Serological detection of ‘Candidatus Liberibacter asiaticus’ in citrus, and identification by GeLC-MS/MS of a chaperone protein responding to cellular pathogens

We describe experiments with antibodies against ‘Candidatus Liberibacter asiaticus used to detect the pathogen in infected plants. We used scFv selected to bind epitopes exposed on the surface of the bacterium in tissue prints, with secondary monoclonal antibodies directed at a FLAG epitope included at the carboxyl end of the scFv. Unexpectedly, the anti-FLAG secondary antibody produced positive results with CaLas diseased samples when the primary scFv were not used. The anti-FLAG monoclonal antibody (Mab) also identified plants infected with other vascular pathogens. We then identified a paralogous group of secreted chaperone proteins in the HSP-90 family that contained the amino acid sequence DDDDK identical to the carboxy-terminal sequence of the FLAG epitope. A rabbit polyclonal antibody against one of the same epitopes combined with a goat anti-rabbit secondary antibody produced very strong purple color in individual phloem cells, as expected for this pathogen. These results were entirely specific for CaLas-infected citrus. The simplicity, cost and ability to scale the tissue print assay makes this an attractive assay to complement PCR-based assays currently in use. The partial FLAG epitope may itself be useful as a molecular marker for the rapid screening of citrus plants for the presence of vascular pathogens.


Results
Detection of CaLas with ScFv antibodies by direct tissue blot immuno assay (DTBIA). Single chain variable fragment antibodies scFv 748 (OmpA) and scFv 932 (FlgI) were selected for testing in the DTBIA. These scFv had been used successfully when expressed from phagemid vector pKM19 to detect purified protein antigens and in dot blot assays against crude extracts from infected sweet orange 21 . In the present work the scFv were expressed from the plasmid vector pKM16 in E. coli BL21 and purified by taking advantage of the 6X His tag provided by the vector. The scFv preparations were shown to be pure by SDS-PAGE and successfully detected their purified antigenic targets in dot blot and western blot assays (data not shown).

Recognition of CaLas antigen by ScFv antibody in citrus samples.
Petioles of sweet orange and rough lemon leaves from HLB-symptomatic trees were used to prepare tissue prints to detect CaLas using the FLAG epitope incorporated into the scFv and a secondary anti-FLAG Mab conjugated with alkaline phosphatase. All petioles used for DTBIA were also tested by qPCR after the tissue prints had been prepared. Typical results of qPCR assays of the petiole sections used for tissue printing showed that the concentration of CaLas DNA was consistently high in these petioles, typically with a Cq value of ~21 (Table 1). This represents about 10 7 CaLas genomes per g of petiole 26 . The results of the DTBIA appeared to show that scFv 748 and scFv932 specifically recognized CaLas in DTBIA from petioles of rough lemon and sweet orange. Color development was strong and limited to only some cells in the phloem ring in infected petioles of both rough lemon and sweet orange as expected for the phloem limited pathogen (Fig. 1). Diffuse color development, not focused in the phloem cells, was often observed in the vascular ring of the healthy samples as well, though the diseased and healthy samples could easily be distinguished.

