Production of additional materials for education and outreach – HLB awareness activity book (50 copies) – HLB awareness/detection poster calendars 2011 (1250 copies) + 2012 (1500 copies) – Promotional magnifiers (500 units) These materials are distributed at training and workshop events, and also at Extension county agent organized events and professional trade meetings (ornamental industry). Various events where the team conducted educational and outreach programs reach an approximate 270 people. This include master gardener volunteers, Extension county agents, nursery and landscape professionals, citrus enthusiasts. All produced publication material (as well as video materials) is posted at http://plantclinic.tamu.edu/citrusgreening . All materials are available for download.
The objective of this project is to characterize the hypI (renamed as hyvI) gene and determine its effects on insect transmission and/or virulence in host plants. Transient expression with alternative expression systems and RT-qPCR, etc., will be used to elucidate the function of the hypI (hyvII) gene of Las and shed light on the molecular mechanism of this “phase variation” phenomenon; thereby developing a novel control strategy for citrus HLB. In addition, antibodies and probes along with standardized protocols developed during this project can be applied for better detection and differentiation of the HLB bacteria. The hyvI and hyvII within two Las prophages were further characterized. Sequencing analysis of the hyvI/hyvII genes from 40 Las DNA samples collected globally revealed sequence conservation within the individual repeats but an extensive variation regarding repeat numbers, their rearrangement, and the sequences outside of repeat region. These differences are found not only in samples with distinct geographical origins but also from a single origin and even from a single Las-infected sample. The Florida isolates contain both hyvI and hyvII while all other global isolates contain only one of the two. Interclade assignments of the putative HyvI/II proteins from Florida isolates with other global isolates in the phylogenic trees imply a multi-source introduction of the Las bacterium into Florida. We have developed a real-time PCR using SYBR Green 1 (LJ900fr) and TaqMan’ (LJ900fpr) protocols with primers and probe targeting nearly identical tandem repeats of 100bp hyvI and hyvII. Because of higher copy number of the tandem repeats per bacterial genome, these methods significantly improved detection capacity of the HLB bacterium. In comparable samples relative to 16S HLBaspr real-time PCR, these new methods reduced the relative detectable threshold by approximate 9 and 3 cycles for LJ900fr and LJ900fpr, respectively. From HLB samples with extreme low titer of Las bacterium, both LJ900fr and LJ900fpr detected Las from otherwise non-detectable samples by APHIS standard, HLBaspr. To determine the cellular localization of the HyvI protein in plant cells and the role of the two putative NLSs in hyvI gene, full-length hyvI and C-terminal region including two putative NLSs were amplified and cloned into pCX-DG vector with GFP driven by CaMV35S promoter, and transformed Agrobacterium tumefaciens strain GV 2660 with the pCX-DG-hyvI constructs. The results indicate that the HyvI protein did not target in plant nucleus but located in cytoplasm when transient expression in tobacco. In addition,the results show that hyvI gene did not induce hypersensitive response(HR) on tobacco leaves. RT(Reverse transciptase) PCR confirmed the hyvI gene expression both in the host plant and psyllids though the HyvI protein was not detected with Western blot. The obstacles for detection of the antigen from HLB-infected plants remains to be further investigated. When the full length (12 full repeats) of hyvI gene was cloned into pUF047plasmid, and replicated in heterologous hosts, the repeat number of hyvI gene remained the same in E. coli, but varied in Xanthomonas citri (citrus canker bacterium), ranging from 1.0 to 10.0 full repeats. Clones of X. citri containing the hyvI gene displayed different degree of growth retardation, indicating potential toxic effect of hyvI gene to X. citri.
