DEFINING THE INTERACTIONS WITHIN THE CITRUS MICROBIOME DURING HUANGLONGBING DISEASE PROGRESSION AND IN RESPONSE TO PHYTOPATHOGEN REMOVAL

By

RYAN ANDREW BLAUSTEIN

A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY

UNIVERSITY OF FLORIDA

2017

1

© 2017 Ryan Andrew Blaustein

2

To my wife, parents, siblings, and grandmother for constant encouragement and unconditional love

3

ACKNOWLEDGMENTS

Earning my Ph.D. was an arduous and exciting journey that would not have been possible without the help of many others. I recognize and thank the following individuals for their contributions and assistance.

I express gratitude to my major professor, Dr. Max Teplitski, for inviting me to the

University of Florida and providing the constant mentorship and essential tools needed to complete this research. I also wish to thank my supervisory committee for their support. I am grateful to Dr. Graciela Lorca for giving me the opportunity to lead projects within her interdepartmental collaborations and for providing supervision that that has been instrumental to my progress. Thanks also goes to Dr. Ana Conesa and Dr. Kelly

Morgan for their insights that helped improve my dissertation projects.

Within the UF Genetics Institute, I wish to thank Dr. Julie Meyer for teaching me the molecular biology methods and bioinformatics analyses that were a major component of this research. Also, thanks goes to Dr. Claudio Gonzalez for advising field experimentation and providing feedback. I am grateful to Dr. Fernando Pagliai and Mr.

Christopher Gardner for their oversight to the antimicrobial treatment study. Additionally,

I recognize my lab colleagues, Dr. Marcos de Moraes, Dr. Andree George, and Mrs.

Hanh Nguyen, for their advice and friendship. Thanks also goes to my undergraduate mentees, Mr. Javier Silfa-Cifuentes and Ms. Sarah Stavros, for their tremendous help with sample and data processing.

Outside of the laboratory, I am thankful for the guidance provided by Mr. Michael

Sisk, the Academic Program Specialist in the UF Soil and Water Sciences Department.

Thanks also goes to Mr. Robert Adair, Dr. José Chaparro, Dr. Nick Comerford, Dr.

4

Yong-Ping Duan, and Dr. Kelly Morgan for allowing access to the various field sites sampled in this work.

This research was funded by the National Institute of Food and Agriculture, U.S.

Department of Agriculture grant (no. 2015-70016-23029) awarded to my supervisors. I am grateful to the University of Florida Graduate School Fellowship for financial support.

The contents of this dissertation is solely my responsibility and do not necessarily represent the official views of the granting agencies.

5

TABLE OF CONTENTS

page

ACKNOWLEDGMENTS ...... 4

LIST OF TABLES ...... 9

LIST OF FIGURES ...... 10

LIST OF OBJECTS ...... 12

LIST OF ABBREVIATIONS ...... 13

ABSTRACT ...... 16

CHAPTER

1 CHALLENGES FOR MANAGING HUANGLONGBING DISEASE: CURRENT CONTROL MEASURES AND FUTURE DIRECTIONS ...... 18

Summary ...... 18 Introduction ...... 18 Evaluating Treatment Effects: Phytopathogen Detection ...... 22 Effects of Chemical Treatments ...... 24 Broad-range Antibiotics ...... 25 Antimicrobials that Specifically Target Liberibacter spp...... 31 Additional Measures to Control Liberibacter spp...... 33 Plant Microbiota Implications for Citrus Health ...... 37 Ongoing Challenges Associated with Field-scale Treatment ...... 39 Aims and Hypotheses ...... 42

2 MATERIALS AND METHODS: PROTOCOL OPTIMIZATION FOR 16S rRNA GENE SEQUENCING OF CITRUS SAMPLES ...... 53

Sample Collection ...... 53 DNA Extraction ...... 53 16S rRNA Gene Amplicon Preparation and Sequencing: An Initial Attempt ...... 54 16S rRNA Gene Amplicon Preparation and Sequencing: PNA Clamps ...... 57 Bioinformatics Pipelines ...... 60 Chemical Therapeutics ...... 60

3 DEFINING THE CORE CITRUS LEAF- AND ROOT-ASSOCIATED MICROBIOTA: FACTORS ASSOCIATED WITH COMMUNITY STRUCTURE AND IMPLICATIONS FOR MANAGING HUANGLONGBING (CITRUS GREENING) DISEASE ...... 71

Summary ...... 71

6

Introduction ...... 72 Methods ...... 75 Sample Collection and Processing ...... 75 16S Illumina Sequencing ...... 76 Data Analysis ...... 77 Results ...... 79 Citrus-associated Microbial Community Structure and Core Microbiota ...... 79 Spatiotemporal Variability of HLB ...... 81 Biotic and Abiotic Factors Associated with Community Structure ...... 82 Putative Interactions Between Native Bacteria and Liberibacter spp...... 85 Discussion ...... 86

4 NOVEL, TARGET-SPECIFIC ANTIMICROBIALS SUPPRESS Liberibacter PHYTOPATHOGEN WITHOUT DISRUPTING NATIVE MICROBIOTA ...... 100

Summary ...... 100 Methods ...... 103 Antimicrobial Treatments ...... 103 Microbial Community Analyses ...... 104 Results ...... 106 Treatments Do Not Disrupt Microbiota Diversity ...... 106 Treatments Suppress Liberibacter in Root-associated Microbial Communities ...... 108 Discussion ...... 109

5 CITRUS METAGENOME ASSOCIATIONS WITH HUANGLONGBING DISEASE: PHYTOPATHOGEN REMOVAL CORRELATES WITH SHIFTS IN FUNCTIONAL DIVERSITY ...... 119

Summary ...... 119 Introduction ...... 120 Methods ...... 123 Datasets Obtained ...... 123 Metagenome Predictions ...... 124 Comparative Data Analyses ...... 125 Results ...... 126 Quality Control ...... 126 Metagenomes of Citrus Microbiota Associated with HLB Disease ...... 126 Phytopathogen Removal May Shift Root Metagenomes to “Healthier” State . 128 Discussion ...... 129

6 CONCLUSIONS AND FUTURE DIRECTIONS ...... 141

APPENDIX

A BIOINFORMATICS PIPELINE ...... 155

Pre-processing ...... 155

7

Taxonomic Assignments with QIIME ...... 158 Open-reference OTU Picking ...... 158 Closed-reference OTU Picking ...... 160 Metagenomic Predictions with PICRUSt ...... 160

B SUPPLEMENTAL MATERIAL FOR CHAPTER 3 ...... 163

LIST OF REFERENCES ...... 170

BIOGRAPHICAL SKETCH ...... 184

8

LIST OF TABLES

Table page

1-1 Results from previous studies that have reported effects of antimicrobial treatments on Liberibacter asiaticus and HLB progression ...... 49

1-2 Results from previous studies that have reported effects of non-antimicrobial control measures on Liberibacter asiaticus and HLB progression ...... 51

2-1 DNA extractions with different kits ...... 61

2-2 List of index barcodes on primers that were used for making amplicon libraries ...... 62

2-3 Treatments for optimizing amplicon preparation protocol ...... 64

3-1 Metadata for samples collected and used in the molecular survey ...... 92

3-2 Associations between microbiota structure and location and cultivar ...... 93

4-1 Relative abundances of members of Alphaproteobacteria that were impacted by antimicrobial treatment ...... 114

5-1 List of samples used in metagenome prediction study ...... 134

5-2 KO terms that had relative abundances that were most associated with the differences in metagenomes of asymptomatic and HLB-diseased trees ...... 135

B-1 Differences in leaf microbiota from Valencia trees by location sampled ...... 163

9

LIST OF FIGURES

Figure page

2-1 Relative abundances of dominant taxa in leaf microbiota detected in initial sequencing event ...... 65

2-2 Effects of PNA treatment on bacterial enrichment in sequencing procedure ...... 66

2-3 Effects of PNA treatment on microbiota structure ...... 67

2-4 Effects of PNA treatment on bias in taxa detected ...... 68

2-5 Effects of PNA treatment on characterization depth of microbial communities ... 69

2-6 Effects of PNA treatment on alpha diversity detected for microbial communities ...... 70

3-1 Rank abundance curves for citrus microbiota ...... 95

3-2 Spatiotemporal variation in Liberibacter spp...... 96

3-3 NMDS plots demonstrating differences in microbial community structure based on plant organ, location, cultivar, and HLB symptom severity ...... 97

3-4 Relative abundances of dominant classes and genera within citrus- associated microbial communities ...... 98

3-5 Citrus leaf microbiota interaction network ...... 99

4-1 Microbiota structure not impacted by treatment; Leaf microbiota structure varies based on season ...... 115

4-2 Alpha diversity of microbiota is not grossly disrupted by treatment, suggesting target-specificity ...... 116

4-3 Effects of treatment and season on fold-change in Liberibacter in leaf and root microbiota ...... 117

4-4 Bacterial families that experienced a fold change in relative abundance in response to Liberibacter reduction within root microbiota ...... 118

5-1 PCA plots indicate differences in metagenomic profiles of microbiota of asymptomatic and HLB-diseased tree taken from asymptomatic and HLB- diseased trees ...... 136

5-2 OTU contributions to the KO functions that were differentially abundant relative abundances in asymptomatic and HLB-diseased trees ...... 137

10

5-3 Heatmap of KEGG (Level 2) pathways encoded in metagenomes of citrus microbiota of asymptomatic and HLB-diseased trees ...... 138

5-4 PCA showing clustering for root metagenomes in molecular survey and treatment study; asymptomatic trees are more similar to HLB-diseased trees that were treated with antimicrobials than untreated controls ...... 139

5-5 Heatmap of KEGG (Level 3) pathways encoded in metagenomes of citrus microbiota of asymptomatic and HLB-diseased trees that were treated with antimicrobials ...... 140

B-1 Locations sampled in molecular survey ...... 164

B-2 Photos illustrating the range in categories of HLB symptom severity for trees sampled in the molecular survey ...... 165

B-3 Distributions of core and accessory members of citrus-associated microbial communities ...... 166

B-4 Photos showing HLB symptom progression in severely diseased trees over time ...... 167

B-5 Correlation between Liberibacter spp. relative abundance and microbiota alpha diversity ...... 168

B-6 Relative abundances of dominant families in citrus root microbiota are stable over time ...... 169

11

LIST OF OBJECTS

Object page

Object 3-1. List of core citrus microbiota ...... 94

12

LIST OF ABBREVIATIONS

AA ascorbic acid

ANOSIM analysis of similarity

ANOVA analysis of variance bp base pair cm centimeter

Cas9 CRISPR associated protein 9 cmpd compound

CRISPR clustered regularly interspaced short palindromic repeats

CTV citrus tristeza virus

CVC citrus variegated chlorosis

Cu copper cv cultivar diam diameter

DMSO dimethyl sulfoxide

DNA deoxyribonucleic acid dNTPs deoxynucleotide triphosphates

ELF electrophoretic lateral fractionation

FDR false discovery rate g gram gDNA genomic DNA

GMO genetically modified organism

HF high fidelity

HLB Huanglongbing (i.e., “Citrus greening” disease)

13

KEGG Kyoto Encyclopedia of Genes and Genomes

KO KEGG orthology

L liter mL milliliter mm millimeter

Mn manganese mPNA mitochondrial PNA mRNA messenger RNA

NCBI national center for biotechnology information nm nanomolar

NMDS non-metric data scaling

NSTI nearest sequenced taxon index

ºC degree Celsius

OTU operational taxonomic unit

PAST Paleontological Statistics (software)

PICRUSt Phylogenetic Investigation of Communities by Reconstruction of

Unobserved States (software)

PCA principle component analysis

PCR polymerase chain reaction

PNA peptide nucleic acid pPNA plastid PNA

QIIME Quantitative Insights Into Microbial Ecology (software) qPCR quantitative PCR

14

RNA ribonucleic acid rRNA ribosomal RNA

SD standard deviation

SE standard error sec seconds

TRIS tris(hydroxymethyl)aminomethane

2-DDG 2-deoxy-D-glucose

µL microliter

µM micromolar uncl unclassified

Zn zinc

15

Abstract of Dissertation Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy

DEFINING THE INTERACTIONS WITHIN THE CITRUS MICROBIOME DURING HUANGLONGBING DISEASE PROGRESSION AND IN RESPONSE TO PHYTOPATHOGEN REMOVAL

By

Ryan Andrew Blaustein

December 2017

Chair: Max Teplitski Major: Soil and Water Sciences

Huanglongbing (HLB; “citrus greening”) continues to spread and cause significant damages to global citrus production. Among potential treatment options, biocontrol of the disease through the manipulation of the phytomicrobiota could be considered. Prior to exploring this, it is critical to understand the stable state(s) of the citrus microbiome and to elucidate key interactions between native microbiota and

Liberibacter (i.e., HLB pathogen), as well as associations between host health and the structure and function of the microbiota. In this work, the core citrus microbiota were defined across a number of variables, including disease symptom severity, location, cultivar, season, and time. The relative abundance of Liberibacter among leaf microbiota negatively correlated with alpha diversity and positively correlated with HLB symptom progression, suggesting that community diversity decreases as symptoms progress. Network analysis identified mutually exclusive relationships between

Liberibacter and Burkholderiaceae, Micromonosporaceae, and Xanthomonadaceae, which may mediate the inverse association between the phytopathogen and the stable microbial community. Next, the effects of two small molecule antimicrobials on citrus

16

microbiota were determined in a long-term field study. Although the treatments did not impact the relative abundance of the Liberibacter in leaves, they significantly suppressed its proliferation in roots and induced removal in many cases. The compounds did not grossly disrupt microbiota structure, suggesting target-specificity.

Lastly, metagenomic functions of the microbiota in both studies, as predicted from taxonomic assignments, were compared. Metagenomes of HLB-diseased trees contained enrichments in functions involved in cell repair and motility, while those of asymptomatic trees had enrichments in functions involved in diverse metabolic pathways and cell signaling with the surrounding environment. The functional profiles of root microbiota of asymptomatic trees were more similar those of HLB-diseased trees that had received treatment than untreated controls, suggesting that the antimicrobials caused the microbiome to shift towards a “healthier” functional state. Overall, we (i) identified plant-beneficial microbiota and microbial interactions that have implications for biocontrol of HLB, and (ii) determined that the novel treatments tested were target- specific against Liberibacter, warranting further investigation. Our findings advance the understanding of microbiota-pathogen-host relationships, which not only have applications for HLB management, but also broader disease control.

17

CHAPTER 1 CHALLENGES FOR MANAGING HUANGLONGBING DISEASE: CURRENT CONTROL MEASURES AND FUTURE DIRECTIONS*

Summary

Huanglongbing (HLB; "citrus greening” disease) has caused significant damages to the global industry as it has become well established in leading citrus-producing regions and continues to spread worldwide. In recent years, there have been substantial efforts to develop practical strategies for managing Liberibacter spp., the HLB pathogen; however, a literature review on the outcomes of such attempts is still lacking. This work summarizes the greenhouse and field studies that have documented the effects and implications of chemical-based treatments (i.e., applications of broad-spectrum antibiotics, small molecule compounds) and non-chemical measures (i.e., applications of plant-beneficial compounds, applications of inorganic fertilizers, biological control, thermotherapy) for phytopathogen control. The ongoing challenges associated with mitigating HLB at the field-scale, such as the seasonality of symptoms and resiliency of

Liberibacter spp. populations, limitations for therapeutics to contact the phytopathogen in planta, adverse impacts of broad-spectrum treatments on plant-beneficial microbiota, and potential implications on public and ecosystem health, are also discussed.

Introduction

Huanglongbing (HLB), commonly known as “citrus greening”, has become the most destructive citrus disease worldwide. It is caused by the psyllid-transmitted, phloem-limited bacteria Liberibacter asiaticus, Liberibacter americanus, and Liberibacter

* Reprinted (151) with permission from Blaustein et al., 2017, Phytopathology, Challenges for managing Candidatus Liberibacter spp. (huanglongbing disease pathogen): current control measures and future directions, Phytopathology, doi:10.1094/PHYTO-07-17-0260-RVW.

18

africanus (1). Infection by any one of these organisms, which were named for their geographic distributions when confirmed with molecular-based reinfection studies (i.e.,

Koch’s postulates cannot be fulfilled since they are not culturable) (2, 3), causes similar symptoms of fruit malformation and altered host phenotype (4). Diseased trees develop leaves with characteristic blotchy mottle (i.e., asymmetric leaf yellowing) and produce undersized, lopsided, and greenish-colored fruit that often drops prematurely (4). HLB- induced host limitations for photoassimilate transport and nutrient uptake, and the associated dieback of canopy and fibrous roots, lead to tree death, typically, within a decade of initial symptom development (1). All varieties of cultivated citrus are susceptible to HLB, though to varying degrees (5, 6). From an economic standpoint, the severity of HLB is underscored by an estimated $4.5 billion in damages to the Florida citrus industry attributed to the disease since L. asiaticus was first was detected in the state in 2005 (7). With Liberibacter spp. having become established in over 40 countries

(8), and considering that citrus is one of the most important fruits in the world in terms of its commercial production market, processing, and global trade (9), there is an urgent need for the development of effective strategies to mitigate HLB.

Despite the presence of HLB-like symptoms dating back over a century (in

China) and the ongoing efforts to control the disease, there is still a lack of treatment options that are technically feasible, sustainable, and environmentally safe (4, 10).

Bimonthly foliar applications of brassinosteroids on HLB-diseased trees over the course of one year were recently reported to induce expression of plant defense genes and significantly lessen, though not eliminate, titers L. asiaticus within leaves (11).

Alternatively, while sprays applications of plant beneficial compounds (i.e., L-arginine

19

and 6-benzyl-adenine combined with gibberellins) and defense regulators (i.e., ascorbic acid [AA], β-aminobutyric acid [BABA], 2,1,3-benzothiadiazole [BTH], 2-deoxy-D- glucose [2-DDG], and 2,6-dichloroisonicotinic acid [INA]) were reported to somewhat slow HLB symptom progression in trees, they lacked inhibitory effects on L. asiaticus in planta and did not facilitate disease symptom recovery (12, 13) . Similarly, inorganic fertilizer applications that have been tested in efforts to boost citrus production in HLB- affected groves were reported to be generally inconsequential (6). In fact, applying zinc- based compounds to treat HLB-associated zinc deficiency-like symptoms has been reported to actually harm citrus plants by promoting the growth of L. asiaticus, perhaps due to the high-affinity zinc uptake system encoded in the genome of the phytopathogen (14, 15). Others reported nutrient amendments combined with insecticide applications for vector control to improve citrus yields in HLB-affected groves, though these forms of management, which do not target the phytopathogen, did not limit the titer of L. asiaticus nor disease symptom progression in the infected trees

(16). Alternatively, applications of several types of chemical therapeutics have demonstrated capabilities for L. asiaticus suppression and HLB symptom recovery in small-scale greenhouse and growth chamber studies (17–19). However, there is still much work to be done before these types of treatments may be practical for broad HLB control, since there have been mixed results, with only moderate success at impacting phytopathogen titer and disease symptoms, that have been reported for antibiotic-based treatments applied to HLB-diseased trees in the field (19–22).

On the horizon, there are prospects for utilizing an engineered version of citrus tristeza virus (CTV) in planta to attack Liberibacter spp. or to produce a spinach defense

20

gene that may be effective against HLB (23). Researchers are also attempting to use the CRISPR-Cas9 gene editing tool to create citrus cultivars that are less susceptible to

Liberibacter spp. and/or express genes that may prevent vector transmission (23–25).

Similar genome editing methods were previously used to confer citrus resistance to another important phytopathogen, Xanthomonas citri, the causative agent of citrus canker (26). However, these novel CTV-based and gene editing approaches are still several years away from producing desired results and, ultimately, meeting regulatory requirements that would make them feasible and marketable for commercial production operations (23). In the meantime, the most viable option for managing HLB in regions where it has become problematic may be the use of antimicrobials. Compared to other forms of disease control, small molecules with antibacterial activities, which have the capacity to kill Liberibacter spp. or, at least, suppress its growth, hold significant promise (17, 27, 28). Finding and utilizing novel compounds that are target-specific to

Liberibacter spp. is essential for improving treatment efficacy, especially since the use of broad-spectrum antibiotics in the open environment is less than desirable due to implications for dissemination of antibiotic resistant bacteria and associated genes that have implications for public health (29–31).

Due to logistical and economic challenges associated with controlling

Liberibacter spp., the potential for implementing biological control has also been gaining attention (32). Plant microbiota are known to support host health by providing protection against pathogens through a variety of mechanisms (33, 34). Several recent studies suggested that managing the taxonomic and/or functional diversity of microbiota may have positive effects on crop yields at the field scale (35, 36). Interestingly, clone library

21

sequencing and PhyloChip surveys have shown that the citrus leaf microbiota re- structures during HLB symptom development, and it has been suggested that the population dynamics of native bacteria may affect titers of L. asiaticus (15, 20, 37, 38).

At this point, however, applications of plant-beneficial bacteria have only been suggested to contribute to reducing, but not stopping, L. asiaticus infection and associated HLB symptom development in citrus trees at early stages of disease (32).

Further investigation of possible interactions between native citrus-associated bacteria that may mediate the potential for microbiome-associated Liberibacter spp. suppression is needed to better understand the role of the microbiota in HLB progression and, ultimately, for formulating strategies to manipulate microbiota to control disease.

This literature review attempts to summarize the studies that have evaluated the impacts of various treatments (e.g., antimicrobials) and other control measures (e.g., plant defense inducers, inorganic fertilizers) that have been used against Liberibacter spp. The potential for microbiome-associated biological control and the ongoing challenges for mitigating HLB at the field-scale (e.g., resiliency of Liberibacter spp. populations, mode of therapeutic delivery, antimicrobial resistance) are also discussed.

Evaluating Treatment Effects: Phytopathogen Detection

Despite recent advancements made in comparative genomics of Liberibacter spp., the HLB pathogens cannot be cultured for laboratory study (39). With regard to monitoring their populations and evaluating for HLB infection, several nucleic-acid- based tests are available for detection and quantification of the (viable) pathogen in plant tissue samples (e.g., leaf, root). Polymerase chain reaction (PCR) is traditionally used to detect the DNA of any one of these phytopathogens in a sample, indicating presence, and quantitative PCR (qPCR) is often used to measure the number of copies

22

of a specific DNA sequence of the phytopathogen corresponding to its titer (1, 40). In addition, the number of cells of Liberibacter spp. per gram of plant tissue sample can be estimated following qPCR analyses, as described in the work of Zhang et al. (19). From these values, or based on the Ct values obtained from qPCR, the relative change in the contents of the phytopathogen in response to treatment can be quantified. That is, for studies that only reported Ct values (and not population densities) of Liberibacter spp. when assessing the impacts of control measures on HLB (20, 21), the relative change in phytopathogen titer in response to treatment can still be determined with the 2−∆∆Ct method (41). Moreover, reverse-transcriptase qPCR (RT-qPCR), which can quantify the contents of a gene that has been transcribed (i.e., reflecting transcriptional activity indicative of the cell’s viability), is also used for the detection and/or quantification of the

HLB pathogens, as described in works that have focused on the viability of Liberibacter spp. in response to chemical treatments (28). In addition, the relative abundance percentage of Liberibacter spp. within microbial communities (i.e., the proportion of the amount of Liberibacter spp. to that of all bacteria in the sample) can be obtained from

DNA microarrays (e.g., PhyloChip) (20), 16S rRNA gene clone library sequencing (37), and high-throughput 16S rRNA gene amplicon sequencing (e.g., Illumina). Overall, the various studies that have reported on the effects of control measures on Liberibacter spp. have used a variety of detection methods, and the interpretation of results is dependent on the respective method used. Throughout this chapter, I present data from these studies in the context of relative changes in Liberibacter spp. in response to the treatment, as computed from changes in population densities that have been reported

23

for the phytopathogen, when available, or from raw data for Ct values using the Livak and Schmittgen (41) method.

Effects of Chemical Treatments

Antimicrobials have played an important role in plant agriculture for controlling a variety of bacterial phytopathogens. The most common examples are chemical therapeutics with streptomycin or oxytetracycline as active ingredients, which have been marketed and used for over 50 years in the USA to mitigate Erwinia amylovora, the pathogen responsible for fire blight of apple, pear and related ornamental trees (31). In fact, it has been suggested that without incorporating antimicrobial treatments into certain apple and pear production operations to prevent and/or combat fire blight, several popular cultivars and even entire orchards would have been abandoned due to excessive losses (42). These compounds, as well as gentamicin and oxalic acid, are also approved for use in several countries in Europe, the Middle East, Central America, and Mexico in order to control bacterial diseases of fruit and vegetable crops caused by members of Erwinia, Pectobacterium, Pseudomonas, Ralstonia, and Xanthomonas genera (31, 42). Commercial usage of these treatments varies by location and is subject to extensive regulation based on potential efficacy and associated risk (e.g., fate of the chemicals in the environment, selective pressures associated with evolution and spread of of antibiotic-resistant pathogens). In 2016, streptomycin sulfate, oxytetracycline hydrochloride, and oxytetracycline calcium complex were approved for use as foliar sprays to treat HLB in Florida, USA; however, the potential benefits of these compounds are still unclear (24). While judicious use of narrow-spectrum antimicrobials (e.g., small molecules) may be desirable for broadening the toolset of pro-active solutions for managing Liberibacter spp., the use of antibiotics that are closely related to those used

24

in human and veterinary medicine (e.g., streptomycin, oxytetracycline, and other broad- spectrum compounds) should be closely evaluated for the balance of potential benefits and also risks to public health (29, 30).

Broad-range Antibiotics

When HLB was hypothesized to be of bacterial origin in the 1970s, antibiotic trunk injections, using mainly tetracycline-based compounds, were tested throughout

Asia and Africa for their impacts on disease progression and severity (43). However, due to limited success, along with phytotoxic effects and labor costs for annual applications of these compounds in order to limit reinfection (i.e., tetracyclines are bacteriostatic not bactericidal), these treatments were not considered feasible at the time and were abandoned (20). However, due to the substantial impacts that HLB has had on leading citrus production industries during the past 10-15 years, the development of antimicrobial-based strategies to mitigate Liberibacter spp. is, once again, gaining interest. Several recent greenhouse and field studies have documented the responses of the titer of L. asiaticus within leaves of HLB-diseased trees to antibiotic foliar sprays, root drenches, and/or trunk injections (18–22). Within these studies, broad-spectrum compounds within antibiotic classes that have varying activities, such as aminoglycosides (streptomycin, kasugamycin), tetracyclines (oxytetracycline), β- lactams (penicillin), and sulfonamides (sulfadimethoxine, sulfathiazole), were utilized

(Table 1-1).

Penicillin G trunk injections (1.0 and 6.0 g per tree) were recently tested on 2- year-old Ray Ruby grapefruit seedlings in the greenhouse and 7-year-old Ray Ruby grapefruit trees in the field (21). Within 24 hours, the chemical became distributed throughout the canopy and roots of the seedlings, yet only throughout the canopy, but

25

not roots, of the trees (21). This suggests possible limitations for thorough chemical transport through the vascular system of citrus trees in the field. According to qPCR analyses from the field study, at 3 months following the trunk injections, the mature leaves from trees receiving the low and high rates of penicillin G had, on average, a 6- fold and 12-fold decreases in L. asiaticus titer, respectively, compared to those of the untreated control trees. Alternatively, the phytopathogen was more than 15-fold lower in the new flushes of trees receiving the higher dose of treatment than within those of control trees (p<0.05) (21). Furthermore, the authors reported the chemical applications to have little or no side effects on native bacterial populations or on the proliferation of penicillin-resistant bacteria, as measured by a crude assessment of the bacteria that could grow on generic growth media. Thus, although penicillin G slowed the progression of HLB, long-term effects related to Liberibacter spp. resilience in older leaves and issues with chemical transport to roots, as well as long-term side effects remain largely unknown. Even though no effects on promoting penicillin resistance within plant- associated microbiota were detected in this study (21), the world-wide concern associated with the spread of virulent β-lactam resistant bacteria (31) raises questions about the feasibility of commercial applications of this β-lactam at the field-scale.

The effects of oxytetracycline hydrochloride trunk injections on 5-year-old HLB- infected Hamlin trees (2 g per tree) were also recently documented (22). The titers of L. asiaticus (as measured by qPCR) within both leaves and roots were reported to have dropped precipitously during the month following treatment, varying between 1-3 log unit decreases from the starting density of more than 3 x 106 cells per gram tissue, depending on the amount of trunk injection ports used (22). The population density of L.

26

asiaticus remained significantly lower in treated trees than untreated controls over the course of the 9-month study. However, during months 2-9, there was about 1-log of phytopathogen regrowth, perhaps a reflection of the bacteriostatic activity of the antibiotic. In addition, although improvements in HLB symptoms in response to treatment were described (i.e., no HLB symptoms were present on new growth, so the overall canopy appeared healthier), there were also moderate side effects (i.e., brown discoloration to “leaf burn”) on young leaves (22). The concentration-dependent phytotoxicity of oxytetracycline has been reported in several other works (19, 44, 45).

Therefore, although oxytetracycline was able to suppress L. asiaticus and help alleviate the progression of HLB symptoms, the longevity of these impacts remains unclear because of the phytopathogen regrowth, and an optimal application rate must be used in order to limit phytotoxicity.

Treating HLB with multiple antibiotics at the same time has also been attempted.

Using combination therapy was suggested in a greenhouse study that investigated the effects of drenching the roots of HLB-infected Ray Ruby grapefruit seedlings with penicillin G (1.0 g L-1), streptomycin (0.1 g L-1), or both of the antibiotics these rates

(19). At 3 months after treatment, the population density of L. asiaticus (as estimated based on qPCR) had decreased by about 2-4 log units, depending on treatment, from the starting value around 5 x 105 cells per gram of leaf tissue, while that of the untreated controls remained relatively stagnant. However, by 6 months, the reduced level of L. asiaticus was only maintained in the leaves of seedlings that had received the combination treatment, while, alternatively, the seedlings that were treated with only one compound had leaves with the number of L. asiaticus cells that was comparable to that

27

at pre-treatment (19). Thus, synergistic effects of two or more compounds may allow for prolonged antimicrobial activity against the phytopathogen.

A field study on the effects of penicillin and streptomycin via trunk injections on 7- year-old HLB-infected Lee mandarin x Orlando tangelo trees (5 g penicillin G + 0.5 g streptomycin per tree; 10 g penicillin G + 1 g streptomycin per tree) was also described in Zhang et al. (19). The lower rate did not demonstrate any phytotoxicity, and compared to untreated control trees, those receiving this rate experienced a 3.5-5 log unit reduction in L. asiaticus within leaves, as measured by qPCR (19). However, there were still trends for up to 2 log units of increase in L. asiaticus titer in treated trees between 4 and 14 months following treatment, which suggested that continued chemotherapy with these compounds would be necessary in order to manage HLB over the long-term. Moreover, Zhang et al. (20) built on that field work and attempted trunk injections of penicillin G and streptomycin (5 g penicillin G + 0.5 g streptomycin per tree), as well as trunk injections of oxytetracycline and kasugamycin (2 g oxytetracycline hydrochloride + 1.0 g kasugamycin per tree) on other HLB-diseased trees at the same site. The titer and relative abundance of L. asiaticus in leaves were monitored bimonthly for up to 14 months with qPCR and PhyloChip analyses, respectively. Although L. asiaticus was suppressed by combination therapies, being anywhere from 3- to 30-fold lower in leaves of trees receiving treatment than in untreated controls, the phytopathogen demonstrated seasonality and regrowth over time (20). For example, even though the titer was lower in leaves of trees receiving kasugamycin and oxytetracycline than it was in untreated controls during the fall months of the study, all trees experienced increases in titer by about 10-fold between spring and fall months

28

(20). Moreover, both types of combination treatments induced a change in phyllosphere bacterial populations, as the numbers of bacterial OTUs detected in the leaves of treated trees comprised only 79.7% of all OTUs detected in the study (20). Thus, applications of penicillin with streptomycin and applications of oxytetracycline with kasugamycin were able to suppress the HLB pathogen in a season-dependent manner, which coincided with subtle adverse effects on native microbiota.

In addition to β-lactams (e.g., penicillin G), tetracyclines (e.g., oxytetracycline), and aminoglycosides (e.g., streptomycin, kasugamycin), there are a few other broad- spectrum antibiotics that have been used to treat HLB. Yang et al. (18) reported root drenches of sulfonamides (i.e., 1.0 g sulfathiazole sodium, or 1.0 g sulfadimethoxine sodium) to have moderate effects on the relative abundance, but not the titer, of L. asiaticus in leaves of HLB-infected Ray Ruby grapefruit seedlings in a recent greenhouse study. Specifically, the average relative abundance of the phytopathogen

(as measured by PhyloChip) was about 7-9% lower, yet the average titer (as measured by qPCR) was 1.5-2 times higher, in leaves of treated seedlings than in those of controls (18). This may be explained by considering an increase in the numbers of all bacteria within the community, along with the detected increases in numbers of L. asiaticus. That is, if the cumulative number of bacteria within leaf communities increased, which could have occurred in response to the antibiotics disrupting a key taxon or taxa that may have kept others in check, then the content of the phytopathogen among the broader community could have still slightly decreased even though the numbers of L. asiaticus nearly doubled. Accordingly, there was up to a 35% decline in total bacterial OTUs within leaves following treatments; several bacterial populations

29

other than L. asiaticus, possibly those associated with the fitness of the phytopathogen, may have been selected for during treatment (18). Furthermore, the authors noted enhanced effects of both sulfonamides against L. asiaticus when they were used in combination with thermotherapy (i.e., when the seedlings were placed in a growth chamber at 40oC or 45oC for 1 week prior to the root drench treatment). The lack of impacts of sulfonamides on the HLB pathogen at room temperature (i.e., thermotherapy less logistically feasible in the field), along with subsequent adverse effects on native microbiota, make them relatively ineffective and not a feasible option for controlling L. asiaticus.