Identification of a protein with a partial FLAG epitope in response to cellular pathogens.
Further experiments were carried out to optimize the concentration of scFv and detection antibody in these experiments, and the results looked promising, with strong purple color in the phloem sieve cells of CaLas infected, but not healthy petioles. As a control, the secondary anti-FLAG Mab was incubated with tissue prints of CaLas infected petioles without the use of any primary scFv that contained the FLAG epitope. Surprisingly, the results with the secondary Mab alone were difficult to distinguish from those when both the primary scFv and secondary Mab were used together (Fig. 2). Color development was sharply localized in phloem sieve cells of CaLas infected petioles and more weakly and without sharp foci in the area of the vascular rings of the healthy plant controls.   petioles from trees infected with other phloem and xylem limited pathogens. Similar results were obtained when petioles infected with two genotypes of CTV, as well as with CCDV and Xylella fastidiosa. Strong color developed in the vascular ring of petioles from mature leaves from the infected, but not from the healthy control plants, with sharp foci in the phloem cells (Fig. 3). A protein with a partial FLAG epitope was also shown to be present in infected leaves of C. sinensis by western blot analysis of proteins extracted from both diseased and healthy leaves (Fig. 4). These results showed that the protein with the partial FLAG epitope was not detectable in healthy, mature citrus leaves, though it was weakly expressed in healthy, immature citrus leaves. However, the protein with a partial FLAG epitope was strongly expressed in leaves infected by CaLas, CTV or CCDV.
Identification of the partial FLAG epitope. A list of C. sinensis peptides expressed in diseased petioles of mature leaves was compiled by massively parallel identification of peptides by GeLC-MS/MS. An automated p-BLAST search of the C. sinensis proteome identified several peptides that were portions of a paralogous group of C. sinensis proteins ( Table 2). The full-length proteins shared a common amino acid sequence of DDDDK, the carboxy terminal five amino acids of the FLAG octapeptide. A similar search of the C. clementina proteome returned a similar set of proteins (Table 3; Figure S2). The paralogous proteins from C. sinensis were all members of the heat shock protein-90 (HSP-90) family of proteins.  Detection of CaLas with an anti-OmpA polyclonal antibody by DTBIA. We also tested a polyclonal rabbit antibody (Pab) prepared against the same portion of the major outer membrane protein of CaLas, OmpA. The anti-OmpA Pab was then detected with a secondary goat anti rabbit Pab conjugated with alkaline phosphatase. When this antibody system was used for DTBIA from CaLas infected and healthy sweet orange petioles, strong color foci were observed in the phloem sieve cells, as expected for CaLas, without similar color development in the DTBIA from petioles of healthy sweet orange (Fig. 5). When the secondary goat anti-rabbit Pab was used without the primary rabbit anti OmpA antibody, no color was observed in the immuno tissue prints (Fig. 6). The specificity of the polyclonal rabbit anti-OmpA antibody was tested against the panel of phloem-limited citrus pathogens. Color in the vascular ring was observed in DTBIA of petioles from trees infected with CaLas, but not in DTBIA of petioles from trees infected with phloem limited viruses or Xylella fastidiosa (Fig. 7). Purple color in the phloem sieve tubes also was not observed in the phloem sieve cells in the DTBIA prepared from petioles from trees infected with the other phloem and xylem limited pathogens when probed with only the secondary goat anti rabbit Pab conjugated with alkaline phosphatase in the absence of the primary rabbit anti OmpA antibodies (Fig. 8).

