strains used in this study are listed in Supplementary Table S1. Contigs of phyllogen sequences were created from draft genomes of the HP, RhY, and PaWB-Japan strains sequenced in this study (see Supplementary Methods), as well as from phytoplasma genomic sequences deposited in GenBank (Supplementary Table S1); we performed tBLASTn searches using amino acid sequences from phyllogen homologs (Iwabuchi et al., 2020). To determine upstream sequences for the RhY, PaWB-Japan, and PEY strain contigs, we performed polymerase chain reaction (PCR) using primer pairs that targeted each contig and conserved regions of phytoplasma fliA, which is frequently found in the first or second open reading frame (ORF) of PMUs (Arashida et al., 2008a; Wang et al., 2018; Supplementary Table S2). The flanking regions of the RhY and PaWB-Japan strain contigs were determined by PCR using primer pairs designed to match the corresponding regions of the HP and PaWB-China genomes, respectively (Supplementary Table S2). The downstream region for the PEY strain was determined by PCR using primer pairs that targeted the phyllogen sequence and dnaG sequence in the AY-WB strain PMU (Supplementary Table S2). PCR was performed using LA Taq (TaKaRa Bio, Inc., Shiga, Japan) and 0.3 µM of each primer, in accordance with the manufacturer’s instructions. Amplified fragments (>1.5 kbp) were purified and sequenced using PCR primers, followed by primer walking. Amplified fragments (<1.5 kbp) were purified, cloned into a pCR2.1-TOPO vector (Invitrogen, Carlsbad, CA, United States), and sequenced by Sanger sequencing. DNA samples from phytoplasma strains containing phyl-A group phyllogens (Supplementary Table S1) were subjected to PCR amplification of the region between hflB-like genes with similarity to hflB and a region downstream of phyllogens using primers described previously (Jomantiene et al., 2007; Supplementary Table S2). The phyllogen flanking sequences were added to the draft genomes and deposited in DDBJ under the accession numbers listed in Supplementary Table S1.

The genomic regions that flanked the phyllogens were annotated for protein coding genes using MetaGeneAnnotator (Noguchi et al., 2008) in DFAST (Tanizawa et al., 2018) and analyzed using BLASTp searches restricted to sequences >8,000 bp (Altschul et al., 1990; Camacho et al., 2009). Homologous genes were identified when the E-value, identity, and query-cover between two genes were < 1e−6, >30%, and >75%, respectively. To identify genomic locations of phyllogen flanking regions in strains related to ‘Ca. P. asteris,’ genomic regions were subjected to comparative analyses with other genomes of strains related to ‘Ca. P. asteris’ using BLASTp. The results of these comparative analyses were displayed using Clinker ver. 0.0.21 software to enable gene cluster comparisons (Gilchrist and Chooi., 2021). Pairwise sequence identities were calculated using the Sequence Demarcation Tool (SDT; ver. 1.2) (Muhire et al., 2014). Pseudogenes split into multiple ORFs due to premature stop codons were identified by combining the ORFs.

The nucleotide sequences of phyllogen, fliA, himA, and hflB were picked up from the list in Supplementary Table S3 and were aligned using the MUSCLE algorithm in MEGA X software (Kumar et al., 2018). The neighbor-joining (NJ) method (Saitou and Nei, 1987) was used to create phylogenetic trees of these genes using MEGA X software (Kumar et al., 2018). The whole/draft genome-based phylogenetic analysis of phytoplasma strains related to ‘Ca. P. asteris’, except for NJAY strain (GenBank accession number MAPF00000000; completeness of only 91%) and TW strain (GenBank accession number QGKT00000000; draft genome reportedly contaminated with two different strains; Cho et al., 2020), was performed as follows. Single-copy genes shared by the ‘Ca. P. asteris’-related phytoplasma genomes (Supplementary Table S3) were identified using SonicParanoid software ver. 1.3.5 (Cosentino and Iwasaki, 2019). The identified homologs were aligned as described above. Aligned sequences were combined for each strain in the same order, and were subjected to phylogenetic analyses using the NJ method as described above.