Some of the aforementioned treatment studies were performed following screening tests for a variety of broad-spectrum compounds that were hypothesized to demonstrate activity against Liberibacter spp. (19, 38, 40, 44, 45). These screens were done by soaking either cuttings or budsticks/scions from HLB-infected plants in a solution containing antimicrobials and then rooting and planting the cuttings or grafting the budsticks/scions onto a healthy stock and monitoring the outcomes. For example, using the grafting approach, Zhang et al. (45) screened 31 compounds from a variety of antibiotic classes (i.e., aminoglycosides, ansamycins, β-lactams, cephalosporins, glycopeptides, lincosamides, oxazolidinones, polypeptides, quinolones, sulfonamides, and tetracyclines) for effectiveness and phytotoxicity. From the diverse set of compounds tested, ampicillin, carbenicillin, penicillin, cefalexin, rifampicin and sulfadimethoxine were reported to be most effective in eliminating or, at least, suppressing HLB, based on L. asiaticus titers measured (by qPCR) in leaves both the treated scion and the inoculated rootstock (45). This method was also implemented to

30

screen for potential impacts of chemical treatments on citrus-associated microbiota.

Zhang et al. (38) soaked budsticks taken from HLB-diseased lemon trees in ampicillin- or gentamycin-containing solutions and graft-inoculated healthy grapefruit seedlings.

The leaf microbiota were then monitored over six months with PhyloChip analyses.

Although the ampicillin treatment eliminated L. asiaticus, it also induced a 15% reduction in community richness, compared to the diseased control (38). Alternatively, while the gentamycin treatment did not impact L. asiaticus, it, interestingly, caused the communities to become less stable as a plethora of low abundance OTUs appeared

(38). Disruption of keystone species in response to antimicrobial solution treatment may have caused that spike in diversity. In summary, the screening methods have also provided insight to the efficacy and potential side effects of several antibiotics that could be considered for use in treating HLB.

Overall, although the usage of antibiotics to control Liberibacter spp. has had promising developments, challenges and consequences associated with using broad- range compounds include seasonality of the pathogen and re-establishment over time, minimal inactivation of the pathogen in older flushes, phytotoxicity, and adverse impacts on richness and diversity of native microbiota (19–22). Optimization of field treatments with compounds that may suppress Liberibacter spp. or combinations of such antibiotics warrants further investigation.

Antimicrobials that Specifically Target Liberibacter spp.

Developing novel therapeutics that may target Liberibacter spp. is challenging due to difficulties associated with identifying molecular targets, since the phytopathogens cannot be cultured. Efforts have still been made to search for compounds with activity against L. asiaticus based on the information from the

31

pathogen’s sequenced genome (14) and, also, by using close phylogenetic relatives as model organisms in culture-based experiments (17, 27, 28).

Several small molecule compounds that may target molecular pathways essential to the survival of L. asiaticus have been identified (17, 27, 28). For example,

Akula et al. (27) identified twenty small molecule compounds that could inhibit SecA activity of the HLB pathogen via molecular docking in silico and then confirmed the antimicrobial activity of five of these compounds in vitro using Agrobacterium tumefaciens as a culturable model. In addition, Pagliai et al. (28) characterized a regulon of L. asiaticus involved in cell wall remodeling, which contains a transcription factor (ldtR) and a transpeptidase (ldtP), that is essential to its survival. Small molecules that bind and inactivate LdtR were identified and their activity was validated with thermal shift assays and DNA binding assays (28). Several of these compounds (i.e., benzbromarone, hexestrol, phloretin) were further reported to down-regulate the expression of ldtR and ldtP in vitro in culturable α-proteobacteria that are closely related to L. asiaticus (i.e., Liberibacter crescens and Sinorhizobium meliloti) and, ultimately, inhibit the viability of L. asiaticus (measured with RT-qPCR) in citrus leaves in leaf- soaking assays, suggesting their applicability for future HLB treatment studies (28).

Moreover, another L. asiaticus transcription factor (PrbP), which is putatively involved in pathogenesis, persistence, cell viability, and environmental stress resistance, was recently characterized, and compounds that may interact with this protein were identified in vitro using L. crescens and a model γ-proteobacterium Escherichia coli (17).

One of these compounds (i.e., tolfenamic acid) was also shown to inhibit the viability of

L. asiaticus (measured with RT-qPCR) in citrus leaves in leaf-soaking assays (17).

32

Tolfenamic acid was further tested for activity against L. asiaticus in HLB-infected

Valencia seedlings in a long-term greenhouse experiment (10 µM tolfenamic acid applied by foliar spray and root soaking) (Table 1-1). Based on RT-qPCR analyses, at

11 months following treatment, there had been approximately an 80-95% reduction in the expression of L. asiaticus genes rplJ and gyrA in 75% of the treated seedlings, suggesting that the compound can lead to substantial inactivation of the metabolically active phytopathogen (17). Importantly, the authors noted that the plants responded to the treatment with substantial improvements in fibrous root growth/development and enhanced foliage appearance. Thus, the target specificity of tolfenamic acid, as well as that of other small molecules that have been suggested to have activity against

Liberibacter spp. (27, 28), would be an interesting avenue of research for future field studies. Compared to broad-spectrum antibiotics, small molecule compounds may be able to more effectively control HLB and minimize adverse ecological impacts (i.e., phytotoxicity, disruption of native microbiota).

Additional Measures to Control Liberibacter spp.

Managing L. asiaticus with approaches other than antimicrobial treatment has also been the focus of several studies (Table 1-2). For example, thermotherapy appears to be somewhat effective at suppressing phytopathogen titer, at least in greenhouse/growth chamber settings (15, 18, 46, 47). Hoffman et al. (47) demonstrated that exposing HLB-infected Ray Ruby grapefruit seedlings to temperatures of 40oC for

24 h or to 42oC for 19 h, followed by 30oC for 5 h, in growth chambers for only two consecutive days resulted in the titer of L. asiaticus dropping by more than 40-fold to a level below the detection limit of qPCR within two months. Importantly, although it was noted that a 45oC exposure caused severe plant tissue damage, any heat stresses

33

imposed by the 40oC or 42oC treatment were overcome by the plants (47). Alternatively,

Zhang et al. (15) reported thermotherapy of HLB-infected Ray Ruby seedlings via exposure to 40oC or 42oC for 8h a day for 1 week to be ineffective at lessening the titer of L. asiaticus in leaves, while exposure to 45oC resulted in greater than an 5-fold decreases in phytopathogen titer by the 2-month time point. In another study, in which

HLB-infected tangerine seedlings were exposed to 45oC or 48oC once a week for three consecutive weeks, the titer of L. asiaticus within leaves declined by about 30% and

55%, respectively, while that of the controls increased by over 300% (46). Moreover,

Yang et al. (18) performed thermotherapy on HLB-infected Ray Ruby seedlings in growth chambers that were set for a 12 h exposure of 40oC or 45oC per day for one week, and the higher temperature treatment resulted in a decrease in titer of L. asiaticus within leaves by more than 3 log units to a level that was below the qPCR detection limit. The pathogen titers remained below detection throughout ten months of monitoring post-treatment (18). Although thermotherapy is an extensively time-consuming and expensive process in the field, there have been efforts to develop commercial equipment that could make it more feasible at the larger scale (48).

Other non-antimicrobial HLB treatments include applications of compounds that may boost plant growth and stimulate plant defenses. Canales et al. (11) reported bimonthly foliar sprays of a brassinosteroid (i.e., 0.084 µM epibrassinolide) over the course of one year to Valencia orange trees in the field to induce the expression of several key plant defense genes and lower L. asiaticus titers in leaves by approximately

7-fold at the end of that year. The impacts of the routine applications of the plant steroid on seedlings in the greenhouse were even greater, as L. asiaticus titer was reported to

34

have reduced from about 106 cells g-1 to 104 cells g-1 leaf tissue in as soon as three months (11). In addition, spray applications of several plant defense regulators (i.e.,

BABA, BTH, INA, AA, 2-DDG)) were reported to somewhat slow down the proliferation of L. asiaticus in leaves of 7-year-old HLB-infected Navel orange trees (12). While the numbers of L. asiaticus cells per gram of leaf tissue from treated trees were always within 1 log unit from that of untreated controls, these values were still sometimes significantly different (p<0.05) at time points of 1 year or beyond. In other words, the phytopathogen titer in all trees increased over time, but the rates of increase were sometimes slightly lower in treated trees than controls. The treatments were also described to suppress HLB symptom severity by up to 30% and have positive impacts on fruit yield and quality (12). Moreover, spray applications of several other plant- beneficial compounds, including L-arginine and 6-benzyl-adenine combined with gibberellins, were tested on HLB-infected Valencia trees in greenhouse and field studies

(13). RT-qPCR analyses of plant mRNA indicated that the treatments had positive impacts on the abundances of citrus gene transcripts involved in biotic stress responses, starch metabolism, and systemic acquired resistance. Alternatively, the treatments did not have inhibitory effects on L. asiaticus or HLB symptom progression

(13). The authors suggested that while certain plant-beneficial treatments can modulate the expression of key citrus defense genes, they might only be able to have mitigating effects against HLB if they are applied frequently before or immediately at the onset of visible HLB symptoms, which was not the case in their study (i.e., HLB in the trees was already at a more progressed stage) (13). In summary, it appears that plant-beneficial compounds can somewhat help with slowing HLB progression by having positive effects

35

on citrus defenses and growth; however, these forms of treatment may not have strong deleterious impacts on Liberibacter spp.

Furthermore, since micronutrient deficiencies, especially zinc deficiency, are often associated with HLB symptoms, it can be speculated that applications of zinc- based compounds may help alleviate disease symptoms. However, a recent greenhouse study demonstrated that treating HLB-infected Ray Ruby grapefruit seedlings with ZnSO4 or Zineb (i.e., a zinc salt-based pesticide) actually induced an increase in titer of L. asiaticus within leaves within 4 months; this increase was 1.5x greater than the increase experienced in untreated controls during the same time period

(15). Thus, receiving zinc-based compounds was harmful to the plants as it expedited proliferation of the HLB pathogen. In a separate field study that attempted treating HLB- diseased citrus trees with zinc, spray applications of zinc-metalosate combined with phosphite, as well as other micronutrient-based sprays or trunk injections, were reported to have no significant effects on L. asiaticus titer, fruit yield, or juice quality (6).

Collectively, these studies suggest that micronutrient amendments cannot mitigate

Liberibacter spp., at least when used alone. Although enhanced nutrient programs may promote citrus tree growth in asymptomatic trees or in those in groves with poor soil fertility at early stages of HLB infection, they lack significant effects on HLB-diseased trees that have developed advanced stages of infection (24).

Overall, with regard to non-antimicrobial approaches to control HLB, thermotherapy appears to be the most effective against Liberibacter spp., though this has only been confirmed in small-scale settings. Applications of plant growth or defense regulator compounds appear to be beneficial for supporting overall tree health and

36

might be able to slow disease progression; however, they do not lessen the existing pathogen titer. Micronutrient amendments are either inconsequential or harmful to the plant. Utilizing some these holistic approaches for phytopathogen control in combination with antimicrobial treatment would be an interesting avenue for future research.

Plant Microbiota Implications for Citrus Health

Plant microbiota play critical roles in plant development by providing support for disease control and stress tolerance, among other mechanisms (33, 34). A growing number of studies have focused on understanding how plant-associated microbial communities change in richness and diversity during disease development (49–54), including that of HLB (37, 55, 56). Harnessing the beneficial potential of native microbiota may be one of the few logistically and economically feasible solutions for controlling certain phytopathogens that are otherwise difficult to manage, such as

Liberibacter spp.

Implementing biological control, either alone or in combination with other disease management strategies, may be feasible for treating emergent woody plant diseases that have proven challenging to control otherwise (e.g., HLB) (57). Perhaps Liberibacter spp. may be treatable in ways similar to how fungi, phytopathogen-specific viruses, and beneficial bacteria and secondary metabolites have been used to treat Fusarium oxysporum (i.e., causative agent of bayoud disease) in date palm, Heterobasidion annosum (i.e., causative agent of root rot) in conifers, and multiple pathogens of grapevine (58–61). Moreover, endophytic microbes that possess biocontrol capabilities offer an especially interesting potential for the development of novel agricultural biotechnologies (57). For example, inoculating stems of citrus variegated chlorosis

(CVC)-diseased Catharanthus roseus with Curtobacterium flaccumfaciens, which was a

37

biocontrol agent originally isolated from asymptomatic citrus, reduced CVC symptoms over time (62). This effect was mediated by interactions between C. flaccumfaciens and the xylem-limited CVC pathogen (Xylella fastidiosa) (62). Similarly, Lacava et al. (63) suggested that interactions between X. fastidiosa and the biocontrol bacterium

Methylobacterium mesophilicum may also affect CVC progression in citrus. Strategies to manage the phloem-limited HLB pathogen with beneficial microbes are an interesting avenue for future research. Preliminary data shows that applications of certain beneficial microbes may limit, though not prevent, L. asiaticus infection when applied on

HLB-symptomatic trees that are still at an early stage of disease symptom development

(32).

Aside from applications of plant beneficial microbiota, using organic amendments to promote the growth and diversity of resident microbiota are another form of biological control. Although research on organic amendment effects on Liberibacter spp. is lacking, there are studies that suggest adverse impacts on the insect vector. For example, Rao et al. (64) reported applications of vermicompost and several organic manures (i.e., farmyard manure, poultry manure, green manure) to induce significant decreases in populations of the Diaphorinia citri (i.e., Asian citrus psyllid; vector for L. asiaticus and L. americanus) among Nangpur mandarin trees, compared to controls that received inorganic fertilizers. Although the mechanism for such activity was unclear, the authors suggested that D. citri may have been sensitive to organic amendment- associated increases in phenol contents and enzyme (i.e., polyphenol oxidase, peroxidase) activities of the trees (64). Nonetheless, organic amendments may have indirect effects on Liberibacter spp. in citrus groves by limiting vector transmission. In a

38

separate study, it was reported that the diversity of microbiota in soils of citrus groves increased when using a fertilizer regime that incorporated a mix of composted manure with lower rates of the inorganic fertilizers that were traditionally used at the site, compared to using only the full rate of inorganic fertilizers (65). The microbial diversity of soils may be important for combatting HLB due to implications that native microbiota may have for citrus health. After all, soil microbial diversity is positively associated with crop production (35, 36).

Ongoing Challenges Associated with Field-scale Treatment

Despite ongoing efforts made by growers to modify components of citrus production operations (e.g., fertilizer regime, irrigation schedule, usage of pesticides to manage Liberibacter spp. and D. citri) to control HLB, the phytopathogen has spread to infect the vast majority of citrus groves. While chemical compounds with activity against

Liberibacter spp. are becoming recognized (17, 20, 22, 28), there are a number of factors that continue to make treatment at the field-scale relatively challenging.

Upon transmission to citrus trees, Liberibacter spp. proliferates within and translocates throughout phloem tissue, reaching high, unevenly distributed numbers in leaves and roots (e.g., more than 106 cells per gram leaf tissue) (22, 66). Thus, in order for antimicrobials to be effective against HLB, they must come in contact with

Liberibacter spp. by entering the phloem and becoming widely dispersed throughout the vascular system. Although foliar sprays may be the most logical way to deliver treatment at the field-scale, this mode of delivery may not allow for chemical absorbance into phloem due to physical barriers of waxy leaf surfaces and the robustness of plant cell walls. Alternatively, although applying treatment via trunk injection may provide direct transfer of antimicrobials into the plant vascular system, this

39

form of delivery is highly labor intensive in the field. Even with trunk injections, there are still limitations with chemical transport throughout trees to reach the roots (21). In addition, while relatively low doses of antimicrobials may be ineffective at suppressing pathogens, higher doses of at least some of the tested antimicrobials are known to have phytotoxic effects (20). Thus, with inherent challenges for chemical dispersal within trees, the added issue of phytotoxicity when using broad-range antibiotics makes phytopathogen control even more difficult.

There is a latency period between infection and early symptom expression (e.g., blotchy mottle on leaves), during which time infected trees appear asymptomatic (16).

While HLB treatments are expected to have highest efficacy if administered at the onset of infection (13), this timing is not easily anticipated. By the time HLB is recognized and treatments are implemented, Liberibacter spp. populations may already be well established and capable of, at least partial, withstanding of antimicrobial treatment effects. For example, when chemical treatments suppress Liberibacter spp. within diseased trees, populations of residual Liberibacter spp. that survive the treatment have been reported to regrow over time (20, 22). Thus, the usage of antibiotics may need to continue over time in order to control HLB over the long-term, which introduces problems associated with operation costs and potentially adverse environmental effects.

Although it remains largely unknown, there may also be issues associated with resilient

Liberibacter spp. populations and native microbiota becoming resistant to antibiotics that are used, which is concerning with regard to the dissemination of antibiotic resistant bacteria and associated genes through the environment.

40

Treatments with broad-spectrum antibiotics can have adverse effects on native microbiota (18, 20, 38), which are known to play an important role in supporting tree health. Under the right circumstances, the plant microbiota can bolster plant productivity by providing protection against phytopathogens, improving tolerance of environmental stresses, priming immune response signaling pathways, and assisting with acquisition of nutrients from soil, among other mechanisms (33, 34). Finding a treatment with minimal effects on native microbiota, or even developing strategies for biological control

(32), may be needed for improving field-scale HLB management.

Although antimicrobials can suppress Liberibacter spp. within infected trees, high levels of pathogen inoculum are still widespread in areas where HLB has become endemic. For example, in Florida, over 130,000 acres of land previously used for commercial citrus production have become unmanaged or abandoned due to logical issues associated with maintaining groves infected with HLB (67). These areas remain reservoirs for L. asiaticus and D. citri. Therefore, even if HLB is able to be managed within a specific field site, there remains the possibility for HLB re-introduction via vector migration. Controlling populations of D. citri (i.e., the psyllid vector of L. asiaticus and L. americanus) and Trioza erytreae (i.e., the psyllid vector of L. africanus) are also critical for field-scale control of Liberibacter spp. For an extensive review on HLB vector management, see Grafton-Cardwell et al. (10).

Overall, the need for feasible measures to control Liberibacter spp. remains an urgent issue. Future studies that focus on developing integrated approaches to use antimicrobial applications in combination with other forms of disease control (e.g., plant

41

beneficial compounds, biological control) may be essential for improving efforts to mitigate HLB.

Aims and Hypotheses

Uncovering the dynamic interactions between phytopathogens and native plant microbiota is needed for developing effective and environmentally safe strategies to control certain plant diseases (68). This is especially relevant to HLB disease of citrus, as there is still a lack of logistically feasible options for managing Liberibacter spp. and its devastating consequences (4, 10). This work tested the overall hypothesis that the microbiome plays a fundamental role in supporting citrus tree health through interactions and associations between the HLB pathogen and native microbiota of citrus trees. A series of experiments was conducted to investigate key microbial interactions, as well as the important changes in microbiota taxonomic and functional diversity, that may occur during both HLB symptom progression and in response to treatment with novel antimicrobial compounds. The aims and hypotheses for these studies are described below.

Illumina sequencing of 16S rRNA genes was used throughout this work to characterize the microbial communities. A pilot study was conducted to optimize this protocol, since our initial sequencing data indicated substantial contamination in the citrus leaf sample amplicon libraries by eukaryotic gene sequences (i.e., those of plastids and mitochondria) (Chapter 2). Peptide nucleic acid (PNA) clamps were recently developed to block the amplification of plastid and mitochondrial 16S rRNA gene sequences in PCR reactions (69). Thus, I tested whether they could be incorporated into the amplicon library preparations to enrich for the amplification of bacterial/archaeal 16S rRNA gene sequences. Due to PNA clamp sequence specificity

42

and based on the previous success reported for using these blockers to enhance

Arabidopsis root microbiota characterizations (69), I hypothesized that they would allow for a broader characterization of citrus-associated microbial communities without creating bias in the dominant taxa that would be detected and a variety of concentrations was tested. In addition, our initial sequencing data indicated issues with quality control that were likely related to the sequencing settings that were used (i.e., as described in Chapter 2, when using 2x300 bp settings there were substantial decreases in the phred scores of raw sequenced reads after about 150-200 bp). I investigated if modifying the sequencing protocol to use shorter read lengths (i.e., 2x150 bp) would improve the phred scores of sequenced reads and, in turn, increase the total number of sequenced reads that would pass quality control. The hypotheses of this preliminary study were confirmed, and the protocols for citrus sample processing and Illumina sequencing were improved and used throughout the remainder of this work (Chapter 2).

A molecular survey was conducted throughout Florida to define the core citrus microbiota across a number of factors, including HLB symptom severity, location, cultivar, and season/time, and elucidate interactions between native microbiota and

Liberibacter spp. (Chapter 3). It has become well established that plant microbiota can bolster plant productivity by providing protection against pathogens, among other mechanisms, and several studies have suggested that the population dynamics of native bacteria may affect L. asiaticus titers within HLB-diseased citrus (15, 20, 33, 34,

37, 38). Thus, I hypothesized that members of the citrus microbiota that are differentially abundant at different stages of HLB progression, especially those within the microbiota interaction network that may co-occur or be mutually exclusive with the phytopathogen,

43

may have important implications for host health and biological control. While some members of the microbiota may fluctuate in a response to host physiological changes, certain taxa may, alternatively, have the ability to directly impact (e.g., resource competition or antibiosis with Liberibacter spp.) or indirectly impact (e.g., host health promotion via assisting with nutrient acquisition, priming plant defenses) HLB outcomes.

For example, since niche similarities can impact co-existence (70), one can speculate that there may be resource-related competition between Liberibacter spp. and other members of its phylogenetic class, Alphaproteobacteria (e.g., Methylobacterium,

Sphingomonas). Moreover, L. asiaticus has been reported to be commonly present in asymptomatic trees in regions where the HLB is endemic, though at relatively low titers, suggesting that a specific threshold in titer may be critical for disease establishment

(37). Considering that the relative abundance of the phytopathogen among native bacteria is a function of its titer and that any member of a microbial community co-varies with overall microbiota diversity (i.e., increases/decreases in an organism are inversely proportional to the cumulative sum of abundances of all other organisms), I hypothesized that citrus microbiota diversity is significantly associated with HLB disease progression. Confirming this would have additional implications for biological control in that the efficacy of potential treatments for HLB may depend on promoting microbiota diversity or, at the very least, not disrupting it when targeting the phytopathogen. Aside from having applications for HLB control, this study also aimed to investigate broader microbiota-phytopathogen-host relationships. For example, it has been theorized that commonly occurring organisms across similar microbiomes comprise a core microbial community that plays crucial roles in ecosystem functioning within that type of microbial

44

habitat (36, 71). Given the limited number of studies that have discerned the core members of plant-associated microbial communities (51, 72, 73), much remains unknown about the structure of the core community among all microbiota and its importance with regard to plant health. Our plant host/disease model allowed us to investigate associated questions, such as, what are the proportion of numbers and abundances of taxa that are core among the microbiota, how does this differ across a large spatiotemporal scale, are the host-health associated members of the microbiota predominately core members or site-specific members, how does core microbiota structure vary when assessing at different taxonomic levels, what abiotic/biotic factors are most associated with core microbiota structure (e.g., location, cultivar, disease, time)? Overall, this study would provide characterizations of the network of interactions between native bacteria and phytopathogens, such as Liberibacter spp., that are needed for advancing the understanding of the roles of (core) microbiota in plant disease progression.

Finding and utilizing novel compounds that are target-specific to Liberibacter spp. is essential for improving HLB treatment efficacy, especially since the use of broad- spectrum antibiotics in the open environment is less than desirable due to implications for dissemination of antibiotic resistant bacteria and associated genes that have implications for public health (29–31). In an 18-month field study, two novel antimicrobials were tested as treatments for HLB, and the impacts on the citrus microbiota were evaluated (Chapter 4). Since the ligand binding areas of these small molecule compounds in Liberibacter spp. are not identical to those in other bacteria, I hypothesized that the treatments would be target-specific and suppress the

45

phytopathogen without having adverse effects on the diversity of native microbiota.

Although it is possible that the small molecules may interact with close phylogenetic relatives of Liberibacter spp. (e.g., members of Rhizobiaceae, such as S. meliloti (28)), the activity of the compounds may, also, inherently support their environmental safety.

That is, they are known to disrupt a regulon involved in osmotic stress tolerance; although this may be essential for Liberibacter spp. survival because it resides in citrus phloem tissue, it may not be an issue for microbiota that live in other types of citrus- associated microenvironments and, thus, may not thwart their viability. Moreover, while a growing number of studies has focused on understanding how plant microbiomes change during disease establishment and progression (37, 49–56), the critical shifts that may occur in response to phytopathogen removal, perhaps during plant disease treatment and recovery, remain largely unexplored. Using target-specific compounds to treat HLB-infected citrus provides a compelling system to address such fundamental questions and infer directional associations within the microbiome network. I hypothesized that chemically-induced removal of the phytopathogen will coincide with increases in abundances of members of the citrus microbiota that are associated with citrus health (i.e., those that would be identified in the preceding experiment). I further predicted that the taxa that positively respond to Liberibacter spp. removal will be members of the core citrus microbiota, as opposed to opportunistic site-specific taxa, since core members may more closely associate with the citrus host as a result of co- evolutionary relationships. In summary, this study would advance the understanding of whether novel antimicrobials may provide an effective and environmentally safe option

46

for HLB treatment, as well as answer important questions regarding microbiota transitions that follow the targeted removal of an invading phytopathogen.

Since microbiota functional profiles are inherently more stable and contain less noise than taxonomic profiles (i.e., genomes of different taxa may encode many similar functions), characterizing microbial communities by the former might actually provide more discriminatory power for determining the biological importance of samples with regard to host health (74). I aimed to build on the two aforementioned field studies and determine the differences in the metagenomes of microbiota of healthy and HLB- diseased citrus trees and investigate whether treatment may lead to plant-beneficial shifts in the metagenome (Chapter 5). Due to functional redundancies being common across prokaryotes (e.g., all require essential housekeeping genes), especially for closely related organisms (e.g., metabolic pathways may be similar within a phylogenetic class), I hypothesized that there would be a core functional profile conserved across metagenomes, though some key pathways would be enriched based on citrus tree health-state. I posited that asymptomatic/healthy trees, which were previously linked to having more diverse microbiota (Chapter 4), would be associated with encoding more diverse metabolic pathways. Alternatively, while the microbiota of

HLB-diseased trees may be transitioning and facing environmental stresses as symptoms progress, I hypothesized that enrichments in genes encoding mechanisms for cellular machinery repair and, perhaps, virulence (e.g., secretion system) would be characteristic of these metagenomes. I further speculated that the metagenomes of healthy/asymptomatic trees would be more similar to those of the HLB-diseased trees that received antimicrobial treatments than to the HLB-diseased controls, which may

47

suggest treatment-induced shifts towards a “healthier” functional state and link the the importance of microbiome structure and function.

Collectively, the series of experiments that comprised this work aimed to elucidate microbial interactions and host-associations, as well as outcomes of novel treatments for Liberibacter spp., that may be critical for controlling the devastating HLB disease. At a broader level, this work aimed to advance the understanding of: (i) plant microbiome selection across multiple variables and (ii) changes in community structure and function that are associated with disease establishment, symptom progression, and phytopathogen removal, which have important implications for microbiota- phytopathogen-host relationships and, ultimately, plant disease management.

48

Table 1-1. Antimicrobials that have been tested against HLB infection in studies that incorporated quantification of the phytopathogen. Antimicrobial Field/ Compound Greenhouse Impact on Liberibacter asiaticus (Detection Impact on HLB Potential Side Ref. (Antimicrobial (Application method) Symptoms Effects Class) method) Depending on the amount of injection ports used Moderate for application, the population density in leaves New flushes did not phytotoxicity – slight Oxytetracycline Field (trunk decreased by 1-3 log units within 1-month after display chlorosis, so brown discoloration (22) (Tetracycline) injection) treatment (qPCR). It remained lower in treated the overall canopy to leaf burning on trees than controls for 9 months, although appeared healthier some young flushes population re-growth occurred during this time. No phytotoxicity; Concentration-dependent reduction in titer by 6- to little or no impact on Penicillin G Field (trunk 12-fold in leaves of treated trees compared to Slight increases in native bacterial (21) (β-lactam) injection) untreated controls at the 3-month time point after canopy size populations and treatment (qPCR) penicillin resistance within populations Reduction in population density in leaves by more Penicillin G Greenhouse than 3 log units within 3 months after treatment, Not discussed No phytotoxicity (19) (β-lactam) (root drench) yet re-growth to a level close to the starting content by the 6-month time point (qPCR) Reduction in population density in leaves by more Streptomycin Greenhouse than 3 log units within 3 months after treatment, Not discussed No phytotoxicity (19) (Aminoglycoside) (root drench) yet re-growth to a level close to the starting concentration by the 6-month time point (qPCR) Slightly less chlorosis Approximately a 9% lower relative abundance Partial deleterious development in Sulfadimethoxine Greenhouse (PhyloChip), but about twice as high titer (qPCR), effects on relative canopy of treated (18) (Sulfonamide) (root drench) in leaves of treated seedlings than controls at the abundances of seedlings than in that 2-month time point after treatment native bacteria of controls Slightly less chlorosis Approximately a 7% lower relative abundance Partial deleterious development in Sulfathiazole Greenhouse (PhyloChip), but about twice as high titer (qPCR), effects on relative canopy of treated (18) (Sulfonamide) (root drench) in leaves of treated seedlings than controls at the abundances of seedlings than in that 2-month time point after treatment native bacteria of controls

49

Table 1-1. Continued. Antimicrobial Field/ Compound Greenhouse Impact on Liberibacter asiaticus (Detection Impact on HLB Potential Side Ref. (Antimicrobial (Application method) Symptoms Effects Class) method) Approximately an 80-95% reduction in the Substantial Tolfenamic Acid Greenhouse expression of L. asiaticus genes rplJ and gyrA in improvements in (N/A; small (foliar spray; root 75% of the treated seedlings, indicating fibrous root No phytotoxicity (17) molecule drench) substantial reduction in viable population (RT- development and compound) qPCR) foliage appearance

The titer in leaves of treated trees, which Combination: Fluctuating titer was correlated with relative abundance (PhyloChip), Partial deleterious Kasugamycin described to Field (trunk was anywhere from 3- to 30-fold lower than in effects on relative (Aminoglycoside) somewhat correlate (20) injection) those of controls during 14 months of monitoring; abundances of + Oxytetracycine with symptom however, seasonal fluctuations indicated re- native bacteria (Tetracycline) appearance growth (qPCR)

In the greenhouse study, there was reduction in population density by more than 2 log units within 2 months following treatment, which continued to Combination: Greenhouse slightly decrease during the 6-month monitoring Penicillin G (root drench) and period (qPCR). In the field study, the population Slight phytotoxicity (β-lactam) + Not discussed (19) Field (trunk density in leaves of treated trees was anywhere in both studies Streptomycin injection) from 3.5-5 log units lower than in those of controls (Aminoglycoside) during 14 months of monitoring; however, there were about 2 log unit increases between the 4- and 14-month time points (qPCR).

