Horizontal Gene Acquisition of Liberibacter Plant Pathogens from a Bacteriome-Confined Endosymbiont of Their Psyllid Vector

he Asian citrus psyllid Diaphorina citri is a notorious agricultural pest that transmits the phloem-inhabiting alphaproteobacterial ‘Candidatus Liberibacter asiaticus’ and allied plant pathogens, which cause the devastating citrus disease called Huanglongbing or greening disease. D. citri harbors two distinct bacterial mutualists in the symbiotic organ called bacteriome: the betaproteobacterium ‘Candidatus Profftella armatura’ in the syncytial cytoplasm at the center of the bacteriome, and the gammaproteobacterium ‘Candidatus Carsonella ruddii’ in uninucleate bacteriocytes. Here we report that a putative amino acid transporter LysE of Profftella forms a highly supported clade with proteins of L. asiaticus, L. americanus, and L. solanacearum. L. crescens, the most basal Liberibacter lineage currently known, lacked the corresponding gene. The Profftella-Liberibacter subclade of LysE formed a clade with proteins from betaproteobacteria of the order Burkholderiales, to which Profftella belongs. This phylogenetic pattern favors the hypothesis that the Liberibacter lineage acquired the gene from the Profftella lineage via horizontal gene transfer (HGT) after L. crescens diverged from other Liberibacter lineages. K A/K S analyses further supported the hypothesis that the genes encoded in the Liberibacter genomes are functional. These findings highlight the possible evolutionary importance of HGT between plant pathogens and their insect vector’s symbionts that are confined in the symbiotic organ and seemingly sequestered from external microbial populations.


Introduction
The Asian citrus psyllid Diaphorina citri (Hemiptera: Psyllidae) is an important agricultural pest that transmits a serious citrus disease, Huanglongbing (HLB) or greening disease. This insect is widely distributed in Asia, and is spreading into other citrus growing regions worldwide [1]. The causative agents of HLB are considered to be three species of a fastidious phloem-inhabiting alphaproteobacterial lineage of the genus Candidatus Liberibacter: L. asiaticus, L. americanus, and L. africanus [2,3]. D. citri vectors L. asiaticus and L. americanus in Asia and the Americas, and the African citrus psyllid Trioza erytreae (Hemiptera: Triozidae) vectors L. africanus in Africa [1,2,3,4]. Similar diseases have been found in potatoes, tomatoes and other solanaceous crops infected with L. solanacearum (also known as L. psyllaurous) [5]. These Liberibacter species are very fastidious, but L. crescens, the species recovered from mountain papaya, has recently been reported to be readily culturable [6]. Complete genome sequences have been determined for L. asiaticus [7], L. solanacearum [5], and L. crescens [6], whereas draft genome sequence is available for L. americanus [8].
In its abdomen, D. citri possesses a large yellow symbiotic organ called the bacteriome, where two distinct symbionts are harbored [9]. The betaproteobacterium 'Candidatus Profftella armatura' is located in the syncytial cytoplasm at the center of the bacteriome, whilst the gammaproteobacterium 'Candidatus Carsonella ruddii' is found in uninucleate bacteriocytes on the surface of the bacteriome. Our previous study revealed that Profftella is a toxinproducing defensive symbiont that potentially protects D. citri from natural enemies, while Carsonella_DC is a nutritional symbiont that provides the host with essential amino acids, which are scarce in the psyllid's diet of phloem sap [10].
Here we report that the Liberibacter lineage horizontally acquired a putative transporter gene from a bacterium closely related to the extant Profftella.

