Gene content, metabolic pathway comparison statistic comparison

As shown in Fig. 13, according to the genome annotation by BlastKOALA (see materials and methods), ~63.3% of the protein in ‘Ca. P. cynodontis’ GY2015 genome could be assigned to specific function characterized COG categories. Among the poorly characterized proteins, 111 (26.1%) were annotated as hypothetical protein, including 46 (10.8%) were predicted as putative effectors (see below). Comparing to the non-host-restricted bacteria Escherichia coli genome (Hayashi et al. 2001), GY2015 genome (obligate parasite) has reduced proportion of proteins assigned to COG metabolism categories (GY2015: 16.9%; E. coli: 26.1%), which is consistent with the result in Chung et al. 2013 and Lo et al. 2016. Among them, 24 proteins (33.4% of the metabolism categories) are transporting related proteins, which is also more than E. coli (394 out of 1359, 29.0%). The relatively higher proportion of transporter genes indicates the high reliance of phytoplasma to uptake nutrients (maybe even high-energy compounds) from its host plant and insect. In addition, the [M] category, which is correlated to cell wall/membrane/envelope biogenesis, takes up a much smaller portion in GY2015 genome than in E. coli genome. The only three proteins assigned to the [M] category are all membrane proteins. By contrast, the [J], [K], [L]

categories take up a relatively large portion in GY2015 genome than in E. coli genome. Most of the genes assigned to these three categories are correlated to ribosome biogenesis, DNA replication and RNA synthesis, those which are important and indispensable for basic cell function. Therefore, these genes are comparatively conserved among different species and take up a large portion of small genomes like phytoplasma’s.

For the distribution patterns of homologous gene clusters (Fig. 14), genes of GY2015 were clustered into 424 homologous groups. In ‘one-to-one’ homologous shared comparison, GY2015 shared only 71.2% of total gene clusters with the sister species genome ‘Ca. P.

oryzae’ Mbita1, and 69.6% with PnWB NTU2011, the other two available genomes also in the yellow clade. Interestingly, the percentages of shared homologous with phytoplasmas in other clades are also around 70% (69.3% with ‘Ca. P. mali’ AT, 68.9% with ‘Ca. P.

australiense’ PAa, 71.0% with ‘Ca. P. asteris’ AY-WB, and 69.8% with ‘Ca. P. asteris’ OY-M). The ratios of shared homologous within the yellow clade are a bit lower than expected:

We expected the closer related species should share higher ratio of homologous gene clusters.

The draft status of these genomes may be one of the reasons. It may also due to the high evolutionary diversity among the yellow clade. Nonetheless, the core gene clusters that shared with all seven phytoplasma are only 200, that is 47.2% of GY2015 and 26.6% of the biggest genome ‘Ca. P. asteris’ OY-M. These low ratios of conserved genes among phytoplasmas indicate the high divergence of phytoplasma genomes, for both gene content distribution and genome size. The variations may be owing to the host adaptation genes for their inconstant environment, and the large portion of repeat DNA fragment (i.e., PMUs) among their genomes.

For a more detailed gene content comparison among these seven phytoplasmas, we further generated the table-like figure (Fig. 15) according to the annotation, to show the gene presence or absence in each genome. In general, the genes related to transporting functions, including the Sec secretion system and spermidine/putrescine transporters, are relatively more conserved than genes related to metabolism and genome recombination and repair, which is not surprising. The Sec secretion system is indispensable for phytoplasma to secrete weapons (i.e., effector proteins) to attack or interact with their host. Spermidine and putrescine are polyamines, which are associated with biosynthesis of ethylene (plant hormone)

on plant physiology. The quantities of polyamines could affect plant growth and development.

The conserved spermidine/putrescine transporters would not only help phytoplasmas to absorb these compounds as the source for other necessary amino acids, but might also affect growth and development of their host plants. It is worth noting that the phytoplasmas belonging to yellow clade, including ‘Ca. P. cynodontis’ GY2015, ‘Ca. P. oryzae’ Mbita1, and PnWB NTU2011, have a reduced gene set for glycolysis pathway. As described in Bai et al. 2006, the ATP synthesis in phytoplasmas was limited to glycolysis since the absence of TCA cycle and other energy biosynthesis enzyme genes in their genomes. Instead, the yellow clade phytoplasmas have cit genes (citCDEFGSX) and sfcA for citrate fermentation, indicating that they may have the capability to utilize the organic acid (e.g., citrate, malate) as carbon source to produce energy or to transform to other essential compounds.

Another interesting finding is the pseudogenization of pdhC gene specific to yellow clade (Fig. 1). The product of pdhC gene is the E2 component of pyruvate dehydrogenase (dihydrolipoamide acetyltransferase), together with E1 and E3 would form the giant pyruvate dehydrogenase complex (PDC). All of these three subunits are involved in the conversion of pyruvate to acetyl-CoA and generating NADH, thus is fundamentally important for most, if not all, organisms. However, we found that the pdhC gene were broken into two segments in

‘Ca. P. cynodontis’ GY2015, ‘Ca. P. oryzae’ Mbita1, and PnWB NTU2011 (Fig. 16).

According to the amino acid alignment and conserved domain search (CD-search) (Marchler-Bauer et al. 2017), both two segments of these three phytoplasma are lower than 70% of the whole domain. Comparing to the most integral pdhC, which belongs to ‘Ca. P. australiense’

PAa, the two segments in ‘Ca. P. cynodontis’ GY2015 covered 19.7% and 53.8% of the whole domain, respectively. The breaking points in these three phytoplasmas are very conserved (the variations are less than five amino acids), indicating that the pseudogenization event has occurred once at the root of yellow clade (Fig. 1). As to the function of the

pseudogenized pdhC and if there is any other functional copy in the unassembled regions, we were not able to make any conclusion in this study. To further confirm their function in the future study, we could try mutating the pdhC in other culturable bacterium such as E. coli, then express the complete or pseudogenized phytoplasmal pdhC to test their function.