The titer in leaves of treated trees, which Combination: Fluctuating titer was correlated with relative abundance (PhyloChip), Partial deleterious Penicillin G described to Field (trunk was anywhere from 3- to 30-fold lower than in effects on relative (β-lactam) + somewhat correlate (20) injection) those of controls during 14 months of monitoring; abundances of Streptomycin with symptom however, seasonal fluctuations indicated re- native bacteria (Aminoglycoside) appearance growth (qPCR)

50

Table 1-2. Control measures other than antimicrobials that have been tested against HLB in studies that incorporated quantification of the phytopathogen. Field/ Impact on Liberibacter Treatment Treatment Details Greenhouse asiaticus (Detection Impact on HLB symptoms Side Effects Ref. (Applic. method) method) Anywhere from no change Some instances of chlorosis- to >40-fold reduction in titer Some cases like symptoms being (qPCR), depending on the of moderate (15, 18, 40∘C Greenhouse mitigated over time, though study. About a 9% leaf tissue 47) some cases of normal reduction in relative damage symptom progression abundance (PhyloChip) Some instances of chlorosis- Anywhere from no Some cases like symptoms being reduction to >40-fold of moderate (15, 42∘C Greenhouse mitigated over time, though reduction in titer (qPCR), leaf tissue 47) some cases of normal depending on the study. damage symptom progression Thermotherapy Anywhere from 5-fold to 1000-fold reduction in titer Some cases in response to treatment Chlorosis-like symptoms of severe (15, 18, 45∘C Greenhouse (qPCR). Over 80% generally mitigated over time leaf tissue 46) reduction in relative damage abundance (PhyloChip)

Reductions in titer in leaves of treated trees by about Chlorosis-like symptoms Not 48∘C Greenhouse 55%, while that in untreated (46) mitigated over time discussed controls increased by over 300% (qPCR) No reduction in titer within AA, BABA, BTH, Chemical leaves; however, the rate of Trees receiving treatment INA, 2-DDG (used Not inducers of Field (spray) increased was slowed had greater fruit yield and (12) individually or in discussed plant defenses compared to controls quality than controls combination) (qPCR) No phenotypic effects; Plant- L-arginine; 6- Greenhouse No reduction in titer within however, genes involved in Not regulating benzyl-adenine + (spray) and leaves of trees receiving plant metabolism and (13) discussed compounds gibberellins Field (spray) treatments (RT-qPCR) immune response were up- regulated

51

Table 1-2. Continued. Field/ Impact on Liberibacter Treatment Treatment Details Greenhouse asiaticus (Detection Impact on HLB symptoms Side Effects Ref. (Applic. method) method) Concentration-dependent No HLB symptoms on new Greenhouse reduction in titer within flushes; genes involved in Not Brassinosteroids epibrassinolide (spray) and leaves by about 160- and 7- (11) plant defense response were discussed Field (spray) fold in greenhouse and field up-regulated study, respectively (qPCR)

Increases in titer (qPCR) Greenhouse No effects; symptoms Not ZnSO4 and relative abundance (15) (root drench) progressed discussed (PhyloChip) in leaves

Increases in titer (qPCR) Greenhouse No effects; symptoms Not Zineb and relative abundance (15) (root drench) progressed discussed Micronutrient- (PhyloChip) in leaves based phosphite compounds combined with Mn-carbonate, Mn-, Cu-, or Zn- Field (spray; No reduction in titer in No effects; symptoms Not (6) metalosate; trunk injection) leaves (qPCR) progressed discussed soluble copper or silver combined with a polymer

52

CHAPTER 2 MATERIALS AND METHODS: PROTOCOL OPTIMIZATION FOR 16S rRNA GENE SEQUENCING OF CITRUS SAMPLES

Sample Collection

Within the series of original research experiments that comprise this dissertation, leaf and root samples were taken from citrus trees throughout Florida that had varied by location, cultivar, Huanglongbing (HLB) disease symptom expression, season, time, and antimicrobial treatment. While Chapters 3-5 provide these specific metadata, the same sampling methods for leaves and roots were applied in each study. To sample leaves, the canopy of each tree was divided into four quadrants and 10 leaves from each quadrant were randomly selected. The leaves were collected aseptically by cutting at the stem and placing into sterile Whirl-Pak® stomacher bags (Nasco, Fort Atkinson, WI).

To sample roots, 3 soil cores (5 cm diam., 25 cm depth) were taken within the root zone of each tree and the fibrous roots were screened from the soil and combined. A composite root sample for each tree was collected in a sterile 50 mL polypropylene tube. Leaf and root samples were transported to the lab in coolers on dry ice.

DNA Extraction

All samples were frozen at -80C and lyophilized. The leaf samples were crushed into <1 cm fragments in their respective stomacher bags. The root samples were crushed into <1 cm fragments by adding a sterile steel grinding ball (9.5 mm diam.) to each tube and shaking vigorously by hand. The crushed leaf and root samples, along with a sterile steel grinding ball (9.5 mm diam.), were placed in 5 mL polyethylene vials

(Spex, Metuchen, NJ) and homogenized via bead beating with three 1-min bursts in the

GenoGrinder 2000 (Spex, Metuchen, NJ). Genomic DNA for Illumina sequencing was extracted from each homogenized sample with the Isolate II Plant DNA kit (Bioline,

53

Taunton, MA). This kit was chosen after a preliminary assessment indicated it, compared to several other DNA extraction kits, provided relatively high quality DNA extracted from citrus leaf samples (i.e., 260/280 values closest to the high quality of 1.8-

2.0 (75)) (Table 2-1).

Moreover, the Isolate II extraction protocol that was utilized was slightly modified from that of the product manual (http://www.bioline.com/sg/downloads/dl/file/id/1216/ isolate_ii_plant_dna_kit_protocol.pdf) in order to tailor the DNA extraction to what may be best for the citrus samples. After testing several combinations of plant tissue and chemical reagent amounts utilized in the procedure (i.e., trying those other than the recommended 0.2 mg lyophilized tissue and 400 µL lysis buffer), it was determined that

0.15 mg sample tissue and 525 µL lysis buffer generally provided the best quality and quantity of DNA extracted from leaf samples and that 0.25 mg sample tissue and 450

µL lysis buffer generally provided that from root samples (data not shown). These slightly modified steps were used in all further leaf and root DNA extractions.

16S rRNA Gene Amplicon Preparation and Sequencing: An Initial Attempt

Our initial amplicon preparation and DNA sequencing run was performed on leaf samples collected from the 24 C. sinensis cv. Valencia trees sampled in Gainesville on

9/23/15, as described in Chapter 4 on novel HLB treatments that were tested. Amplicon libraries were prepared for the V4 region of 16S rRNA genes with PCR reactions that incorporated Illumina-compatible universal bacteria/archaeal primers 515f/806r, with a

12-bp indexing barcode added to the forward primer (76, 77). Each forward primer contained the Illumina 5’ adapter, 12-bp index barcode, primer pad, primer linker, and

515f forward primer; the sequences were each

AATGATACGGCGACCACCGAGATCTACACGCTXXXXXXXXXXXXTATGGTAATTGT

54

GTGYCAGCMGCCGCGGTAA (the X’s indicate the unique barcode region). The forward primer barcodes used are listed in Table 2-2. The universal reverse primer contained the reverse construct of the Illumina 3’ adapter, primer pad, primer linker, and

806rB reverse primer; the sequence was CAAGCAGAAGACGGCATACGAGATAGTCA

GCCAGCCGGACTACNVGGGTWTCTAAT. Samples were amplified in 25 µL reactions that contained 0.5 units of Phusion High-Fidelity (HF) DNA Polymerase (M0530S, New

England Biolabs), 1x Phusion HF Reaction Buffer, 3% DMSO, 200 µM dNTPs, 0.25 µM of each primer, and 1 µL of template DNA (approx. 45 ng). PCR reactions were performed using a SimpliAmp™ Thermal Cycler (Applied Biosystems, Foster City, CA) with an initial denaturation stage at 94C for 3 minutes, followed by 35 cycles of denaturation at 94C for 45 sec, annealing of primers at 50C for 60 sec, and elongation at 72C for 90 sec, and completed with a final elongation stage at 72C for 10 min (77).

Triplicate amplifications for each sample were pooled together and cleaned with the

MinElute PCR Purification Kit (Qiagen, Valencia, CA). The amplicons were quantified for DNA concentration and 260 nm/280 nm ratio using a NanoDrop 1000 (ThermoFisher

Scientific, Waltham, MA). Final pools containing 500 ng of DNA from each amplicon library were prepared and submitted to the NextGen DNA Sequencing core at the

Interdisciplinary Center for Biotechnology Research at the University of Florida. Paired- end sequencing (2 x 300 cycles) was performed on an Illumina MiSeq platform using custom sequencing primers (Read 1 primer:

TATGGTAATTGTGTGYCAGCMGCCGCGGTAA; Read 2 primer:

AGTCAGCCAGCCGGACTACNVGGGTWTCTAAT, Index primer:

AGTCAGCCAGCCGGACTACNVGGGTWTCTAAT (www.earthmicrobiomeproject.org)).

55

The 300-cycle read length was chosen in order to be able to read completely through the primer sequences of the reads.

Sequenced reads were de-multiplexed based on the index barcodes at the sequencing center. The reads were processed with Cutadapt

(https://github.com/marcelm/cutadapt) and Sickle (https://github.com/najoshi/sickle) to remove any residual Illumina adapters and primer sequences, truncate sequences at the first N position, trim sequences at a bp with a phred score below 30, and remove reads that were shorter than 120 bp. Paired-end reads were joined with Eautils

(https://github.com/ExpressionAnalysis/ea-utils) with the requirements of having a minimum overlap of 30 bp and a 3% maximum difference in the overlap region. Sample names were added to the definition lines of sequencing reads using the sed command and concatenated into one fasta file in order to make them compatible for analysis in

QIIME v.1.8 (77). Clustering of OTUs at 97% similarity, with no removal of singletons, was performed in QIIME with the open-reference OTU picking method (78), and taxonomy assignments were made by mapping to the Greengenes reference database version 13.8 (79). The total counts of OTUs and assigned taxa for each taxonomic rank were transformed to relative abundance values.

Microbial community structure was analyzed with phyloseq (80) and plotted with ggplot2 (81) in R. v.3.2.1. Results indicated associations between the relative abundance of Liberibacter and the composition and diversity of the leaf microbiota, as presented in Blaustein et al. (82). There was greater variation in microbiota structure across trees than within trees, suggesting that the four leaf samples taken from each tree may be able to be pooled together to provide a sample representative of the leaf-

56

associated communities of each tree (Figure 2-1). These findings, however, were based on a largely limited number of 16S sequences that were actually assigned to bacteria

(i.e., < 3% of total sequenced reads). Thus, the major issue with the data collected from this initial sequencing run was that the vast majority of total sequenced reads that had passed the quality control cutoffs were actually assigned to 16S plastid DNA (~70%),

16S mitochondrial DNA (~25%), or were unassigned (2%) (Figure 2-2). An additional issue was that a significant portion of the sequenced reads had low-quality phred scores

(<30) after approximately 200 bp, which lead to many (about 1/3) sequences being discarded for downstream analyses. Thus, it was determined that a modified protocol may be needed to optimize the methods for sequencing the citrus microbiota (described below).

16S rRNA Gene Amplicon Preparation and Sequencing: PNA Clamps

The characterization of microbial communities with 16S rRNA gene sequencing can be hindered by the presence of contaminant 16S rRNA gene sequences from a eukaryotic host, especially for plant-associated samples that are known to be rich in chloroplast and mitochondrial DNA, all of which generate rRNA gene sequences amplified with the 16S-specific primers (69). Peptide nucleic acid (PNA) clamps were recently developed to be incorporated into PCR reactions during amplicon library preparation in order to enrich the amplification of 16S rRNA gene sequences of bacteria/archaea among eukaryotic contaminant sequences (69). PNA acts by binding to targeted template sequences within a PCR reaction and, subsequently, blocking the

PCR primers from annealing. Lundberg et al. (69) designed mPNA

(GGCAAGTGTTCTTCGGA) and pPNA (GGCTCAACCCTGGACAG) that have high affinity to mitochondrial and plastid sequences, respectively.

57

I designed a study to determine whether PNA would resolve the issues regarding the limited sequence data obtained for bacteria described above. I opted to perform

Illumina sequencing using 2 x 150 cycles in order to, also, determine whether these cycle settings, compared to 2 x 300 cycles, would yield higher quality data. Moreover, I tested whether combining the leaf samples from 4 quadrants of each tree into a composite sample would provide a sound representation of the broad leaf-associated community.

The optimization study utilized samples taken from two trees that were earlier described in the initial sequencing event (Tree D and Tree H in Figure 2-1). DNA was extracted from the 4 leaf samples and 1 root sample from the two trees, as described above. In addition, DNA was extracted from a composite leaf sample for each tree, which had contained an equal volume of the leaf tissue from each of the 4 leaf samples per tree. Amplicon libraries were prepared from each DNA extract with PCR reactions that had incorporated 9 different combinations of mPNA/pPNA (Table 2-3). The PCR reactions were performed under the same settings described above with an added stage for PNA clamp annealing at 78C for 10 sec that was set between each cycle’s denaturation and primer annealing stages (69). The PCR reactions were performed in triplicate to remove amplification bias and the products from each sample were combined and cleaned with the MinElute PCR Purification Kit (Qiagen). The DNA concentration of each library was measured with NanoDrop and all libraries were pooled together at equimolar concentrations and sent to Interdisciplinary Center for

Biotechnology Research at the University of Florida for sequencing with the Illumina

58

MiSeq set for 2 x 150 cycles. The same bioinformatics pipeline described above was used to process the sequence data.

Sequencing with settings of 2 x 150 cycles yielded 89,774 ± 20,433 joined reads, which was a significant improvement to the initial run under settings for 2 x 300 cycles that had yielded 38,320 ± 6,756 joined reads (mean ± SD) (p<0.001). It was evident that using the PNA clamps in PCR reactions yielded a substantial enrichment in bacterial

16S rRNA gene sequences (Figure 2-2) without having adverse effects on microbiota structure (Figure 2-3). In fact, the proportion of sequences assigned to taxa shown in

Figure 2-1 was a subset of those that were most abundant in the microbial communities that were characterized when incorporating PNA clamps into the amplicon preparation protocol; that is, there was generally no bias in dominant taxa that detected when using the modified method (Figure 2-4). Obtaining a deeper coverage for the microbial community by limiting contaminant sequence amplification allowed for increases in the detected richness and diversity of the microbiota (Figure 2-5; Figure 2-6). Furthermore, the method of mixing leaves from the 4 quadrants of each tree to analyze a composite sample of leaves representative of all communities in the tree seemed to be appropriate. For the microbial communities described in Figure 2-1, variation in

Shannon measure of alpha diversity within trees (i.e., variation by quadrant) was less than the variation across trees; an ANOVA indicated that the sum of squares was 5.55 within groups and 10.07 between groups. In agreement, the composite leaf microbiota sample (i.e., “Tree 10; Leaf Composite”) appeared to cluster in the middle of that from the each of the 4 quadrant samples (i.e., “Tree 10; Leaves in Quad. 1-4”) in the PCA

59

plot shown in Figure 2-3. That is, microbiota structure of the combined leaf samples was likely a robust representation of the four leaf samples taken from each tree.

Overall, these results suggested that using 0.75 µM pPNA and 0.75 µM mPNA in the PCR reactions for amplicon library preparation, along with performing Illumina sequencing at 2 x 150 cycles, would provide optimal sequencing data for citrus- associated microbial communities. Thus, I proceeded with the series of studies described in this dissertation using these methods.

Bioinformatics Pipelines

The aforementioned pre-processing steps and QIIME (with open-reference OTU picking) pipelines are presented in Appendix A. The pipeline for closed-reference OTU- picking in QIIME and that for all analyses in PICRUSt (83), which are described in

Chapters 4-5, are also documented (Appendix A). All of these bioinformatics analyses were performed in HiPer-Gator 2.0, the University of Florida research computing cluster.

Chemical Therapeutics

Chapters 4-5 of this dissertation describe the effects of antimicrobial treatments that were tested on HLB-diseased trees in a field study. The compounds that were used in the study are currently pending patent review (at the time of submission of this dissertation to the University of Florida) and have not yet been publicly disclosed.

Therefore, they are referred to throughout the text as Compound A and Compound B.

60

Table 2-1. Quantity (concentration) and quality (260/280) of DNA extracted from citrus leaf samples collected from two Citrus sinensis cv. Valencia trees in Gainesville, Florida in January 2015. Three different DNA extraction kits were tested: MoBio Biofilm (MO BIO Laboratories, Inc., Carlsbad, CA), MoBio PowerPlant (MO BIO Laboratories, Inc., Carlsbad, CA), Isolate II Plant DNA (Bioline, Taunton, MA). DNA extraction kit Tree No. Extractions DNA Concentrationa 260/280 ratioa MoBio Biofilm A 1 45.8 n/a 1.60 n/a MoBio Biofilm B 1 54.4 n/a 1.69 n/a MoBio PowerPlant A 3 184.4 4.9 1.68 0.01 MoBio PowerPlant B 3 154.6 6.9 0.88 0.06 Isolate II Plant DNA A 3 21.5 10.1 1.78 0.02 Isolate II Plant DNA B 3 25.5 5.3 1.81 0.02 aMean SE determined with a NanoDrop 1000 (ThermoFisher Scientific, Waltham, MA)

61

Table 2-2. Index barcodes for the forward primers that were used for amplicon library preparation for 16S rRNA gene sequencing throughout this dissertation (www.earthmicrobiome.org). Primer Name Index Barcode Primer Name Index Barcode 515rcbc0 AGCCTTCGTCGC 515rcbc36 GTGGAGTCTCAT 515rcbc1 TCCATACCGGAA 515rcbc37 ACCTTACACCTT 515rcbc2 AGCCCTGCTACA 515rcbc38 TAATCTCGCCGG 515rcbc3 CCTAACGGTCCA 515rcbc39 ATCTAGTGGCAA 515rcbc4 CGCGCCTTAAAC 515rcbc40 ACGCTTAACGAC 515rcbc5 TATGGTACCCAG 515rcbc41 TACGGATTATGG 515rcbc6 TACAATATCTGT 515rcbc42 ATACATGCAAGA 515rcbc7 AATTTAGGTAGG 515rcbc43 CTTAGTGCAGAA 515rcbc8 GACTCAACCAGT 515rcbc44 AATCTTGCGCCG 515rcbc9 GCCTCTACGTCG 515rcbc45 AGGATCAGGGAA 515rcbc10 ACTACTGAGGAT 515rcbc46 AATAACTAGGGT 515rcbc11 AATTCACCTCCT 515rcbc47 TATTGCAGCAGC 515rcbc12 CGTATAAATGCG 515rcbc48 TGATGTGCTAAG 515rcbc13 ATGCTGCAACAC 515rcbc49 GTAGTAGACCAT 515rcbc14 ACTCGCTCGCTG 515rcbc50 AGTAAAGATCGT 515rcbc15 TTCCTTAGTAGT 515rcbc51 CTCGCCCTCGCC 515rcbc16 CGTCCGTATGAA 515rcbc52 TCTCTTTCGACA 515rcbc17 ACGTGAGGAACG 515rcbc53 ACATACTGAGCA 515rcbc18 GGTTGCCCTGTA 515rcbc54 GTTGATACGATG 515rcbc19 CATATAGCCCGA 515rcbc55 GTCAACGCTGTC 515rcbc20 GCCTATGAGATC 515rcbc56 TGAGACCCTACA 515rcbc21 CAAGTGAAGGGA 515rcbc57 ACTTGGTGTAAG 515rcbc22 CACGTTTATTCC 515rcbc58 ATTACGTATCAT 515rcbc23 TAATCGGTGCCA 515rcbc59 CACGCAGTCTAC 515rcbc24 TGACTAATGGCC 515rcbc60 TGTGCACGCCAT 515rcbc25 CGGGACACCCGA 515rcbc61 CCGGACAAGAAG 515rcbc26 CTGTCTATACTA 515rcbc62 TTGCTGGACGCT 515rcbc27 TATGCCAGAGAT 515rcbc63 TACTAACGCGGT 515rcbc28 CGTTTGGAATGA 515rcbc64 GCGATCACACCT 515rcbc29 AAGAACTCATGA 515rcbc65 CAAACGCACTAA 515rcbc30 TGATATCGTCTT 515rcbc66 GAAGAGGGTTGA 515rcbc31 CGGTGACCTACT 515rcbc67 TGAGTGGTCTGT 515rcbc32 AATGCGCGTATA 515rcbc68 TTACACAAAGGC 515rcbc33 CTTGATTCTTGA 515rcbc69 ACGACGCATTTG 515rcbc34 GAAATCTTGAAG 515rcbc70 TATCCAAGCGCA 515rcbc35 GAGATACAGTTC 515rcbc71 AGAGCCAAGAGC

62

Table 2-2. Continued. Primer Name Index Barcode Primer Name Index Barcode 515rcbc72 GGTGAGCAAGCA 515rcbc93 CGCACCCATACA 515rcbc73 TAAATATACCCT 515rcbc94 GTGCCATAATCG 515rcbc74 TTGCGGACCCTA 515rcbc95 ACTCTTACTTAG 515rcbc75 GTCGTCCAAATG 515rcbc96 CTACAGGGTCTC 515rcbc76 TGCACAGTCGCT 515rcbc97 CTTGGAGGCTTA 515rcbc77 TTACTGTGGCCG 515rcbc98 TATCATATTACG 515rcbc78 GGTTCATGAACA 515rcbc99 CTATATTATCCG 515rcbc79 TAACAATAATTC 515rcbc100 ACCGAACAATCC 515rcbc80 CTTATTAAACGT 515rcbc101 ACGGTACCCTAC 515rcbc81 GCTCGAAGATTC 515rcbc102 TGAGTCATTGAG 515rcbc82 TATTTGATTGGT 515rcbc103 ACCTACTTGTCT 515rcbc83 TGTCAAAGTGAC 515rcbc104 ACTGTGACGTCC 515rcbc84 CTATGTATTAGT 515rcbc105 CTCTGAGGTAAC 515rcbc85 ACTCCCGTGTGA 515rcbc106 CATGTCTTCCAT 515rcbc86 CGGTATAGCAAT 515rcbc107 AACAGTAAACAA 515rcbc87 GACTCTGCTCAG 515rcbc108 GTTCATTAAACT 515rcbc88 GTCATGCTCCAG 515rcbc109 GTGCCGGCCGAC 515rcbc89 TACCGAAGGTAT 515rcbc110 CCTTGACCGATG 515rcbc90 TGAGTATGAGTA 515rcbc111 CAAACTGCGTTG 515rcbc91 AATGGTTCAGCA 515rcbc112 TCGAGAGTTTGC 515rcbc92 GAACCAGTACTC

63

Table 2-3. Treatments for amplicon library preparation. The concentration of pPNA or mPNA in the PCR reactions is indicated. Treatment pPNA mPNA P1 none none P2 0.25 µM none P3 0.75 µM none P4 none 0.25 µM P5 none 0.75 µM P6 0.25 µM 0.25 µM P7 0.75 µM 0.75 µM P8 0.25 µM 0.75 µM P9 0.75 µM 0.25 µM

64

Figure 2-1. Relative abundance of the top 15 bacterial genera (* indicates unclassified genus within family listed, ** indicates unclassified genus within order listed) detected within the leaf microbiota. Note that these microbial communities were characterized with the initial amplicon preparation procedure (i.e., PNA clamps were not used).

65

Figure 2-2. Percentages of sequenced reads assigned to bacteria, chloroplast, or mitochondria 16S DNA for different PNA treatments in leaf- (left) and root- associated (right) communities. The number of green and brown circles below each bar corresponds to 0.25 µM pPNA and 0.25 µM mPNA, respectively (e.g., 3 circles = 0.75 µM).

66

Figure 2-3. Effects of PNA treatment on clustering of top 90 genera of citrus-associated microbial communities. Colors are of the same DNA sample and letters/numbers represent PNA treatment (9 treatments per sample) that corresponds to Table 2-3. Labels: L-leaf, T-tree number (note that Tree 10 and Tree 24 correspond to Tree D and Tree H in Figure 2-1), Q-quadrant number, comp-composite sample, P-PNA treatment number. Note that the “Tree 10; Leaf composite” sample is the leaf sample that was prepared from combining the leaves from Tree 10 Quadrants 1-4 at equal volume.

67

Figure 2-4. Heatmaps of the relative abundances of the 100 most abundant genera in leaf samples using the P1 (no clamps), P6 (low pPNA and mPNA concentrations), and P7 (high pPNA and mPNA concentrations) treatment during amplicon preparation. Note: top row is Liberibacter asiaticus (i.e., the HLB pathogen). Column labels: R-root, L-leaf, T10-tree D, T24-tree H, Q- quadrant number, comp-composite sample, P-PNA treatment number.

68

Microbiota of leaves sampled from the four quadrants of canopy in Tree D

Microbiota of leaves sampled from the four quadrants of canopy in Tree H

Tree D – Roots Microbiota Tree H – Roots Microbiota

Fig. 7. Rarefaction curves representative of OTUs assigned to each sample when preparing Figureampl 2ic-5.on Rarefactionlibraries w icurvesth no PrepresentativeNA clamps ( Pof1) OTUs, using assigned low pP toN Aeachand sample mPNA whenconc preparingentrations amplicon (P6) libraries with no PNA clamps (P1), using low pPNA and mPNA concentrations (P6) and high pPNA and mPNA concentrations (P7)

69

Figure 2-6. Alpha diversity (Chao1 and Shannon measures) for leaf microbiota (left) and root microbiota (right) when preparing amplicon libraries with no PNA (P1), with 0.25 µM pPNA and mPNA (P6), and with 0.75 µM pPNA and mPNA (P7).

70

CHAPTER 3 DEFINING THE CORE CITRUS LEAF- AND ROOT-ASSOCIATED MICROBIOTA: FACTORS ASSOCIATED WITH COMMUNITY STRUCTURE AND IMPLICATIONS FOR MANAGING HUANGLONGBING (CITRUS GREENING) DISEASE*

Summary

Stable associations between plants and microbes are critical to promoting host health and productivity. The objective of this work was to test the hypothesis that restructuring of the core microbiota may be associated with the progression of

Huanglongbing (HLB), the devastating citrus disease caused by Liberibacter asiaticus,

Liberibacter americanus, and Liberibacter africanus. The microbial communities of leaves (n=94) and roots (n=79) from citrus trees that varied by HLB symptom severity, cultivar, location, and season/time were characterized with Illumina sequencing of 16S rRNA genes. The taxonomically rich communities contained abundant core members

(i.e., detected in at least 95% of the respective leaf or root samples), some overrepresented site-specific members, and a diverse community of low-abundance variable taxa. The composition and diversity of the leaf and root microbiota were strongly associated with HLB symptom severity and location; there was also an association with host cultivar. The relative abundance of Liberibacter spp. among leaf microbiota positively correlated with HLB symptom severity and negatively correlated with alpha diversity, suggesting that community diversity decreases as symptoms progress. Network analysis of the microbial community time series identified a mutually exclusive relationship between Liberibacter spp. and members of Burkholderiaceae,

Micromonosporaceae, and Xanthomonadaceae. This work confirmed several previously

* Reprinted (118) with permission. Copyright © American Society for Microbiology, Applied Environmental Microbiology, 83, 2017, e00210-17, doi:10.1128/AEM.00210-17.

71

described plant disease-associated bacteria, as well as identified new potential implications for biological control. Our findings advance the understanding of (i) plant microbiota selection across multiple variables and (ii) changes in (core) community structure that may be a precondition to disease establishment and/or may be associated with symptom progression.

Introduction

The co-evolution of microorganisms with plant hosts has given rise to finely tuned symbiotic and parasitic relationships that are central to plant growth and health. In addition to tightly co-evolved relationships between individual partners, recent studies have demonstrated that plant-associated microbial communities play critical roles in plant development. Under the right circumstances, the microbiome (i.e., the totality of the metagenomes of the host-associated microbiota) can bolster plant productivity by providing protection against pathogens, among other mechanisms (33, 34, 84, 85).

Several studies have indicated that managing microbial taxonomic/functional diversity at the field scale can have positive effects on crop production (35, 36, 86, 87).

Commonly occurring organisms across similar microbiomes comprise a core microbial community that is hypothesized to play key roles in ecosystem functioning within that type of microbial habitat (36, 71). While numerous deep sequencing studies have revealed thousands of bacterial OTUs to comprise plant microbiomes, there is typically a small number of taxa that dominate the broader community (37, 51, 72, 73,

88–92). Some of the highly abundant taxa within these studies are noticeably conserved across microbiomes of similar plant species, even under variable experimental conditions. This suggests that a core microbial community forms stable associations with particular hosts across temporal and geographic scales. Nonetheless, the structure

72

of plant microbiota is known to be associated with a suite of biotic factors (e.g., plant developmental stage, phytopathogens) and abiotic factors (e.g., soil type, climate, season) (51, 72, 73, 91, 93). Given the limited number of studies that have discerned the core members of plant-associated microbial communities (51, 72, 73), much remains unknown about the structure of the core community among all microbiota and its importance with regard to plant health.

A growing number of studies have focused on understanding how the plant microbiome changes during disease development (37, 49–56). While results from these studies suggest that plant-associated microbial communities undergo perturbation during environmental change or disease progression, one may alternatively argue that the disease itself is a consequence of complex changes to the communities. For example, infections of tomato plants with Rhizoctonia solanacearum led to changes in relative proportions of several bacterial classes that were consistently dominant among all classes detected in the rhizosphere (i.e., changes in abundances of core classes)

(50). Similar trends were observed with another Rhizoctonia species in beets; however, the presence of a complex rhizosphere community that had included potentially beneficial bacterial groups (i.e., Pseudomonadaceae, Burkholderiaceae,

Xanthomonadales, and Lactobacillaceae) was linked to suppressing the development of beet root rot caused by the pathogen (94). In addition, several plant diseases are caused, or at least enhanced, by synergistic interactions among multiple bacterial pathogens (e.g., tomato pith necrosis, mulberry wilt, broccoli head rot) (95). Collectively, these studies suggest that plant disease establishment may be associated with changes in the microbiome that involve suppressing key, possibly core, members of the native

73

community. The interactions between native bacteria that may mediate the potential for microbiome-related pathogen suppression are still not well documented. These characterizations are needed for better understanding of the role of (core) native bacteria in disease progression and, ultimately, for formulating novel strategies to manipulate plant-associated microbial communities to mitigate disease.

While plant biocontrol bacteria have shown great promise in the lab (96–98), their efficacy under field conditions has been only moderately successful (99, 100). However, harnessing the beneficial potential of native microbiota is still one of the few logistically and economically feasible solutions for controlling certain phytopathogens. This is likely the case for the devastating citrus greening disease or Huanglongbing (HLB), which is caused by the phloem-limited alphaproteobacteria L. asiaticus, L. americanus, and L. africanus (1). Infection by any one of these organisms causes host callose depositions and inhibition of nutrient uptake and transport, which lead to a series of symptoms that culminate in tree death (1). The severity of HLB is underscored by billions of dollars in economic damages and thousands of jobs lost throughout Florida, where L. asiaticus is widespread, that occurred during the first few years of its documented emergence (7).

Interestingly, clone library sequencing and PhyloChip surveys have shown that citrus- associated microbiota re-structure during HLB symptom development and it has been suggested that the population dynamics of native bacteria can affect the titer of L. asiaticus (15, 20, 37, 38). In the present study, the microbial communities of citrus trees were defined across a number of variables, including HLB symptom severity, geographic location, citrus cultivar, season, and time. We used citrus and HLB as a plant host/disease model to investigate: (a) the relationship between leaf- and root-

74

associated microbial communities and host plant, (b) the interactions between native microbiota and a phytopathogen, and (c) the core microbiota, their prevalence within the broader microbial communities, and correlations between community structure and plant disease, among other factors.

Methods

Sample Collection and Processing

Leaves and roots were collected from citrus trees across Florida (Gainesville, Ft.

Pierce, Immokalee, Quincy, Vero Beach) (Figure A-1), where Huanglongbing (HLB) disease symptoms ranged from asymptomatic to severely symptomatic, during 2015-

2016. Valencia (Citrus sinensis L. Osbeck), Navel (Citrus sinensis (L.) Osbeck),

Honeybell (Citrus x tangelo), Owari (Citrus unshiu Marcovitch), and Ray Ruby (Citrus paradisi Macfadyen) were sampled in the study (Table 3-1). The HLB symptom appearance for each set of replicate trees was classified by 4 categories (I, II, III, or IV) corresponding to asymptomatic, symptomatic-mild, symptomatic-moderate, or symptomatic-severe, respectively. Figure A-2 provides a detailed description of each category.

To sample leaves, the canopy of each tree was divided into four quadrants and

10 leaves from each quadrant were randomly selected. The leaves were collected aseptically by cutting at the stem and placing into sterile Whirl-Pak® stomacher bags

(Nasco, Fort Atkinson, WI). To sample roots, 3 soil cores (5 cm diam., 25 cm depth) were taken within the root zone of each tree and the fibrous roots were screened from the soil and combined. A composite root sample for each tree was collected in a sterile

50 mL polypropylene tube. The leaf and root samples were transported to the lab in a cooler on dry ice.

75

All samples were frozen at -80C and lyophilized. The leaf samples were crushed into <1 cm fragments in their respective stomacher bags. Each set of leaf samples that was taken from the same tree was combined at equal volume in sterile 15 mL polypropylene tubes. The root samples were crushed into <1 cm fragments by adding a sterile steel grinding ball (9.5 mm diam.) to each tube and shaking vigorously by hand.

The crushed leaf and root samples, along with a sterile steel grinding ball (9.5 mm diam.), were placed in 5 mL polyethylene vials (Spex, Metuchen, NJ) and homogenized via bead beating with three 1-min bursts in the GenoGrinder 2000 (Spex, Metuchen,

NJ).

16S Illumina Sequencing

Genomic DNA for Illumina sequencing was extracted from each homogenized sample with the Isolate II Plant DNA kit (Bioline, Taunton, MA). Amplicon libraries were prepared for the V4 region of 16S rRNA genes with PCR reactions that incorporated

Illumina-compatible universal bacteria/archaeal primers 515f/806r, with a 12-bp indexing barcode added to the forward primer (76, 77). Samples were amplified in 25 µL reactions that contained 0.5 units of Phusion High-Fidelity DNA Polymerase (M0530S,

New England Biolabs), 1x Phusion HF Reaction Buffer, 3% DMSO, 200 µM dNTPs,

0.25 µM of each primer, and 1 µL of template DNA (approx. 45 ng). PNA clamps (0.75

µM pPNA, 0.75 mPNA) (69) were included in each reaction in order to limit amplification of contaminant 16S plastid and 16S mitochondrial sequences. PCR reactions were performed using a SimpliAmp™ Thermal Cycler (Applied Biosystems, Foster City, CA) with an initial denaturation stage at 94C for 3 minutes, followed by 35 cycles of denaturation at 94C for 45 sec, annealing of PNA clamps at 78C for 10 sec, annealing

76

of primers at 50C for 60 sec, and elongation at 72C for 90 sec, and completed with a final elongation stage at 72C for 10 min (69). Triplicate amplifications for each sample were pooled together and cleaned with the MinElute PCR Purification Kit (Qiagen,

Valencia, CA). The amplicons were quantified for DNA concentration and 260 nm/280 nm ratio using a NanoDrop 1000 (ThermoFisher Scientific, Waltham, MA). Final pools containing 500 ng of DNA from each amplicon library were prepared and submitted to the NextGen DNA Sequencing core at the Interdisciplinary Center for Biotechnology

Research at the University of Florida. The samples were quantified at the sequencing center with qPCR and QUBIT for quality control and size-selected with ELF to produce fragment sizes in the desired range for the amplicon sequencing. Paired-end sequencing (2 x 150 cycles) was performed on an Illumina MiSeq platform.