Materials and Methods
HGT candidates in the Profftella genome were extracted by BLASTP searches [11] against NCBI nr database, using deduced amino acid sequences of all protein coding genes on the Profftella genome as queries. Amino acid sequences were aligned using MAFFT 6.847 [12], followed by manual refinement. Amino acid sites corresponding to alignment gap(s) were omitted from the data set. The best fitting amino acid substitution model for the alignment was estimated using ProtTest3 [13]. For the present analysis, ProtTest selected LG with a gamma distribution (+G), a proportion of invariable sites (+I) and empirical base frequencies (+F) as the best fitting substitution model, followed by WAG with the options +I +G +F. Phylogenetic trees were inferred by the Maximum Likelihood (ML) [14] and the Bayesian Inference (BI) [15] methods. ML trees were constructed using RAxML7.2.1 [16] with LG + G + I + F model. The support values for the internal nodes were inferred by 1,000 bootstrap replicates. In the BI, we used the program MrBayes 3.1.2 [15]. Since the LG model is not implemented in MrBayes, WAG as the next best available model was used with the options +I +G +F. In total, 18,000 trees were obtained (Nruns = 2, Ngen = 900000, Samplefreq = 100), and the first 2,000 of each run were considered as the ''burn in'' and discarded. The posterior probability of each node was used as the support value of the node. We checked that the potential scale reduction factor was approximately 1.00 for all parameters and that the average standard deviation of split frequencies converged towards zero.
K S and K A values were calculated as described previously [17]. Statistical significance of the obtained K A /K S value was tested against a bootstrap distribution of K A /K S values, which was generated by 10,000 bootstrap resamplings of codons from the original alignment. When K S values calculated from resampled alignments were close to saturation values (larger than 2.0 per site), the K S values was set as 2.0 for the estimation of K A /K S value.
Molecular phylogenetic analysis demonstrated that the LysE of Profftella forms a highly supported clade with the proteins of Liberibacter spp. (Fig. 2). The Profftella-Liberibacter subclade was placed within a clade that largely consisted of the LysE sequences of betaproteobacteria and gammaproteobacteria that are paraphyletic to Betaproteobacteria [18]. Moreover, this subclade formed a clade with proteins from betaproteobacteria of the order Burkholderiales, to which Profftella belongs [19]. This phylogenetic pattern, together with the presence/absence of the orthologous genes in Liberibacter spp., is most simply explained by the hypothesis that the Liberibacter lineage acquired the transporter gene from the Profftella lineage via horizontal gene transfer (HGT) after L. crescens diverged from other Liberibacter lineages. The structural organizations of the lysE flanking regions were partially conserved among genomes of L. asiaticus, L. americanus and L. solanacearum (Fig. 3), which were all assembled de novo without reference to one another [5,6,7,8], further supporting a single acquisition of this gene in the Liberibacter lineage.
The K A /K S ratio between lysE genes of L. asiaticus and L. solanacearum was significantly lower than 1 (K A = 0.24, K S = 1.61, K A /K S = 0.15, p , 0.0001). Whereas the K S values both between L. asiaticus and L. americanus and between L. solanacearum and L. americanus were saturated (. 3.00), the K A values were still as low as 0.42 and 0.39, respectively. These results support the hypothesis that the lysE genes of Liberibacter spp. are under purifying selection and thus are functional.

Discussion
The present study demonstrated that the Liberibacter lineage horizontally acquired a lysE-type transporter gene from the Profftella lineage, an endosymbiont of their vector insect. K A /K S analyses further supported the hypothesis that the genes encoded in the Liberibacter genomes are functional. Although their true functions are yet to be identified, LysE superfamily proteins of various bacteria are generally involved in exporting substrates, playing important roles in resistance to toxic substances, in maintenance of optimum intracellular concentration of metabolites, and in excretion of regulatory molecules [20,21]. Thus, it is probable that Liberibacter have acquired novel functions through this HGT. Whereas HGTs are rampant among bacteria [22,23], such transfers of genes are rare in intracellular bacteria that are harbored in insects' symbiotic organ and are seemingly sequestered from external microbial populations [24,25,26]. Apparently, Profftella, the putative donor lineage of the lysE gene, is this type of endosymbiont. In this context, infection style of Liberibacter, the putative accepter of the gene, would be noticeable. As Liberibacter spp. are transmitted by psyllids in a persistent manner, exhibiting near systemic infection of various organs and tissues [27], they may also intrude into the bacteriome of the vector psyllids, having opportunity of HGT with endosymbionts therein. The present findings highlight the previously unrecognized possible evolutionary importance of HGT between plant pathogens and their vector's mutualists that are confined in symbiotic organs.