A Small Wolbachia Protein Directly Represses Phage Lytic Cycle Genes in “Candidatus Liberibacter asiaticus” within Psyllids

Host acquisition of a new microbial species can readily perturb the dynamics of preexisting microbial associations. Molecular cross talk between microbial associates may be necessary for efficient resource allocation and enhanced survival. Classic examples involve quorum sensing (QS), which detects population densities and is both used and coopted to control expression of bacterial genes, including host adaptation factors. We report that a 56-amino-acid repressor protein made by the resident psyllid endosymbiont Wolbachia can enter cells of Liberibacter crescens, a cultured proxy for the uncultured psyllid endosymbiont “Ca. Liberibacter asiaticus” and repress “Ca. Liberibacter asiaticus” phage lytic cycle genes. Such repression in “Ca. Liberibacter asiaticus” may be critical to survival of both endosymbionts, since phage-mediated lysis would likely breach the immunogenic threshold of the psyllid, invoking a systemic and nonspecific innate immune reaction.


RESULTS
SC1 lytic cycle genes are repressed in the psyllid host. Comparative expression analyses of select "Ca. Liberibacter asiaticus" phage lytic cycle holin (SC1_gp110), endolysin (SC1_gp035), and tail fiber (SC1_gp025) genes were examined in comparison with "Ca. Liberibacter asiaticus" prfA in three different "Ca. Liberibacter asiaticus"infected hosts: citrus, periwinkle, and psyllids. Transcript abundance in citrus was used as the baseline calibrator. The levels of expression of the holin, endolysin, and tail fiber were significantly lower in D. citri than the levels in citrus (P ϭ 0.00012, 0.000615, and 0.004948, respectively) and dramatically lower than the levels in periwinkle (P ϭ 0.000128, 0.000012, and 0.049213, respectively), corroborating both lack of a productive lytic cycle in psyllids and also the observed presence of phage particles in planta, most particularly in periwinkle (11). In contrast, "Ca. Liberibacter asiaticus" chromosomal gene prfA transcript levels were similar in both plant hosts and the insect host, demonstrating specific repression of bacteriophage late genes in D. citri (Fig. 1).
The holin gene promoter is strongly constitutive in L. crescens but suppressed by exogenously applied psyllid extracts. The holin gene promoter was previously shown to be strongly and constitutively expressed in L. crescens strain BT-1 carrying pLF057 (hol::uidA); however, activity in BT-1 cells was suppressed in a dose-dependent manner by adding crude aqueous extracts from D. citri to the media (12). The inhibition of GUS reporter expression by exogenously supplied psyllid extracts was found to be both heat labile and susceptible to proteinase K inactivation, indicating involvement of a proteinaceous inhibitor; furthermore, the inhibitor appeared to be greater than 3 kDa but less than 30 kDa (Fig. 2). Together, these data led us to hypothesize that either D. citri and/or one of its bacterial endosymbionts produced one (or more) heat-sensitive proteinaceous inhibitors capable of regulating the SC1 lytic cycle activation through repression of the holin gene promoter.
A holin promoter repressor protein is made by Wolbachia. To identify the hypothetical protein able to suppress SC1 holin gene promoter activity, crude aqueous extracts of D. citri were size fractionated, and potential DNA-binding proteins from the 3-to 30-kDa fraction were affinity captured and subjected to liquid chromatographytandem mass spectrometry (LC-MS/MS). These analyses resulted in peptide fingerprints representing 25 different proteins identified from multiple protein databases that most closely matched the MS/MS spectra of proteins captured in the sample. Two of these putative proteins were of endosymbiotic bacterial origin (Wolbachia and "Ca. Profftella armatura"), and the remaining proteins were from psyllids. Of the 25 proteins, the top FIG 1 Relative expression of SC1 phage lytic cycle holin (SC1_gp110), endolysin (SC1_035), and tail fiber (SC1_gp025) genes and "Ca. Liberibacter asiaticus" (Las) prfA gene in psyllids, citrus, and periwinkle. The relative transcript abundance of each gene was normalized against expression levels of gyrB within each sample, and transcript abundance of psyllid and periwinkle samples was calibrated against expression levels in "Ca. Liberibacter asiaticus"-infected citrus. Values are averages (bars) plus standard errors of the means (error bars) (n ϭ 3).