A modified tobacco rattle virus (TRV)-based gene expression vector system (Iwabuchi et al., 2019) was used to express phyllogen in planta. In brief, the PHYL_RP166 and PHYL_NCHU2019 sequences were optimized for plant codons and synthesized by ThermoFisher Scientific (Waltham, MA, United States; Supplementary Table S4). The fragments were cloned into pTRV2-2A-sGFP with the primers shown in Supplementary Table S2 by replacing the sGFP region using the NEBuilder HiFi DNA Assembly Cloning kit (New England Biolabs, Ipswich, MA, United States). PHYL_HP and PHYL_OY ^E33 were cloned with the primers shown in Supplementary Table S2 by adding a single amino acid mutation into pTRV2-cloned PHYL_JWB (Iwabuchi et al., 2020) and PHYL_OY (Iwabuchi et al., 2019), respectively, using the GeneArt site-directed mutagenesis system (Invitrogen). PHYL_PEY was cloned with the primers shown in Supplementary Table S2 by incorporating a two-amino acid mutation into pTRV2-cloned PHYL_RP166, amplifying the insert by PCR, and amplifying the rest of the plasmid by inverse PCR. The amplified fragments were ligated using the NEBuilder HiFi DNA Assembly Cloning kit (New England Biolabs). Arabidopsis thaliana was maintained as described previously (Iwabuchi et al., 2020). Agrobacterium tumefaciens EHA105 cells containing pTRV1 and pTRV2-empty, pTRV2-PHYL_OY, pTRV2-PHYL_RP166, pTRV2-PHYL_PEY, pTRV2-PHYL_NCHU2019, pTRV2-PHYL_HP, pTRV2-PHYL_JWB, or pTRV2-PHYL_OY ^E33 were adjusted to an OD₆₀₀ of 0.1, mixed at a ratio of 1:1, and co-infiltrated into 2- and 3-week-old A. thaliana as described previously (Takahashi et al., 2006).

To understand the mechanisms involved in the HGT of phyllogens across phytoplasmas, we compared phyllogen flanking regions among the genomes of 17 phytoplasma strains related to six species (Figure 1; Supplementary Table S1). Strains harboring the phyl-B group phyllogens were not analyzed because the corresponding genomes have not been characterized. Prior to this analysis, we generated draft genome data for the ‘Ca. P. asteris’ HP, RhY, and PaWB-Japan strains (Supplementary Figure S1; Supplementary Table S5). The draft genome data were used to generate full-length phyllogen sequences for the RhY and PaWB-Japan strains, as well as sequences for phyllogen flanking regions from the HP, RhY, and PaWB-Japan strains. The detailed results about genome sequencing were described in Supplementary Results. The phyllogen flanking regions for the ‘Ca. P. phoenicium’ PEY strains were also determined using PCR. Phylogenetic analyses of phyllogens showed that the genes of the 17 strains could be separated into three clades (Figure 1A) corresponding to the phyl-A (6 strains), phyl-C (1 strains), and phyl-D groups (10 strains) described by a previous study (Iwabuchi et al., 2020); the RhY and PaWB-Japan phyllogens belong to the phyl-D group. In most phyllogen flanking regions (all phyl-A and phyl-C, and seven phyl-D group phyllogens), there were several PMU-associated genes (fliA, ssb, dam, himA, hflB, smc, tmk, dnaB, dnaG, and tra5; Figure 1B; Arashida et al., 2008a). This observation is consistent with previous reports of phytoplasma strains with phyllogens within PMUs (Sugio and Hogenhout, 2012; Maejima et al., 2014).

Except for those from the WBD strain, phyllogen flanking PMUs frequently contained genes (e.g., fliA, himA, and hflB) in their upstream regions. These PMUs could be categorized as type 1 or 2 based on some characteristic genes and synteny (Figure 1B). PMU typing was already conducted based on the order of tmk and dnaB in a previous paper (Huang et al., 2022). Since these genes were not all located in the phyllogen flanking PMUs, we classified them based on other genes as described below.