Discussion
The simplicity of tissue printing on nitrocellulose membranes has made it a widely used technique. Others have used immune detection to localize the distribution of rubisco and catalase in tissue sections 27 , of (1-5)-α -L-arabinan, a component of pectin, in leaf sections 28 and cysteine rich proteins 29 . Enzymatic activities can also be localized to specific locations by tissue printing followed by histochemical detection of the reaction products, for example both pectin methylesterase and acetylesterase in citrus fruits 30 , and peroxidase 31 , and myrosinase activity produced during active defense by plants against pathogens 32 . Immuno-tissue printing has also been applied to study the distribution of pathogens within plants. Examples include Apple chlorotic leaf spot virus and Apple stem grooving virus in apple shoots 33   virus 34 . The distribution of RNA associated with pathogens can also be followed by tissue printing using either chemiluminescent or colorigenic assays 35,36 . Recently this approach has also been applied to CaLas in citrus 37 . Tissue printing has also been applied recently for the detection of Spiroplasma citri, another phloem-limited bacterial pathogen of citrus. In this case, rather than detect the pathogen directly, the authors identified a protein    secreted by S. citri which is transported systemically in the plant. Polyclonal antibodies against this protein provided an effective detection method for the pathogen in infected citrus, in a tissue printing format that is intended to overcome the problem of uneven distribution of the pathogen in the host, a problem that also exists with CaLas 38 .
In the present study we prepared tissue prints from sweet orange and rough lemon tissues that were infected with CaLas. The DTBIA was first performed to detect scFv that contained a FLAG epitope at the carboxyl terminus of the scFv. We used scFv that recognized a portion of OmpA, a protein expected to be present in high concentration on the surface of the CaLas cells, and FlgI, a component of the flagellar apparatus encoded by CaLas 39 . As a control, mock DTBIA was performed with only the secondary anti-FLAG Mab on immuno tissue prints from healthy and diseased petioles. No scFv specific for CaLas were used in these DTBIA, yet color development was restricted to the phloem ring as observed previously with the complete assay. Reactions tended to be stronger with diseased samples. Similar detection of the partial FLAG epitope in diseased citrus was also observed with an anti-FLAG Mab obtained from a different commercial source (not shown). Taken together, this suggests that the anti-FLAG Mabs were recognizing a host epitope in the phloem region of CaLas infected plants in addition to the FLAG epitope attached to the carboxyl terminus of the scFv.
We therefore performed DTBIA of citrus petioles from trees infected with other systemic pathogens restricted to the vascular system, including mild and severe strains of CTV (Closteroviridae), CCDV (Geminviridae), and Xylella fastidiosa. The pattern of color development was similar in all cases, suggesting that the anti-FLAG Mab was binding to an epitope produced by the plants in response to vascular infection, perhaps a pathogenesis-related protein (PR). Similar levels of purple color were observed by DTBIA in tissue prints made from sweet orange petioles infected with either CTV-B2 (mild strain) or CTV-B6 (severe strain). Therefore the severity of disease symptoms in the plants tested by DTBIA was not a factor in the detection of a partial FLAG epitope in diseased citrus, but the presence of a pathogen was specifically required. The age of the leaves tested was found to be an important factor related to the presence or absence of the partial FLAG epitope in HLB-diseased and healthy samples. When extracts of CaLas-infected mature symptomatic leaves were tested by western blotting, much more partial FLAG epitope was found in diseased leaves than in those from healthy controls consistent with the results of tissue printing.
Thus, we have observed a partial FLAG epitope to be present in the phloem tissue in response to several plant pathogens. The DTBIA with the anti-FLAG secondary antibody alone identified plants infected by CaLas as well as other vascular pathogens, and may itself be a useful general assay for the presence of CaLas and these pathogens. A p-Blast search of the C. sinensis proteome identified a paralogous family of proteins that contained the five amino acid sequence DDDDK, which is the carboxy terminal sequence of the FLAG octapeptide. These proteins were present in an SDS-PAGE gel in a region that reacted with the anti-FLAG mAB and were members of the HSP-90/endoplasmin group. HSP-90s proteins play very important role in plant development and are involved in the resistance to pathogens 40,41 . Members of the HSP-90 group are secreted chaperone proteins that bind various proteins in response to different developmental programs as well as stress and disease 42-44 . Thus, this group of paralogous proteins has the expected size, biological properties and cellular location as well as the amino acid motif needed to bind the FLAG-antibody in our experiments. No longer sequence identical to the FLAG epitope was observed. The DDDDK sequence was also observed in a single protein of 49.23 kD. This protein is much too small to be responsible for the FLAG epitope that we have observed.
When polyclonal rabbit antibodies against OmpA of CaLas were used in the DTBIA, the assay produced a strong reaction in the phloem ring of petioles taken from HLB diseased, but not healthy tissue prints, as expected. The color development was localized in individual phloem cells as expected for the phloem limited pathogen. In contrast with the results of the scFv detected with the FLAG epitope, no color developed in the phloem cells of tissue prints from healthy plants, or when tissue prints from either healthy or HLB diseased trees were tested with the secondary antibody alone. Also in contrast to the detection of the FLAG epitope, when the anti-OmpA Pab with alkaline phosphatase based detection was tested with tissue prints from trees infected with the other citrus pathogens, the color reaction in the phloem ring was specific for the CaLas sample. The goat anti-rabbit Pab did not produce any color reaction at all, as expected, when used in the absence of a primary rabbit antibody.
Thus, the combination of a rabbit polyclonal antiserum against the major outer membrane protein of CaLas and a secondary goat anti-rabbit Pab provided specific detection of CaLas in the DTBIA. We then tried to use the same anti OmpA polyclonal antibodies in a DAS-ELISA assay for the detection of CaLas but without success (not shown). We interpret this to mean that the non-uniform distribution of CaLas in phloem tissues provides insufficient density of antigen for the ELISA assay, at the concentration of CaLas found in diseased plant samples (10 7 genomes/g). However, the natural foci of CaLas and the OmpA antigen are retained in tissue prints at much higher local concentrations, providing for a successful detection result. We have shown that this technique can be used to provide very sensitive and detailed information about the distribution of CaLas in citrus petioles, stems, roots, seeds, and fruit 45 . The DTBIA is inexpensive and simple to perform and is readily scalable, and will provide a very useful complement to current PCR based diagnostic tests for CaLas. The Hsp-90 secreted chaperone protein family we identified may serve as a useful molecular marker for the diagnosis of cellular pathogens infecting citrus with the convenience of using a commercial anti-FLAG secondary antibody. The evaluation of other antigens for use in similar tissue print applications and the further validation of Hsp-90 proteins as earlier diagnostic molecular marker for high-flux accurate detection of CaLas is currently under investigation in our laboratory.