The putative Wolbachia repressor protein binds "Ca. Liberibacter asiaticus" holin promoter DNA. The WDIAC_RS0101550 gene was PCR amplified from Wolbachiainfected psyllids, sequenced, cloned in the pEXP5-CT expression vector, and used for in vitro cell-free protein synthesis. The in vitro-translated protein was size fractionated, concentrated, and used for examining its potential DNA binding activity. Electrophoretic mobility shift assay (EMSA) analysis of the interaction between biotinylated holin promoter DNA and the in vitro-translated Wolbachia wDi repressor protein revealed DNA binding activity (Fig. 3). Holin promoter DNA mobility was sharply decreased in the presence of the purified repressor, and the mobility shift was quenched by spiking the sample with unlabeled competitive DNA, confirming the specificity of DNA binding by this Wolbachia protein.
The Wolbachia DNA-binding protein functionally represses the "Ca. Liberibacter asiaticus" holin promoter. The in vitro-translated Wolbachia protein was added to liquid cultures of L. crescens BT-1 cells expressing the GUS reporter (hol::uidA) on pLF057, which replicates stably in BT-1. The repressor protein significantly inhibited GUS reporter activity driven by the holin promoter (Fig. 4). However, inhibition by the in vitro-expressed Wolbachia repressor was less than that observed using crude aqueous D. citri extracts. Addition of heat-inactivated psyllid extracts failed to enhance suppression of holin promoter activity mediated by the Wolbachia repressor protein (Fig. 4). Together, these observations indicate that complete suppression of the holin promoter requires either an additional interacting partner or posttranslational modification of the repressor by the psyllid. The Wolbachia DNA-binding protein was found only in the Wolbachia wDi genome from D. citri. Comparative sequence analyses failed to uncover any homologs of the Wolbachia wDi repressor protein from any other Wolbachia-infected insects, including the complete genomic sequence of Wolbachia endosymbiotic strain wNo of Drosophila simulans (GenBank accession number NC_021084). As expected, crude aqueous extracts from Wolbachia-infected Drosophila failed to inhibit the GUS reporter in L. crescens cells (Fig. 4).
The Wolbachia repressor protein is constitutively expressed in D. citri. Quantitative reverse transcriptase PCR (RT-PCR) was carried out in "Ca. Liberibacter asiaticus"infected and healthy ("Ca. Liberibacter asiaticus"-free) psyllids to determine whether the presence of "Ca. Liberibacter asiaticus" caused changes in the expression levels of the Wolbachia wDi gene-encoded repressor protein. Relative abundance of Wolbachia repressor protein (normalized against the expression of endogenous Wolbachia surface protein gene wsp) in six independent psyllid extracts (each with 10 insects) indicated  that Wolbachia appeared to maintain constitutive expression of the repressor protein irrespective of the presence of "Ca. Liberibacter asiaticus" (Fig. 5).

DISCUSSION
A productive lytic cycle is marked by transcriptional activation of phage late genes, including wall-degrading enzymes called endolysins and bacterial cell inner membrane-permeabilizing proteins called holins, allowing host cell lysis and phage particle egress (19). Functional holin and endolysin genes were confirmed carried on SC1 (SC1_gp110 and SC1_gp035, respectively), with the holin gene promoter encoding holin as the lead gene in the first operon of the late gene region (12). Consistent with lack of an observed "Ca. Liberibacter asiaticus" phage lytic cycle in psyllids and a single report of phage particles observed in citrus (13), comparative expression data revealed that transcription of the SC1 holin gene in "Ca. Liberibacter asiaticus" was significantly repressed in its insect host compared to citrus and dramatically suppressed compared to periwinkle (Fig. 1). To further investigate this repression, L. crescens was developed as a culturable proxy for uncultured "Ca. Liberibacter asiaticus," allowing development of reporter gene assays using "Ca. Liberibacter asiaticus" promoter elements. The holin promoter was strongly and constitutively active in L. crescens and repressed in a dose-dependent manner by aqueous extracts of D. citri and plant hosts but to a lesser extent by extracts from plant hosts (12).