The type 1 PMUs were found in six strains (‘Ca. P. asteris’ AY-WB, DY 2014, OY, RP166, WBD, and ‘Ca. P. phoenicium’ PEY), all of which contained phyl-A group phyllogens. These type 1 PMUs also contained hypothetical protein 1 (hp1), hp2, and an hflB-like gene in their upstream regions (Figure 1B). Additionally, the type 1 PMUs exhibited conserved synteny of fliA, ssb, himA, hflB, hp1, hp2, hflB-like, phyllogen, and hp3, except for the PMUs of the WBD strain. In the WBD strain, only the synteny of hp2, hflB-like, and the phyllogens was conserved. In the ‘Ca. P. pruni’ CP and ‘Ca. P. trifolii’ CPS strains, as well as in some strains related to ‘Ca. P. asteris,’ only the region between the hflB-like genes and the intergenic regions downstream of the phyllogens were characterized; these regions were conserved in all of these strains, and in the OY and WBD strains (Supplementary Figure S2). This result suggests that the type 1 PMUs that harbor phyl-A group phyllogens are conserved in ‘Ca. P. pruni,’ ‘Ca. P. trifolii,’ ‘Ca. P. asteris,’ and ‘Ca. P. phoenicium’.

Type 2 PMUs were found in seven strains that have phyl-D group phyllogens (‘Ca. P. asteris’ HP, RhY, PaWB-Japan, PaWB-China, ‘Ca. P. luffae’ NCHU2019, ‘Ca. P. ziziphi’ JWB-nky, and Hebei-2018). Type 2 PMUs were also found in the ‘Ca. P. pruni’ Vc33 strain, which has a phyl-C group phyllogen. The type 2 PMUs contained hp4, hp5, hp6, hp7, smc, and smc-like genes with similarity to smc and hp8 genes in the upstream region and hp9 genes in the downstream region. In the PaWB-Japan, PaWB-China, HP, and RhY strains, the type 2 PMUs exhibited conserved synteny of fliA, himA, hp4, hp5, hp6, hp7, hflB, smc, smc-like genes, hp8, phyllogen, and hp9. The JWB-nky and Hebei-2018 strains had another phyllogen, and its flanking region also contained genes that were characteristic of type 2 PMUs, although synteny was only partly conserved (Supplementary Figure S3A). The NCHU2019 strain had three type 2 PMUs with the same gene synteny (Supplementary Figure S3B). These results suggest that type 2 PMUs harboring phyl-D or phyl-C group phyllogens are conserved in four different species (‘Ca. P. asteris,’ ‘Ca. P. luffae,’ ‘Ca. P. ziziphi,’ and ‘Ca. P. pruni’). Additionally, phylogenetic trees based on genes shared by type 1 and 2 PMUs (i.e., fliA, himA, and hflB) indicate that the evolutionary history of these genes differs from that of phytoplasmas. In particular, the hflB tree formed two clades, with type 1 and 2 PMU strains (Supplementary Figure S4). Thus, our data indicate that the phylogeny of phyllogens is consistent with PMU type (Figure 1).

On the other hand, several phyllogens were not located within PMUs (indicated by asterisks; Figure 1). No PMU-associated genes were found in the 20 kbp upstream or downstream of the phyl-D group phyllogen in ‘Ca. P. aurantifolia’ PnWB, or in the ‘Ca. P. asteris’ MD-China strains (except for a dam gene in the MD-China strain; Supplementary Figure S6). Additionally, the phyl-D group phyllogen in the ‘Ca. P. aurantifolia’ EPWB strain was located near the PMU-associated genes, but outside of PMU regions (Supplementary Figure S5).