Data Analysis

Sequenced reads were de-multiplexed based on the indexing barcodes at the sequencing center. Parsed raw sequencing reads are publicly available through NCBI’s

Sequence Read Archive under the BioProject accession number PRJNA362723. We processed the reads with Cutadapt (https://github.com/marcelm/cutadapt) and Sickle

(https://github.com/najoshi/sickle) to remove any residual Illumina adapters and primer sequences, truncate sequences at the first N position, trim sequences at a bp with a phred score below 30, and remove reads that were shorter than 120 bp. Paired-end reads were joined with Eautils (https://github.com/ExpressionAnalysis/ea-utils) with the requirements of having a minimum overlap of 30 bp and a 3% maximum difference in the overlap region. Sample names were added to the definition lines of sequencing reads using the sed command and concatenated into one fasta file in order to make

77

them compatible for analysis in QIIME v.1.8 (101). Clustering of OTUs at 97% similarity, with no removal of singletons, was performed in QIIME with the open-reference OTU picking method (78), and taxonomy assignments were made by mapping to the

Greengenes reference database version 13.8 (79). Unassigned OTUs and those that were identified as 16S mitochondrial or plastid DNA were removed from further analyses. The total counts of OTUs and assigned taxa for each taxonomic rank were transformed to relative abundance values.

The core members of the citrus leaf and root microbiota were defined as those that had at least 1 sequencing read in at least 95% of the respective samples (102).

Rank abundance curves were generated to inspect the distribution of core members and non-core/variable members (i.e., those found in less than 95% of samples). The ratios of core members to non-core/variable members were compared at the phylum, class, order, family, and genus levels.

In order to evaluate the spatiotemporal variability of the relative abundance of

Liberibacter spp. and how it correlated with HLB disease symptom severity, the statistical software PAST (103) was used to perform ANOVA and Tukey’s post hoc tests to determine the differences in the relative abundance of the pathogen in microbiomes of citrus trees based on: (a) location and citrus cultivar (trees sampled in Fall 2015), (b) season/time (trees sampled in Gainesville during 2015-2016), and (c) HLB symptom severity (all trees).

Microbial community structure was analyzed with phyloseq (80) and plotted with ggplot2 (81) in R. v.3.2.1. Regression analysis was used to determine the relationship between microbiota alpha diversity and Liberibacter spp. relative abundance. The

78

associations of microbial community structure with plant organ (i.e., leaf, root), location, cultivar, HLB symptom severity, season, and time were evaluated with non-metric multidimensional scaling (NMDS) plots and/or analysis of similarity (ANOSIM)

(significance based on 999 permutations) in Vegan v.2.3.2 (104). Note that an ANOSIM

R statistic of 0 means the communities are identical; whereas, R of 1 means the communities have no overlap. The Student’s t-test and Power calculation were applied to determine the differences in relative abundances of core bacterial genera that were associated with asymptomatic trees (i.e., category I) and HLB symptomatic trees (i.e., category II-IV).

Network analysis of interactions between the 20 most abundant families, with

Liberibacter spp. separate from other members of its family (Rhizobiaceae), of leaf microbiota from the Valencia trees sampled in Gainesville during 2015-2016 (n=32) was performed using the CoNet app (105) in Cytoscape v.3.0.2 (106). Significant positive and negative correlations between taxa were determined by support of three separate measures: Pearson's correlation, Spearman's correlation, or Bray-Curtis dissimilarity.

Networks from the three measures were merged by intersection, keeping only significant interactions (α set at 0.05) with support from all methods. In addition, an

ANOVA was applied to determine differences in the relative abundances of these taxa at the four different time points.

Results

Citrus-associated Microbial Community Structure and Core Microbiota

Approximately 13.77% 1.18% and 91.22% 1.22% (mean standard error

[SE]) of all quality-filtered reads from leaf and root samples, respectively, were classified as 16S rRNA genes of bacteria/archaea. We used peptide nucleic acid (PNA) clamps

79

(69) during library preparation in order to limit the amplification of contaminant sequences (i.e., 16S rRNA of mitochondria and chloroplasts). In a preliminary study without the use of PNA, libraries prepared from leaf samples (n=9) and root samples

(n=3) yielded 0.90% 0.15% and 16.78% 6.58% (mean SE) of reads corresponding to bacteria/ archaea (data not shown). Thus, the PNA clamps allowed for substantial enrichment of desired sequences. Moreover, after removing the residual contaminant sequences from the quality-filtered libraries, there were 94 microbial communities of leaves and 79 microbial communities of roots with more than 1,000 16S rRNA gene sequences assigned to 16S rRNA of bacteria/archaea (Table 3-1). From these 173 microbial communities, the 7,767,069 assigned reads, with an average of 13,725 730 sequences per leaf sample and 81,986 5,168 sequences per root sample (mean

SE), were included in further analyses.

Several hundred genera were detected in the leaf- and root-associated microbial communities. Approximately 6% of the leaf microbiota and 12% of the root microbiota accounted for 90% of microbial abundance within the respective communities (Figure 3-

1). There was a larger consortium of core bacteria associated with roots (183 genera,

137 families, 83 orders, 45 classes, 8 phyla) than leaves (28 genera, 28 families, 21 orders, 13 classes, 8 phyla) and most of the relatively dominant members within the microbial communities were core (Figure 3-1). The ratios of core members to all members within the microbiomes consistently decreased as taxonomic rank became finer (Figure A-3). Accordingly, certain non-core taxa at lower taxonomic ranks comprised broader core groups. For example, there were more core members at the family-level in leaf microbiomes (n=28) than there were total families of core genera in

80

leaf microbiomes (n=21) (Object 3-1) (i.e., any non-core genera within the other 7 core families must have been present across samples). With regard to the non-core citrus- associated microbiota, most taxa appeared to be sample-specific rather than well- replicated (Figure A-3). In summary, the microbial communities were comprised of the key abundant core members and some overrepresented site-specific members, with a diverse community of low-abundance variable taxa.

Spatiotemporal Variability of HLB

In leaf-associated communities (n=94), the relative abundance of Liberibacter spp. was 25.60 3.14 % (mean SE), which was, on average, the most abundant out of all genera detected. Alternatively, in root-associated communities (n=79), its relative abundance was 0.038 0.012 % (mean SE). ANOVAs indicated that the relative abundance of the pathogen among leaf microbiota significantly differed across locations

(p<0.001) and among citrus cultivars (p<0.001). Figure 3-2A illustrates pairwise comparisons between specific treatment groups. Also, as expected, there was a positive correlation between the severity of HLB symptoms displayed by citrus trees and the relative abundance of Liberibacter spp. among leaf microbiota (Figure 3-2C).

To test the hypothesis that the prevalence of the HLB pathogen within the microbiomes of HLB-symptomatic trees varies by season and increases over time, we compared the relative abundances of Liberibacter spp. within leaves from Valencia trees sampled in

Gainesville in Spring 2015 (4/1/15), Summer 2015 (6/1/15), Fall 2015 (9/23/15), and

Spring 2016 (3/29/16). Over the course of the year, the abundance of the pathogen fluctuated, with a drop during the Fall season. There was still a net increase between

Spring 2015 to Spring 2016 (Figure 3-2B). Interestingly, HLB symptoms also

81

progressed between Spring 2015 and Spring 2016, as the chlorosis-like appearance of leaves, as well as branch die-back, became more widespread throughout the canopy

(Figure A-4). Thus, while seasonal effects contributed to changes in the abundance of the HLB pathogen, there were increases over time that concurred with disease symptom progression.

Biotic and Abiotic Factors Associated with Community Structure

There were fundamental differences in the structure of leaf and root microbiota

(ANOSIM R statistic = 0.950, significance based on 999 permutations = 0.001), which was illustrated by clustering in NMDS plots (Figure 3-3). Interestingly, while the proportions of taxa within broader phylogenetic groups (i.e., at the class-level) of respective leaf- and root-associated communities were generally conserved across location and cultivar, there were trends for differences in relative abundances of dominant microbiota at the genus-level based these variables (Figure 3-4). Accordingly, the composition and diversity of leaf and root microbiota were substantially associated with location (leaves: ANOSIM R statistic = 0.655, significance based on 999 permutations = 0.001; roots: ANOSIM R statistic = 0.577, significance based on 999 permutations = 0.001); there was also an association with citrus cultivar (leaves:

ANOSIM R statistic = 0.318, significance based on 999 permutations = 0.001; roots:

ANOSIM R statistic = 0.289, significance based on 999 permutations = 0.001) (Figure 3-

3). Additional ANOSIMs were performed for subsets of data in order to determine (a) whether the microbial communities of specific cultivars collected at different sites varied and (b) whether the microbial communities of different cultivars varied within the sites where multiple cultivars were sampled. Interestingly, the structure of leaf microbiota from each cultivar that had multiple collection sites (i.e., Valencia, Navel, and Ray Ruby)

82

substantially varied by location (Table 3-2). Alternatively, the differences in root microbiota structure across location were substantial for Navel trees, but only moderate for Valencia trees (Table 3-2). Furthermore, the differences in microbial community structure by cultivar were location-specific. At Vero Beach, the microbiota of the different different cultivars (i.e., Valencia, Ray Ruby) substantially varied; whereas, at

Immokalee and Quincy, only moderate differences were detected among the cultivars sampled (Table 3-2).

In order to determine effects of location and cultivar on microbial community structure that were free from confounding effects of HLB, subsets of data from trees within the same HLB symptom severity category were compared. Regarding location, the leaf microbiota of the moderately symptomatic Valencia trees in both Ft. Pierce and

Immokalee still substantially differed (ANOSIM R statistic = 0.749, significance based on

999 permutations = 0.001) (Table A-1). In addition, for the three asymptomatic varieties sampled in Quincy, there were substantial differences between the community structures of Owari and Honeybell leaf microbiota (ANOSIM R statistic = 0.959; significance based on 999 permutations = 0.023) and root microbiota (ANOSIM R statistic = 0.867; significance based on 999 permutations = 0.016). Alternatively, little or no differences were detected between the microbiota of Navel trees with that of either

Owari or Honeybell trees. Overall, the structure of the leaf and root microbiota strongly correlated with location and moderately correlated with cultivar. Even though these associations were partially attributable to HLB symptom severity, examples of correlations between microbiota and location, as well as microbiota and cultivar, that are free from confounding effects of each other and HLB were evident.

83

Most importantly, according to ANOSIM, HLB symptom severity of trees (i.e., asymptomatic, mildly symptomatic, moderately symptomatic, and severely symptomatic) was also strongly associated with community structure of leaf microbiota

(ANOSIM R statistic = 0.572; significance based on 999 permutations = 0.001) and root microbiota (ANOSIM R statistic = 0.592; significance based on 999 permutations =

0.001), regardless of the other variables being considered. At the genus-level, the leaf- associated communities of asymptomatic trees had relatively low abundances of

Liberibacter and relatively high abundances of Enterobacter, Hymenobacter, and/or

Methylobacterium, compared to those from HLB-symptomatic trees (Figure 3-4 I-B).

Regression analysis showed a negative correlation between the alpha diversity of leaf microbiota and the relative abundance of Liberibacter spp. (Figure A-5). Furthermore, the root-associated microbial communities of HLB-symptomatic trees had: (a) higher abundances of Bradyrhizobiaceae and lower abundances of Agrobacterium within

Alphaproteobacteria, (b) higher abundances of Burkholderia and lower abundances of

Comamonadaceae within Betaproteobacteria, and (c) higher abundances of

Rhodanobacter, Sinobacteraceae, or Xanthomonadaceae and lower abundances of

Comamonadaceae or Pseudomonas within Gammaproteobacteria, compared to that of asymptomatic trees (Figure 3-4 II-B-D). Many of the taxa whose relative abundances differed based on HLB symptoms were members of the core citrus microbial community

(Figure 3-4; Object 3-1).

Moreover, to test the hypothesis that season and time were associated with community structure, we compared microbiota from Valencia trees sampled in

Gainesville on 4/1/15, 6/1/15, 9/23/15, and 3/29/16. By comparing the microbiota

84

sampled in 2015, we determined that season correlated with the structure of the leaf- associated communities (ANOSIM R statistic = 0.282; significance based on 999 permutations = 0.001), but not with that of root-associated communities (ANOSIM R statistic = 0.064; significance based on 999 permutations = 0.091). By comparing the samples from Spring 2015 and Spring 2016, we determined that there were also differences in leaf microbiota based on time (ANOSIM R statistic = 0.218; significance based on 999 permutations = 0.012), which were less than those by season, and no differences in root microbiota (ANOSIM R statistic = 0.012; significance based on 999 permutations = 0.337) based on time. These results are illustrated by the sample date- specific trends observed for proportions of dominant taxa within the leaf microbiota

(Figure 3-5A), but not root microbiota (Figure A-6).

Putative Interactions Between Native Bacteria and Liberibacter spp.

To test the hypothesis that microbe-microbe interactions could potentially influence the progression or suppression of the HLB pathogen, we examined co- occurrence patterns of the 20 most abundant bacterial families (with Liberibacter spp. separate from other members of Rhizobiaceae), most of which were core microbiota

(Figure 3-5A), over the course of the year. Family-level was chosen because it was the lowest taxonomic rank where almost all of the top 20 taxa had full classification (i.e., only 1/20 families was unclassified within an assigned bacterial order). A total of 51 significant interactions (p<0.05) were detected between 19 of these taxa, one of which was Liberibacter spp. All taxa in the network, except for the pathogen, had at least one positive interaction (co-occurrence) (p<0.05) with another taxon and two cluster groups were observed (Figure 3-5B). The smaller cluster was linked to negative interactions

(mutual exclusion) (p<0.05) with Liberibacter spp. (Figure 3-5B). The negative

85

interactions between Liberibacter spp. and Burkholderiaceae, Micromonosporaceae, and Xanthomonadaceae appeared to mediate the negative relationship between the

HLB pathogen and the majority of native bacteria, since the top 20 families constituted

>75% of all family-level members of the microbial community at any time point (Figure

3-5A), during HLB disease progression.

Discussion

Harnessing the beneficial potential of the plant microbiome to mitigate major crop diseases, such as HLB, is gaining interest as a sustainable approach to improve agricultural production (35, 36, 99). It has been suggested that the abundances of core members within plant-associated microbial communities, which may stably associate with particular hosts (51, 72, 73), are a determinant factor in whether the community is functioning in a ‘healthy’ state (71). Understanding the importance of the core group with regard to its implications for plant health and productivity is still at a stage of infancy. Plant diseases are known to be associated with changes in the microbiome (37,

49–56), but much remains to be learned about potential interactions between phytopathogens and native bacteria, notably core microbiota, that could play key roles in disease progression. The present study investigated these associations and interactions through a comprehensive analysis of the core leaf- and root-associated communities of citrus hosts that varied in the extent of HLB disease progression.

The core compositional members of the microbial communities were defined at each taxonomic rank and as rank became more specific (i.e., phylum to genus), the ratios of the numbers of core taxa to all taxa consistently decreased. Certain non-core members at finer levels of taxonomy collectively comprised some of the taxa that were defined as core at higher taxonomic ranks. This could be attributable to plant

86

microbiome selection being dependent on microbial resource requirements and functional genes that are redundant within broader phylogenetic groups (71). Moreover, the vast majority of all genera detected in the communities were members of the rare biosphere (107) and were highly variable in terms of occurrence across samples, even among replicates. Although the numbers of core members comprised a small fraction of all citrus-associated taxa, these organisms made up the majority of the relatively dominant microbiota.

Microbial community structure was strongly associated with HLB symptom severity. The positive correlation between the relative abundance of the Liberibacter spp. and symptom severity, as well as its negative correlation with alpha diversity, suggests that the diversity of leaf microbiota decreases as HLB progresses. While

Liberibacter spp. was significantly more abundant within leaves of HLB-symptomatic trees, other Alphaproteobacteria (Methylobacterium, Sphingomonas, and

Methylocystaceae) were present in greater proportions in those of asymptomatic trees.

Since niche similarities can adversely impact co-existence (70), there may have been resource-related competition between the pathogen and other members of its bacterial class. Previous studies have also indicated Methylobacterium and Sphingomonas to be involved in protecting host plants from various pathogens (108–111). Furthermore, by examining the year-long co-occurrence patterns for the 20 most abundant families and

Liberibacter spp., we uncovered negative correlations between the pathogen and three bacterial families (Burkholderiaceae, Xanthomonadaceae, and Micromonosporaceae) that have previously been described to have plant-beneficial properties (55, 92, 94,

112). These taxa may have, to an extent, been involved in mediating the suppressive

87

relationship between the pathogen and the stable microbial community. Future work is needed to validate such microbe-pathogen interactions.

With regard to root microbiota, even though the abundance of Liberibacter spp. was miniscule, there were significant differences in the structure of communities of asymptomatic and HLB-symptomatic trees. The low detection of the pathogen in root- associated communities was probably due to it being phloem-limited (i.e., an endophyte) (1), while the majority of bacteria in the root samples were likely present as epiphytes or on soil particles from the surrounding rhizosphere that had been in contact with the root surface (73). Several dominant core genera of citrus roots (Kaistobacter and unclassified genera in Bradyrhizobiaceae and Xanthomonadaceae) were found to be associated with asymptomatic trees. Bradyrhizobiaceae are known soil bacteria and root endophytes involved in symbiotic relationships with plants as nitrogen fixers (113,

114). While some members of Xanthomonadaceae are citrus pathogens (e.g., citrus canker is caused by Xanthomonas axonopodis) (115), others have been implicated in biological control (94). Alternatively, the relative abundances of other core genera, such as Steroidobacter and unclassified genera in Comamonadaceae, Hyphomicrobiaceae,

MND1, IS-44, and Rhizobiales, were significantly greater in their respective classes among the root microbiota of HLB-symptomatic trees. It is unclear whether these disease-associated taxa may have synergistically impacted HLB progression or if they were opportunistic colonizers of as the microbiomes were re-structuring in response to disease progression having caused changes in the local microenvironments. In addition to the core taxa associated with roots, several dominant non-core/variable members were also linked to host health. For example, Rhodanobacter and Burkholderia were

88

common within their bacterial classes in the root communities of asymptomatic trees, yet absent in those of diseased trees. The former has previously been shown to exhibit biological control activity towards phytopathogens both in vitro and in vivo (116), while the latter is known to colonize the root surface, gain entry into the internal tissue, translocate in xylem, and is capable of inducing host defense signaling (112).

Interestingly, there were consistencies between some potentially beneficial taxa associated with both leaves and roots (e.g., Burkholderiaceae, Xanthomonadaceae), which may have implications for biological control of Liberibacter spp. Developing novel approaches to treat or manage HLB by bolstering the stability of the native microbial community or by using treatments that are target-specific to Liberibacter spp., so that the native community is not disrupted, warrants further investigation.

Separate from the correlations between microbial community structure and HLB symptom expression, there were unique associations between the plant microbiota and location, cultivar, season, and time, which have been previously discussed (51, 72, 73,

91, 93). A limitation to this study was that the analyses were performed at the taxonomic level of genus or above, so the strain-specificity of potential microbial interactions or associations was not able to be determined. Although strain level diversity of L. asiaticus (i.e., the HLB pathogen in Florida) is largely unexplored, perhaps due to the fact that the pathogen is not culturable (37), it cannot be ruled out that the differences in

HLB symptom severity and microbial community structure that were noted for the different sampling locations could have been related to differences in pathogenicity of the Liberibacter spp. strains that were present at the site. It is possible that, compared the more severely symptomatic trees, the asymptomatic trees that were positive for the

89

pathogen or the mildly symptomatic trees may have been colonized by benign or, at least, relatively less-virulent Liberibacter spp.

We emphasize that the findings from this work indicate correlation, not causation, between microbial community structure and HLB progression, among other variables.

Although the directional associations between pathogen, microbiota, and host remain unclear, several hypotheses can be posited. For example, the differences in microbial community structure between healthy and diseased trees may have been a response to changes in host-associated microhabitats that were affected by disease progression.

HLB is known to cause carbohydrate partitioning imbalances and deficiencies of essential micronutrients (e.g., Fe, Mn, Zn) throughout tree canopy (66, 117). These factors could have impacted the leaf-associated microbial communities. As for root microbiota, the declining health-state of the trees may have altered the composition and amounts of root exudates produced. Trivedi et al. (56) reported HLB-infection to significantly alter the abundances of functional genes in the citrus rhizosphere that were involved in carbon fixation, nitrogen cycling, phosphorus utilization, and metal homeostasis, which reflected a state of environmental change. On the other hand, since the relative abundance of Liberibacter spp. within leaf-associated microbial communities was found to correlate with disease symptom severity, which was in agreement with a previously made suggestion that a minimal titer of the pathogen is required for HLB symptoms to develop (37), certain changes within the microbial communities that could have involved a pathway for Liberibacter spp. proliferation among native bacteria may have been a critical factor for disease establishment. Potential plant-beneficial bacteria that were differentially abundant within the trees that had varied in HLB symptom

90

severity could have been involved in: (a) competing with Liberibacter spp. for nutrient resources, limiting its ability to grow and spread, (b) antibiosis towards the pathogen or other disease-associated bacteria, (c) assistance with host nutrient acquisition, which could help alleviate disease symptom-related nutrient deficiencies, or (d) induction of host-plant signaling pathways that stimulated plant defense responses. Distinguishing the changes within core plant-associated microbial communities that are associated with disease establishment from those that are associated with consequences of disease symptoms is an interesting avenue for future research.

Overall, our results advance the understanding of: (1) plant microbiome selection across multiple variables and (2) changes in community structure that are associated with disease establishment or symptom progression, having implications for biological control. We not only confirmed previously described bacterial associations with plant health (e.g., potentially beneficial bacteria), but also demonstrated the importance of core taxa within the broader community, as well as identified new associations and potential interactions between certain bacteria and the economically important phytopathogen, Liberibacter spp.

91

Table 3-1. The microbial communities of leaf and root samples from 73 citrus trees across Florida were characterized. For each sample set, the location, citrus cultivar, sampling date, number of trees sampled, numbers of leaf and root microbial communities included in data analyses (i.e., the subset of total samples that passed quality control and contained >1,000 16S sequences with assigned taxonomy), and Huanglongbing (HLB) symptom appearance are listed. For the latter, the categories I, II, III, or IV correspond to asymptomatic, symptomatic- mild, symptomatic-moderate, or symptomatic-severe, respectively (see Fig. S2 for detailed description. No. Leaf No. Root HLB Symptom Location Date(s) surveyed Citrus tree species and cultivar No. Trees Communities Communities Appearance 4/1/15, 6/1/15, Gainesville, FL Citrus sinensis L. Osbeck cv. Valencia 8* 8/date 8/date# IV 9/22/15, 3/29/16 Fort Pierce, FL 10/13/15 Citrus sinensis L. Osbeck cv. Valencia 15 14 9 III Immokalee, FL 11/3/15 Citrus sinensis L. Osbeck cv. Valencia 5 5 0 III Immokalee, FL 11/3/15 Citrus paradisi Macfadyen cv. Ray Ruby 5 5 5 IV Immokalee, FL 11/3/15 Citrus sinensis (L.) Osbeck cv. Navel 5 5 5 II Quincy, FL 10/20/15 Citrus sinensis (L.) Osbeck cv. Navel 10 9 10 I Quincy, FL 10/20/15 Citrus unshiu Marcovitch cv. Owari 5 5 5 I Quincy, FL 10/20/15 Citrus x tangelo cv. Honeybell 3 3 3 I Vero Beach, FL 11/2/15 Citrus sinensis L. Osbeck cv. Valencia 5 4 2 II Vero Beach, FL 11/2/15 Citrus paradisi Macfadyen cv. Ray Ruby 12 12 9 I *The same 8 trees were sampled on each date # On 6/1/15, only 7 replicates were included in the analysis

92

Table 3-2. Differences in microbial community structure: (a) across locations for cultivars that were sampled at multiple sites and (b) across cultivars within a location where multiple cultivars were sampled. The ANOSIM R statistic and the significance based on 999 permutations are listed for each ANOSIM test. An R statistic of 0 means the communities are identical; whereas, R of 1 means the communities have no overlap. R values greater than 0.6 with significance less than 0.05 indicate substantial difference and are listed in bold font. Microbial Factor Cultivar(s) Location(s) ANOSIM R Significance Community Location Leaf Navel Immokalee/ Quincy 0.606 0.002 Location Leaf Ray Ruby Vero Beach/ Immokalee 0.897 0.001 Ft. Pierce/ Gainesville, Location Leaf Valencia 0.663 0.001 Immokalee/ Vero Beach Location Root Navel Immokalee/ Quincy 0.766 0.001 Ft. Pierce/ Gainesville/ Location Root Valencia 0.226 0.015 Vero Beach Cultivar Leaf Navel/ Ray Ruby/ Valencia Immokalee 0.296 0.034 Cultivar Leaf Ray Ruby/ Valencia Vero Beach 0.625 0.046 Cultivar Leaf Honeybell/ Navel/ Owari Quincy 0.365 0.006 Cultivar Root Navel/ Ray Ruby Valencia Immokalee 0.308 0.033 Cultivar Root Ray Ruby Valencia Vero Beach 0.758 0.003 Cultivar Root Honeybell/ Navel/ Owari Quincy 0.319 0.016

93

Object 3-1. The list of genera that comprised the core microbiomes associated with citrus leaves and roots. Core taxa identified down to genus-level are listed in alphabetical order by taxonomic group (a blank cell or a cell that contains "Other" means that it is an unclassified taxon within an assigned higher group). The relative abundance of the microbiota in all trees, in only asymptomatic trees, or in only HLB-symptomatic trees sampled in Fall 2015 are listed. Results from a Student’s t-test and Power calculation to evaluate differences in asymptomatic and HLB-symptomatic trees are listed. Significant differences (p<0.05) with a Power > 0.9 are indicated in bold font; green bold font indicates higher abundance in asymptomatic trees and red indicates higher abundance in diseased trees. (.xlsx file 78 KB)

94

Figure 3-1. Rank abundance curves for the genera assigned to 16S sequences detected in the root- and leaf-associated microbial communities. Color indicates whether the genus was a member of the core microbiome (i.e., present in >95% of respective samples (102)). The dotted line represents the threshold for the rare biosphere (i.e., 0.1% relative abundance (107)).

95

) ) ) % % % ( ( 100 c

b (

(C) ab (A) (B) ) ) e e e c c % c %

a Cultivar ( ( n

n n abc ab . e a a a c d d d d n n Ray Ruby n n

abc n a u u 75 u d u

b ab b n b b

Honeybell u A A A

a b b

ab . e A e l e

v Owari v v e e i i i t t r v t

i t a a r a l l l

Valencia a l e e

e 50 e t e R R c R

a R

. .

.

Navel a r p p p e b t p p i abc p c r s s s a e r r b r i b r e e e i 25 t t t e L c c b c

bc i . a a a L

b b a b . i i i r r a r a c C e

c e e c c C b b b i i i L L

L 0

Asymp. Mild Moderate Severe Immokalee Quincy Vero Beach Sampling Date HLBHLB Sy Symptommptom AppearAppeaancerance Location

Figure 3-2. Relative abundance of Liberibacter spp. (i.e., the HLB pathogen) in citrus leaf samples by location and cultivar (panel A), by sampling date for Valencia trees at the Gainesville site (panel B), and by severity of HLB symptoms (panel C). See Fig. S2 for description of symptom severity categories. Letters indicate significant difference based on Tukey’s post-hoc test following ANOVA (α set at 0.05).

96

A.

B. C.

1.0 0

2 0.5 2 S S D D

M M −1 N N 0.0

−0.5 −2

−1.0 −0.5 0.0 0.5 1.0 −1 0 1 2 NMDS1 NMDS1

Figure 3-3. NMDS plots for differences in the structure of microbial communities that were associated with the citrus trees sampled in Fall 2015 (A: all samples; B: only root samples; C: only leaf samples). Closed and open symbols correspond to root and leaf microbiota, respectively. Symbols correspond to tree location (color) and cultivar (shape); the category for HLB symptom severity (I:asymptomatic; II:symptomatic-mild; III:symptomatic-moderate; IV:symptomatic-severe) is indicated.

97

Level of taxonomy: Class

s Level of taxonomy: Genus e i t i n u

m Other Other s m ) ) ) # #

e Sphingobacteriia Methylosinus o % % % ( ( ( # m

C Acidobacteriia * Uncl. in Blattabacteriaceae

e e e o l c c c i Planctomycetia# n n Enterobacter n a b a a i a # # d d d o Betaproteobacteria * Hymenobacter b n n n r u u o u # # c b b b Flavobacteriia Halomonas * r i A c A A

# # i Cytophagia Uncl. in Methylocystaceae * M e e e

v v v f i i i # # M t t t Actinobacteria Uncl. in Rhizobiales

a a a a f l l l

e # # e e e Gammaproteobacteria a Sphingomonas * R R R L e Alphaproteobacteria#* CaLibe. Lriibbaericbtaecrt#e*r#* L CaLi.be Lribibaericbtaecrt#e*r# * Methylobacterium# *

Other

) Uncl. in IS-44* % ( Achromobacter* s e c e Uncl. in Oxalobacteraceae (A)# * n i t a

i # d Uncl. in MND1 * n n u

u #

Other b Uncl. in Oxalobacteraceae (B) s m A e #

) Acidobacteria-6 * Cupriavidus# e m % m v i ( # t

o Deltaproteobacteria *

o Salinispora a e i l

c #

C # e b Uncl. in Comamonadaceae (A) * [Saprospirae] n l R a o # # d a r Th ermophilia Uncl. in Comamonadaceae (B) * n i c u

i # b Planctomycetia * b Burkholderia* o

A # M

r Betaproteobacteria *

e t c v

i # i t

o Gammaproteobacteria * a l o

M # e Alphaproteobacteria R t R

o Ca. Liberibacter Liberibacter Other o Actinobacteria#

R # ) Uncl. in Gammaproteobacteria * % (

En terobacter e

c Aquicella# n a

d Rhodanobacter* n

u Uncl. in Coxiellaceae# b

A # Dokdonella e

v # Sample Legend (x-axis) i

t Pseudomonas * a l #

e Uncl. in Xanthamonadaceae * R Steroidobacter#* Uncl. in Sinobacteraceae# *

Other #

) Kaistobacter *

% # ( Uncl. in Hyphomicrobiaceae *

e # c Uncl. in Rhizobiaceae n

a Uncl. in Sphingomonadaceae# d n

u Devosia# b

A #

Uncl. in Rhizobiales * e

v #

i Uncl. in Rhodospirillaceae * t a l Agrobacterium# * e

R Uncl. in Bradyrhizobiaceae# * Rhodoplanes#

Figure 3-4. Relative abundances of dominant taxa (A: classes, with Liberibacter spp. separate from other Alphaproteobacteria, B-D: genera) assigned to 16S sequences detected in leaves (I) and roots (II) of the citrus trees sampled in Fall 2015. Symbols correspond to tree location (color) and cultivar (shape); the category for HLB symptom severity (I:asymptomatic; II:symptomatic-mild; III:symptomatic-moderate; IV:symptomatic-severe) is indicated. In each individual legend, the # indicates member of core microbiota and the * indicates significant difference, with statistical Power > 0.9, from Student’s t- test for differences in relative abundance in asymptomatic trees (category I) and HLB-symptomatic trees (category II-IV) (α set at 0.05).

98

A. B.

100 OtherOther BacterialClasses SinobacterSinobactaceaeeraceae#* Actinobacteria MicromonosporMicromonosaceaeporaceae #* Alphaproteobacteria HyphomicrobiaceaeHyphomicrobiaceae #* Betaproteobacteria # Beijerinckiaceae BeijerBeijeinckiaceaerinckiaceae Halomonadaceae 75 Gammaproteobacteria BlattabacterBlattabaciaceaeteriaceae Methylobacteriaceae Sphingobacteria PseudomonadaceaePseudomonadaceae #* ) Methylocystaceae ) % # ( HalomonadaceaeHalomonadaceae % (

e # e c XanthomonadaceaeXanthamonadaceae* Pseudonocardiaceae c n n a a Burkholderiaceae d

d Burkholderiaceae* n n u u 50 Pseudonocardiaceae # Rhizobiales*

b Pseudonocardiaceae b A

A # e OxalobacterOxalobacaceaeteraceae * Liberibacter v e i t v

i Sphingomonadaceae a # l Microbacteriaceae t Microbacteriaceae e a l R # e EnterobacterEnterobaciaceaeteriaceae * R Microbacteriaceae StreptomStreptoycetaceaemycetaceae #* 25 RhizRhiobiales*zobiales #* Pseudomonadaceae Xanthomonadaceae # MethMetylocystaceaehylocystaceae Cytophagaceae SphingomonadaceaeSphingomonadaceae #* Enterobacteriaceae Oxalobacteraceae Micromonosporaceae CytophagaceaeCytophagaceae #* Streptomycetaceae MethMetylobacterhylobaiaceaecteraceae # 0 CandidatusLiberibac Liberter #ibacter* Burkholderiaceae Hyphomicrobiaceae 4/1/15 6/1/15 9/23/15 3/29/16 Sampling Date SamplingDate Sinobacteraceae

Figure 3-5. (A): Relative abundances of the 20 most abundant families, with Liberibacter spp. separate from other members of Rhizobiaceae, in leaf-associated microbial communities of Valencia trees sampled in Gainesville, FL on 4/1/15, 6/1/15, 9/23/15, and 3/29/16. The # indicates the taxon was a member of core microbiota and the * indicates significant difference (p<0.05) for different dates based on ANOVA. (B): Correlation-based network analysis of the relative abundances of these top 20 taxa reveals significant interactions (i.e., dotted lines indicate co-occurrence; solid lines indicate mutual exclusion) (p<0.05). The size of each circle corresponds to the average relative abundance and the color corresponds to the bacterial class represented.