The ultimate goal of the present project is the identification of inhibitory compounds to inactivate potential virulence determinants in Liberibacter asiaticus, the citrus greening causative agent. Consequently, it was proposed to evaluate a chemical library of compounds with a variety of chemical scaffolds. The library will be screened to identify ‘ligands’ for the selected targets. Five transcription factors and two periplasmic proteins involved in nutrients transport were selected. The first logical step toward our goal is the cloning and purification of the selected targets. The genes encoding three transcription factors (Clibasia_00835; 03370 and 01180) were successfully amplified, cloned and the proteins successfully expressed. The protein 01180 was already purified to homogeneity and it demonstrated excellent qualities to be screened using thermomelt assays. The proteins product of the genes identified as Clibasia_02535 and 01510 are currently being cloned. The periplasmic proteins were, as well, successfully amplified and cloned. The correct sequence of both genes was confirmed by sequencing reactions. The original clones, His6X tagged at the amino terminal, had very low solubility being found mostly as inclusion bodies. Consequently, the gene was re-cloned into different vectors carrying different Tags (like GST and S-tag) for affinity chromatography (to allow 1 step purification) and to be expressed using different hosts. The His6X was changed from the amino terminus to the carboxy terminus to allow better expression and higher purification yields. The new recombinant proteins are being tested in new hosts such as Bacillus subtilis or the yeast Pichia pastoris.
Several classification and spectral mapping methods were investigated using multispectral (MS) and hyperspectral (HS) airborne images of citrus groves taken in 2010. Spectral features derived from both ground reflectance measurement and airborne images were analyzed to find some notable differences between HLB infected and healthy canopies. With better ground truth records, more precise library for HLB infected and healthy canopies were collected that increased the classification accuracy. Spectral feature fitting achieved highest accuracy (90%) in the HS image with a reasonable false positive rate, but it didn’t work well with the MS image. Mixture tuned matched filtering had good accuracy and also the lowest rate of false positive. Mahalanobis distance showed a relatively high performance (76%). Maximum likelihood, neural network and support vector machine (SVM) couldn’t work properly in HLB detection, but SVM provided a fast, easy and adoptable way to build a mask for tree canopy to eliminate background that interfered with classification. We are carrying out a year-long (2010-2011) study to account for weather and seasonal variations in volatile organic compounds of HLB-infected trees using multiple methods at Lake Alfred, FL. Additional SPME GC/MS, and Twister GC/TOFMS samples were collected at USDA facility in Beltsville, MD from green house trees infected with three different strains of HLB (Las, Lam, African). Data obtained from all of these studies were subjected to a variety of data mining and signal processing methods to define the accuracy of disease incidence predictions. The diagnostic accuracy up to 82% was achieved. 14 robust biomarkers were singled out, with the diagnosis accuracy up to 98%. An issue of co-infection with CTV was identified and currently being addressed. Moreover, we have optimized the purge and trap procedures to concentrate leaf volatiles and have now sampled several HLB and control trees. Leaf volatiles are separated on a polar column and detected with a flame ionization detector. The HLB infected trees have 30 to 50% more volatiles than similar control leaves. Have observed almost 100 peaks from leaves on some HLB infected trees, but have only be able to identify half of them at this point. They are mainly aldehydes, hydrocarbon terpenes, alcohols and acids. Attempts to couple the purge and trap system to our mass spectrometer have been unsuccessful to date. Data analysis of fluorescence using the feature extraction method revealed that yellow fluorescence (UV excitation) always contributed to the classification of healthy, nutrient-deficient and HLB infected leaves. Among the different classifiers, Na’ve-Bayes model yielded high classification accuracy under lab conditions (> 85%); while bagged decision tree yielded high overall classification accuracy under both lab and field conditions (> 94%). In regard to visible-near infrared spectroscopy, we are transferring the knowledge from ground-based sensors to remote sensing applications to detect stress from top of the canopy. Multiple sensing platforms, retractable mast in agricultural vehicle for ground and multi-rotor remote sensing platform for aerial detection are being used. Multiple spectral camera (MCA, Tetracam Inc.) with selective visible-near infrared bands and thermal camera (Tau640, FLIR Systems) were purchased and a sensor system is being developed to evaluate the efficiency of the sensors to detect stress/disease from top of the canopy. Publications: (1) Li, X., W. S. Lee, M. Li, R. Ehsani, A. Mishra, C. Yang, and R. Mangan. 2011. Comparison of different detection methods for citrus greening disease based on airborne multispectral and hyperspectral imagery. ASABE Paper No. 1110570. St. Joseph, Mich.: ASABE. (2) Sankaran,S. and R. Ehsani. 2011. Visible-near infrared spectroscopy based citrus greening detection: Evaluation of spectral feature extraction techniques. Crop Protection, In Press.