Some PMUs contain fewer genes than PMUs reported to exist as probably transposable circular extrachromosomal elements such as AY-WB PMU1 (Toruño et al., 2010) and some PMU-associated genes are truncated (Bai et al., 2006; Ku et al., 2013). To understand the evolution of PMUs harboring phyllogens, we analyzed the conservation of genes in type 1 and 2 PMUs using pairwise amino acid sequence comparisons. In the type 1 PMUs, phyllogens were highly conserved (87%–100%) between ‘Ca. P. asteris’ and ‘Ca. P. phoenicium.’ However, genes in the phyllogen flanking region (fliA, ssb, himA, hflB, hp1, hp2, and hflB-like) exhibited low sequence identities (>57%), except in the closely related OY and DY2014 strains (Supplementary Table S6; Figure 2A). In the type 2 PMUs, phyllogens were also highly conserved (91%–100%) among ‘Ca. P. asteris’, ‘Ca. P. luffae,’ and ‘Ca. P. ziziphi,’ except for the Vc33 strain phyllogen that belongs to a different phyl-C group. However, the genes in the phyllogen flanking regions (fliA, himA, hp6, hp7, hflB, smc, smc-like, and hp9) exhibited low sequence identities (>32%), except for the closely related HP and RhY strains, and JWB-nky and Hebei-2018 strains (Supplementary Table S7; Figure 2B). Additionally, several PMU-associated genes, including hflB in the OY strain and smc in the HP, RhY, and PaWB-Japan strains, were truncated by premature stop codons due to frameshifts or single-nucleotide polymorphisms (Figure 1B). Moreover, the RP166, OY, PaWB-Japan, PaWB-China, RhY, and HP strains lacked PMU-associated genes downstream of phyllogens (e.g., tmk, dnaB, dnaG, and tra5; Figure 1B). These results indicate that although phyllogens are highly conserved in both types of PMU, phyllogen flanking genes have accumulated mutations, undergone pseudogenesis, and sometimes been lost altogether.

Next, we focused on relationships among phyllogens in ‘Ca. P. asteris’ because sequence information and genetic diversity were richer in ‘Ca. P. asteris’ than in other phytoplasma species (Iwabuchi et al., 2020). Although many strains related to ‘Ca. P. asteris’ had phyl-A group phyllogens, several strains had phyl-D group phyllogens.

First, intraspecies evolutionary relationships among strains with phyllogens were assessed by analyzing single-copy genes present in all whole/draft genome sequences. In total, 16 strains related to ‘Ca. P. asteris’ were analyzed, with a ‘Ca. P. meliae’ strain used as an outgroup. A WBD strain, which had recently been proposed for reclassification to ‘Ca. P. tritici’ (Zhao et al., 2021), was also used in this analysis. The resulting phylogenetic tree showed that strains related to ‘Ca. P. asteris’ formed two clades (Figure 3A). The minor clade comprised the AY-WB and WBD strains, each of which contained phyl-A group phyllogens. The major clade comprised four subclades and the TBZ1 strain. Two subclades comprised the strains with phyl-D group phyllogens; one of these subclades comprised the HP, RhY, and MD-China strains, and the other comprised the PaWB-Japan and PaWB-China strains. No phyllogens were found in the draft genome of the SW86 strain, which belonged to the former subclade. Another subclade comprised the strains containing phyl-A group phyllogens (i.e., the CYP, RP166, DY 2014, and OY strains). The final subclade comprised strains with complete (De Villa and M3 strains) or draft (LD1 strain) genomes that lacked phyllogens. The TBZ1 strain that contained a phyl-A group phyllogen did not belong to any of these subclades. These results indicate that the strains with phyl-D group phyllogens formed two subclades in ‘Ca. P. asteris’, while the strains with phyl-A group phyllogens formed one clade and one subclade. They also indicate that although strains related to ‘Ca. P. asteris’ retained the same group of phyllogens, at least at the subclade level, the overall relationships were complex.