99

CHAPTER 4 NOVEL, TARGET-SPECIFIC ANTIMICROBIALS SUPPRESS Liberibacter PHYTOPATHOGEN WITHOUT DISRUPTING NATIVE MICROBIOTA

Summary

Over the past decade, Huanglongbing disease (HLB) has caused billions of dollars in damages to worldwide citrus production. With no foreseeable solutions, there remains an urgent need to develop effective strategies to mitigate Liberibacter asiaticus

(i.e., the HLB pathogen). Here, a field study was conducted to evaluate the effects of trunk injections of two novel antimicrobial compounds on this phytopathogen and citrus- associated microbial communities. Roots and leaves were collected at 0, 2, 6, 12, and

18 months of treatment and the associated microbial communities were characterized via 16S rRNA gene sequencing. In both types of samples, the relative abundances of the genus Liberibacter demonstrated seasonality. Interestingly, although the treatments did not appear to impact Liberibacter abundance in leaves, they significantly suppressed its proliferation in roots. In fact, at the 18-month time point, while the root microbiota of all control trees contained the phytopathogen at abundance levels at or above their starting value, approximately half of the treated trees had experienced net reductions. Regression analyses showed that the chemically-induced reduction in

Liberibacter was significantly associated with the proliferation of several core taxa, some of which were previously identified to be associated with citrus health. Moreover, the treatments did not have adverse impacts on microbiota alpha diversity or the abundance of Alphaproteobacteria (i.e., the taxonomic class of Liberibacter).

Collectively, these results suggest target-specific activity of the compounds against the

HLB pathogen, at least within roots. This work will be extended to further investigate these compounds as sustainable strategies to control HLB.

100

Introduction

Manipulating plant microbiomes with the intention to control crop diseases and improve yields may be an essential component to the future of sustainable agriculture

(68). While many phytopathogens are normal members of plant microbiota that happen to escape host restrictions or attack when the host is in a weakened state, others clearly act in an invasive manner. In cases of the latter, interactions that commence within the complex network of native microbiota and the introduced phytopathogen(s) may, to an extent, govern microbiome re-structuring that is associated with disease. For example, with 16S rRNA gene sequencing analyses, it was recently shown that the proliferation of

Liberibacter asiaticus, the causative agent of Huanglongbing (HLB) disease, among the leaf microbiota in Citrus sinensis trees negatively correlates with community alpha diversity and positively associates with HLB symptom progression (118). Based on network analyses, mutually exclusive interactions between L. asiaticus and members of

Xanthomonadaceae, Burkholderiaceae, and Micromonosporaceae may mediate the negative relationship between the phytopathogen and the broader native community

(118). While a growing number of studies has focused on understanding how plant microbiomes change during disease establishment and progression (37, 49–56), the critical shifts that may occur in response to phytopathogen removal, perhaps during plant disease treatment and recovery, remain largely unexplored. It is unclear whether pathogen reduction would cause certain suppressed microbiota with beneficial properties to proliferate and/or microbial diversity to transition to a considerably

“healthier” state. Citrus and L. asiaticus provide a compelling system to address such fundamental questions and infer directional associations within the microbiome network.

101

HLB threatens to destroy the global citrus industry as it continues to rapidly spread throughout major citrus-producing regions (e.g., São Paolo, Brazil and Florida,

USA). Citrus trees that become infected by L. asiaticus (or another one of the HLB pathogens: Liberibacter africanus, Liberibacter americanus) develop a series of symptoms (e.g., blotchy mottle, coupled with chlorosis of leaves; production of undersized, green fruit; tree death) that result in substantial decreases in fruit quality and yield (1). All citrus varieties are susceptible to HLB, although to varying degrees, and there are still no standard curative methods (5, 6, 40). For example, the use of agronomic approaches (i.e., micronutrient amendments, applications of plant defense inducers and plant growth regulators) to try to correct nutrient deficiencies associated with HLB have been largely unsuccessful (6, 12, 13), which may be due to a lack of activity against the phytopathogen. While measures to control the spread of HLB are still limited to the destruction or quarantine of infected trees, the need for the development of novel strategies to mitigate Liberibacter spp. remains an urgent issue.

Antimicrobials have played an important role in plant agriculture for controlling a variety of other bacterial phytopathogens (e.g., Erwinia, Pectobacterium, Pseudomonas,

Xanthomonas) (31, 42). Incorporating antimicrobial applications into citrus production operations may be a viable option for pro-active HLB management. However, in several recent field studies, applications of broad-range antimicrobials, including aminoglycosides (streptomycin, kasugamycin), beta-lactams (penicillin), and/or tetracyclines (oxytetracycline), to HLB-infected trees were reported to have little success at slowing disease symptom progression (19–22). Challenges and consequences associated with using broad-range compounds included seasonality of

102

the pathogen and re-establishment over time, minimal inactivation of the pathogen in older flushes, phytotoxicity, and adverse impacts on richness and diversity of native microbiota (19–22, 119). The latter may be of critical importance since the diversity of citrus-associated microbial communities is known to be strongly associated with tree health and HLB symptom progression (118). To our knowledge, there are no chemical therapeutics that have been confirmed to inactivate the HLB pathogen without disrupting native citrus-associated microbial communities.

Since L. asiaticus, L. americanus, and L. africanus are not culturable, the development of novel antimicrobial treatments for HLB has inherent challenges. Efforts have still been made to identify compounds with activity against L. asiaticus based on information from the pathogen’s sequenced genome (14) and, also, by using close phylogenetic relatives (i.e., Liberibacter crescens, Sinorhizobium meliloti) as model organisms in culture-based experiments (17, 27, 28). These approaches have led to the identification of several small molecule compounds that target transcription factors produced by L. asiaticus, essential to its survival (17, 28). Here, a long-term field study to evaluate two of these compounds as potential HLB treatments was conducted. The objectives were to determine their effects on (i) the relative abundance pathogen and (ii) the composition and diversity of native microbiota within citrus leaves and roots over time. This design allowed for investigation of fundamental questions regarding microbial community re-structuring in response to the removal of an invasive phytopathogen.

Methods

Antimicrobial Treatments

The study took place on the main campus of University of Florida, Gainesville, FL

(29.38 N 82.21 W) between the Spring 2015 and Fall 2016 growing seasons. Twenty-

103

four Citrus sinensis L. Osbeck cv. Valencia (Valencia) trees, all of which were approximately 20 years old and were characteristically symptomatic for HLB (i.e., blotchy mottle appearance of leaves coupled with chlorosis, canopy dieback), were included in the study. Randomized block design was used for the three treatment groups (n=8 trees per treatment). Treatments included trees receiving trunk injections of two small molecules compounds (referred to as Compound A and Compound B; see

Chapter 2 of this dissertation) dissolved in TRIS-buffer solution (pH 7.5), as well as controls (i.e., injected with only buffer solution). Trunk injections often were performed over two to three days. The initial treatment of 100 µM of either compounds took place on 4/1/15 and the subsequent treatments of 1 g of either compound occurred on 6/4/15,

9/31/15, and 4/26/16.

Leaf and root tissue samples were collected at time 0 (4/1/15; before treatment was initiated) and then at 2 months (6/1/15), 6 months (9/23/15), 12 months (3/29/16), and 18 months (9/7/16). To sample leaves, the canopy of each tree was divided into four quadrants and 10 leaves from each quadrant were randomly selected. The leaves were collected by cutting at the stem and placing into sterile Whirl-Pak® stomacher bags

(Nasco, Fort Atkinson, WI). To sample roots, 3 soil cores (5 cm diam., 25 cm depth) were taken within the root zone of each tree and the fibrous roots were screened from the soil and combined. A composite root sample for each tree was placed in a sterile 50 mL polypropylene tube. All leaf and root samples were brought back to the lab in a cooler on dry ice.

Microbial Community Analyses

The root and leaf samples were processed for amplicon sequencing using methods described in Chapters 2-3 of this dissertation. In brief, genomic DNA was

104

extracted from each sample and amplicon libraries were prepared for the V4 region of

16S rRNA genes in PCR reactions that incorporated PNA clamps to enrich the amplification of 16S bacterial/archaeal sequences (69). Final pools containing 500 ng of

DNA from each library were submitted to the NextGen DNA Sequencing core at the

Interdisciplinary Center for Biotechnology Research at the University of Florida, where paired-end sequencing (2 x 150 cycles) was performed on an Illumina MiSeq platform and the sequenced reads were de-multiplexed based on the index barcodes.

Taxonomic assignments were made in QIIME v.1.8 (101) after using the pre- processing steps described in Blaustein et al. (118). Clustering of OTUs at 97% similarity was performed in with the closed-reference OTU picking method, and taxonomy assignments were made by mapping to the Greengenes reference database version 13.5 (79). OTUs that were identified as 16S mitochondrial or plastid DNA were removed for further analyses. The total counts for OTUs and those for assigned genera and families were transformed to relative abundance values.

Data analyses were performed in R v.3.2.1. Microbial community alpha diversity

(Chao1 index and Shannon measure) was computed in phyloseq (80). Two-way

ANOVAs were utilized to determine the effects of time and treatment on alpha diversity measures, as well as on the relative abundance of all Alphaproteobacteria (i.e., the taxonomic class of the HLB pathogen). Effects of treatments on the relative abundances of specific genera within Alphaproteobacteria were also determined with this method.

Moreover, the associations of microbial community structure with treatment and with time were were evaluated with Non-Metric Data Scaling (NMDS) plots and analysis of similarity (ANOSIM) (significance based on 999 permutations) in Vegan v.2.3.2 (Dixon,

105

2003). Note that an ANOSIM R statistic of 0 means the communities are identical; whereas, R of 1 means the communities have no overlap. Changes in the relative abundance of Liberibacter, as well as other taxa, were calculated as log2 ratios for all time points relative to time 0 (i.e., log2 fold change). Since log transformation of zero values is undefined, samples with a 0% relative abundance were tallied as present with a relative abundance equivalent to 50% that of a singleton in the sample, as described in Amend et al. (120). The time series for alpha diversity and that of the log2 fold changes in Liberibacter relative abundance were plotted with ggplot2 (81). A two-way

ANOVA was utilized to determine the effects of time and treatment on the log2 fold changes in Liberibacter. Associations between log2 fold changes in relative abundances of the pathogen and those of bacterial families that had average relative abundances greater than 0.1% throughout the study (i.e., to focus on taxa that were consistently present) were determined with regression analyses.

Results

Treatments Do Not Disrupt Microbiota Diversity

There were 115 leaf samples and 119 root samples that yielded microbial communities with more than 1,000 16S rRNA gene sequences assigned to 16S rRNA of bacteria/archaea. From these 234 samples, the 3,929,605 assigned reads, with an average of 15,598 584 sequences per leaf sample and 32,891 1,092 sequences per root sample (mean SE), were included in further analyses.

There were 6,987 total OTUs that comprised the leaf microbiota; 501 31 OTUs were detected in each leaf sample (mean SE). There were 12,162 total OTUs that comprised the root microbiota; 2,577 39 OTUs were detected in each root sample

106

(mean SE). Sampling time was substantially associated with the structure of leaf microbiota (ANOSIM R statistic = 0.415, significance based on 999 permutations =

0.001), though not with that of root microbiota (ANOSIM R statistic = 0.105, significance based on 999 permutations = 0.001) (Figure 4-1). Treatments did not have any detectable effects on the overall structure of the microbiota (leaves: ANOSIM R statistic

= -0.013, significance based on 999 permutations = 0.972; roots: ANOSIM R statistic =

0.022, significance based on 999 permutations = 0.016) (Figure 4-1). Accordingly, the

Chao1 indices significantly differed at different time points (leaf p<0.001; root p=0.010), but not based on treatment (leaf p=0.505; root p=0.940) (Figure 4-2). The same trends were evident for the Shannon measure of alpha diversity, in which there were also significant differences based on time (leaf p<0.001; root p=0.006), but not treatment

(leaf p=0.165; root p=0.535). Interestingly, at the 18-month time point, there were trends for higher diversity in root microbiota of treated trees compared to that of controls

(Figure 4-2). Moreover, the relative abundance of Alphaproteobacteria in leaves and roots did not significantly differ based on time (leaf p=0.720; root p=0.362) or treatment

(leaf p=0.307; root p=0.626). Within Alphaproteobacteria, there were 89 and 98 genera detected throughout the study in leaf and root samples, respectively. Those with relative abundances that significantly differed based on treatment are listed in Table 4-1. Thus, while treatments did not grossly disrupt alpha diversity or adversely impact cumulative

Alphaproteobacteria abundance, there were a few cases in which treatments may have been detrimental to specific genera within the phytopathogen’s taxonomic class (Table

4-1).

107

Treatments Suppress Liberibacter in Root-associated Microbial Communities

At time 0, the relative abundance of Liberibacter (i.e., the genus of the HLB pathogen) among leaf microbiota was 54.7 ± 5.5 % (mean SE), which was, on average, the most abundant out of the 491 genera detected in the leaf samples.

Alternatively, in the root-associated communities, the relative abundance of the pathogen was 0.07 0.01 % (mean SE), which was, on average, the 152nd most abundant out of the 760 genera detected in the root samples. For leaf communities, the log2 fold change in the relative abundance of the pathogen significantly differed over time (p=0.048), but not in response to treatment (p=0.116) (Figure 4-3A). Alternatively, for root communities, the changes in relative abundance of Liberibacter not only differed over time (p=0.005), with major proliferation of Liberibacter in control trees, but the treatments also had suppressive effects on the log2 fold change in pathogen abundance

(p=0.001) (Figure 4-3B).

At the 18-month time point, the differences in the log2 fold changes of

Liberibacter relative abundance in root communities of controls and treated trees were most drastic (Figure 4-3B). Interestingly, about half of the root microbiota of treated trees at that time had experienced a net reduction in the relative abundance of the pathogen (n=7 out of 16 trees), while all control trees contained levels at or above their starting values (Figure 4-3C). Thus, the treatments not only suppressed the proliferation of the Liberibacter in roots, but they caused net reductions in pathogen abundance a decent proportion of the trees tested.

108

Root Microbiota Response to Removal of Liberibacter

To test the hypothesis that the chemically-induced removal of Liberibacter was associated with increases in contents of other key taxa, for the 7 treated trees that experienced reduction in Liberibacter in the root microbiota at the 18-month time point

(Figure 4-3C), the log2 fold changes in relative abundance of Liberibacter at this time were compared to those of each bacterial family with regression analyses. The families that underwent significant increases in log2 fold change in relative abundance (p<0.05) in response to the Liberibacter reduction and also had a > 0.1% relative abundance in root microbiota throughout the study (i.e., they were continuously present) were

Coxiellaceae, MLE1-12, Oxalobacteriaceae, and Rhodospirillaceae (Figure 4-4A).

Moreover, in our previous study, negative associations (p<0.05) between Liberibacter and Burkholderiaceae, Micromonosporaceae, and Xanthomonadaceae within leaf microbiota were identified (118). Here, it was tested whether these potential interactions held true in the root microbiota by, once again, focusing on the data for trees that had experienced a reduction in Liberibacter abundance. However, of these three families, only the log2 fold change in relative abundance of only Xanthomonadaceae appeared to increase in response to pathogen removal, although these increases were not quite significant (p=0.095) (Figure 4-4B). There were no associations between log2 fold change in relative abundance of Liberibacter and Burkholderiaceae (p=0.488), and those between the pathogen and Micromonosporaceae were, unexpectedly, positively correlated (p=0.011) (Figure 4-4B).

Discussion

The need for a curative option for HLB remains an urgent issue. Since plant microbiomes are known to play fundamental roles in supporting host health and

109

productivity (33), and due to the fact that Liberibacter relative abundance and citrus microbiota diversity are strongly associated with HLB progression (118), optimal HLB management may require control measures that do not disrupt the native microbiota.

The small molecule compounds used in the field study appeared to demonstrate target- specific antimicrobial activity against the pathogen by suppressing its proliferation among root microbiota, even causing reductions in some cases, without adversely impacting overall microbial community structure. Aside from this study confirming that the novel compounds have implications for citrus health management at the field-scale, the unique design also allowed for the investigation of microbiota transitions that coincide with the controlled removal of a key pathogen.

The chemically-induced removal of Liberibacter correlated with increases in relative abundances of OTUs assigned to Coxiellaceae, Oxalobacteriaceae,

Rhodospirillaceae, and an unclassified family of MLE-12. In agreement, bacterial genera within Oxalobacteriaceae (Oxalobacter formigenes) and Rhodospirillales

(unclassified) were previously identified in root-associated communities of healthy citrus trees, though absent in HLB-diseased trees in the same respective studies (55, 56).

Interestingly, our previous work identified each of the aforementioned four bacterial families as core members of the citrus root microbiota, as they were commonly detected in roots of both healthy and diseased citrus trees across a broad spatiotemporal scale

(118). Collectively, this suggests that microbiota that colonize during transitions of waning pathogens are possibly more likely to be those that stably associate with the host (i.e., core microbiota) rather than those that may be invasive or site-specific.

Moreover, Oxalobacteraceae, along with 17 other core bacterial families, were part of

110

an interacting network of microbiota that may be mutually exclusive to Liberibacter

(118). Other studies have also reported Coxiellaceae to be plant-beneficial indicators

(121) and Oxalobacteriaceae to be successful root colonizers during microbial community transitions (122). Taxa that may interact or compete with Liberibacter, particularly those that may have the potential to replace it within the citrus microbiota, should be further investigated for implications in biological control.

Alpha diversity within leaf- and root-associated microbial communities did not differ based on treatment, which supports the notion of the compounds were target- specific. In addition, although one may speculate that the chemicals might impact members of Alphaproteobacteria due to phylogenetic similarities with the targeted organism, it was determined that the relative abundance of this bacterial class was not affected by the treatments and, similarly, only a small fraction of genera within it were suppressed in leaves (i.e., unclassified genera of Aurantimonadaceae and

Sphingomonadaceae) and roots (i.e., Agrobacterium, Aminobacter, Rubellimicrobium,

Sphingomonas) during treatment. Of these taxa, only Sphingomonas was previously identified as a core member of the citrus microbiota that, based on its relative abundances in healthy and diseased trees across Florida, may be associated with citrus health (118). Alternatively, in other field and greenhouse studies that have monitored the microbiome when attempting to use antimicrobials to treat HLB-infected citrus, adverse impacts of those treatments on community richness and diversity were common (18, 20, 38). For example, the phylum diversity in leaf communities of citrus trees that received trunk injections of kasugamycin and oxytetracycline or penicillin and streptomycin throughout one year was lower than in that of control trees receiving

111

water, and the OTUs in treated trees represented, on average, only 38.9% of the OTUs in the control populations (20). With relevance to citrus health, both forms of treatment induced upwards of 50% declines in OTUs within Bacteroidetes (20), and some members of this phylum are core citrus leaf microbiota that are negatively associated with Liberibacter (e.g., Cytophagaceae) (118). Likewise, in greenhouse studies that performed root drenches or graft-inoculated citrus seedlings with HLB-diseased scions that had been soaked in antibiotic-containing (i.e., ampicillin, sulfathiazole, sulfadimethoxine) solutions, alpha diversity was consistently greater in controls (i.e., scions soaked in water) than treatment groups (18, 38). In the unique case in which leaf microbial communities became more diverse in response to antimicrobial treatment (i.e., when using gentamycin), the communities became much less stable as a plethora of low abundance OTUs appeared, and Liberibacter was still not suppressed (38).

Disruption of keystone species in response to antimicrobial solution soaking may have caused that spike in diversity. Moreover, while phytotoxic effects have been reported for several broad-range compounds (i.e., oxytetracycline, penicillin, streptomycin) (19, 22), the chemical therapeutics used in this work did not cause damage to the trees. Thus, these compounds proved to be environmentally safe, especially when compared to the broad-range antimicrobials used in previous studies.

Although Liberibacter was significantly reduced in root-associated communities of the infected trees, the treatments did not appear to impact pathogen abundance in leaves. Accordingly, the trees did not recover from HLB symptoms throughout the study.

This may have, at least partially, been due to seasonality and re-establishment of the pathogen over time, which is consistent with all prior studies that have documented the

112

use antimicrobials for HLB in the field (19–22). Optimization of the therapeutic delivery method to better distribute antimicrobial compounds across the canopy (e.g., using pressurized injections, coupling with spray applications) might be needed to improve treatment efficacy. Moreover, at the start of this study, the trees were approximately 20 years old and were severely symptomatic in appearance, perhaps already in state of substantial health decline. Whether or not the antimicrobials would be more effective against HLB in younger or less severely-diseased trees remains unknown and warrants investigation. In theory, treatment efficacy would be highest if performed at the early onset of phytopathogen infection (13). Future research that focuses on developing integrated approaches that include applications of the novel compounds used in this work, or others with similar target-specific activity against Liberibacter, in combination with other measures for plant disease control (e.g., applications of plant defense regulators, biological control agents, micronutrient amendments) may be essential for mitigating HLB.

113

Table 4-1. Genera that significantly differed (p<0.05) in relative abundance based on treatment received, and corresponding p-values from a two-way ANOVA evaluating differences based on treatment (Control vs. Compound A or Compound B), time (0, 2, 6, 12, 18 months), and interactions of treatment x time. For each genus listed, there were no differences for the three treatment groups at the start of the study (i.e., p>0.05 in two-tailed Student’s t-test). Relative Abundance Relative Abundance Sample Treatment Genus in Microbiota of in Microbiota of p(treat.)b p(time)b p(inter.)b Type Control Treesa Treated Treesa Leaf Cmpd A Uncl. in Sphingomonadaceae 0.467 ± 0.089 % 0.301 ± 0.069 % 0.023 <0.001 0.119 Leaf Cmpd B Uncl. in Sphingomonadaceae 0.467 ± 0.089 % 0.257 ± 0.054 % 0.019 <0.001 0.16 Leaf Cmpd B Uncl. in Aurantimonadaceae 0.226 ± 0.053 % 0.100 ± 0.024 % 0.027 <0.001 0.089 Root Cmpd A Agrobacterium 2.535 ± 0.355 % 1.607 ± 0.197 % 0.017 0.332 0.584 Root Cmpd A Uncl. in Alphaproteobacteria 0.942 ± 0.046 % 1.056 ± 0.043 % 0.013 0.653 0.736 Root Cmpd A Sphingomonas 0.595 ± 0.074 % 0.404 ± 0.039 % 0.04 0.659 0.574 Root Cmpd B Aminobacter 0.0089 ± 0.0015 % 0.0044 ± 0.0011 % 0.009 0.388 0.257 Root Cmpd B Rubellimicrobium 0.0026 ± 0.0009 % 0.0003 ± 0.0002 % 0.044 0.676 0.29 aMean ± SE for relative abundance at all times beyond initial treatment (i.e., 2, 6, 12, 18 months) bAnalysis had been performed on the subset of genera within Alphaproteobacteria that had significant differences (p<0.05) based on treatment in a two-way ANOVA for differences based on treatment and time (i.e., the values may have differed between Compound A and Compound B or between either compound and the Control)

114

Figure 4-1. NMDS plots for differences in the structure of microbial communities (A-Leaf; B-Root) as a factor of sampling time (shape) and treatment (color).

115

Leaf Microbiota Root Microbiota

2000 Control

Cmpd A 4000

Cmpd B x x e

e 1500 d d n n I I

1 1 o o a a h h C C 1000 3500

500 A) B) 0 5 10 15 0 5 10 15 Time (Month) Time (Month)

B

4 6.2 e e r r u u

s 3 s a a e e 6.0 M M

n n o o n n

n 2 n a a h h S S 5.8 1 C) D)

0 5 10 15 0 5 10 15 Time (Month) Time (Month)

Figure 4-2. Alpha diversity of microbiota throughout the study. Panel A: Chao1 index for leaf microbiota; Panel B: Chao1 index for root microbiota; Panel C: Shannon measure for leaf microbiota; Panel D: Shannon measure for root microbiota. ANOVA and Tukey’s post-hoc test was performed for each time point; differences between values associated with Control trees and trees that received Compound A or Compound B (if present) are indicated by letters above the respective point (“a” or “b” indicates p<0.1; “A” or “B” indicates p<0.05).

116

%

e Leaf Microbiota c

n Root Microbiota a 2 d n u b A

e v i t t a n l

e 1 e

R 10

m n t i

a e g e n r a T h Control f

C 0

o d l o

Cmpd A s F h − t 2 n g Cmpd B o

A) o L −1 M

0 5 10 15

Time (Month) 8

1 5

r e t Ab f a %

e Root Microbiota e c n g a n d 5.0 a n u h b A C

e d v l i t o a l F e 2.5

R 0

2 n

i a g

e o g L n a h 0.0 C

d l o F − 2

g C) o B) L −2.5 0 5 10 15 ControlControl CmBenzpd A CmTolfpd B Time (Month) Treatment

Figure 4-3. Log2 fold changes in the relative abundance of Liberibacter relative to time 0. Panel A: Leaf microbiota at all sampling time points; Panel B: Root microbiota at all sampling time points. ANOVA and Tukey’s post-hoc test was performed for each time point; differences between values associated with Control trees and trees that received Compound A or Compound B (if present) are indicated by letters above the respective point (“a” or “b” indicates p<0.1; “A” or “B” indicates p<0.05). Panel C: Root microbiota at the 18-month time point, in which the differences between values for the control and treatment groups were most drastic. Points that fall below the dotted line correspond to trees that had a net reduction in Liberibacter in root communities.

117

Figure 4-4. Root microbiota response to the chemically-induced removal of Liberibacter in root-associated communities. Panel A: Bacterial families that experienced significant increases (p<0.05) in relative abundance in response to the reduction of Liberibacter. Panel B: Response of the three bacterial families that were previously identified to negatively interact with Liberibacter in leaf microbiota (118) to the removal of Liberibacter in roots.

118

CHAPTER 5 CITRUS METAGENOME ASSOCIATIONS WITH HUANGLONGBING DISEASE: PHYTOPATHOGEN REMOVAL CORRELATES WITH SHIFTS IN FUNCTIONAL DIVERSITY

Summary

The metagenomes of plant microbiota affect plant fitness by providing beneficial functions to the host that are involved in growth promotion and pathogen suppression, among other mechanisms. The objectives of this work were to determine: (i) the fundamental differences in the metagenomic profiles of microbiota of healthy and

Huanglongbing (HLB)-diseased citrus trees and (ii) whether the treatment-induced suppression of Liberibacter (i.e., the HLB pathogen) within citrus-associated microbial communities impacts these profiles. PICRUSt was utilized to predict functional metagenomes, based on microbiota composition, of leaves and roots from asymptomatic and HLB-diseased citrus trees that were sampled across Florida. These profiles were then compared to those of HLB-diseased citrus trees that were treated with novel Liberibacter-specific antimicrobials. While the rank abundances of KEGG pathways encoded in the respective leaf and root metagenomes were highly conserved, indicating a core functional profile, there was significant variation in abundances of certain pathways based on citrus tree health-state. The KO functions encoded in the metagenomes that mostly influenced these differences, as well as contributing OTUs, were identified. Moreover, beta diversity analyses indicated that the root metagenomes, predicted by PICRUSt, of asymptomatic trees were more similar to those of the HLB- diseased trees that received antimicrobial treatment than the HLB-diseased controls.

While the root metagenomes of asymptomatic trees had enrichments in a variety of metabolic pathways (e.g., nitrogen metabolism, glycolysis, tryptophan metabolism,

119

benzoate degradation, butanoate metabolism, propanoate metabolism) and those of diseased controls had enrichments in several housekeeping, motility, and potentially virulent pathways (e.g., flagellar assembly, bacterial secretion system, chromosome,

DNA replication proteins), the root metagenomes of diseased trees that had received extensive Liberibacter treatment generally encoded these functions at abundances in a mid-range of the two extremes. Collectively, this suggests that phytopathogen removal within diseased plant hosts may cause the metagenome to transition towards a relatively “healthier” functional state, linking the importance of microbiome structure and function.

Introduction

Plant-associated microbial communities promote host health by enhancing nutrient acquisition from the surrounding environment and providing support for disease control and stress tolerance, among other mechanisms (33). Indigenous microbiota of leaves and roots likely compete with invading phytopathogens for space and resources.

They may also “prime” plant immune response pathways by stimulating the production of plant growth promoting and/or plant defense-inducing hormones (e.g., auxin, ethylene, jasmonic acid) (84, 123). Several studies have shown that the taxonomic diversity of native plant microbiota is strongly associated with plant disease development (37, 49–56, 94, 118). However, relatively little is known about the molecular functions encoded in the microbial communities that may actually confer host protection, as well as those that may shift during disease progression. In one study that had addressed this with pyrosequencing and shotgun sequencing analyses, populations of root microbiota were found to re-assemble following root-knot infection by

Meloidogyne spp. and the succeeding communities encoded an enrichment of

120

functional genes involved in degradation of plant polysaccharides, carbohydrate and protein metabolism, and biological nitrogen fixation (124). Thus, the authors suggested that bacteria at the infection sites may have actually assisted with nematode pathogenesis, to an extent, by breaking down the plant cell wall and allowing for phytopathogen nutrient procurement (124). Since functional profiles are inherently more stable and contain less noise than taxonomic profiles (i.e., genomes of different taxa may encode many similar functions), characterizing microbial communities by the former might actually provide more discriminatory power for determining the biological importance of samples with regard to host health (74). Thus, understanding how the functional potential of plant microbiota fundamentally differs for healthy and diseased plant hosts (e.g., in terms of metabolism, catabolism, plant-microbe interaction, pathogenicity factors, etc.), which has been largely unexplored, is needed for formulating novel strategies to manipulate plant-associated microbial communities to mitigate disease (68).

Phylogeny and biomolecular function are strongly correlated, which provides rationale for using computational approaches to infer microbial functional capabilities from taxonomic assignments. Langille et al. (83) developed the program PICRUSt

(Phylogenetic Investigation of Communities by Reconstruction of Unobserved States) to predict the composition of metagenomes using marker gene data (i.e., 16S rRNA genes of bacteria and archaea) and a database for reference genomes. Within the program, the nearest sequenced taxon index (NSTI), or the phylogenetic distance between each taxon detected in a sample and the taxon’s closest relative with a reference genome, can be quantified as a measure of uncertainty that correlates with prediction accuracy

121

(83). In fact, metagenome predictions for soil samples have been reported to be approximately 80% accurate based on comparing predicted metagenomes to actual metagenomes obtained from shotgun sequencing (83). Moreover, there has been a growing interest to characterize plant metagenomes and determine their associations with various biotic and abiotic factors (e.g., host species and cultivar, host age, location, season) with PICRUSt (125–133). For example, Yuan et al. (132) reported rhizospheric and endophytic bacteria associated with the halophytic seepweed Suaeda salsa to, importantly, encode genes contributing to salt stress acclimatization (e.g., osmoprotectant transport system, ion anti-transporters) and nutrient solubilization (e.g., phosphatase, nitronate monooxygenase, nitrite reductase). Similarly, PICRUSt may be useful for investigating plant metagenome associations with phytopathogen infection and disease progression.

In a previous study and chapter 3 of this dissertation, the core citrus leaf and root microbiota were defined across a number factors associated with community structure, including Huanglongbing (HLB) disease symptom severity (118). HLB is caused by unculturable members of Liberibacter, a phloem-limited genus within

Alphaproteobacteria (1). The relative abundance of this phytopathogen was found to increase during HLB disease progression and negatively correlate with microbiota alpha diversity, and negative interactions (mutual exclusions) between Liberibacter and several members of the native microbiota were detected (118). In a separate study, the suppression of Liberibacter among root microbiota was induced by performing long-term applications of target-specific antimicrobial treatments (Chapter 4 of this dissertation).

Several members of the core root microbiota, some with plant-beneficial associations,

122

increased in abundances as a response (Chapter 4 of this dissertation). Both of these studies were limited to describing the taxonomic diversity of citrus microbiota based on

16S rRNA gene sequencing. Here, the objectives were to expand on these studies to analyze the datasets with PICRUSt in order to determine (i) the differences in the metagenomes of microbiota of healthy and HLB-diseased citrus trees and (ii) how

Liberibacter control impacts these profiles. This design allowed for the investigation of key functions associated with plant health and whether the removal of an invasive phytopathogen may lead microbial community shifts to a, perhaps, healthier functional state.

Methods

Datasets Obtained

The present study was a comparative analysis of citrus metagenomes that were predicted from 16S rRNA gene sequences of samples that had been collected in two previous field studies (Chapters 3-4 of this dissertation). The first study was a molecular survey in which the core citrus leaf- and root-microbiota were defined across variables including location in Florida, host cultivar, HLB symptom severity, and season/time. The second study defined the effects of trunk injections of two novel antimicrobials on the microbiota of HLB-diseased Citrus sinensis trees in an 18-month field study. Since the treatment-induced suppression of Liberibacter among root microbiota was documented in the latter, and the differences in pathogen levels in treated trees and untreated controls were largest at the final time point, the root sample data from that 18-month time point (n=24) were used in the present study. Meta-data for all samples used in this work are shown in Table 5-1.