A comparative field trial involving two copper loaded silica nanogel (CuSiNG) formulations (Cankicide-opaque and Cankicide-transparent) and a series of commercially available Cu biocides (such as Kocide 2000, Kocide 3000, Champ 30DP, Cuprofix Ultra 40, Kentan DF, Badge X2, NuCop HB, Nordox and Magna-Bon) has been initiated by James H. Graham group (CREC, Lake Alfred, FL). During this trial period, data are being collected to evaluate canker incidence and retention of copper residue on fruit surface. Additional particle size characterization was conducted using Atomic Force Microscopy (AFM). AFM data showed the formation of ultra-small particles as small as 10 nm. These particles were interconnected to form “gel-like” micron-size structures. The AFM data matches with high-resolution transmission electron microscopy (HRTEM) and dynamic light scattering (DLS) data as reported earlier. The antibacterial efficacies of the two formulations (transparent and opaque) were tested against X. alfalfae (a citrus canker surrogate), in nutrient broth under laboratory conditions. Kocide 3000 was used as the positive control and silica nanoparticle/nanogel (SiNP/NG) as the negative control. The metallic copper concentration was varied from 866 microgram to 43.3microgram per milliliter. The minimum inhibitory concentration (MIC) of metallic copper was estimated to be 100 microgram per milliliter. Additional antibacterial tests and fluorescence imaging study are being conducted to evaluate the interaction of the cankicide material with bacteria cells.
The goal of this project is the in vitro culture of Ca. Liberibacter asiaticus (LAS) associated with the Citrus greening syndrome. The strategy consists of primo-cultures of the bacteria in insect cells cultures used as feeder cells. Co-culture and axenisation LAS/Drosophila co-cultures: LAS can be maintained in co-culture with drosophila cells but are quickly lost due to the high growth rate of drosophila cell lines used. We previously observed that sodium pyruvate and ethanol amine had a positive effect on LAS fitness and a negative effect on insect cells growth. We tested those additives on drosophila cells: sodium pyruvate, at the concentration tested, had a lethal effect on drosophila cells but ethanol amine had a less deleterious effect and we are currently monitoring this cell culture before inoculation with LAS. LAS/Aedes co-cultures: Different co-cultures were continuously grown over several months with successive transfers. After successive dilutions we could get rid of the Aedes cells without losing the LAS detection. To verify that the culture was axenic, we checked if LAS was the only bacteria in our cultures. Other bacteria strains were found in co-culture with LAS (Delftia acidovorans, actinobacteria). Antibiotic selection was applied to get rid of the contaminants but resulted in the loss of the LAS signal as a consequence. Enforced antibiotic selection and additional sterilization steps were applied to the infected plant material surface for our new inoculations and the co-cultures are currently monitored. LAS/ Diaphorina citri co-cultures: Two Diaphorina citri cell lines from the Keyhani’s lab are now stably maintained. LAS was detected after 3 transfers but not more. D. citri cells are progressively adapted to an alternative medium that gave good results for LAS/ Aedes co-cultures. Medium improvement and LAS fitness: To improve LAS fitness in culture, we complemented the primo-cultures with various nutrients based on data from healthy and HLB infected Citrus phloem or on potential metabolic pathways deduced from the LAS genome sequence available. Several additives were selected that improved LAS growth in co-cultures. We did inoculations of LAS in medium with different combinations of those additives, with and without Aedes cells. LAS could be detected no more than 4 transfers in medium only but still much less than in conjunction with Aedes cells. A solid version of the medium was tested to try to isolate LAS clones with no success. Citrus inoculation The very next step is to inoculate the cultured bacteria into healthy Citrus in order to verify its pathogenicity and re-isolate it (Koch’s postulates). We set up a protocol of mechanical inoculation of healthy citrus and will test further the diffusion in the citrus phloem with a fluorescent dye. Another way to inoculate the pure LAS culture would be by acquisition through membrane by the insect vector and inoculation of healthy citrus via the infected insect. We are setting up experiments of acquisition through membranes. Our co-cultures conditions prove to be very reproducible in getting LAS in culture and our major line of improvement is to get no ‘contaminant’ bacteria along with liberibacter.