The positions of the mobile elements within genomes may provide important evolutionary information. Therefore, we analyzed the genomic locations of PMUs in strains related to ‘Ca. P. asteris,’ based on their flanking genomic regions. First, we analyzed the genomic locations of the type 1 PMUs. The PMUs and surrounding genomic regions in the OY, DY 2014, RP166, and AY-WB strains are shown in Figure 3B; Supplementary Figure S6B. The type 1 PMUs of the OY, DY 2014, and RP166 strains, which belong to the same subclade as ‘Ca. P. asteris’ (Figure 3A), are flanked by several single-copy genes. These genes include dppD, oppF, ibpA, and pacL in the upstream region, and nrdE, nrdF, mscL, and mdlB in the downstream region (Figure 3B). The syntenies of these single-copy genes were also conserved in the other ‘Ca. P. asteris’ genomes, but there were no phyllogens in the corresponding regions, except in the HP and RhY strains. However, type 2 PMUs harboring phyllogens were found in these corresponding genomic regions. The type 1 PMU in the AY-WB strain was in a different genomic region, and was flanked by several single-copy genes including pdhC, acoL, tatD, and plsX in the upstream region and rpsD, mgtA, degV, and tsaD in the downstream region (Supplementary Figure S6B). The syntenies of these genes were conserved in other ‘Ca. P. asteris’ genomes, although no type 1 PMUs or phyllogens were found. These results indicate that the type 1 PMUs were inserted into different regions in the AY-WB and OY/DY2014/RP166 strains. Next, we analyzed the genomic locations of the type 2 PMUs. The PMUs and surrounding genomic regions in the PaWB-Japan and PaWB-China strains, which belong to the same subgroup, are shown in Figure 3C. These PMUs were delimited by a complete and truncated tra5 gene downstream and upstream of the phyllogens, respectively (Bai et al., 2006). Several single-copy genes were also found near the PMUs, but these genes were different from those surrounding the type 1 PMUs; they included rpsP, prfB, secA, and ypgC in the upstream region and tengu, glnQ, and artl in the downstream region (Figure 3C). Although the syntenies of these single-copy genes were also conserved in other ‘Ca. P. asteris’ genomes, there were no type 2 PMUs or phyllogens. The genomic region between the ypgC and tengu genes in the OY, DY 2014, De Villa, M3, and WBD strains was <13 kbp, which was much shorter than in the PaWB-Japan (35 kbp) and PaWB-China (67 kbp) strains. Several PMU-associated genes were found in this genomic region in the RP166 and AY-WB strains; however, the syntenies of these genes differed from those of the type 2 PMUs (Figure 3C). On the other hand, the type 2 PMUs in the HP and RhY strains were flanked by several single-copy genes different from those present in the PaWB strains (Figure 3B). The phyl-D group phyllogen present in the MD-China strain was also flanked by single-copy genes different from the genes flanking other phyllogens (Supplementary Figure S6). These results indicate that type 2 PMUs were inserted into different regions in the PaWB-Japan/PaWB-China, MD-China, and HP/RhY strains.

To correlate the inserted type 2 PMUs with other genomes at the nucleotide level, the upstream and downstream sequences of the PMUs in the PaWB strains were aligned with the corresponding genomic regions in the OY, De Villa, and DY2014 strains. The alignments identified regions of approximately 350 bp at both ends of the insertions that contained inverted repeat sequences, which is characteristic of transposon insertion (Szeverényi et al., 2003), and these inverted repeats were conserved between the PaWB strains (Figure 4; Supplementary Figure S7). Although these inverted repeat sequences were not identical in the other three strains, there were nearby sequences with strong similarity among these genomes (Figure 4), suggesting that PMU boundaries were located near the inverted repeats. These results indicate that a type 2 PMU was inserted at this position in the common ancestor of the PaWB strains.