Methods
Plant materials and pathogens. Isolates of CaLas were propagated by bud inoculation onto either rough lemon (Citrus jambhiri Lush.) or sweet orange (Citrus sinensis L.) seedlings propagated on rough lemon rootstocks. Two isolates of CTV were used: CTV-B2 (very mild symptoms, T30 genotype, Florida) and CTV-B6 (very severe symptoms, VT genotype, California). Isolates of CTV and Xylella fastidiosa were maintained by bud inoculation onto 'Madame Vinous' sweet orange. Citrus chlorotic dwarf virus (CCDV) 46 was maintained by bud inoculation Scientific RepoRts | 6:29272 | DOI: 10.1038/srep29272 of alemow (Citrus x macrophylla Wester). Trees were grown in Metro Mix 510 potting mix. The greenhouse was maintained at 65-80 °F (18-27 °C) and ambient light was supplemented with high pressure sodium vapor lighting on cloudy days and throughout the winter season to extend the photoperiod. Trees were watered as needed with water containing nitrogen/phosphorus/potassium (100/25/100 ppm), copper (2 ppm) and iron (6 ppm).

Production and purification of scFv.
A library of single chain antibodies (scFv) with 2 × 10 7 unique members was created using the phagemid vector pKM19 and Escherichia coli DH5α F' for screening and selection, and plasmid pKM16 for propagation of the scFv 21,25 . This scFv library was screened to select scFv that bound to segments of proteins expressed on the surface of CaLas, including the major outer membrane protein, OmpA (scFv748; YP_003065185) and a membrane component of the flagellar apparatus, FlgI (scFv932; YP_003064789). The sequences that encode these scFv were cloned into the vector pKM16, which was designed to add both a His Tag and a FLAG epitope 47 to the carboxyl end of the scFv to facilitate purification and screening of the scFv. Single chain antibodies were produced and purified as described 48 by chromatography with the Ni-NTA purification system (Invitrogen, Frederick, MD). Clones of E. coli DH5α (pKM16) with inserts that encoded scFv were inoculated into of LB medium with ampicillin (100 mg/l) and glucose (2%) and grown at 37 °C overnight. Bacteria were diluted 1:10 into fresh media and grown with vigorous shaking until the culture density reached OD 600 0.8. The cells were collected by centrifugation at 8000 RPM for 15 min and suspended in 100 ml LB with ampicillin (100 μ g/ml), MgCl 2 (20 mM), and IPTG (1 mM). The bacteria were incubated at 32 °C overnight with shaking and then collected by centrifugation at 8000 RPM for 15 min. The bacterial pellet was suspended in 8 ml of native binding buffer and lysozyme (12 mg) was added followed by incubation on ice for 30 min. The pH of the native binding buffer was adjusted to 8.0 immediately before the purification was started. The cells were broken by six cycles of sonication at high power on ice (Virsonic 300, Virtis Scientific). Each cycle of sonication was for 10 s followed by 30 s of rest. Cellular debris was removed by centrifugation at 8000 RPM and the supernatant was transferred to a clean tube and the manufacturer's protocol was followed to purify the scFv. Aliquots of the purified scFv were removed for SDS-PAGE and quantification with the Bradford assay (Dojindo Molecular Technologies, Rockville, MD). A proteinase inhibitor cocktail (Sigma P9599) was added and the purified scFv were stored at 4 °C.