The multilocus simple sequence repeat (SSR) marker system effectively detects and differentiates ‘Candidatus Liberibacter asiaticus’ strains from various geographic and environmental locations worldwide. This genotyping system has improved resolution capabilities in fingerprinting genetically close ‘Ca. L. asiaticus’s strains. HLB-associated ‘Ca. L. asiaticus’ strains vary in pathogenicity. Various types and severities of HLB symptom were observed both in field and greenhouse grown citrus plants. Some of these strains cause only mild symptom while others show more severe symptom. These differences were found even within the same variety of citrus plants, suggesting that the genetic variations of the pathogen could play important roles in these host-pathogen interactions. The objective of this study is to extend from DNA-based genotyping to functional phenotyping. The goal of this research is to develop a gene-based marker system that links to phenotypes of ‘Ca. L. asiaticus’s strains. In silico genome analyses revealed that ‘Ca. L. asiaticus’ contained a number of genes associated with pathogenicity. Some of these genes seem to be of phage origin, suggesting that the bacterium has been actively involved in multiple phage integration events via the horizontal gene transfer. We have identified the SSR loci in this bacterial genome. Some of SSR loci were located in prophage gene sequence regions. Given the fact that the Las has a significantly reduced genome (1.2 Mbp), many of these repetitive regions located within or near coding or regulatory regions could have profound effects on gene expression and function. It has been reported that the repeat number variation in SSR seems to be intimately related to modulation of the expression of virulent factors. Based on the repetitive regions in Las genome, we identified pathogenic genes that are located in or near these regions. These pathogenic genes were classified into four groups based their putative functions: 1) Phage-derived genes which are often involved in pathogenesis through their transfer of genes and/or genetic rearrangement. 2) Genes defined as two component response regulators. This group of genes is important for sensing environmental signals and producing a transcriptional response from those signals. 3) Genes identified with hydrolase activity. One of the interesting genes in this group is salicylic acid hydrolase which could play an important role in hydrolysis of salicylic acid, a signaling molecule expressed in host in defense responses pathogen attack. However, to confirm its hydrolase function, the substrate for this gene-coded enzyme needs to be determined. 4) Zinc binding proteins, znuABC genes identified as putative high-affinity zinc uptake system that belongs to the ATP-binding component of ABC transporter protein. Zn uptake system ZnuABC is required for obligate bacterial zinc homeostasis in intracellular environments which contributes to the virulence of Las in citrus plants. Twenty primer pairs were designed that flank on the upstream and downstream of ORF to produce a full length for each gene. Ten HLB-affected samples representing virulent and weak virulent strains of ‘Ca. L. asiaticus’s were selected and amplified during this period of study. Research is in the process of conducting sequence association analyses to determine the linkage disequilibrium of the gene-based markers associated with the virulent traits of Las strains.
Citrus greening disease also known as Huanglonbing (HLB) is caused by a phloem-limited Gram-negative alpha proteobacterium, Candidatus Liberibacter asiaticus (CLas). It is the most destructive disease of citrus that could potentially compromise the productivity of Florida industry. In general phytopathogenic bacteria and alter the physiology of the host by injecting effector proteins into host cells to establish infection. . The sequence information has identified number of unique features of the CLas and most relevant to this proposal identification of number of potential effectors as part of hypothetical proteins which do not have homologues in other bacterial systems. Based on this supposition and as elucidated in the previous progress report we have cloned number of these effcetor genes into Citrus Tristeza Virus (CTV) vector to express the in the phloem of citrus plants which is the natural niche of the pathogen. The transfer of effectors using CTV vector is facilitated through prior agroinoculation of Nicotiana benthamiana and purification of CTV virions from agroinoculated N. benthamiana carrying CLas effetors before inoculation of citrus. This has been the most time consuming part of research till now. However, we have made good progress and several effectors have been ready for inoculation of citrus. One of the important effectors or virulence factors is hemolysin, a 50 kDa protein secreted by type I secretion system. The hypothesis is that the hemolysin might interfere with metal ion transportation leading to host metabolic imbalance potentially resulting in disease symptoms. Although it is identified as a toxin, it does not induce any visible phenotypic symptoms in N. benthamiana. We had submitted an abstract entitled ‘Functional studies of hemolysin (exotoxin) of ‘Candidatus Liberibacter asiaticus’expression in citrus using Citrus Tristeza Virus vector’ to the 2011 American Phytopathological Society conference. We plan to do the transcriptome analysis of the citrus following its transfer to citrus to understand the potential role as CLas toxin. We also tried to introduce another potential toxin, Serralysin, but did not succeed to clone into CTV mainly because of it s size. But we will make further attempts to introduce this potentially important toxin to citrus. We are also using a different approach to screen for virulence genes of HLB using Potato virus x (PVX) vector system which has been used extensively to screen for virulence genes herbaceous hosts. We have noticed that in the genome of CLas, there is only one copy of flagellin-encoding gene in contrast to four copies of them in Sinorhizobium meliloti. Transient expression of the Las-flagellin using PVX expression vector induces an immunity response with cell death in Nicotiana benthamiana. Although a full flagellum is not observed with HLB, rudimentary flagellar apparatii exist in CLas based on the sequence information. We have submitted an abstract identifying such genes in CLas and inoculating tobacco plants. We are towards the end of year one of this reserach and we have really made good progress which will be presented in the annual report and make a request for extension for the second year.