Preliminary EMSAs revealed binding of a factor(s) in the crude aqueous extracts from psyllids to the holin promoter DNA (not shown). Following affinity capture of proteins contained in the crude aqueous psyllid extracts, a predicted small protein encoded only by a gene carried by Wolbachia wDi found in psyllids was identified by LC-MS/MS. The gene encoding the 56-aa Wolbachia wDi protein was expressed in vitro and was functional in repressing a holin::GUS reporter in L. crescens when applied outside L. crescens cells (Fig. 4). The Wolbachia wDi repressor protein did not repress the GUS reporter in L. crescens in a dose-dependent manner and not to the level obtained by use of crude aqueous extracts from psyllids. Complete repression appeared to require (i) an enzymatic posttranslational modification in the psyllid, (ii) a cognate proteinaceous interacting partner of the repressor, or (iii) another independent proteinaceous repressor that is yet to be identified from psyllids and/or one of their endosymbionts.
In contrast to many insects, annotation of the D. citri draft genome revealed the complete absence of putative genes for the Imd pathway and known antimicrobial peptides (AMPs) in an attempt to accommodate symbioses with primary Gram-negative endosymbionts (25,26). However, D. citri may be capable of detecting and resisting Gram-negative bacteria via the Toll and Janus kinase (JAK)/signal transducer and activator of transcription factor (STAT) pathways, and clearing any microbial infection through cellular immune responses (26,27). The "Ca. Liberibacter asiaticus" genome has been annotated as encoding genes for several stereotypical pathogen-associated molecular patterns (PAMPs) including flagellin, diaminopimetic acid (DAP) in its peptidoglycan layer, and lipopolysaccharide (LPS) (28,29). Several lines of evidence indicate a potential for recognition of "Ca. Liberibacter asiaticus" infection by the psyllid host immune system (30,31). In contrast, Wolbachia bacteria are thought to neither activate nor repress the immune system in their native hosts (32) and appear to completely evade host immune recognition by sequestration within intracellular host-derived membranous vesicles (33). However, evasion by stealth does not safeguard Wolbachia against the risk of collateral damage as a consequence of other bacterial infections (34).
It is safe to presume that in an event involving activation of phage lytic cycle genes in "Ca. Liberibacter asiaticus"-infected psyllids, the resulting bacterial lysate would likely breach the immunogenic threshold and invoke an immune response as a consequence of release of additional immunogenic determinants that were previously sequestered within the intact cells. For example, phage particles and DNA, when delivered intracellularly in mice, altered the expression of innate immune genes (35), and transfection of bacterial DNA in Drosophila cells elicited an innate immune response (36). In addition, both purified LPS and LPS contained in outer membrane fragments are similarly more immunogenic than intact bacterial LPS (37). The Wolbachia repressorencoding gene identified in this work from psyllids has not been found with any significant level of similarity at the protein level in any of the other sequenced Wolbachia genomes. We speculate that in response to the significant immunogenic potential posed in an event of SC1 lytic cycle activation in "Ca. Liberibacter asiaticus," this repressor is a specific adaptation by Wolbachia in response to the presence of SC1, either originally in "Ca. Liberibacter asiaticus" or in another member of the psyllid endosymbiont community. Either way, it likely enhances survival prospects of both "Ca. Liberibacter asiaticus" and Wolbachia in psyllids.
"Ca. Liberibacter asiaticus" is clearly much better adapted to its psyllid host as an endosymbiont than to its citrus host as a pathogen. "Ca. Liberibacter asiaticus" crosses multiple cellular membranes of different organs (salivary glands, filter chamber, midgut, muscle tissue, and ovaries) in psyllids (38), lives in evident equilibrial symbioses with several other psyllid endosymbionts, and although it can cause localized apoptosis in psyllid midgut tissue, causes few overt disease symptoms (10). In contrast in citrus, "Ca. Liberibacter asiaticus" is limited to phloem and causes a debilitating citrus decline due to HLB, often resulting in tree death (7). Repression or loss of the SC1 phage lytic cycle may be critical for growth of "Ca. Liberibacter asiaticus" in psyllids.