Comparative analyses revealed conservation of amino acid sequences of the PMU-associated phyllogens. Thus, we examined functional conservation of the phyllogens by testing the capacity of various phyllogens to induce phyllody. We tested the phyl-A group phyllogens of RP166 (PHYL_RP166) and the PEY strains (PHYL_PEY), as well as the phyl-D group phyllogens of NCHU2019 (PHYL_NCHU2019) and the HP strains (PHYL_HP). We used the PHYL_OY phyllogen of the OY strain (phyl-A group) and the PHYL_JWB phyllogen of the JWB strain (phyl-D group) as controls known to induce phyllody (Iwabuchi et al., 2020). Each phyllogen was expressed in A. thaliana using a TRV-based gene expression vector system. All of the tested phyllogens converted sepals, petals, and stamens into leaf-like structures with stellate trichomes, changed pistils into secondary vegetative shoot-like structures, and enlarged flowers (Figure 5A), as did the PHYL_JWB phyllogen control. These results showed that the functions of the type 1 and 2 PMU-associated phyllogens were as highly conserved as their sequences.

Interestingly, in the PHYL_OY expression test, the stamens of most flowers were not converted into leaf-like structures, the other floral organs had fewer trichomes, and the flowers did not enlarge (Figure 5A). This suggested that PHYL_OY exhibited a reduced capacity to induce phyllody. Sequence comparisons revealed that PHYL_OY had a unique polymorphism at position 33 (lysine; Figure 5B). A PHYL_OY mutant with a reciprocal substitution at this position (glutamate; PHYL_OY ^K33E) induced phyllody to the same extent as the other homologs (Figure 5A). Therefore, this amino acid is important for strong induction of phyllody.

The apparent differences between the evolutionary histories of phyllogens and 16S rRNA genes in phytoplasmas suggest that phyllogens may have undergone HGT (Iwabuchi et al., 2020; Kalla et al., 2021). However, the mechanisms involved were previously unclear due to the lack of phyllogen flanking-region sequence data. In this study, we determined draft genomes for three strains and analyzed phyllogen flanking regions from 17 phytoplasma strains related to six different species. We found that most phyllogens were associated with PMUs (Figure 1). Furthermore, phyllogen groupings correlated closely with the types of PMU (Figure 1). These results indicate that PMUs drive the horizontal transfer of distinct groups of phyllogens between phytoplasma species and strains. This study demonstrates that the symptom-determinant effectors of phytoplasmas are evolutionarily correlated with the types of PMU rather than the phytoplasma genomes.

In addition to phyllogens, three other effectors (TENGU, SAP11, and SAP05) have been linked experimentally with phytoplasma virulence (Hoshi et al., 2009; Sugawara et al., 2013; Minato et al., 2014; Cho et al., 2019; Huang et al., 2021). Among these, the SAP11 and SAP05 genes are also located near PMUs; they are conserved in various species (Sugio and Hogenhout., 2012; Cho et al., 2019; Huang et al., 2022) and have phylogenetic trees that differ from those of phytoplasmas (Chang et al., 2018; Huang et al., 2022). These observations suggest that PMU-mediated HGT may play a major role in the transfer of virulence genes across phytoplasma species.

Many studies have reported HGT of bacterial virulence genes based on genomic analyses of individual species (Ma et al., 2006; McCann and Guttman, 2008; Drew et al., 2021), but few have shown how HGT can occur via transposable elements (such as transposons) by comparing the genomes of different bacterial species. Thus, this study provides new insight into the mechanisms of HGT of virulence genes.

Genetic elements may become immobilized due to inactivation or deletion of genes necessary for their transfer (Dobrindt et al., 2004). Degeneration of PMUs has also been reported in several phytoplasma strains, including AY-WB (Bai et al., 2006; Ku et al., 2013). In this study, some PMUs harboring phyllogens lacked tra5 (Figure 1B). In addition, some PMU genes had accumulated mutations, were truncated, or had been lost altogether (Figures 1B, 2), including genes with putative roles in DNA recombination, replication, and transposition (Arashida et al., 2008a; Ku et al., 2013). Therefore, most of these PMUs may begin to lose their capacity for transposition after the phyllogens have been acquired. However, the type 2 PMUs of the PaWB-Japan and PaWB-China strains retained their tra5 genes and inverted repeat-like sequences at both ends (Figures 3, 4), suggesting that they had also retained their capacity for transposition.