In the present study, performed in collaboration with Dr. Nelson Wulff at Fundecitrus, we have continued sequencing a fosmid DNA library of Ca. L. americanus (Lam) strain ‘S’o Paulo’ isolated from infected periwinkle from Brazil. Nearly all contigs have now been organized on a predicted “scaffold” and gaps continue to be closed by PCR. The overall gene organization and structure of the Lam genome is more similar to Ca. Liberibacter solanacearum (Lso) than to Las. Lam phage sequences corresponding to both SC1 and SC2 of Las were found, although SC1 (and not SC2 as in Las) appears to replicate as an integrative plasmid prophage. Las phage were found to become lytic in plant infections (Zhang et al. 2011); Lam appears to have same potential (ie., Las and Lam both carry “suicide program” or lytic cycle genes). SC2 in Lam Sao Paulo has clearly degenerated to a nearly nonfunctional state (in terms of lytic cycle, but its replication was detected in circular form. SC1 in Lam appears to more assume the role played by SC2 in Las, since its replication is high, and it carries two peroxidase genes that are likely to be needed for pathogenicity of plants and possibly transmission in psyllids. Notably, 2 of 4 peroxidase genes were missing on the SC2 phage, as well as 2 adhesins (surface anchor proteins).
Thus far, 496 medium formulations have been tested for support of Candidatus Liberibacter asiaticus ( Las). None have supported continuous axenic culture of the bacterium. A basal medium developed with the aid of the bacterium from mountain papaya or babaco that is closely related to Las has been modified in attempts to grow Las. different complex ingredients, such as bovine heart infusion, have been tested as replacements for the brain heart infusion broth in the medium. Supplementing the medium with different concentrations of insect prepared culture media, such as TMN-FH medium, CMRL 1066 medium, medium 199, and DS2 medium, has been tried. Replacing the ACES buffer with HEPES buffer to facilitate testing the effects of higher pH (from pH 7.2 to pH 7.8) than the usual pH that has been used (pH 6.8 to pH 7.0), and pH 7.4 was selected for further tests. Various supplements, such as mucin, zinc sulfate, potassium nitrate taurine, betaine, folic acid, catalase, yeastolate, cyclic AMP, arabinose, sodium pyruvate, and lactoalbumin hydrolysate, have been tried. Some of the medium modifications appear to promote Las biofilm formation around the psyllid alimentary tracts that have been used as a source of inoculum, but planktonic growth has been minimal. Attempts to culture Las continue.