One alternative to repression of the lytic cycle is selection for loss of either the phage or the lytic cycle genes of the phage. Defective prophage variants are thought to arise as a result of such selection pressure (39). Indeed, the majority (90.4%) of "Ca. Liberibacter asiaticus" strains surveyed in southern China carried only one of the SC1 or SC2 prophage variants, and the majority of those that carried one prophage variant were missing the lytic SC1 prophage (40). The SC2 prophage is notable for its lack of lytic cycle genes and a lysogenic conversion factor favorable for colonization of plants to encode an active peroxidase (14).
Plants and psyllids may or may not have an innate immune response adequate to both survive and clear Liberibacter infections, and clearly liberibacters have developed different strategies to suppress (14) or avoid triggering (29) such responses. Curing the psyllid of Wolbachia wDi carrying the repressor or mutating or deleting the repressor from the Wolbachia wDi genome might result in elimination and clearing of "Ca. Liberibacter asiaticus" or in lethality of the psyllid host. Since the holin reporter gene is constitutively active in L. crescens cells in the absence of a repressor, it is likely that the "Ca. Liberibacter asiaticus" holin and other lytic cycle genes would also be active in "Ca. Liberibacter asiaticus" as soon as axenic cultures are attempted without addition of host plant or psyllid extracts. The inability to culture "Ca. Liberibacter asiaticus" has been a significant obstacle to research progress in understanding and controlling this devastating citrus disease.
Although there are reports of (i) peptide-mediated quorum sensing in Gram-positive bacteria (41), (ii) antimicrobial lytic peptides that bind DNA (42), (iii) cysteine-rich plant peptides that can enter rhizobia and dramatically affect bacterial morphology (43), and (iv) homeodomain proteins (as long as 60 residues) being able to cross insect cellular membranes (44), to our knowledge, this is the first report of a protein repressor able to permeate a bacterial cell and directly affect gene expression. This also represents a new and unusual form of bacterial cross talk between genera that does not involve quorum sensing.

MATERIALS AND METHODS
Plant material and insect samples. "Ca. Liberibacter asiaticus"-infected leaf samples were excised from the curated citrus (Citrus paradisi) and periwinkle (Citrus roseus) plants maintained in a quarantine greenhouse at the University of Florida, Gainesville. Healthy mixed sex adult psyllids (reared on "Ca. Liberibacter asiaticus"-free orange jasmine, Murraya paniculata) were provided by Eric Rohrig, Florida Department of Agriculture. "Ca. Liberibacter asiaticus"-infected insects (reared on "Ca. Liberibacter asiaticus"-infected sweet orange [Citrus sinensis]; approximately 80% infection density) were provided by David Hall, USDA, ARS. Drosophila cultures were obtained from the lab of Marta Wayne, Biology Department, University of Florida.
Bacterial growth conditions and transformation. The relevant characteristics and source and/or reference for the bacterial strains and plasmids used in this study are listed in Table 1. Escherichia coli was grown in Luria-Bertani (LB) medium at 37°C. L. crescens strain BT-1 was maintained in liquid BM7 medium containing 2 g alpha-ketoglutarate, 10 g N-(2-acetamido)-2-aminoethanesulfonic acid (ACES) buffer, and 3.75 g KOH in 550 ml water (pH 6.9) followed by addition of filter-sterilized 300 ml fetal bovine serum (HyClone Laboratories, Logan, UT, USA) and 300 ml modified Grace's insect culture medium (TNM-FH; HyClone Laboratories), with moderate aeration at 150 rpm at 28°C. Electrocompetent BT-1 cells were prepared and transformed as previously described (12,14). The following antibiotics were used as needed at the indicated concentrations: ampicillin (Amp), 100 &micro;g/ml, kanamycin (Kn), 50 &micro; g/ml, and gentamicin (Gm), 2 &micro;g/ml.