123

From both studies, the leaf and root samples were processed for 16S rRNA gene amplicon sequencing as described in Blaustein et al. (118). In brief, genomic DNA was extracted from each sample and amplicon libraries were prepared for the V4 region of

16S rRNA genes in PCR reactions that incorporated PNA clamps to enrich the amplification of 16S bacterial/archaeal sequences (69). Final pools containing 500 ng of

DNA from each library were submitted to the NextGen DNA Sequencing core at the

Interdisciplinary Center for Biotechnology Research at the University of Florida, where paired-end sequencing (2 x 150 cycles) was performed on an Illumina MiSeq platform and the sequenced reads were de-multiplexed based on the index barcodes.

Metagenome Predictions

The 16S rRNA gene sequence data were pre-processed with the same methods as Blaustein et al. (118). Microbiota characterizations that were compatible with

PICRUSt v.1.1.0 (83) were made in QIIME v.1.8 (101). Specifically, clustering of OTUs at 97% similarity was performed in with the closed-reference OTU picking method, and taxonomy assignments were made by mapping to the Greengenes reference database version 13.5 (79). OTUs that were identified as 16S mitochondrial or plastid DNA were removed for further analysis.

PICRUSt was used to predict the Kyoto Encyclopedia of Genes and Genomes

(KEGG) Orthology (KO) (134) functions for metagenomes from the 16S rRNA gene taxonomic data. For quality control, the weighted Nearest Sequenced Taxon Index

(NSTI) value was computed for each sample. The KO predictions were collapsed into broader KEGG pathways and the KO term and KEGG pathway counts were converted to relative abundances for further analyses.

124

Comparative Data Analyses

All data analyses were performed in R v.3.2.1. First, the metagenomes of samples taken from the asymptomatic and HLB-diseased citrus trees from the molecular survey were compared. In order to identify differences in metagenome profiles at the most specific functional level, principle component analysis (PCA) was used to evaluate the variation in samples based on KO term relative abundances. The

KO functions that most strongly influenced citrus-health-associated differences were determined from the PCA scores, and the OTU contributions that corresponded to these notable functions were found using PICRUSt. Moreover, the relative abundances of broader KEGG functional pathways (Level 2) in asymptomatic and HLB-diseased trees were compared with heatmaps made in Gplots v.3.0.1 (135). Two-tailed Student’s t- tests, with false discovery rate (FDR) correction for significance, were utilized to determine which pathways were differentially abundant based on tree health-state.

To test the hypothesis that Liberibacter removal from the microbiota of HLB- diseased trees may be associated with changes in the functional profile of the microbiota, the root metagenomes of the trees at the 18-month time point in the treatment study were compared with those of the asymptomatic and symptomatic C. sinensis trees from the molecular survey. For the cross-study comparisons, PCA was performed and heatmaps were generated as described above. Beta diversity between root metagenomes was determined based on bray-curtis dissimilarity in Vegan v.2.3.2

(104).

125

Results

Quality Control

The weighted NSTI values for leaf and root metagenomes were 0.045 0.010

(mean sd; n=62) and 0.136 0.029 (mean SD; n=72), respectively. Thus, based on the PICRUSt algorithm, the metagenomes that were predicted for leaf and root samples were made from full genome data for taxa that were, on average, 95.5% and 86.4% phylogenetically similar to the OTUs that were characterized in this work.

Metagenomes of Citrus Microbiota Associated with HLB Disease

The predicted metagenomes of the asymptomatic and HLB-diseased citrus trees from the molecular survey were compared. In total, there were 5,993 and 6,139 KO functions detected in the leaf and root samples, respectively. HLB symptom presence explained significant variation in the metagenomes of both types of samples (Figure 5-

1). The top 10 KO functions that had influenced these differences for the leaf and root metagenomes (i.e., those with the highest scores from the corresponding PC axis in the respective PCAs (Figure 5-1)) are listed in Table 5-2. Additionally, the OTU contributions for this set of differentially abundant functions, which were a factor of taxon relative abundance and taxon gene count per genome, were determined (Figure

5-2). Within leaf metagenomes, Liberibacter, Xanthomonadaceae, Methylobacterium, and Sphingomonas largely contributed to the functional profile differences based on host health-state as the two former and two latter were more associated with key functions in the metagenomes of HLB-diseased and asymptomatic trees, respectively

(Figure 5-2A). Within roots, there was a more diverse set of taxa that contributed to such differences (Figure 5-2B). Catellatospora, Streptomyces, Bradyrhizobiaceae, and

Burkholderiaceae played an important role in contributing to functions encoded in the

126

root metagenomes of asymptomatic trees, while Agrobacterium and Patulibacteraceae did so in those of HLB-diseased trees (Figure 5-2B). In summary, the citrus metagenomes significantly differed based on host health-state and several key functions that were mostly attributable to this variation, as well as their contributing

OTUs, were identified.

To compare the metagenomes of the asymptomatic and HLB-diseased trees at the pathway level, the KO functions associated with the leaf and root metagenomes were collapsed to gene families that were represented by KEGG pathways. As expected, several KEGG pathways dominated the metagenomes and the rank abundances of all pathways were generally conserved, regardless of HLB (Figure 5-3).

Interestingly, the relative abundances of many pathways still significantly differed based on host health-state (Figure 5-3). For example, within leaf metagenomes, while pathways for protein folding/sorting/degradation, translation, DNA replication/repair, glycan biosynthesis, and nucleotide metabolism had higher relative abundances in the diseased trees, pathways for cell transport and catabolism, signaling molecules and interaction, signal transduction, membrane transport, and amino acid metabolism had higher relative abundances in the asymptomatic trees (Figure 5-3). In addition, within root metagenomes, while pathways for cell growth and death, cell motility, signal transduction, and cofactor and vitamin metabolism had higher relative abundances in the diseased trees, pathways for transport and catabolism, signaling molecules and interaction, transcription, biosynthesis of secondary metabolites, and carbohydrate metabolism had higher relative abundances in the asymptomatic trees (Figure 5-3).

Overall, although the citrus metagenomes of asymptomatic and HLB-diseased trees

127

were comprised of core functional pathways, many pathways were still differentially abundant based on host health-state.

Phytopathogen Removal May Shift Root Metagenomes to “Healthier” State

Within the treatment study, I had demonstrated trunk injections of two compounds to essentially remove Liberibacter from root microbiota of C. sinensis trees over the course of 18 months (Chapter 4 of this dissertation). In order to determine whether the phytopathogen transitions were associated with changing the functional profiles of the metagenomes, the root metagenomes from the final time point of the treatment study were compared to those of the asymptomatic and HLB-diseased from

C. sinensis trees in the molecular survey. Interestingly, even though beta diversity within studies was lower than that across studies (i.e., the root metagenomes of trees from the treatment study were more similar to each other than to those from the molecular survey), the root metagenomes of asymptomatic trees clustered closer to those of the trees that received Liberibacter-specific treatment than than to the controls (Figure 5-4).

Accordingly, Bray-Curtis dissimilarity between the functional profiles for roots of asymptomatic trees and the untreated controls was 0.0388, while that between asymptomatic trees and the trees receiving Compound A and Compound B was 0.0286 and 0.0321, respectively. That is, the diversity of the metagenomes of the HLB- diseased trees that received treatment, compared to those of the HLB-diseased controls, was less distant from the diversity of the root metagenomes of asymptomatic trees after 18 months of treatment. To explain these trends, certain pathways involved in bacterial motility (e.g., bacterial chemotaxis, flagellar assembly) and genetic or environmental information processing (e.g., ribosome biogenesis, bacterial secretion system, chromosome, translation proteins, DNA replication proteins, protein folding and

128

associated processing) were commonly associated with the untreated controls, while certain metabolic pathways (e.g., nitrogen metabolism, glycolysis, tryptophan metabolism, benzoate degradation, butanoate metabolism, propanoate metabolism) were commonly associated with the asymptomatic trees (Figure 5-5). Interestingly, the metagenomes of trees that received Compound A and/or those of the trees that received Compound B appeared to encode each of those aforementioned pathways at a relative abundance between that of the untreated controls and that of asymptomatic trees (Figure 5-5).

Discussion

Managing plant-microbiome interactions in efforts to improve crop production will require a thorough understanding of not only the taxonomic composition and diversity of plant-associated microbial communities, but, also, the functions provided by these microbiota to their host plants (68). Linking structure and function of microbiomes has been widely studied with regard to the health of humans (136–139), but this topic has not been well documented for that of plants. The present study aimed to elucidate critical differences in metagenome functions encoded by microbial communities associated with asymptomatic and HLB-diseased citrus trees and investigate whether a

“healthier” functional state may be achieved following the removal of the invasive HLB pathogen.

At all levels for KEGG functional pathways, ranging from specific KO functions to collapsed gene families, the metagenomes of healthy and diseased citrus trees drastically differed. Interestingly, leaf metagenomes of diseased trees contained enrichments of functions involved mostly in genetic and physical repair (e.g., protein folding/sorting/degradation, DNA replication and repair, nucleotide metabolism,

129

translation, glycan biosynthesis), while those of healthier trees encoded greater abundances of functions typically involved in interactions with the host, oxther microbes, and the environment (e.g., signal transduction, signaling molecules and interaction, membrane transport, transport and catabolism, amino acid metabolism). In addition, root metagenomes of diseased trees encoded higher abundances of genes involved in housekeeping and motility (e.g., cell growth and death, cell motility, signal transduction, and cofactor and vitamin metabolism), while those of healthier trees contained higher abundances of pathways largely associated with critical metabolic pathways and cell signaling (e.g., transport and catabolism, signaling molecules and interaction, biosynthesis of secondary metabolites, carbohydrate metabolism). Thus, in a mutualistic manner, the root communities of healthier trees were likely more efficient at consuming host-provided carbon resources and, in turn, producing important metabolic byproducts.

Moreover, despite the characteristic differences for metagenomes of healthy and diseased hosts, the rank abundances for many functions were highly conserved, indicating the presence of a core functional profile. Likewise, other studies observed similar sets of functions encoded in metagenomes associated with plants that differed by cultivar and age, which may be attributable to functional redundancies across variable plant-associated microbiota (129, 130). This suggests that abundances of key functions within the core metagenome may play a fundamental role in supporting host health, and the exploitation of certain plant-beneficial functions identified here and in other works (127, 129) would be an interesting avenue for future research.

The functional diversity of plant metagenomes depends on both the relative abundances of taxa encoding the functions and the number of genes associated with a

130

given function that they encode. Although Liberibacter was a highly prevalent member of leaf microbiota of HLB-diseased trees (118), the metagenomic functions that were most important in defining the health-state of citrus trees were actually encoded by other members of the microbiota that had been more prevalent in asymptomatic trees, including Methylobacterium and Sphingomonas (Figure 5-2A). This suggests that the invasive phytopathogen may be associated with limiting the abundances of these key microbiota that provide important functions in a healthier environment, such as those involved in chemotaxis, transporters, iron receptors, and glutathione metabolism (Table

5-2). It is possible that Liberibacter outcompeting these native Alphaproteobacteria is critical for disease symptoms to progress. Other studies have also indicated

Methylobacterium and Sphingomonas to be involved in protecting host plants from various pathogens (108–111). Moreover, for the citrus root microbiota, our findings suggest that functions involved in antibiotic production, transporters, and metabolism encoded by Catellatospora, Streptomyces, Bradyrhizobiaceae, and Burkholderiaceae may be critical for citrus health. Future studies are needed to validate the importance of these predicted functional pathways, in terms of gene expression in the associated taxa, within citrus metagenomes to further establish the links between microbiota structure and function with host health.

It is well known that microbial communities change in response to invading phytopathogens (50, 53, 118) and, to our knowledge, this is the first work to assess coinciding shifts in the functional profiles of the microbiomes. In addition, trends for metagenome responses to phytopathogen removal were established. The removal of

Liberibacter from root microbiota correlated with shifts towards a “healthier” functional

131

state, as beta-diversity was less between the root metagenomes of asymptomatic trees and HLB-diseased trees that received Liberibacter-specific treatment than between those of asymptomatic trees and untreated HLB-diseased controls. Accordingly, while the root metagenomes of asymptomatic trees had enrichments in a variety of metabolic pathways (e.g., nitrogen metabolism, glycolysis, tryptophan metabolism, benzoate degradation, butanoate metabolism, propanoate metabolism) and those of diseased controls had enrichments in several motility-associated, housekeeping, and potentially virulent pathways (e.g., flagellar assembly, ribosome biogenesis, bacterial secretion system, chromosome, DNA replication proteins), the root metagenomes of trees that had received extensive Liberibacter treatment generally encoded these functions at abundances in a mid-range of the two extremes. This suggests that selecting for organisms that contribute to diverse metabolic pathways within the microenvironment, possibly important for host interaction, may be associated with controlling phytopathogens. Moreover, unlike the trends for changes in functional potential that had occurred, taxonomic alpha diversity of the root microbiota did not differ between treated trees and controls. Thus, the functionally-redundant organisms that succeeded the waning pathogen collectively encoded an enrichment of genes associated that were more associated with healthy (asymptomatic) citrus trees. This presents a unique case where functional diversity, rather than taxonomic diversity, may actually be a better biomarker for plant disease-state.

A limitation to this study was that rather than using a shotgun sequencing approach to characterize the microbiota functional profiles, the metagenomes were predicted based on 16S rRNA taxonomic assignments. The PICRUSt developers

132

demonstrated NSTI values for predicted metagenomes under 0.05 to typically have about 80% accuracy based on spearman correlation with actual metagenomes identified by shotgun sequencing (83). Indeed, the leaf metagenomes in this work were associated with NSTI values fell below this threshold. Although environmental samples

(e.g., soil, root) typically have NSTI values above this threshold, perhaps a reflection of there being less reference genomes available for non-human-associated taxa, an accuracy around 80% was also reported for soil samples with NSTI values in the range of 0.11-0.19 (83). Likewise, Zarraonaindia et al. (133) reported significant correlation between actual metagenomes of root samples and PICRUSt predictions for those metagenomes that had NSTI values around 0.2. Thus, since the NSTI values associated with our leaf and root metagenome predictions were comparable or lower than those of previous studies (127, 128, 133), it can be concluded that the metagenomic predictions in this work were highly accurate.

Overall, our results advance the understanding of the connection between plant microbiome structure and function with regard to phytopathogen infection and disease.

Fundamental differences between the metagenomes of microbiota associated with healthy and HLB-symptomatic citrus trees were identified and Liberibacter suppression was shown to select for microbiota that encode an enrichment of functional pathways that may be involved in supporting citrus health. This work should be extended to try and better understand how the beneficial potential of these molecular pathways, as well as those for other plant/disease model systems, may be able to be harnessed to improve crop yields.

133

Table 5-1. List of samples used in this work. For each sample set within the Molecular Survey (from Chapter 3 of this dissertation) or Treatment Study (from Chapter 4 of this dissertation), the HLB disease condition, location, tree cultivar, sampling date, and the number of metagenomes from root and leaf samples (each representative of the root/leaf microbiota from one tree) are listed. Disease Date No. Root No. Leaf Study Cultivar Location Condition Sampled Metagenomes Metagenomes

Molecular Survey Asymptomatic Citrus sinensis (L.) Osbeck cv. Navel Quincy, FL 10/20/15 10 9

Molecular Survey Asymptomatic Citrus unshiu Marcovitch cv. Owari Quincy, FL 10/20/15 5 5

Molecular Survey Asymptomatic Citrus x tangelo cv. Honeybell Quincy, FL 10/20/15 3 3

Citrus paradisi Macfadyen cv. Ray Molecular Survey Asymptomatic Vero Beach, FL 11/2/15 9 12 Ruby Citrus sinensis L. Osbeck cv. Molecular Survey HLB-symptomatic Fort Pierce, FL 10/13/15 9 14 Valencia Citrus sinensis L. Osbeck cv. Molecular Survey HLB-symptomatic Immokalee, FL 11/3/15 0 5 Valencia Citrus paradisi Macfadyen cv. Ray Molecular Survey HLB-symptomatic Immokalee, FL 11/3/15 5 5 Ruby

Molecular Survey HLB-symptomatic Citrus sinensis (L.) Osbeck cv. Navel Immokalee, FL 11/3/15 5 5

Citrus sinensis L. Osbeck cv. Molecular Survey HLB-symptomatic Vero Beach, FL 11/2/15 2 4 Valencia Citrus sinensis L. Osbeck cv. Treatment Study HLB-symptomatic Gainesville, FL 9/7/16 24# N/A Valencia #These trees were all sampled at the 18-month time-point of the treatment study. There were 3 treatment groups: those that received Compound A, Compound B, or untreated controls (n=8/treatment).

134

Table 5-2. List of KO terms with the top 10 scoresa associated with metagenomic differences among asymptomatic and HLB-diseased trees. Metagenome KO PC Score Rel. Abund. % Rel. Abund. % in Function Description for KO Term p-value Sample Term (Rank)b in Asym. Treesc Diseased Treesc Leaf K03406 116.36 (1) methyl-accepting chemotaxis protein 0.498 ± 0.023 0.342 ± 0.020 <0.001 Leaf K02014 99.30 (2) iron complex outer membrane receptor protein 0.582 ± 0.016 0.470 ± 0.017 <0.001 sulfonate/nitrate/taurine transport system substrate- Leaf K02051 79.56 (3) 0.385 ± 0.016 0.276 ± 0.014 <0.001 binding protein branched-chain amino acid transport system substrate- Leaf K01999 77.78 (4) 0.345 ± 0.015 0.229 ± 0.013 <0.001 binding protein Leaf K07497 63.30 (5) putative transposase 0.278 ± 0.013 0.198 ± 0.009 <0.001 Leaf K00799 62.75 (6) glutathione S-transferase 0.385 ± 0.013 0.303 ± 0.012 <0.001 branched-chain amino acid transport system permease Leaf K01997 49.02 (7) 0.224 ± 0.010 0.150 ± 0.008 <0.001 protein Leaf K02651 43.80 (8) pilus assembly protein Flp/PilA 0.065 ± 0.002 0.129±0.013 <0.001 sulfonate/nitrate/taurine transport system ATP-binding Leaf K02049 43.14 (9) 0.309 ± 0.009 0.248 ± 0.010 <0.001 protein Leaf K02035 42.16 (10) peptide/nickel transport system substrate-binding protein 0.285 ± 0.008 0.221 ± 0.010 <0.001 Root K02025 59.18 (1) multiple sugar transport system permease protein 0.259 ± 0.007 0.236 ± 0.014 0.15 Root K02027 55.88 (2) multiple sugar transport system substrate-binding protein 0.241 ± 0.006 0.220 ± 0.014 0.19 Root K09687 48.51 (4) antibiotic transport system ATP-binding protein 0.302 ± 0.009 0.251 ± 0.010 <0.001 Root K13787 42.26 (7) geranylgeranyl diphosphate synthase, type I 0.104 ± 0.005 0.070 ± 0.006 <0.001 Root K08884 38.41 (9) serine/threonine protein kinase, bacterial 0.182 ± 0.006 0.211 ± 0.013 0.05 Root K09686 32.76 (10) antibiotic transport system permease protein 0.219 ± 0.007 0.186 ± 0.007 0.001 Root K02529 28.97 (12) LacI family transcriptional regulator 0.259 ± 0.005 0.229 ± 0.008 0.002 Root K00924 28.25 (13) MSHA pilin protein MshA 0.048 ± 0.002 0.080 ± 0.005 <0.001 Root K00059 26.58 (14) 3-oxoacyl-[acyl-carrier protein] reductase 0.380 ± 0.004 0.356 ± 0.004 <0.001 Root K03559 24.33 (15) biopolymer transport protein ExbD 0.118 ± 0.003 0.132 ± 0.004 0.008 a Redundant KO term, uncharacterized functions, or functions with p>0.2 were omitted from the “top 10” b PC1 scores for leaf metagenomes and PC2 scores for root metagenomes shown in Figure 5-1 c Mean ± SE

135

Figure 5-1. PCAs for the leaf and root metagenomes indicate clustering of samples taken from asymptomatic and HLB- diseased trees.

136

Figure 5-2. OTU contributions to KO functions that most explained metagenomic differences based on host health-condition (Table 2) for leaf metagenomes (panel A) and root metagenomes (panel B). The OTUs displayed in each figure are those that consistently had the most contributions across all 10 functions shown above.

137

RelativeAbundance(%) RatiosforAsym./Symp. Symptomatic Asymptomatic

Kegg Pathways(L1) KeggPathways(L2)

Cell CommCell Communicationunication *Transpor* Transport and Catabolismt and Catabolism CellularProcesses Cell* GroCellwth Gro andwth Death and Death Cell* MotilityCell Motility *CellularCellular Processes Processes and Signaling and Signaling (Other) *Signaling* Signaling Molecules Molecules and Inter andaction Inter action EnvironmentalInfo.Processing *Signal* SignalTransduction Transduction *MembrMembrane Transporane Transport t *Folding,Folding, Sorting Sor andting Degr andadation Degradation Genetic* Genetic Information Information Processing Processing (Other) GeneticInfo.Processing Transcr* Tiptionranscription *TranslationTranslation *ReplicationReplication and Repair and Repair Biosynthesis* Biosynthesis of Other of SecondarOther Secondary Metabolitesy Metabolites EnzymeEnzyme Families Families *GlycanGlycan Biosynthesis Biosynthesis and Metabolism and Metabolism *MetabolismMetabolism of Other of AminoOther AminoAcids Acids MetabolismMetabolism of Ter penoidsof Terpenoids and P olykandetides Polyketides MetabolismMetabolism (Other) Metabolism * *NucleotideNucleotide Metabolism Metabolism Lipid MetabolismLipid Metabolism Metabolism* Metabolism of Cof actorsof Cof actorsand Vitamins and Vitamins XenobioticsXenobiotics Biodegr Biodegradationadation and Metabolism and Metabolism EnergyEnergy Metabolism Metabolism Carboh* Carbohydrate ydrMetabolismate Metabolism Amino AminoAcid Metabolism Acid Metabolism *Poorly PCharoorlyacter Charizactered ized ) ) ) ) s s s . . . . s t e t e y y o v y y s s m m o v s s o a t m m s s y y R e e o a A A y y o L ( S ( S v R e A A ( ( o f L S S a t f _ _ a t f t o _ _ a R e e o f t o e a o L L o L R a o e o R e o

L R L R Figure 5-3. Heatmap showing the average relative abundances of KEGG Level 2 pathways, ordered from least abundant to most abundant within the broader KEGG Level 1 groups, for the leaf and root metagenomes of asymptomatic and HLB-symptomatic citrus trees. The ratios for average relative abundance within the metagenomes of HLB-symptomatic and asymptomatic trees are illustrated in the red/blue panel on the right. An * indicates significant difference in relative abundances within the metagenomes based on a two- tailed Student’s t-test with FDR correction (q<0.05).

138

KEGG Pathways in Root Communities 2 . 0 . r a v . l p x e 0

. 0 % 1 1

: 3 C 2 . P 0 − 4 . 0 − −0.3 −0.2 −0.1 0.0 0.1 0.2 0.3

PC2: 19% expl.var.

Figure 5-4. PCA showing variation in the root metagenomes (KEGG Pathway Level 3) of C. sinensis trees from the molecular survey (circles) and those from the 18- month time point in the treatment study (triangles).

139

Figure 5-5. Heatmap showing relative abundances of KEGG Level 3 pathways in roots of asymptomatic C. sinensis trees from the molecular survey that had > 0.5% average relative abundances within the metagenome, compared to those same pathways in the root metagenomes of C. sinensis trees at the 18-month time point in the treatment study.

140

CHAPTER 6 CONCLUSIONS AND FUTURE DIRECTIONS

Huanglongbing (HLB) disease continues to rapidly spread and cause significant damages to the global citrus industry. There is a lack of logistically feasible treatment options for the phytopathogen (Liberibacter spp.), and there remains an urgent need for the development of effective strategies to improve citrus crop management in efforts to mitigate the devastating consequences of the disease. Any successful effort to manage

HLB will almost certainly require a holistic, systems approach, where manipulation of the citrus microbiome will be an important component of the management strategy. It is well established that plant-associated microbial communities are actively involved in supporting host health through suppressing phytopathogens, stimulating plant defense pathways, enhancing nutrient mobilization, and promoting resistance to environmental stresses, among other mechanisms (140, 141). The manipulation of plant microbiomes may have the potential to lessen the incidence of plant disease and enhance crop production (142). Thus, the theoretical premise of this work was that before the beneficial potential of citrus-associated microbiota can be exploited for biological control of HLB, it is critical to first define the "healthy" microbiome of the plant.

After improving the methods for deep sequencing characterization of citrus- associated microbial communities, a series of experiments was conducted to test the central hypothesis that key interactions between native microbiota and Liberibacter spp., as well as important associations between host health and the structure and function of the microbiota, play a critical role in HLB establishment and progression.

Building on the importance of these interactions and associations, it was then investigated whether novel antimicrobial treatments (i.e., small molecule compounds)

141

may suppress the HLB pathogen without disrupting microbial diversity. Findings from this work advanced the understanding of several overarching questions regarding microbiota-pathogen-host relationships, which not only have applications for HLB management, but also broader disease control.

1. How can understanding plant microbiota-host associations and microbe-

phytopathogen interactions be applied to control plant disease?

Describing plant microbiomes through DNA sequencing and extensive bioinformatics analyses is a rapidly expanding area of plant biology that has applications for developing strategies to mitigate plant disease (143). Microbe-microbe interactions are known to play a critical role in shaping microbiota structure in both plants and animals, and elucidating interaction patterns may reveal how individual members of the microbiota contribute to the stability of the system and have beneficial or antagonistic effects on host health (144). For example, using a germ-free experimental system, it was shown that a synthetic consortia could inhibit the fungal phytopathogen Fusarium verticilliodes in maize roots and that keystone species within the community mediated this activity (i.e., removing one particular species disrupted the community interactions that were essential for pathogen suppression) (145). In another study, Santhanman et al. (146) demonstrated that an interacting consortium of five bacteria was required to suppress the phytopathogen responsible for wilt disease of tobacco; disease attenuation was dependent on synergistic activity. In this work, by examining year-long co-occurrence patterns between dominant bacterial families and

Liberibacter spp. in leaf microbiota, mutual exclusions between the phytopathogen and three bacterial families (Burkholderiaceae, Xanthomonadaceae, and

142

Micromonosporaceae) that have previously been described to have plant-beneficial properties were uncovered (55, 92, 94, 112). These potentially beneficial bacteria co- occurred with each other and were part of a network of positively-interacting microbiota.

Since Liberibacter spp. was on the periphery of the microbiota network, the taxa that were negatively associated with it may have, to an extent, been involved in mediating the suppressive relationship between the phytopathogen and the stable microbial community. After all, taxa that frequently co-occur with each other within complex networks may have regulatory effects on the system (147). This work and that of others support the notion that understanding the interactive roles of a consortia of microbiota may be essential for promoting plant disease resistance (68, 148).

Indigenous microbiota of leaves and roots that are found to be differentially abundant in healthy and diseased hosts may also have important implications for biological control. This may be especially true for close phylogenetic relatives of phytopathogens, since niche similarities can adversely impact co-existence through competition for space and resource (70). In agreement, several members of

Alphaproteobacteria (i.e., the phylogenetic class of Liberibacter spp.) that were significantly more abundant within the leaf microbiota of asymptomatic trees than HLB- diseased trees (e.g., Methylobacterium, Sphingomonas, and Methylocystaceae) were identified in this work. Previous studies have also suggested Methylobacterium and

Sphingomonas to be involved in protecting host plants from various pathogens (108–

111). Additionally, several members of the root microbiota that were found to be more abundant in asymptomatic trees than diseased trees (e.g., Bradyrhizobiaceae,

Burkholderia, Kaistobacter, and Xanthomonadaceae) were also described to have plant

143

disease suppression-associated properties in other works (94, 114). Several functions

(described in next question) underscore the beneficial potential of these microbiota.

On the other hand, disease-associated microbiota may also affect the host in a negative way by assisting the success of phytopathogens. For example, populations of root microbiota of tomatoes were found to re-assemble following root-knot infection by

Meloidogyne spp. and the succeeding communities encoded an enrichment of functional genes that may have assisted with pathogenesis via roles in breaking down the plant cell wall to allow for phytopathogen nutrient procurement (124). In the present study, several taxa that were more abundant in leaves and roots of diseased citrus trees than asymptomatic trees were also identified (e.g., Comamonadaceae,

Hyphomicrobiaceae, MND1, Steroidobacter). It is unclear whether these disease- associated taxa may have impacted HLB progression and could serve as additional targets for HLB control, or if they were opportunistic colonizers of the microbiomes that may have been re-structuring in response to disease progression having caused changes in the local microenvironments. Nonetheless, the microbiota interactions and associations with citrus health that were identified have interesting implications for biological control and should be further investigated in manipulative studies. For an extensive review on advances in formulating bacterial inoculant strategies for biological control to enhance crop production, see Bashan et al. (149).

2. What functions encoded by microbiota may play critical roles in conferring plant

disease/stress resistance?

The genetics of microbe-plant partnerships that drive microbial colonization and impact plant phenotype (e.g., disease symptom expression) are still at an early stage

144

(148). Native microbiota may “prime” plant immune response pathways by stimulating the production of plant growth promoting and/or plant defense-inducing hormones (84).

For example, using a germ-free experimental system, Zamioudis et al. (123) demonstrated the importance of rhizobacteria, particularly Pseudomonas, in stimulating

Arabidopsis root development and eliciting systemic immune response (i.e., production of ethylene and jasmonic acid). Regarding microbiota support of plants that are under stress, it was recently shown that rhizospheric and endophytic bacteria associated with the halophyte Suaeda salsa encode genes contributing to nutrient solubilization (e.g., phosphatase, nitronate monooxygenase, nitrite reductase) and salt stress acclimatization (e.g., osmoprotectant transport system, ion anti-transporters) (132).

Moreover, gene-probe surveys have demonstrated differential abundances in functional pathways – carbon fixation, nitrogen cycling, phosphorus utilization, and metal homeostasis – in the rhizosphere for HLB-diseased and asymptomatic citrus trees (56).

In agreement, the present study further described metagenomic functions encoded by citrus microbiota to be significantly associated with host health state. Interestingly, the microbiota of HLB-diseased trees were predicted to be enriched in functions involved in cell repair, motility, and, possibly, virulence (e.g., protein folding, DNA replication, glycan biosynthesis, cell growth and death, cell motility, bacterial secretion system), while that of asymptomatic trees were enriched for functions involved in metabolic pathways and cell signaling with the surrounding environment (e.g., signaling molecules and interaction, membrane transport, transport and catabolism, amino acid metabolism, biosynthesis of secondary metabolites, carbohydrate metabolism, nitrogen metabolism).

These fundamental differences suggest that the microbiota of diseased hosts may have

145

been largely combatting environmental stresses, while those of the heathier hosts were likely more efficient at consuming carbon resources and, in turn, producing important metabolic byproducts that could have been involved in supporting a mutualistic relationship with the host. Perhaps focusing on crop health management strategies that select for organisms that encode diverse metabolic functions, which may be beneficial for host interaction, may be important for controlling phytopathogens.

Furthermore, the taxa that predominantly contributed to the microbial functional differences associated with host health were identified. For example, few distinct organisms in leaves (i.e., Methylobacterium, and Sphingomonas) and roots (i.e.,

Catellatospora, Streptomyces, Bradyrhizobiaceae, and Burkholderiaceae) accounted for the enriched functions in communities of asymptomatic trees. It is notable that while some of these organisms were also associated with plant heath based on taxonomic assessment alone (described in previous question), the functional analyses revealed additional taxa that may also play key roles as well. In other words, based on functional capabilities, taxa that may not be dominant in microbial communities can have disproportional effects on the system. Since it is becoming recognized that functional profiles are inherently more stable and contain less noise than taxonomic profiles (i.e., genomes of different taxa may encode many similar functions), characterizing microbial communities by the former may be more important for determining the biological importance of samples in terms of host health (74). Future studies are needed to test the importance of the key functional pathways, in terms of gene expression in the associated taxa, that this work identified to further establish the links between microbiome functions and host health.

146

3. Are changes in microbiota structure a precursor to disease establishment or a

response to disease consequences?

A growing number of studies have focused on understanding how plant microbiome composition and diversity shift during disease development (37, 49–56).

While results from some studies have suggested that microbial communities may undergo perturbation during environmental change or disease progression, one may alternatively argue that the disease itself could be a consequence of complex changes to the communities. For example, a diverse rhizosphere community of beets that had included potentially beneficial bacterial groups (i.e., Pseudomonadaceae,

Burkholderiaceae, Xanthomonadales, and Lactobacillaceae) was linked to suppressing the development of root rot caused by Rhizoctonia (94). In addition, several plant diseases are known to be caused, or at least enhanced, by synergistic interactions among multiple bacteria taxa (e.g., tomato pith necrosis, mulberry wilt, broccoli head rot), which could become prevalent within the microbial community if the taxa involved are introduced and selected for in a changing environment (95). In the present study, the spatiotemporal variation in the relative abundance of Liberibacter spp. in leaf microbiota of citrus trees was negatively correlated with microbiota alpha diversity and positively associated with HLB symptom progression. Since the citrus trees sampled ubiquitously contained Liberibacter spp. (i.e., the pathogen was present at lower abundances in asymptomatic trees), although it can be postulated that some changes in citrus microbiota diversity that coincided with disease progression may have been a response to physiological changes to the host, it is likely possible that microbiota composition and diversity may have, to an extent, conferred resistance to development

147

of HLB disease. This “microbiota protection hypothesis” may, at least partially, explain why HLB has spread throughout all major citrus-producing areas worldwide except for in the Mediterranean Basin or Australia (24). That is, perhaps the native microbiota may be involved in host resistance, although this hypothesis remains to be robustly tested.