The main objective of the project is to select single chain variable fragment antibodies (scFv) against Huonglongbing (HLB), Citrus Verigated Chlorsis (CVC) and the development of immunoassays that incorporate these scFvs. We have successfully constructed one scFv library from chicken spleens that were immunized with bacterial proteins. The immunization protocol and delivery of spleens were performed in 100 days, and mRNA was immediately extracted to construct the library, using recombinant PCR, into an M13 vector. To validate the library, we have selected through Phage Display several antibodies with high immunoreactivity to their antigen targets for three pathogens: Xylella fastidiosa, Agrobacterium tumefaciens and Xanthomonas axonopodis. For pre-validation of selections, we have performed ELISA assays for each of the selected recombinant antibodies and on the basis of this have selected 3 candidate scFvs. Since these scFv’s were selected using intact whole bacteria we attempted to identify the candidate outer membrane proteins (OMPs) whose epitopes form the basis of their reactivity. We isolated and purified the proteins present in the membrane fraction of the Xylella and Xanthomaonas pathogens and performed cross-reaction analysis of selected scFvs against these proteins, which confirmed three highly reactive clones: the Xf.02, Xf.03 and Xf.05. We then sequenced the genes that encode these three scFvs by sequencing the DNA from the M13 vectors, and performed bioinformatic analysis for structural prediction of the fragment antibodies. The corresponding scFv protein were expressed and secreted in culture in a TOP10 vector, and HPLC-purified for further validation with ELISA and Western blots. All three antibodies strongly recognize a 42 kDa MopB protein but via different epitopes (we speculate). Interestingly, the same antibodies could also recognize the related protein in a related pathogen Xanthomonas axonopodis OMPs, providing further evidence that these antibodies might detect the MopB in CVC and HLB. The detection of MopB in Xylella was comparable with the reactivity of a polyclonal antibody against purified Xylella MopB (made by Prof. Bruening). Proteomic analysis was carried out on the reactive protein band by excising it out of the gel, treating it with trypsinization and then functional identification of the protein by sequencing fragments on a ESI:MS:MS machine. Additional evidence of scFv reactivity was obtained through fluorescence microscopy showing the binding of the selected scFvs to the Xylella surface. We hypothesize that one of these antibodies recognizes multiple epitopes with common motifs and may become a general antibody for citrus bacterial pathogen detection. We have established a very simple conventional ELISA with the antibodies against bacterial OMPs, and the assay parameters will be translated to a lateral flow and fluorescence spectroscopy assays in the near future to improve sensitivity of the test.
The main objective of the project is to select single chain variable fragment antibodies (scFv) against Huonglongbing (HLB), Citrus Verigated Chlorsis (CVC) and the development of immunoassays that incorporate these scFvs. We have successfully constructed one scFv library from chicken spleens that were immunized with bacterial proteins. The immunization protocol and delivery of spleens were performed in 100 days, and mRNA was immediately extracted to construct the library, using recombinant PCR, into an M13 vector. To validate the library, we have selected through Phage Display several antibodies with high immunoreactivity to their antigen targets for three pathogens: Xylella fastidiosa, Agrobacterium tumefaciens and Xanthomonas axonopodis. For pre-validation of selections, we have performed ELISA assays for each of the selected recombinant antibodies and on the basis of this have selected 3 candidate scFvs. Since these scFv’s were selected using intact whole bacteria we attempted to identify the candidate outer membrane proteins (OMPs) whose epitopes form the basis of their reactivity. We isolated and purified the proteins present in the membrane fraction of the Xylella and Xanthomaonas pathogens and performed cross-reaction analysis of selected scFvs against these proteins, which confirmed three highly reactive clones: the Xf.02, Xf.03 and Xf.05. We then sequenced the genes that encode these three scFvs by sequencing the DNA from the M13 vectors, and performed bioinformatic analysis for structural prediction of the fragment antibodies. The corresponding scFv protein were expressed and secreted in culture in a TOP10 vector, and HPLC-purified for further validation with ELISA and Western blots. All three antibodies strongly recognize a 42 kDa MopB protein but via different epitopes (we speculate). Interestingly, the same antibodies could also recognize the related protein in a related pathogen Xanthomonas axonopodis OMPs, providing further evidence that these antibodies might detect the MopB in CVC and HLB. The detection of MopB in Xylella was comparable with the reactivity of a polyclonal antibody against purified Xylella MopB (made by Prof. Bruening). Proteomic analysis was carried out on the reactive protein band by excising it out of the gel, treating it with trypsinization and then functional identification of the protein by sequencing fragments on a ESI:MS:MS machine. Additional evidence of scFv reactivity was obtained through fluorescence microscopy showing the binding of the selected scFvs to the Xylella surface. We hypothesize that one of these antibodies recognizes multiple epitopes with common motifs and may become a general antibody for citrus bacterial pathogen detection. We have established a very simple conventional ELISA with the antibodies against bacterial OMPs, and the assay parameters will be translated to a lateral flow and fluorescence spectroscopy assays in the near future to improve sensitivity of the test.