Preparation of polyclonal antiserum.
OmpA is a protein found in considerable concentration on the outer membrane of gram negative bacteria. A portion of the gene encoding the major outer membrane protein OmpA of CaLas, designated Omp3f, (Supplemental Fig. 1) was cloned into the pET102/D-TOPO vector system for expression in E. coli BL21. The expressed protein was purified under denaturing conditions using the 3′ 6X His tag encoded by the vector. SDS-PAGE analysis was used to confirm purity. The purified protein was adjusted to a concentration of 2 mg/ml and used to immunize a New Zealand white rabbit with four injections over a period of 56 days, 0.3 mg protein/injection (Cocalico Biologicals, Reamstown, PA) using TiterMax adjuvant (Sigma Aldrich, St. Louis, MO).
Direct tissue blot immuno assay. Petioles of symptomatic or asymptomatic leaves of diseased and corresponding healthy greenhouse trees were used to prepare tissue prints, generally following a published protocol 49 . Sections approximately 2 mm thick were cut from petioles and pressed for 10 seconds onto nitrocellulose membranes with 0.45 μ m pore size (Whatman, Thermo Fisher Scientific, Pittsburgh, PA). The printed membranes were transferred to Phosphate Buffered Saline with 0.05% Tween-20 (PBST) and washed two times for five minutes each on a reciprocal shaker. The PBST was removed and replaced with SuperBlock (PBS) Blocking Buffer (Thermo Fisher Scientific) for initial blocking at room temperature for 2 h. The membranes were transferred into a second blocking solution (PBST + 5% fat free skim milk) that contained the scFv diluted 1:100 or 1:1000. The tissue prints were incubated for 90 minutes at 37 °C or overnight at 4 °C and washed three times with PBST for 10 min each. Monoclonal antibodies that bind to the FLAG epitope of the scFv and conjugated with alkaline phosphatase (#A9469, Sigma Aldrich, St. Louis, MO) were diluted 1:10,000 and incubated with the tissue print for 1 h at 37 °C. The tissue prints were washed three times with PBST for 10 min each before substrate was added (33 μ l NBT + 16.5 μ l BCIP; Sigma) in 5 ml of alkaline phosphate assay buffer. Goat Anti Rabbit Pab (#12-448, EMD-Millipore) conjugated with alkaline phosphatase was used to detect Pab bound to OmpA of CaLas in tissue print assays. The tissue prints were incubated until color development could be seen. The tissue prints were photographed with a Zeiss Discovery 20 light microscope equipped with a digital camera. Serial sections of petioles were taken for tissue prints, and alternate sections were either used for DTBIA or set aside for extraction of DNA and qPCR to estimate the amount of CaLas present in each petiole 50 .

Identification of a citrus protein that binds anti-FLAG monoclonal antibodies. An antigen that
bound anti-FLAG monoclonal antibodies was present in the phloem vessels of diseased citrus. HLB-diseased leaves were removed from sweet orange trees. Petioles were ground in liquid nitrogen and the powder was taken up in buffer. Samples were loaded on 4-15% SDS-PAGE gels (Mini Protean TGX, BioRad Labs, Hercules, CA) and subjected to electrophoresis for 30 min at 60 V and 1 h at 100 V. The resolved proteins were then electro-transferred to nitrocellulose membranes. The nitrocellulose membranes were blocked and treated with anti-FLAG monoclonal antibody and developed as described for the tissue prints. SDS-PAGE gels were also prepared as described but not subjected to electro-transfer. The region corresponding to the region that bound the partial FLAG epitope was cut out of the gel and sent for sequence determination by GeLC-MS/MS with a Q-Extractive Orbitrap device (Bioproximity, Springfield VA). This returned a list of 3999 short peptides (5-53 amino acids) present in the excised SDS-PAGE band of interest. These peptides were blasted against the C. sinensis proteome 51 and the corresponding full length protein for each hit with a negative e-value was retrieved using Bioperl. Proteins between 90-115 kD were searched using the Grep program in Linux for the FLAG peptide sequence and variants.