Nucleic acid extractions from psyllid and plant samples. DNA was extracted from plant leaf discs and from whole psyllids using the DNeasy plant minikit and blood and tissue kit, respectively (Qiagen, Valencia, CA, USA). The presence of "Ca. Liberibacter asiaticus" in the infected citrus samples was confirmed using conventional and nested PCR primer sets OI1/OI2c (45) and CG03F/CG05R (F stands for forward, and R stands for reverse) (46). For the extraction of total RNA, the midribs of PCR-confirmed "Ca. Liberibacter asiaticus"-infected leaves and psyllid samples (10 insects pooled together) were ground with a cold mortar and pestle in lysis buffer RLT (provided with Qiagen RNeasy plant minikit). RNA was extracted following the manufacturer's protocol, diluted with nuclease-free water to 200 ng &micro;l Ϫ1 , and cleaned with Turbo DNA-free (DNase) kit (Ambion, Austin, TX, USA). Quantitative reverse transcriptase PCR. The primer pairs used for the quantitative reverse transcriptase PCR (qRT-PCR) analyses are listed in Table 2. Reverse transcription reactions were performed using 1 &micro;g RNA template (iScript Advanced cDNA synthesis kit; Bio-Rad, Hercules, CA, USA). Quantitative RT-PCR analyses were performed using a CFX96 Touch real-time PCR detection system (Bio-Rad, Hercules, CA, USA) by the method of Jain et al. (14). At least three biological replicates and four technical replicates were used with no-template and no-RT controls. Relative expression levels were calculated and normalized by the ΔΔC T method (47) and corrected for amplification efficiency. Data analyses and Student's t tests (␣ ϭ 0.05) were performed using the Bio-Rad CFX Manager Software package 3.0.
Preparation of psyllid and Drosophila extracts. Approximately 50 whole insects were pulverized to a fine powder under liquid N 2 and resuspended in 1 ml deionized water. The resulting suspension was cleared by centrifugation twice at 3,220 ϫ g for 20 min at 4°C, filter sterilized, and stored at Ϫ20°C. The psyllid extracts were heat inactivated by autoclaving (120 lb/in 2 , 20 min). For proteinase treatment, 50 &micro;l psyllid extract was incubated with 2.5 &micro;l (20 mg ml Ϫ1 ) proteinase K (New England Biolabs, Ipswich, MA) for 30 min, and the reaction was terminated by heating at 75°C for 15 min.
DNA affinity capture assay. A 535-bp fragment of the holin promoter region was PCR amplified using Accuprime Taq High Fidelity polymerase (Invitrogen, Carlsbad, CA), 2ϫ Failsafe buffer D (Epicentre, Madison, WI) and the primer pair HPromF/HPromRBio (Table 2) and concentrated using the QIAquick PCR  (48) with Scaffold delta-mass correction. Protein identifications were accepted if they could be established at greater than 80.0% probability and contained at least two identified peptides. Protein probabilities were assigned by the Protein Prophet algorithm (49).
Cloning and expression of Wolbachia repressor protein. The gene encoding the Wolbachia repressor protein was PCR amplified using primers wrpDiF/wrpDiR (Table 2), directly cloned in the pEXP-5-CT/TOPO TA expression vector and transformed into E. coli TOP10 cells (Invitrogen) according to the manufacturer's recommendations. Following sequence verification, purified plasmid DNA was used for in vitro translation (PURExpress in vitro protein synthesis kit; NEB), and the resulting protein was concentrated using Amicon Ultra Centrifugal filter columns (molecular weight of 3,000 [3K] and 30K) (EMD Millipore, Millipore Corp., Billerica, MA). Total protein was quantified by Bradford protein assay using bovine serum albumin (BSA) as the standard (Bio-Rad).

ACKNOWLEDGMENTS
Eric Rohrig and Gloria Lotz (Division of Plant Industry, Gainesville, FL) are gratefully acknowledged for providing a constant supply of healthy "Ca. Liberibacter asiaticus"-free psyllids throughout the course of this investigation. David Hall (USDA-USHRL, Ft. Pierce, FL) and Marta Wayne (Biology Department, University of Florida, Gainesville) are gratefully acknowledged for providing "Ca. Liberibacter asiaticus"-infected psyllids and Drosophila, respectively. We thank Patricia Rayside for excellent technical assistance.
This work was supported by the Florida Citrus Research and Development Founda-