With GC/DMS, SPME GC/MS, and Twister GC/TOFMS, we are carrying out a year-long (2010-2011) sampling plan for the HLB study with collaborators at Lake Alfred, FL to account for weather and seasonal (trees blossoming, fruit harvest, etc.) variations, with four trips having been made in 11/2010, 12/2010, 01/2011 and 02-03/2011. For a CTV study, using the same instruments, we have collected 5 batches of the samples in Lincove and Pauma Valley CA and the USDA facility in Beltsville, MD. Meanwhile, we have also carried out an analysis of PCR positive sweet orange infected with CVC strains 9a5c and 104 also at the USDA facility in Beltsville, MD. Data obtained from all of these studies was subjected to a variety of data mining and signal processing methods to define the accuracy of disease incidence predictions: (1) Student’s t-test selected separable pixels (signals) of the HLB DMS data, we obtained a diagnosis accuracy of 71% with a PLS model; (2) Applying wavelet transformation to the HLB DMS data, we increased the HLB diagnosis accuracy up to 82% and more interestingly the signal within the first 3 minutes yielded an equally good result to the whole time domain; (3) Similarly, by applying a PLS model to the wavelet coefficients of the CTV DMS data, the cross-validation accuracies are 81% for Healthy vs. severe-CTV, 74% for Healthy vs. mild-CTV and 80% for severe-CTV vs. mild CTV; (4) By developing a 3-class classification model for the wavelet coefficients of the CVC DMS data, we systematically generated the following accuracies: 86% for healthy, 46% for mild-CVC, and 97% for severe-CVC. To identify biomarkers associated with specific diseases, our typical results include: (1) For HLB Twister GC/TOFMS data, we detected 62 potential biomarker and obtained a diagnosis accuracy of 94% for HLB using a 4 latent variables based PLS model; and (2) By developing a genetic algorithm oriented searching process which was coupled with a classification model to evaluate the fitness of the selected biomarkers, we not only reduced the number of pertinent biomarkers from 62 to 14 but also increased the diagnosis up to 98%; (3) Based on the HLB SPME GC/MS data, we located 9 potential biomarkers for HLB diagnosis and obtained a diagnosis accuracy of 83%; (4) Based on the CTV Twister GC/TOFMS data, we detected 41 potential biomarker for CTV diagnosis with a diagnosis accuracy of 86%. After narrowing the potential biomarker set down to 31 biomarkers, we generated an accuracy of 85% for a more complex separation between healthy and two CTV related categories: CTV and CTV + ‘stubborn’, indicating the robustness of the selected biomarkers. We are also analyzing gene expression data on HLB and CTV infected plants and found gene based biomarkers focusing on sugar metabolism and soucrce-sink mediated gene regulation. Overall, the results of citrus disease diagnosis and biomarker detection are promising. With more samples to be collected and analyzed and the exploration on machine learning strategies, we are expecting a robust diagnosis model and a set of pertinent biomarkers for each citrus disease in this study.