Despite the deterioration of PMU-associated genes, phyllogens from different species had highly conserved sequences, especially phyllogens from the phyl-A and -D groups (Figure 2). Furthermore, phyllogen functions were conserved (Figure 5). Several phyllogens were not located within PMUs (Figure 1A; Chung et al., 2013; Luo et al., 2022) and their flanking regions exhibited considerable variation (data not shown), suggesting that after phyllogen transfer, PMU sequences may deteriorate completely. Therefore, phyllogens may become fixed in many different phytoplasma genomes after their acquisition, suggesting that they are crucial for their hosts. Previous studies have shown that phyllody symptoms facilitate phytoplasma accumulation/localization within host plants (Arashida et al., 2008b; Su et al., 2011), and phyllogens can attract insect vectors (MacLean et al., 2014). Although further studies are needed to investigate the roles of phyllody symptoms in host adaptability, the fact that phyllogens have been strongly conserved throughout evolution suggests that they are critical for phytoplasma survival.

To date, three different groups of phyllogens have been identified in strains related to ‘Ca. P. asteris’ (phyl-A, -B, and -D; Iwabuchi et al., 2020). In this study, we found that the acquisition of these phyllogens was a complex process, at least for phyl-A and phyl-D. In ‘Ca. P. asteris,’ we identified one subclade comprising strains that retained phyl-A phyllogens and two subclades comprising strains that retained phyl-D phyllogens (Figure 3A). In the former subclade, the PMUs retaining phyl-A were found in the same region of the genomes (Figure 3), suggesting that these PMUs were acquired by their common ancestor. The same region also surrounded PMUs of the HP and RhY strains (Figure 3). Therefore, this region may be a hot spot for genome rearrangement in phytoplasmas, as described by a previous study (Arashida et al., 2008b). PMUs retaining phyl-D were found in different genomic regions in the PaWB-Japan/PaWB-China, MD-China, and HP/RhY strains (Figure 3; Supplementary Figure S6). This observation suggests that PMUs harboring phyllogens either moved frequently within the genome or were acquired separately. The PaWB strains had complete PMUs retaining phyl-D, suggesting that these were acquired more recently than the PMUs in the HP and RhY strains. Therefore, the latter hypothesis seems more plausible. We could not determine the acquisition order of the phyl-A and phyl-D group phyllogens in ‘Ca. P. asteris’ due to insufficient genomic data. Therefore, further studies of phyllogen flanking regions and their genomic positions are needed.

HGT may occur between organisms that share an ecological niche, irrespective of their phylogenetic relationships (Polz et al., 2013). Phylogenetic analyses have shown that all of the phyl-D group phyllogens of ‘Ca. P. asteris’ are most closely related to phyllogens of ‘Ca. P. ziziphi’ (Iwabuchi et al., 2020; Figure 1A). Furthermore, several lines of evidence indicate that strains related to ‘Ca. P. asteris’ and ‘Ca. P. ziziphi’ share common ecological niches. First, the distribution of ‘Ca. P. ziziphi’ is limited to eastern and southern Asia (Jung et al., 2003; Rao et al., 2017), as is the distribution of strains related to ‘Ca. P. asteris’ that retain phyl-D (Kakizawa et al., 2006; Takinami et al., 2013; Luo et al., 2022). Second, co-infection of ‘Ca. P. asteris’ and ‘Ca. P. ziziphi’ has been reported in jujube plants (Sun et al., 2013). Third, Hishimonus sellatus, a vector of ‘Ca. P. ziziphi,’ can transmit two strains related to ‘Ca. P. asteris’ that retain phyl-D (RhY and MD-China; Tanaka et al., 2000; Kusunoki et al., 2002). Therefore, type 2 PMUs in ‘Ca. P. asteris’ may be acquired from ‘Ca. P. ziziphi’ in the same ecological niche. Further accumulation of information on the synteny of PMUs will reveal how PMUs and symptom-determinant genes located in them are shared among phytoplasmas in the same niche.