Collectively, this work and that of others suggest that plant disease establishment may be associated with changes in the microbiome that involve suppression of key members of the native community. Distinguishing the changes within plant-associated microbial communities that may support disease progression from those that are associated with consequences of disease symptom development is an interesting avenue for future research.

4. What is the importance of core microbiota within broader host-associated microbial

communities?

Commonly occurring organisms across similar microbial communities comprise a core microbial community that is hypothesized to play key roles in ecosystem functioning within that type of microbial habitat (36, 71). While numerous deep sequencing studies have revealed thousands of bacterial OTUs to comprise plant microbiomes, there is typically a small number of taxa that dominate the broader community (37, 51, 72, 73, 88–92). Some of the highly abundant taxa within these studies are noticeably conserved across microbiomes of similar plant species, even under variable experimental conditions. This suggests that a core microbial community forms stable associations with particular hosts across temporal and geographic scales, which may be attributable to dependencies that have developed between the co- evolving microbiome and host. However, given the limited number of studies that have

148

discerned the core members of plant-associated microbial communities (51, 72, 73), much remains unknown about the structure of the core community among all microbiota and its importance with regard to plant health.

In the present study, the core citrus leaf and root microbiota (i.e., detected in at least 95% of respective leaf or root samples) were defined at all taxonomic ranks across a number of variables that have been described to impact microbiota structure (i.e., location, cultivar, disease symptom severity, time/season) (51, 72, 73, 91, 93), suggests substantial stability in the core microbe-host relationships. As expected, the ratios of numbers of core members to all members of the microbiota consistently decreased as taxonomic rank became finer, demonstrating phylogenetic redundancy (i.e., it was common for non-core taxa at lower taxonomic ranks to comprise broader core groups).

Regarding community structure, the taxonomically rich communities contained abundant core members, some overrepresented site-specific members, and a diverse community of low-abundance variable taxa. At the genus level (i.e., the lowest level of taxonomic assignment in this work), about 90% of microbial abundance within the microbial communities was accounted by 6% and 12% of the taxa detected in leaves and roots; these small fractions were pre-dominantly comprised of core microbiota. In other words, although the core members comprised a relatively small fraction of the total microbial taxa, these organisms made up the majority of microbiota abundance, which is consistent with previous works (51, 72, 73). In addition to core taxa, while certain KEGG functional pathways were differentially abundant in healthy and diseased hosts, the rank abundances of these functions in the respective leaf and root metagenomes were highly conserved, suggesting a core functional profile. The overall

149

importance of core microbiota is underscored by the fact that most of the core members that were found to (i) be associated with host health state, (ii) part of the interaction network that was mutually exclusive to the phytopathogen, and/or (iii) contribute to the key functions within the microbiota (all described earlier) were core members of the microbiota. Since plant microbiota are taxonomically and functionally diverse, the focus of microbiome research may need to be refined around the relatively stable taxa (i.e., core taxa) that might have a greater likelihood to influence host phenotype in order to connect culture-independent studies to the development of effective biological control

(148).

5. How do antimicrobial treatments impact plant microbiomes and what are the

implications for environmental and human health?

Antimicrobials have played an important role in plant agriculture for controlling a variety of bacterial phytopathogens (e.g., Erwinia, Pectobacterium, Pseudomonas,

Xanthomonas) (31, 42). In 2016, streptomycin sulfate, oxytetracycline hydrochloride, and oxytetracycline calcium complex were approved for use as foliar sprays to treat

HLB in Florida, USA; however, the potential benefits of these compounds are still unclear (24). In several recent field studies, applications of broad-range antimicrobials, including aminoglycosides (streptomycin, kasugamycin), beta-lactams (penicillin), and/or tetracyclines (oxytetracycline), to HLB-infected trees were reported to have little success at slowing disease progression (19–22). Ongoing challenges and consequences include seasonality of symptoms and resiliency of Liberibacter spp. populations, potential phytotoxicity, limitations for therapeutics to contact the

150

phytopathogen in planta, adverse impacts on plant-beneficial microbiota, and potential implications on public and ecosystem health (19–22, 119).

To our knowledge, this is the first work to demonstrate antimicrobial efficacy against Liberibacter spp. without impacting overall microbiota structure. In previous studies, antibiotic applications (e.g., sulfonamides, streptomycin, kasugamycin, oxytetracycline, ampicillin, gentamicin) were reported to disrupt microbiota and/or induce shifts in microbial richness and diversity, a consequence of their broad-spectrum activity (15, 18, 20, 38). Since our separate field study concluded that the diversity of citrus-associated microbial communities is strongly associated with tree health and HLB symptom progression, it is possible that these subsequent impacts of broad-spectrum antimicrobials could have negatively affected the efficacy of those treatments.

Alternatively, in this work, both of the small molecule compounds that were tested suppressed Liberibacter spp. in root microbiota, causing net reductions in many trees over time, without impacting microbial diversity. Based on metagenome predictions, the treatments also caused the root microbiota functional profiles to shift towards a

“healthier” state (i.e., the functional profiles of asymptomatic trees, as described earlier, were more similar to the HLB-diseased trees that received treatment than diseased controls). Moreover, while several broad-spectrum antimicrobials have previously been shown to have phytotoxic effects (Hu 2016; Zhang 2014), that was not an issue with the therapeutics used here. However, despite the beneficial effects, the phytopathogen remained abundant in leaves even after treatment and the severely symptomatic trees never quite recovered. Perhaps the antimicrobial delivery method needs to be optimized to improve the transport through the canopy (i.e., use pressurized injections or foliar

151

sprays). Since treatments are likely most effective if performed at the early onset of phytopathogen infection (13), it is possible that the small molecules could have higher efficacy against HLB in younger or less severely-diseased trees. Due to their target specificity at the field-scale, testing the usage of these compounds in combination or alongside biological control treatment could provide new insights for sustainable HLB management. While judicious use of narrow-spectrum antimicrobials (e.g., small molecules used here) may be desirable for broadening the toolset of pro-active solutions for managing Liberibacter spp., the use of antibiotics that are closely related to those used in human and veterinary medicine (e.g., streptomycin, oxytetracycline, and other broad-spectrum compounds) needs to be closely evaluated for the balance of potential benefits and also risks to public health (29, 30).

6. What microbiota transitions are associated with phytopathogen removal and what

are the implications for controlling plant disease?

The colonization, establishment, and survival or microorganisms in host- associated communities are largely influenced by selective pressures imposed by biotic and abiotic surroundings, including microbe-microbe interactions (e.g., metabolite production), host-associated factors (e.g., plant root exudates), and environmental stresses (e.g., tolerance to desiccation in arid settings). As seen with our data and in that of other studies, plant disease development is associated with shifts in microbiota diversity (37, 49–56), which may reflect outcomes of phytopathogen interactions and/or responses to host phenotypic changes. Likewise, microbiota transitions that occur during phytopathogen removal (e.g., via treatment or through mechanisms involved in plant defense pathways or microbiota suppression) are, at least partially, dependent on

152

the specific role of the organism within the complex microbiome network. For example, if an organism at the center of a microbial network is eliminated, there should be cascading effects throughout the microbial community. This has been demonstrated by removing the fungal pathogen Albugo from the phyllosphere of Arabidopsis (150). In the present study, it was determined that Liberibacter spp. was on the periphery of the citrus leaf microbiota network and that its interactions with three bacteria may mediate the inverse relationship between the phytopathogen and microbiota diversity. Therefore, its removal from citrus microbial communities should not be expected to greatly affect overall community diversity, which was supported in our findings regarding the target- specific removal of Liberibacter spp. from citrus roots. Although alpha diversity did not significantly change, there were several members of the microbiota that positively responded to phytopathogen removal (i.e., Coxiellaceae, Oxalobacteriaceae,

Rhodospirillaceae, and MLE-12). Interestingly, each of these bacterial families was also identified as core members of the root microbiota, which suggests that microorganisms that colonize the host during phytopathogen removal are more likely to be those that stably associate with the host rather than those that may be site-specific or recently introduced. Furthermore, within core microbiota, plant-beneficial members may have an advantage during succession following phytopathogen removal. For example,

Coxiellaceae has been suggested to be a plant-beneficial indicator (121), and members of Oxalobacteriaceae, which was part of the network of microbiota that was mutually exclusive with Liberibacter spp. (described above), and Rhodospirillaceae have been previously associated with citrus root health (55, 56). Taxa that have the potential to replace phytopathogens within host-associated microbial communities have implications

153

for the development of effective strategies for biological control, at least as members of synthetic communities that may be able to combat or prevent disease.

Overall, these studies (i) identified plant-beneficial microbiota and microbial interactions that have implications for biological control of HLB, and (ii) demonstrated that the novel treatments tested were target-specific against Liberibacter spp., warranting further investigation. This work should be extended to advance the understanding of how the beneficial potential of plant microbiomes may be able to be harnessed to mitigate disease and improve the reliability and environmental impacts of crop production.

154

APPENDIX A BIOINFORMATICS PIPELINE

Pre-processing

De-multiplexed reads were cleaned for quality control and paired end reads were joined. Specifically, the raw sequenced reads were processed with Cutadapt

(https://github.com/marcelm/cutadapt) and Sickle (https://github.com/najoshi/sickle) to remove any residual Illumina adapters and primer sequences, truncate sequences at the first N position, trim sequences at a bp with a phred score below 30, and remove reads that were shorter than 120 bp. Paired-end reads were joined with Eautils

(https://github.com/ExpressionAnalysis/ea-utils) with the requirements of having a minimum overlap of 30 bp and a 3% maximum difference in the overlap region. Sample names were added to the definition lines of sequencing reads using the sed command and concatenated into one fasta file in order to make them compatible for analysis in

QIIME v.1.8 (101). Towards this end, the following job was submitted to HiPer-Gator 2.0

(i.e., the University of Florida research computing cluster); inputs were .fastq files for each sample from Illumina sequencing:

###########

#!/bin/bash #SBATCH --job-name=Seq_QC #SBATCH --output=Seq_QC.log #SBATCH --error=Seq_QC.err #SBATCH --mail-type=BEGIN,END,FAIL #SBATCH --mail-user [email protected] #SBATCH --nodes 1 #SBATCH --mem-per-cpu=10gb #SBATCH --time=48:00:00 # pwd; hostname; date cd $SLURM_SUBMIT_DIR

# Unzip files gunzip *fastq.gz

155

########### Cutadapt: remove adapters and trim reads for quality score under 30

# load module module load cutadapt

# make directory for cutadapt output files mkdir cutadapt

# create file names for cutadapt files_cut=`ls | grep "R1.fastq"`

# use loop to run cutadapt function on all original pairs; send output to cut adapt folder for R1 in $files_cut do R1_cut=`echo $R1 | cut -d R -f1`R1_cut.fastq R2=`echo $R1 | cut -d R -f1`R2.fastq R2_cut=`echo $R1 | cut -d R -f1`R2_cut.fastq cutadapt -a AGATCGGAAGAGCACACGTCTGAACTCCAGTCAC -A AGATCGGAAGAGCGTCGTGTAGGGAAAGAGTGTAGATCTCGGTGGTCGCCGTATCATT -q 30 -o cutadapt/$R1_cut -p cutadapt/$R2_cut $R1 $R2 done

# end in cutadapt folder cd /ufrc/teplitski/rblauste/Citrus/SeqD/cutadapt

########### Sickle: trim paired end reads on each read at bp that has quality falling below 30. Remove reads that are shorter than 120 bp. Truncate sequences at first N bp position

# load module module load sickle

# make directory for sickle output files mkdir sickle

# change directory to sickle directory cd /ufrc/teplitski/rblauste/Citrus/SeqD/cutadapt/sickle

# make directory for sickle singles files mkdir singles

# change directory back to cutadapt files cd /ufrc/teplitski/rblauste/Citrus/SeqD/cutadapt/

# create file names for sickle files_sickle=`ls | grep "R1_cut.fastq"`

# use loop to run sickle function on all cut pairs; send sickle output to sickle folder and singles output to singles folder for R1_cut in $files_sickle do R1_sickle=`echo $R1_cut | cut -d R -f1`R1_sickle.fastq R2_cut=`echo $R1_cut | cut -d R -f1`R2_cut.fastq R2_sickle=`echo $R1_cut | cut -d R -f1`R2_sickle.fastq singles=`echo $R1_cut | cut -d R -f1`_singles.fastq sickle pe -f $R1_cut -r $R2_cut -t sanger -o sickle/$R1_sickle -p sickle/$R2_sickle -s sickle/singles/$singles -n -q 30 -l 120

156

done

# end in sickle folder cd /ufrc/teplitski/rblauste/Citrus/SeqD/cutadapt/sickle

########## Join paired end reads: allow a 3% maximum difference and assure the minimum overall overlap is 30 bp

# load module module load eautils

# make directory for joined output files mkdir join

# create file names for joined files files_join=`ls | grep "R1_sickle.fastq"`

# use loop to run fatstq-join function on all sickle pairs for R1_sickle in $files_join do R2_sickle=`echo $R1_sickle | cut -d R -f1`R2_sickle.fastq un1=`echo $R1_sickle | cut -d R -f1`_un1.fastq un2=`echo $R1_sickle | cut -d R -f1`_un2.fastq join=`echo $R1_sickle | cut -d R -f1`_join.fastq fastq-join -p 3 -m 30 $R1_sickle $R2_sickle -o join/$un1 -o join/$un2 -o join/$join done

# move to join folder cd /ufrc/teplitski/rblauste/Citrus/SeqD/cutadapt/sickle/join

# remove un1/un2 files rm *un*

########## Remove primer sequences, if present: min length of 230 and max length of 255; max allowed error rate of 3% due to primer sequence discrepancies

# load module module load cutadapt

# make directory for noprimer output files mkdir noprimers

# create file names for noprimers files files_noprimers=`ls | grep "_join.fastq"`

# use loop to run cutadapt function (remove primers) on all join files for _join in $files_noprimers do noprimers=`echo $_join | cut -d j -f1`noprimers.fastq cutadapt -m 230 -M 255 -e 0.03 -a GGACTACNVGGGTWTCTAAT -g GTGYCAGCMGCCGCGGTAA $_join > noprimers/$noprimers done

# move to noprimers folder cd /ufrc/teplitski/rblauste/Citrus/SeqD/cutadapt/sickle/join/noprimers

157

########## Convert files to fasta format module load fastx_toolkit/0.0.13.2

# make directory for fasta output files mkdir fasta

# create file names for fasta files files_fasta=`ls | grep "_noprimers.fastq"`

# use loop to run convert no primers fastq files to fasta files for _noprimers in $files_fasta do fasta=`echo $_noprimers | cut -d "_" -f1`.fasta fastq_to_fasta -Q33 -i $_noprimers -o fasta/$fasta done

########## Make files compatible with qiime cd /ufrc/teplitski/rblauste/Citrus/SeqD/cutadapt/sickle/join/noprimers/fasta for file in *.fasta ; do name=$(echo "$file" | cut -d"." -f1); cat $file | perl -lane ' if(/^>.*\s(.*)/){s/$1/orig_bc=ATCACG new_bac=ATCACG bc_diffs=0/g;print "$_";}else{print "$_";}' | sed s/://g | perl -lane 'if(/^>([A-Z|0-9]+)(\-).*/){s/$1/'$name'/g;s/\-/ /g; print "$_";}else{print "$_\n";}' | sed /^$/d | awk '/^>/{$1=$1"_"(++i)}1' > $name.modified; done cat *.modified > All_forQiime.fasta

##########

Taxonomic Assignments with QIIME

Pre-processed reads were given taxonomic assignments in QIIME (77). In the data analyses for chapter 3 of this dissertation, open-reference OTU picking was performed. This method was chosen in attempt to provide the deepest classification of citrus-associated microbial communities. In data analyses for chapters 4-5 of this dissertation, closed reference OTU picking was performed. This method was chosen because it was required for making metagenomic predictions from taxonomic assignments in PICRUSt (i.e., all reads must be “assigned”) (83).

Open-reference OTU Picking

Clustering of OTUs at 97% similarity, with no removal of singletons, was performed in QIIME with the open-reference OTU picking method (78), and taxonomic

158

assignments were made by mapping to the Greengenes reference database version

13.8 (79). Unassigned OTUs and those that were identified as 16S mitochondrial or plastid DNA were removed from further analyses. Towards this end, the following job was submitted to HiPer-Gator 2.0; inputs were the .fasta file of the concatenated sequence data for all samples and a (mapping) .txt file that contained sample metadata, per the requirements of QIIME (77):

##########

#!/bin/bash #SBATCH --job-name=Open_qiime #SBATCH --output=Open_qiime.log #SBATCH --error=Open_qiime.err #SBATCH --mail-type=BEGIN,END,FAIL #SBATCH --mail-user [email protected] #SBATCH --nodes 1 #SBATCH --mem-per-cpu=50gb #SBATCH --time=72:00:00 # pwd; hostname; date cd $SLURM_SUBMIT_DIR

#I have added a module purge here as qiime is a wraparound for a lot of other scripts. If you already have that module open and it is a different version of the tool, qiime gets confused. module purge module load qiime

#open reference otu clustering and taxonomy assignment using the greengenes 13.8 database, with no elimination of singletons pick_open_reference_otus.py -i All_forQiime.fasta -r /ufrc/data/reference/qiime/gg/gg_otus-13_8- release/rep_set/97_otus.fasta -o All/ --prefilter_percent_id 0.0 --min_otu_size 1 biom summarize-table -i All/otu_table_mc1_w_tax_no_pynast_failures.biom -o ABs_stats.txt

#remove chloroplast and mitochondria sequences from your dataset filter_taxa_from_otu_table.py -i All/otu_table_mc1_w_tax_no_pynast_failures.biom -o All.biom -n c__Chloroplast,f__mitochondria

#this makes spreadsheets that you can open in excel and do quick plotting, one for each level of taxonomy (class, genus, etc) summarize_taxa_through_plots.py -i All/otu_table_mc1_w_tax_no_pynast_failures.biom -m All_mapping.txt -o All_plots/

#copy the otu table into something you can open in excel, you may need to reference this later biom convert -i All/otu_table_mc1_w_tax_no_pynast_failures.biom -o ABs_otu_table.txt --table- type="OTU table" --to-tsv --header-key taxonomy

159

#add metadata (location, sample types, etc) biom add-metadata -i All.biom -o Allnew.biom -m All_mapping.txt

#slight format change for phyloseq (program for microbiome data analyses in R) biom convert -i Allnew.biom -o All_phyloseq.biom --table-type="OTU table" --to-json

##########

Closed-reference OTU Picking

The same methods described above, except for opting to perform clustering in

QIIME with the closed-reference method and mapping to the Greengenes reference database version 13.5 (79), were used. This version of the database was chosen as it is a requirement for taxonomic data compatibility with PICRUSt (83). The following modifications to the above job were used:

##########

#Pick a closed reference OTU table from concatenated 16S sequences (set OTU table in a folder that can be accessed for subsequent picrust analyses) pick_closed_reference_otus.py -i All_forQiime.fasta -o PICRUSt/OTU_table/ -r /ufrc/teplitski/rblauste/gg13_5/gg_13_5_otus/rep_set/97_otus.fasta -t /ufrc/teplitski/rblauste/gg13_5/gg_13_5_otus/taxonomy/97_otu_taxonomy.txt

#remove contaminant sequences from your dataset (note that there are no “unassigned sequences here) filter_taxa_from_otu_table.py -i PICRUSt/OTU_table/otu_table.biom –o PICRUSt/OTU_table/otu_table_clean.biom -n c__Chloroplast,f__mitochondria

# Convert to classic (tab-delimited) format for picrust compatibility biom convert -i PICRUSt/OTU_table/otu_table_clean.biom -o PICRUSt/OTU_table/otu_table_clean.txt -- to-tsv

##########

Metagenomic Predictions with PICRUSt

In chapter 5 of this dissertation, PICRUSt v.1.1.0 (83) was used to predict the

Kyoto Encyclopedia of Genes and Genomes (KEGG) Orthology (KO) (134) functions for metagenomes from the 16S rRNA gene taxonomic data obtained by closed-reference

OTU picking. For quality control, the weighted Nearest Sequenced Taxon Index (NSTI) value was computed for each sample. The KO predictions were collapsed into broader

160

KEGG pathways and the KO and KEGG pathway counts were converted to relative abundances for further analyses. Towards this end, the following job was submitted to

HiPer-Gator 2.0; inputs were the OTU .biom file and the PICRUSt table files downloaded from :

##########

#!/bin/bash #SBATCH --job-name=picrust_predict #SBATCH --output=picrust_predict.log #SBATCH --error=picrust_predict.err #SBATCH --mail-type=BEGIN,END,FAIL #SBATCH --mail-user [email protected] #SBATCH --nodes 1 #SBATCH --mem-per-cpu=50gb #SBATCH --time=48:00:00 # pwd; hostname; date cd $SLURM_SUBMIT_DIR module purge module load picrust

# Normalize OTUs by copy number normalize_by_copy_number.py -f -i OTU_table/otu_table_clean.txt -o OTU_table/normalized_otus.biom - c /ufrc/data/reference/picrust/16S_13_5_precalculated.tab.gz

# Make metagenome prediction folder mkdir metagenome_prediction

#Predict Metagenomes by KO pathways (obtain NSTI values as well) predict_metagenomes.py -f -i normalized_otus.biom -o NSTI/metagenome_predictionKO_NSTI.txt -a NSTI/nsti_per_sample.tab -c /ufrc/data/reference/picrust/ko_13_5_precalculated.tab.gz

#Collapse KO pathways categorize_by_function.py -i metagenome_prediction/predict_KO.biom -c KEGG_Pathways -l 3 -o metagenome_prediction/predict_KO.L3.biom categorize_by_function.py -f -i metagenome_prediction/predict_KO.biom -c KEGG_Pathways -l 3 -o metagenome_prediction/predict_KO.L3.txt categorize_by_function.py -i metagenome_prediction/predict_KO.biom -c KEGG_Pathways -l 2 -o metagenome_prediction/predict_KO.L2.biom categorize_by_function.py -f -i metagenome_prediction/predict_KO.biom -c KEGG_Pathways -l 2 -o metagenome_prediction/predict_KO.L2.txt categorize_by_function.py -i metagenome_prediction/predict_KO.biom -c KEGG_Pathways -l 1 -o metagenome_prediction/predict_KO.L1.biom categorize_by_function.py -f -i metagenome_prediction/predict_KO.biom -c KEGG_Pathways -l 1 -o metagenome_prediction/predict_KO.L1.txt

161

##########

Additionally, the OTU contributions for several key KO functions (described

Chapter 5 of this dissertation) were determined with the following job; the input was the normalized OTU .biom file:

##########

#!/bin/bash #SBATCH --job-name=picrust_contributions #SBATCH --output=picrust_contributions.log #SBATCH --error=picrust_contributions.err #SBATCH --mail-type=BEGIN,END,FAIL #SBATCH --mail-user [email protected] #SBATCH --nodes 1 #SBATCH --mem-per-cpu=50gb #SBATCH --time=48:00:00 # pwd; hostname; date cd $SLURM_SUBMIT_DIR module purge module load picrust

#predict OTU contributions to a desired KO function (i.e., replace “XXXX”, “YYYY”, “ZZZZ”, etc. with the corresponding KO function number) metagenome_contributions.py -i normalized_otus.biom -l K0XXXX,K0YYYY,K0ZZZZ -o NewContributions/ko_metagenome_contributions.tab -c /ufrc/data/reference/picrust/ko_13_5_precalculated.tab.gz

##########

162

APPENDIX B SUPPLEMENTAL MATERIAL FOR CHAPTER 3

Table B-1. Pairwise comparisons of leaf microbiota from Valencia trees sampled in Immokalee (n=5), Vero Beach (n=4), Ft. Pierce (n=14), and Gainesville (n=8) in Fall 2015. The R statistic for each ANOSIM test is shown above the x-line and the statistical significance based on 999 permutations is shown below the x-line. An R statistic of 0 means the communities are identical; whereas, R of 1 means the communities have no overlap. Location Immokalee Vero Beach Fort Pierce Gainesville Immokalee xxx 0.513 0.749 0.387 Vero Beach 0.034 xxx 0.871 0.555 Fort Pierce 0.002 0.001 xxx 0.732 Gainesville 0.011 0.003 0.001 xxx

163

Quincy

Gainesville

Vero Beach

Ft. Pierce

Immokalee

Figure B-1. Locations where the citrus trees were sampled. The image template was obtained from the Hydrographic Internet Map Service Data Library (USGS. 2009. Florida Counties. . accessed on 12/8/17). The city labels were overlaid in Microsoft PowerPoint (version 15.11.2) based on their approximate locations identified using Map data ©2016 Google.

164

Figure B-2. Each set of replicate trees was categorized based on HLB symptom severity. Photographs illustrate the four categories: asymptomatic (i.e., no observable HLB symptoms), symptomatic-mild (i.e., some blotchy mottle and pointed leaves observed), symptomatic-moderate (i.e., blotchy mottle and some chlorosis/yellowing of leaves observed), and symptomatic-severe (i.e., thinning of canopy, severe chlorosis/yellowing of leaves, dead branches observed). I: Owari tree at the North Florida Research and Education Center in Quincy, FL (NF-REC) (10/20/15). II: Navel tree at the Southwest Florida Research Education Center in Immokalee, FL (SF-REC) (11/3/15). III: Valencia tree at the SF-REC (11/3/15). IV: Valencia tree at the University of Florida IFAS Research Center in Gainesville, FL (4/16/15).

165

s s r I-A. Core-taxa threshold r I-B. e e d d r r O O

f f o o

r r e e b b m m u u N N

Number of Root Microbial Communities Number of Leaf Microbial Communities s s e e

i II-A. i II-B. l l i i m m a a F F

f f o o

r r e e b b m m u u N N

Number of Root Microbial Communities Number of Leaf Microbial Communities a a r r

e III-A. e III-B. n n e e G G

f f o o

r r e e b b m m u u N N

Number of Root Microbial Communities Number of Leaf Microbial Communities

Figure B-3. Frequency of orders (I), families (II), and genera (III) assigned to 16S sequences detected in the leaf samples (A) (n=94) and root samples (B) (n=79). Points on the right of the dotted lines represent the core fraction of the microbiomes (i.e., present in >95% of respective samples (102)).

166

Figure B-4. Photographs that demonstrate progression of HLB symptoms over time (e.g., branch die-back in canopy). Two Valencia trees at the University of Florida IFAS Research Center in Gainesville, FL during Spring 2015 and Winter 2016; the same trees are pictured in each row.

167

R2 = 0.593 R2 = 0.934 p < 0.001 p < 0.001 e e r r u u s s a a e e M M

n n o o n s n p a m i h S S

Liberibacter spp. Relative Abundance (%) Liberibacter spp. Relative Abundance (%)

Figure B-5. Correlation between the relative abundance of Liberibacter spp. among all genera in leaf-associated microbial communities and community alpha diversity (A: Shannon measure; B: Simpson measure) (n=94). The R2 and p- value from each regression are listed.

168

Figure B-6. Relative abundances of the 20 most abundant families assigned to 16S sequences detected in the root-associated microbial communities of Valencia trees sampled in Gainesville, FL on 4/1/15, 6/1/15, 9/23/15, and 3/29/16. All listed taxa were core root microbiota (i.e., detected in >95% of all root samples in the survey). The * indicates significant difference (p<0.05) for different dates based on ANOVA.

169

LIST OF REFERENCES

1. Bové JM. 2006. Huanglongbing: A destructive, newly-emerging, century-old disease of citrus. J Plant Pathol 88:7–37.

2. Jagoueix S, Bové JM, Garnier M. 1996. PCR detection of the two “Candidatus” liberobacter species associated with greening disease of citrus. Mol Cell Probes 10:43–50.

3. do Carmo Teixeira D, Saillard C, Eveillard S, Danet JL, da Costa PI, Ayres AJ, Bové J. 2005. “Candidatus Liberibacter americanus”, associated with citrus huanglongbing (greening disease) in São Paulo State, Brazil. Int J Syst Evol Microbiol 55:1857–1862.

4. Martinelli F, Dandekar AM. 2017. Genetic mechanisms of the devious intruder Candidatus Liberibacter in Citrus. Front Plant Sci 8:904.

5. Folimonova SY, Robertson CJ, Garnsey SM, Gowda S, Dawson WO. 2009. Examination of the responses of different genotypes of citrus to huanglongbing (citrus greening) under different conditions. Phytopathology 99:1346–1354.

6. Gottwald TR, Graham JH, Irey MS, McCollum TG, Wood BW. 2012. Inconsequential effect of nutritional treatments on huanglongbing control, fruit quality, bacterial titer and disease progress. Crop Prot 36:73–82.

7. Hodges AW, Spreen TH. 2012. Economic impacts of citrus greening (HLB) in Florida. IFAS Ext FE903.

8. Croxton SD, Stansly PA. 2014. Metalized polyethylene mulch to repel Asian citrus psyllid, slow spread of huanglongbing and improve growth of new citrus plantings. Pest Manag Sci 70:318–323.

9. Liu Y, Heying E, Tanumihardjo SA. 2012. History, global distribution, and nutritional importance of citrus fruits. Compr Rev Food Sci Food Saf 11:530–545.

10. Grafton-Cardwell EE, Stelinski LL, Stansly PA. 2013. Biology and management of Asian citrus psyllid, vector of the huanglongbing pathogens. Annu Rev Entomol 58:413–32.

11. Canales E, Coll Y, Hernandez I, Portieles R, Rodriguez Garcia M, Lopez Y, Aranguren M, Alonso E, Delgado R, Luis M, Batista L, Paredes C, Rodriguez M, Pujol M, Ochagavia ME, Falcon V, Terauchi R, Matsumura H, Ayra-Pardo C, Llauger R, del Carmen Perez M, Nunez M, Borrusch MS, Walton JD, Silva Y, Pimentel E, Borroto C, Borras-Hidalgo O. 2016. “Candidatus Liberibacter asiaticus”, causal agent of citrus huanglongbing, is reduced by treatment with brassinosteroids. PLoS One 11: e0146223.

170

12. Li J, Trivedi P, Wang N. 2016. Field evaluation of plant defense inducers for the control of citrus huanglongbing. Phytopathology 106:37–46.

13. Martinelli F, Dolan D, Fileccia V, Reagan RL, Phu M, Spann TM, McCollum TG, Dandekar AM. 2016. Molecular responses to small regulating molecules against huanglongbing disease. PLoS One 11:e0159610.

14. Duan Y, Zhou L, Hall DG, Li W, Doddapaneni H, Lin H, Liu L, Vahling CM, Gabriel DW, Williams KP, Dickerman A, Sun Y, Gottwald T. 2009. Complete genome sequence of citrus huanglongbing bacterium, “Candidatus Liberibacter asiaticus” obtained through metagenomics. Mol Plant Microbe Interact 22:1011– 1020.

15. Zhang MQ, Guo Y, Powell CA, Doud MS, Yang CY, Zhou H, Duan YP. 2016. Zinc treatment increases the titre of “Candidatus Liberibacter asiaticus” in huanglongbing-affected citrus plants while affecting the bacterial microbiomes. J Appl Microbiol 120:1616–1628.

16. Stansly PA, Arevalo HA, Qureshi JA, Jones MM, Hendricks K, Roberts PD, Roka FM. 2014. Vector control and foliar nutrition to maintain economic sustainability of bearing citrus in Florida groves affected by huanglongbing. Pest Manag Sci 70:415–426.

17. Gardner CL, Pagliai FA, Pan L, Bojilova L, Torino MI, Lorca GL, Gonzalez CF. 2016. Drug repurposing: Tolfenamic acid inactivates PrbP, a transcriptional accessory protein in Liberibacter asiaticus. Front Microbiol 7:1630.

18. Yang C, Powell CA, Duan Y, Shatters R, Fang J, Zhang M. 2016. Deciphering the bacterial microbiome in huanglongbing-affected citrus treated with thermotherapy and sulfonamide antibiotics. PLoS One 11: e0155472.

19. Zhang M, Powell CA, Zhou L, He Z, Stover E, Duan Y. 2011. Chemical compounds effective against the citrus huanglongbing bacterium “Candidatus Liberibacter asiaticus” in planta. Phytopathology 101:1097–1103.

20. Zhang M, Powell CA, Guo Y, Benyon L, Duan Y. 2013. Characterization of the microbial community structure in Candidatus Liberibacter asiaticus-infected citrus plants treated with antibiotics in the field. BMC Microbiol 13:112.

21. Shin K, Ascunce MS, Narouei-Khandan HA, Sun X, Jones D, Kolawole OO, Goss EM, van Bruggen AHC. 2016. Effects and side effects of penicillin injection in huanglongbing affected grapefruit trees. Crop Prot 90:106–116.

22. Hu J, Wang N. 2016. Evaluation of the spatiotemporal dynamics of oxytetracycline and Its control effect against citrus huanglongbing via trunk injection. Phytopathology 106:1495–1503.