We have completed the target selection and design of MOLigo probes for detection of Candidatus Liberibacter, Xylella fastidiosa, Xanthomonas axonopodis pv. citri, and citrus tristeza virus. Testing has begun on characterization probes for Liberibacter and CTV using synthetic DNA targets, and Liberibacter assays show promise for being able to detect and perform species designation. Probes for CTV detection and discrimination between mild and severe clades are also showing promise on synthetic templates. Because there was recently a large influx of new Liberibacter sequence data since probes were designed, we performed an in silico evaluation of all probes and determined they are still valid for detecting all isolates for which sequence data has been reported. Dr. John Hartung of USDA ARS in Beltsville, MD has provided material for testing Xylella and Xanthomonas, and probes for detecting these pathogens are being tested initially on synthetic targets. Targets for Xylella have been selected as well, and design and testing of these additional probes will be initiated shortly. Once these probes show promise on synthetic targets we will begin testing sample material from Dr. Hartung’s lab. Dr. Bill Schneider of USDA ARS Ft. Detrick has agreed to provide nucleic acids from plants infected with CTV and Liberibacter for the validation panels. We will continue to gather and assemble these samples into our validation panels for testing. We focused our early efforts on developing a full detection and species discriminating assay for Liberibacter due to the urgent need to provide tools for management of HLB. The current assay looks promising when synthetic templates are used, and we look forward to testing them on sample material. We have discussed with Luminex their plans to discontinue in January 2012 their non-magnetic beads that we are using to develop these assays. We plan to continue to develop assays using non-magnetic beads, and will then transfer the assays to magnetic beads once they are performing as expected. We have been assured that Luminex or one of its partner companies will be able to provide xTAG non-magnetic beads into the foreseeable future, but we will evaluate the assays on magnetic beads once development is complete. Retail pricing for their MagPix instrument is currently $50K, and their Luminex 100/200 instruments retail for $75K. A higher throughput instrument (FlexMap 3D) capable of up to 500-plex is also available. Additional accomplishments from computational team: 1) Designed overlapping oligos to generate double-stranded targets. The overlapping oligo pairs will be annealed at 55 degree C and extended with a DNA polymerase to generate double-stranded templates. 2) Universal primers were shortened with reduced TM (45 C) to improve sensitivity and reduce background signal in the overall assay. 3) Redesigned MOLigo probes for targeting negative strands of the 4 target pathogens. 4) Identified signatures specific for the virulent strains of Xylella fastidiosa based presence of certain virulence factors such the gene for pili (XF29-31) and fimbriae (X78, XF80) in the virulent strains. We also identified the SNPs that are capable of distinguishing virulent strains from non-virulent strains.
Over the past two seasons, a total of nine experiments were conducted on ‘Fallglo’, ‘Sunburst’, ‘Ruby’ red grapefruit, and navel oranges evaluating the effect of different washing methods on inhibition of degreening and ease of grading out fruit with canker lesions. The results will likely be applicable to other grade or quarantine defects as well. The washing methods tested included: 1) full wash (brush bed + high-pressure wash [HPW]) plus waxing, 2) full wash (brush bed + HPW), 3) HPW only, 4) HPW without the normal brushes, 5) brush wash only, 6) only running over a PVC roller conveyer, or 7) an untreated control. All wash treatments were conducted on commercial wash lines with wash durations of approximately 70 sec. on the brush bed, and 20 sec. on the HPW. Normal brush rotation speed was ~100 rpm. After washing, early season fruit were degreened under simulated commercial conditions (5 ppm ethylene, 85F, 95% RH) and peel color of the fruit was measured almost daily. The fruit were subsequently stored under ambient conditions (~70-75F) and evaluated weekly for the development of decay and physiological disorders. Results showed that with more brushing on the packingline, the greater the inhibition of degreening. Specifically, degreening was most inhibited by fruit receiving the full wash, followed by brush wash or HPW only, and least in fruit that passed under the HPW nozzles without being brushed. Fruit color development was almost completely inhibited if the fruit were also waxed after washing, whereas fruit that were not washed at all, but simply run on a PVC roller conveyor experienced almost no inhibition of degreening. The method of washing had no significant effect on the development of postharvest decay or disorders. These results were observed on all citrus varieties tested. As expected, the method of washing affected how clean the fruit became and how well graders could identify blemishes on the fruit. On a 1 to 9 scale, with higher scores indicating fruit that were easier to grade, fruit receiving the full wash were the easiest to grade, whereas unwashed fruit (the control) were the most difficult. An experiment with 2400 fruit and another utilizing commercial graders evaluating fruit as it was run over a grading line gave similar results. However, the benefits of washing on gradeability was not significant when the fruit were naturally clean at harvest, i.e., early in the season. Such fruit could be adequately graded even before washing. Therefore, if fruit need to be pre-graded for defects of quarantine significance (i.e., citrus canker, citrus black spot, etc.) early in the season when the fruit is relatively clean, very minimal, if any, washing is needed for adequate grading. Grading the fruit without washing would result in minimal effects on degreening. As the season progresses, fruit become dirtier, but the fruit have much better natural color, and pre-washing can be initiated and intensified as needed to allow adequate pre-grading without adversely affecting color.
(863) 956-8817
(863) 956-5894
700 Experiment Station Road
Lake Alfred, FL 33850
Email Us