171

23. Ledford H. 2017. Engineered virus in line to battle citrus disease old tumours offer rare cancer clues. Nature 545:2004–2005.

24. Wang N, Pierson EA, Setubal C, Xu J, Levy JG, Zhang Y, Li J, Rangel LT, Martins Jr. JM. 2017. The Candidatus Liberibacter–Host Interface : Insights into Pathogenesis Mechanisms and Disease Control. Annu Rev Phytopathol 55:451– 482.

25. Zheng Z, Bao M, Wu F, Chen J, Deng X. 2016. Predominance of single prophage carrying a CRISPR/cas system in “Candidatus Liberibacter asiaticus” strains in southern China. PLoS One 11:e0146422.

26. Jia H, Zhang Y, Orbovic V, Xu J, White FF, Jones JB, Wang N. 2017. Genome editing of the disease susceptibility gene CsLOB1 in citrus confers resistance to citrus canker. J Plant Biotechnol 15: 817–823.

27. Akula N, Trivedi P, Han FQ, Wang N. 2012. Identification of small molecule inhibitors against SecA of Candidatus Liberibacter asiaticus by structure based design. Eur J Med Chem 54:919–924.

28. Pagliai FA, Gardner CL, Bojilova L, Sarnegrim A, Tamayo C, Potts AH, Teplitski M, Folimonova SY, Gonzalez CF, Lorca GL. 2014. The transcriptional activator LdtR from “Candidatus Liberibacter asiaticus” mediates osmotic stress tolerance. PLoS Pathogens 10: e1004101.

29. Chang Q, Wang W, Regev-Yochay G, Lipsitch M, Hanage WP. 2015. Antibiotics in agriculture and the risk to human health: How worried should we be? Evol Appl 8:240–247.

30. Franklin AM, Aga DS, Cytryn E, Durso LM, McLain JE, Pruden A, Roberts MC, Rothrock MJ, Snow DD, Watson JE, Dungan RS. 2016. Antibiotics in agroecosystems: Introduction to the special section. J Environ Qual 45:377–393.

31. Stockwell VO, Duffy B. 2012. Use of antibiotics in plant agriculture. Rev sci tech Off int Epiz 31:199–210.

32. Wang N, Stelinski LL, Pelz-Stelinski KS, Graham JH, Zhang Y. 2017. Tale of the huanglongbing disease pyramid in the context of the citrus microbiome. Phytopathology 107:380–387.

33. Andreote FD, Gumiere T, Durrer A. 2014. Exploring interactions of plant microbiomes. Scientia Agricola 71:528–539.

34. Schlaeppi K, Bulgarelli D. 2015. The plant microbiome at work. Mol Plant- Microbe Int 212:212–217.

172

35. Berg G, Zachow C, Müller H, Philipps J, Tilcher R. 2013. Next-generation bio- products sowing the seeds of success for sustainable agriculture. Agronomy 3:648–656.

36. Gopal M, Gupta A, Thomas G V. 2013. Bespoke microbiome therapy to manage plant diseases. Front Microbiol 4:10–13.

37. Sagaram US, Deangelis KM, Trivedi P, Andersen GL, Lu SE, Wang N. 2009. Bacterial diversity analysis of huanglongbing pathogen-infected citrus, using phyloChip arrays and 16S rRNA gene clone library sequencing. Appl Environ Microbiol 75:1566–1574.

38. Zhang M, Powell CA, Benyon LS, Zhou H, Duan Y. 2013. Deciphering the bacterial microbiome of citrus plants in response to “Candidatus liberibacter asiaticus”-infection and antibiotic treatments. PLoS One 8:e76331.

39. Fagen JR, Leonard MT, McCullough CM, Edirisinghe JN, Henry CS, Davis MJ, Triplett EW. 2014. Comparative genomics of cultured and uncultured strains suggests genes essential for free-living growth of Liberibacter. PLoS One 9: e84469.

40. Zhang M, Duan Y, Zhou L, Turechek WW, Stover E, Powell CA. 2010. Screening molecules for control of citrus huanglongbing using an optimized regeneration system for “Candidatus Liberibacter asiaticus”-infected periwinkle (Catharanthus roseus) cuttings. Phytopathology 100:239–245.

41. Livak KJ, Schmittgen TD. 2001. Analysis of relative gene expression data using real-time quantitative PCR and the 2−ΔΔCT method. Methods 25:402–408.

42. McManus PS, Stockwell VO, Sundin GW, Jones AL. 2002. Antibiotic use in plant agriculture. Annu Rev Phytopathol 40:443–465.

43. Chiu RJ, Tsai MY, Huang CH. 1979. Distribution and retention of tetracyclines in healthy and likubin-affected citrus trees following trunk transfusion. In Proc ROC- US Coop Sci Seminar on Mycoplasma Diseases of Plants, Volume Ser 1. Edited by Su HJ, McCoy RE. Taipei, Taiwan: National Science Council Symposium; 1979:43–152.

44. Zhang M, Powell CA, Guo Y, Doud MS, Duan Y. 2012. A graft-based chemotherapy method for screening effective molecules and rescuing huanglongbing-affected citrus plants. Phytopathology 102:567–574.

45. Zhang M, Guo Y, Powell CA, Doud MS, Yang C, Duan Y. 2014. Effective antibiotics against “Candidatus Liberibacter asiaticus” in HLB-affected citrus plants identified via the graft-based evaluation. PLoS One 9:e111032.

173

46. Fan GC, Xia YL, Lin XJ, Hu HQ, Wang XD, Ruan CQ, Lu LM, Liu B. 2016. Evaluation of thermotherapy against Huanglongbing (citrus greening) in the greenhouse. J Integr Agric 15:111–119.

47. Hoffman MT, Doud MS, Williams L, Zhang M, Ding F, Stover E, Hall DG, Zhang S, Jones L, Gooch M, Fleites L, Dixon W, Gabriel D, Duan YP. 2013. Heat treatment eliminates ‘Candidatus Liberibacter asiaticus’ from infected citrus trees under controlled conditions. Phytopathology 103:15–22.

48. Trotochaud J, Ehsani R. 2016. Commercial-scale thermotherapy for combating citrus greening (huanglongbing). 2016 ASABE Int Meet. doi:10.13031/aim.20162455891.

49. Bulgari D, Casati P, Crepaldi P, Daffonchio D, Quaglino F, Brusetti L, Bianco PA. 2011. Restructuring of endophytic bacterial communities in grapevine yellows-diseased and recovered Vitis vinifera L. plants. Appl Environ Microbiol 77:5018–5022.

50. Li JG, Ren G Di, Jia ZJ, Dong YH. 2014. Composition and activity of rhizosphere microbial communities associated with healthy and diseased greenhouse tomatoes. Plant Soil 380:337–347.

51. Rastogi G, Sbodio A, Tech JJ, Suslow T V, Coaker GL, Leveau JHJ. 2012. Leaf microbiota in an agroecosystem: spatiotemporal variation in bacterial community composition on field-grown lettuce. ISME J 6:1812–1822.

52. Schreiner K, Hagn A, Kyselková M, Moënne-Loccoz Y, Welzl G, Munch JC, Schloter M. 2010. Comparison of barley succession and take-all disease as environmental factors shaping the rhizobacterial community during take-all decline. Appl Environ Microbiol 76:4703–4712.

53. Xu L, Ravnskov S, Larsen J, Nicolaisen M. 2012. Linking fungal communities in roots, rhizosphere, and soil to the health status of Pisum sativum. FEMS Microbiol Ecol 82:736–745.

54. Zhang Y, Du BH, Jin Z, Li Z, Song H, Ding YQ. 2011. Analysis of bacterial communities in rhizosphere soil of healthy and diseased cotton (Gossypium sp.) at different plant growth stages. Plant Soil 339:447–455.

55. Trivedi P, Duan Y, Wang N. 2010. Huanglongbing, a systemic disease, restructures the bacterial community associated with citrus roots. Appl Environ Microbiol 76:3427–3436.

56. Trivedi P, He Z, Van Nostrand JD, Albrigo G, Zhou J, Wang N. 2012. Huanglongbing alters the structure and functional diversity of microbial communities associated with citrus rhizosphere. ISME J 6:363–83.

174

57. Cazorla FM, Mercado-Blanco J. 2016. Biological control of tree and woody plant diseases: an impossible task? BioControl 61:233–242.

58. Compant S, Brader G, Muzammil S, Sessitsch A, Lebrihi A, Mathieu F. 2013. Use of beneficial bacteria and their secondary metabolites to control grapevine pathogen diseases. BioControl 58: 435–455.

59. El Hassni M, J’Aiti F, Dihazi A, Ait Barka E, Daayf F, El Hadrami I. 2004. Enhancement of defence responses against bayoud disease by treatment of date palm seedlings with an hypoaggressive Fusarium oxysporum isolate. J Phytopathol 152:182–189.

60. El Hassni M, El Hadrami A, Daayf F, Chérif M, Barka EA, El Hadrami I. 2007. Biological control of bayoud disease in date palm: Selection of microorganisms inhibiting the causal agent and inducing defense reactions. Environ Exp Bot 59:224–234.

61. Mgbeahuruike AC, Sun H, Fransson P, Kasanen R, Daniel G, Karlsson M, Asiegbu FO. 2011. Screening of Phlebiopsis gigantea isolates for traits associated with biocontrol of the conifer pathogen Heterobasidion annosum. Biol Control 57:118–129.

62. Lacava PT, Li W, Araújo WL, Azevedo JL, Hartung JS. 2007. The endophyte Curtobacterium flaccumfaciens reduces symptoms caused by Xylella fastidiosa in Catharanthus roseus. J Microbiol 45:388–393.

63. Lacava PT, Li WB, Araújo WL, Azevedo JL, Hartung JS. 2006. Rapid, specific and quantitative assays for the detection of the endophytic bacterium Methylobacterium mesophilicum in plants. J Microbiol Methods 65:535–541.

64. Rao CN, Shivankar VJ, Deole S, Dhengre VN. 2013. Effect of organic manures on the incidence of Asian citrus psyllid, Diaphorina citri Kuwayama. Pest Manag Hortic Ecosyst 19:92–94.

65. Jae HJ, Weon HY, Hyun HN, Jeun YC, Koh SW. 2014. Effect of long-term different fertilization on bacterial community structures and diversity in citrus orchard soil of volcanic ash. J Microbiol 52:995-1001.

66. Tian S, Lu L, Labavitch JM, Webb SM, Yang X, Brown PH, He Z. 2014. Spatial imaging of Zn and other elements in huanglongbing-affected grapefruit by synchrotron-based micro X-ray fluorescence investigation. J Exp Bot 65:953–964.

67. Pelz-Stelinski KS, Martini X, Kingdom-Gibbard H, Stelinski LL. 2016. Patterns of habitat use by the Asian citrus psyllid, Diaphorina citri, as influenced by abiotic and biotic growing conditions. Agric For Entomol. doi: 10.1111/afe.12197.

175

68. Busby PE, Soman C, Wagner MR, Friesen ML, Kremer J, Bennett A, Morsy M, Eisen JA, Leach JE, Dangl JL. 2017. Research priorities for harnessing plant microbiomes in sustainable agriculture. PLoS Biol 15:e2001793.

69. Lundberg DS, Yourstone S, Mieczkowski P, Jones CD, Dangl JL. 2013. Practical innovations for high-throughput amplicon sequencing. Nat Methods 10:999–1002.

70. Mordecai EA. 2011. Pathogen impacts on plant communities: Unifying theory, concepts, and empirical work. Ecol Monogr 81:429–441.

71. Shade A, Handelsman J. 2012. Beyond the Venn diagram: The hunt for a core microbiome. Environ Microbiol 14:4–12.

72. Bulgarelli D, Rott M, Schlaeppi K, van Themaat EVL, Ahmadinejad N, Assenza F, Rauf P, Huettel B, Reinhardt R, Schmelzer E, Peplies J, Gloeckner FO, Amann R, Eickhorst T, Schulze-Lefert P. 2012. Revealing structure and assembly cues for Arabidopsis root-inhabiting bacterial microbiota. Nature 488:91–95.

73. Lundberg DS, Lebeis SL, Paredes SH, Yourstone S, Gehring J, Malfatti S, Tremblay J, Engelbrektson A, Kunin V, Rio TG Del, Edgar RC, Eickhorst T, Ley RE, Hugenholtz P, Tringe SG, Dangl JL. 2012. Defining the core Arabidopsis thaliana root microbiome. Nature 488:86–90.

74. Xu Z, Malmer D, Langille MGI, Way SF, Knight R. 2014. Which is more important for classifying microbial communities: who’s there or what they can do? ISME J 8:2357–2359.

75. Kim C, Lee C, Shin J, Chung Y, Hyung N. 1997. A simple and rapid method for isolation of high quality genomic DNA from fruit trees and conifers using PVP. Nucleic Acids Res 25:1085–1086.

76. Apprill A, Mcnally S, Parsons R, Weber L. 2015. Minor revision to V4 region SSU rRNA 806R gene primer greatly increases detection of SAR11 bacterioplankton. Aquat Microb Ecol 75:129–137.

77. Caporaso JG, Lauber CL, Walters WA, Berg-Lyons D, Huntley J, Fierer N, Owens SM, Betley J, Fraser L, Bauer M, Gormley N, Gilbert JA, Smith G, Knight R. 2012. Ultra-high-throughput microbial community analysis on the Illumina HiSeq and MiSeq platforms. ISME J 6:1621–1624.

78. Rideout JR, He Y, Navas-Molina JA, Walters WA, Ursell LK, Gibbons SM, Chase J, McDonald D, Gonzalez A, Robbins-Pianka A, Clemente JC, Gilbert JA, Huse SM, Zhou HW, Knight R, Caporaso JG. 2014. Subsampled open- reference clustering creates consistent, comprehensive OTU definitions and scales to billions of sequences. PeerJ 2:e545.

176

79. DeSantis TZ, Hugenholtz P, Larsen N, Rojas M, Brodie EL, Keller K, Huber T, Dalevi D, Hu P, Andersen GL. 2006. Greengenes, a chimera-checked 16S rRNA gene database and workbench compatible with ARB. Appl Environ Microbiol 72:5069–5072.

80. McMurdie PJ, Holmes S. 2013. Phyloseq: An R package for reproducible interactive analysis and graphics of microbiome census data. PLoS One 8:e61217.

81. Wickham H. 2009. ggplot2: Elegant graphics for data analysis. Media 35: 211.

82. Blaustein R, Meyer J, Conesa A, Lorca G, Teplitski M. 2016. Impacts of abundance of Candidatus Liberibacter on the citrus phytomicrobiome. Phytopathol 10:25-26.

83. Langille MGI, Zaneveld J, Caporaso JG, McDonald D, Knights D, Reyes JA, Clemente JC, Burkepile DE, Vega Thurber RL, Knight R, Beiko RG, Huttenhower C. 2013. Predictive functional profiling of microbial communities using 16S rRNA marker gene sequences. Nat Biotechnol 31:814–821.

84. Blaustein RA, Lorca GL, Meyer JL, Gonzalez CF, Teplitski M. 2017. Defining the core citrus leaf- and root-associated microbiota: Factors associated with community structure and implications for managing huanglongbing (citrus greening) disease. Appl Environ Microbiol 83:e00210-17.

85. Lebeis SL. 2014. The potential for give and take in plant-microbiome relationships. Front Plant Sci 5:287.

86. Mendes R, Garbeva P, Raaijmakers JM. 2013. The rhizosphere microbiome: significance of plant beneficial, plant pathogenic, and human pathogenic microorganisms. FEMS Microbiol Rev 37:634–663.

87. Klein E, Katan J, Minz D, Gamliel A. 2012. Soil suppressiveness to Fusarium disease: Shifts in root microbiome associated with reduction of pathogen root colonization. Phytopathology 103:23–33.

88. Mavrodi OV, Mavrodi DV, Parejko JA, Thomashow LS, Weller DM. 2012. Irrigation differentially impacts populations of indigenous antibiotic-producing Pseudomonas spp. in the rhizosphere of wheat. Appl Environ Microbiol 78:3214– 3220.

89. Bodenhausen N, Horton MW, Bergelson J. 2013. Bacterial communities associated with the leaves and the roots of Arabidopsis thaliana. PLoS One 8: e56329.

177

90. Gottel NR, Castro HF, Kerley M, Yang Z, Pelletier DA, Podar M, Karpinets T, Uberbacher ED, Tuskan GA, Vilgalys R, Doktycz MJ, Schadt CW. 2011. Distinct microbial communities within the endosphere and rhizosphere of Populus deltoides roots across contrasting soil types. Appl Environ Microbiol 77:5934– 5944.

91. Peiffer JA, Spor A, Koren O, Jin Z, Tringe SG, Dangl JL, Buckler ES, Ley RE. 2013. Diversity and heritability of the maize rhizosphere microbiome under field conditions. Proc Natl Acad Sci U S A 110:6548–6553.

92. Redford AJ, Fierer N. 2009. Bacterial succession on the leaf surface: A novel system for studying successional dynamics. Microb Ecol 58:189–198.

93. Weinert N, Piceno Y, Ding GC, Meincke R, Heuer H, Berg G, Schloter M, Andersen G, Smalla K. 2011. PhyloChip hybridization uncovered an enormous bacterial diversity in the rhizosphere of different potato cultivars: Many common and few cultivar-dependent taxa. FEMS Microbiol Ecol 75:497–506.

94. Berg G, Smalla K. 2009. Plant species and soil type cooperatively shape the structure and function of microbial communities in the rhizosphere. FEMS Microbiol Ecol 68:1–13.

95. Mendes R, Kruijt M, de Bruijn I, Dekkers E, van der Voort M, Schneider JHM, Piceno YM, DeSantis TZ, Andersen GL, Bakker PAHM, Raaijmakers JM. 2011. Deciphering the rhizosphere microbiome for disease-suppressive bacteria. Science 332:1097–1100.

96. Lamichhane JR, Venturi V. 2015. Synergisms between microbial pathogens in plant disease complexes: a growing trend. Front Plant Sci 6:385.

97. Andreote FD, Rocha UN Da, Araújo WL, Azevedo JL, Van Overbeek LS. 2010. Effect of bacterial inoculation, plant genotype and developmental stage on root-associated and endophytic bacterial communities in potato (Solanum tuberosum). Antonie van Leeuwenhoek, Int J Gen Mol Microbiol 97:389–399.

98. Dematheis F, Kurtz B, Vidal S, Smalla K. 2013. Multitrophic interactions among western corn rootworm, Glomus intraradices and microbial communities in the rhizosphere and endorhiza of maize. Front Microbiol 4:357.

99. Herschkovitz Y, Lerner A, Davidov Y, Rothballer M, Hartmann A, Okon Y, Jurkevitch E. 2005. Inoculation with the plant-growth-promoting rhizobacterium Azospirillum brasilense causes little disturbance in the rhizosphere and rhizoplane of maize (Zea mays). Microb Ecol 50:277–288.

100. Bakker MG, Manter DK, Sheflin AM, Weir TL, Vivanco JM. 2012. Harnessing the rhizosphere microbiome through plant breeding and agricultural management. Plant Soil 360. doi:10.1007/s11104-012-1361-x.

178

101. Schreiter S, Sandmann M, Smalla K, Grosch R. 2014. Soil type dependent rhizosphere competence and biocontrol of two bacterial inoculant strains and their effects on the rhizosphere microbial community of field-grown lettuce. PLoS One 9:e103726.

102. Caporaso JG, Kuczynski J, Stombaugh J, Bittinger K, Bushman FD, Costello EK, Fierer N, Peña AG, Goodrich JK, Gordon JI, Huttley GA, Kelley ST, Knights D, Koenig JE, Ley RE, Lozupone CA, Mcdonald D, Muegge BD, Pirrung M, Reeder J, Sevinsky JR, Turnbaugh PJ, Walters W a, Widmann J, Yatsunenko T, Zaneveld J, Knight R. 2010. correspondence QIIME allows analysis of high- throughput community sequencing data Intensity normalization improves color calling in SOLiD sequencing. Nat Publ Gr 7:335–336.

103. Huse SM, Ye Y, Zhou Y, Fodor AA. 2012. A core human microbiome as viewed through 16S rRNA sequence clusters. PLoS One 7:e34242.

104. Hammer Ø, Harper DAT, Ryan PD. 2001. PAST: Paleontological Statistics Software Package for Education and Data Analysis. Palaeontol Electron 4:4.

105. Dixon P. 2003. VEGAN, a package of R functions for community ecology. J Veg Sci 14:927–930.

106. Faust K, Raes J. 2012. Microbial interactions: from networks to models. Nat Rev Microbiol 10:538–550.

107. Paul Shannon, Andrew Markiel, Owen Ozier, Nitin S Baliga, Jonathan T Wang, Daniel Ramage, Nada Amin BS. 2003. Cytoscape: A software environment for integrated models of biomolecular interaction networks. Genome Res 13:2498–2504.

108. Lynch MDJ, Neufeld JD. 2015. Ecology and exploration of the rare biosphere. Nat Rev Microbiol 13:217–229.

109. Ardanov P, Sessitsch A, Häggman H, Kozyrovska N, Pirttilä AM. 2012. Methylobacterium-induced endophyte community changes correspond with protection of plants against pathogen attack. PLoS One 7:e46802.

110. Enya J, Shinohara H, Yoshida S, Tsukiboshi T, Negishi H, Suyama K, Tsushima S. 2007. Culturable leaf-associated bacteria on tomato plants and their potential as biological control agents. Microb Ecol 53:524–536.

111. Innerebner G, Knief C, Vorholt J a. 2011. Protection of Arabidopsis thaliana against leaf-pathogenic Pseudomonas syringae by Sphingomonas strains in a controlled model system. Appl Environ Microbiol 77:3202–3210.

179

112. Madhaiyan M, Suresh Reddy BV, Anandham R, Senthilkumar M, Poonguzhali S, Sundaram SP, Sa T. 2006. Plant growth-promoting Methylobacterium induces defense responses in groundnut (Arachis hypogaea L.) compared with rot pathogens. Curr Microbiol 53:270–276.

113. Compant S, Reiter B, Sessitsch A, Clément C, Barka EA, Nowak J, Clement C, Barka EA. 2005. Endophytic Colonization of Vitis vinifera L. by plant growth- promoting bacterium Burkholderia sp. strain PsJN. Appl Environ Microbiol 71:1685–1693.

114. Lugtenberg B, Kamilova F. 2009. Plant-growth-promoting rhizobacteria. Annu Rev Microbiol 63:541–556.

115. Verma, J, Yadav, J, Tiwari, K, Singh, L, Singh V. 2010. Impact of plant growth promoting rhizobacteria on crop production. Int J Agric Res 5:954–983.

116. Das AK. 2003. Citrus canker - a review. J Appl Hortic 5:52–60.

117. De Clercq D, Van Trappen S, Cleenwerck I, Ceustermans A, Swings J, Coosemans J, Ryckeboer J. 2006. Rhodanobacter spathiphylli sp. nov., a gammaproteobacterium isolated from the roots of Spathiphyllum plants grown in a compost-amended potting mix. Int J Syst Evol Microbiol 56:1755–1759.

118. Etxeberria E, Gonzalez P, Achor D, Albrigo G. 2009. Anatomical distribution of abnormally high levels of starch in HLB-affected Valencia orange trees. Physiol Mol Plant Pathol 74:76–83.

119. Puttamuk T, Zhang S, Duan Y, Jantasorn A, Thaveechai N. 2014. Effect of chemical treatments on “Candidatus Liberibacter asiaticus” infected pomelo (Citrus maxima). Crop Prot 65:114–121.

120. Amend AS, Martiny AC, Allison SD, Berlemont R, Goulden ML, Lu Y, Treseder KK, Weihe C, Martiny JBH. 2016. Microbial response to simulated global change is phylogenetically conserved and linked with functional potential. ISME J 10:109–118.

121. Köberl M, Dita M, Martinuz A, Staver C, Berg G. 2017. Members of Gammaproteobacteria as indicator species of healthy banana plants on Fusarium wilt-infested fields in Central America. Sci Rep 7:45318.

122. Ofek M, Hadar Y, Minz D. 2012. Ecology of root colonizing Massilia (Oxalobacteraceae). PLoS One 7:e40117.

123. Zamioudis C, Mastranesti P, Dhonukshe P, Blilou I, Pieterse CMJ. 2013. Unraveling root developmental programs initiated by beneficial Pseudomonas spp. bacteria. Plant Physiol 162:304–318.

180

124. Tian B-Y, Cao Y, Zhang K-Q. 2015. Metagenomic insights into communities, functions of endophytes, and their associates with infection by root-knot nematode, Meloidogyne incognita, in tomato roots. Sci Rep 5:17087.

125. Copeland JK, Yuan L, Layeghifard M, Wang PW, Guttman DS. 2015. Seasonal community succession of the phyllosphere microbiome. Mol Plant- Microbe Int 28:274–285.

126. Kawasaki A, Donn S, Ryan PR, Mathesius U, Devilla R, Jones A, Watt M. 2016. Microbiome and exudates of the root and rhizosphere of Brachypodium distachyon, a model for wheat. PLoS One 11:e0164533.

127. Pii Y, Borruso L, Brusetti L, Crecchio C, Cesco S, Mimmo T. 2016. The interaction between iron nutrition, plant species and soil type shapes the rhizosphere microbiome. Plant Physiol Biochem 99:39–48.

128. Lopes LD, de Cassia Pereira e Silva M, Andreote FD. 2016. Bacterial abilities and adaptation toward the rhizosphere colonization. Front Microbiol 7:1341.

129. Mahoney AK, Yin C, Hulbert SH. 2017. Community structure, species variation, and potential functions of rhizosphere-associated bacteria of different winter wheat (Triticum aestivum) cultivars. Front Plant Sci 8:132.

130. Meaden S, Metcalf CJE, Koskella B. 2016. The effects of host age and spatial location on bacterial community composition in the English Oak tree (Quercus robur). Environ Microbiol Rep 8:649–658.

131. Ruiz-Pérez CA, Restrepo S, Zambrano MM. 2016. Microbial and functional diversity within the phyllosphere of Espeletia sp. in an Andean high mountain ecosystem. Appl Environ Microbiol 82:1807–1817.

132. Yuan Z, Druzhinina IS, Labbé J, Redman R, Qin Y, Rodriguez R, Zhang C, Tuskan GA, Lin F. 2016. Specialized Microbiome of a Halophyte and its Role in Helping Non-Host Plants to Withstand Salinity. Sci Rep 6:32467.

133. Zarraonaindia I, Owens SM, Weisenhorn P, West K, Hampton-Marcell J, Lax S, Bokulich NA, Mills DA, Martin G, Taghavi S, van der Lelie D, Gilbert JA. 2015. The soil microbiome influences grapevine-associated microbiota. MBio 6: e02527-14.

134. Ogata H, Goto S, Sato K, Fujibuchi W, Bono H, Kanehisa M. 1999. KEGG: Kyoto encyclopedia of genes and genomes. Nucleic Acids Res 27:29–34.

135. Warnes GR, Bolker B, Bonebakker L, Gentleman R, Huber W, Liaw A, Lumley T, Maechler M, Magnusson A, Moeller S, Schwartz M, Venables B. 2016. Package ‘gplots’. http://cran.r-project.org.

181

136. Bharwani A, Firoz Mian M, Foster JA, Surette MG, Bienenstock J, Forsythe P. 2016. Structural and functional consequences of chronic psychosocial stress on the microbiome and host. Psychoneuroendocrinology 63:217–227.

137. Rooks MG, Veiga P, Wardwell-Scott LH, Tickle T, Segata N, Michaud M, Gallini CA, Beal C, van Hylckama-Vlieg JE, Ballal SA, Morgan XC, Glickman JN, Gevers D, Huttenhower C, Garrett WS. 2014. Gut microbiome composition and function in experimental colitis during active disease and treatment-induced remission. ISME J 8:1403–1417.

138. Gevers D, Kugathasan S, Denson LA, Vazquez-Baeza Y, Van Treuren W, Ren B, Schwager E, Knights D, Song SJ, Yassour M, Morgan XC, Kostic AD, Luo C, Gonzalez A, McDonald D, Haberman Y, Walters T, Baker S, Rosh J, Stephens M, Heyman M, Markowitz J, Baldassano R, Griffiths A, Sylvester F, Mack D, Kim S, Crandall W, Hyams J, Huttenhower C, Knight R, Xavier RJ. 2014. The treatment-naive microbiome in new-onset Crohn’s disease. Cell Host Microbe 15:382–392.

139. Tong M, Li X, Parfrey LW, Roth B, Ippoliti A, Wei B, Borneman J, McGovern DPB, Frank DN, Li E, Horvath S, Knight R, Braun J. 2013. A modular organization of the human intestinal mucosal microbiota and its association with inflammatory bowel disease. PLoS One 8:e80702.

140. Adam E, Groenenboom AE, Kurm V, Rajewska M, Schmidt R, Tyc O, Weidner S, Berg G, De Boer W, Salles JF. 2016. Controlling the microbiome: Microhabitat adjustments for successful biocontrol strategies in soil and human gut. Front Microbiol 7:1079.

141. Berg G, Grube M, Schloter M, Smalla K. 2014. Unraveling the plant microbiome: Looking back and future perspectives. Front Microbiol 5:148.

142. Turner TR, James EK, Poole PS. 2013. The plant microbiome. Genome Biol 14:209.

143. Ellis J. 2016. Can plant microbiome studies lead to effective biocontrol of plant diseases? Mol Plant-Microbe Int 30:190–193.

144. Hacquard S, Garrido-Oter R, González A, Spaepen S, Ackermann G, Lebeis S, McHardy AC, Dangl JL, Knight R, Ley R, Schulze-Lefert P. 2015. Microbiota and host nutrition across plant and animal kingdoms. Cell Host Microbe 17:603– 616.

145. Niu B, Paulson JN, Zheng X, Kolter R. 2017. Simplified and representative bacterial community of maize roots. Proc Natl Acad Sci 114:E2450–E2459.

146. Santhanam R, Luu VT, Weinhold A, Goldberg J, Oh Y, Baldwin IT. 2015. Native root-associated bacteria rescue a plant from a sudden-wilt disease that emerged during continuous cropping. Proc Natl Acad Sci 112:E5013-E5020.

182

147. van der Heijden MGA, Hartmann M. 2016. Networking in the Plant Microbiome. PLoS Biol 14:e1002378.

148. Finkel OM, Castrillo G, Herrera Paredes S, Salas González I, Dangl JL. 2017. Understanding and exploiting plant beneficial microbes. Curr Opin Plant Biol 38:155–163.

149. Bashan Y, de-Bashan LE, Prabhu SR, Hernandez JP. 2014. Advances in plant growth-promoting bacterial inoculant technology: Formulations and practical perspectives (1998-2013). Plant Soil 378:1–33.

150. Agler MT, Ruhe J, Kroll S, Morhenn C, Kim ST, Weigel D, Kemen EM. 2016. Microbial hub taxa link host and abiotic factors to plant microbiome variation. PLoS Biol 14:e1002352.

151. Blaustein R, Lorca G, Teplitski M. 2017. Challenges for managing Candidatus Liberibacter spp. (huanglongbing disease pathogen): Current control measures and future directions. Phytopathology. doi:10.1094/PHYTO-07-17-0260-RVW.

183

BIOGRAPHICAL SKETCH

Ryan Andrew Blaustein graduated from Sherwood High School in Olney,

Maryland in 2007. He earned his Bachelor of Science in Biology from the University of

Maryland in 2011, graduating with honors as a “UMD Scholar in Life Sciences”. Ryan’s undergraduate experiences as a teaching assistant for microbiology lab and participating in research internships in the Department’s of Biology, Plant Sciences, and

Entomology facilitated his interests in a career in academia.

From 2011-2015, Ryan worked under supervision of Dr. Yakov Pachepsky at the

Environmental Microbial Food Safety Laboratory (EMFSL) of the United States

Department Agriculture: Agricultural Research Service. He directed and assisted various studies on the fate and transport of food- and waterborne pathogens and indicators (Escherichia coli, Salmonella, enterococci, coliforms), which lead to significant discoveries that promote food safety and foster ecosystem and public health.

While collaborating with the University of Maryland, Ryan earned his Master of Science in the field of environmental microbiology in 2014, under direction of Dr. Robert Hill.

Ryan joined Dr. Max Teplitski’s group in the Department of Soil and Water

Sciences at the Genetics Institute of the University of Florida in the summer of 2015. As a UF Graduate School Fellow, Ryan earned his Ph.D. in the fall of 2017. His dissertation incorporated bacterial genomics and bioinformatics to define the key changes in structure and function of the citrus microbiome during the progression and treatment of citrus greening disease. Findings from his work have implications for bio- based and therapeutic strategies that may be used to mitigate the devastating disease.

The research also advanced the understanding of several fundamental questions regarding microbiota-pathogen-host relationships, which have broader applications for

184

disease control. During his Ph.D. program, Ryan enjoyed mentoring several undergraduate students and working as a teaching assistant in Dr. Julie Meyer’s upper- level course on Waterborne Pathogens.

Regarding research accomplishments, Ryan currently has 14 publications in peer-reviewed journals (9 as first author). His work has been presented at 19 national, regional, and local conferences. In addition, he has reviewed for several journals and served as the Graduate Student Board Representative of the American Society for

Microbiology: Florida Branch (2016-2017). Accolades include receiving the Doris and

Earl and Verna Lowe Scholarship for research merit in the UF College of Agricultural and Life Sciences (2016, 2017), the Sam Polston Scholarship for outstanding research and academic performance in the UF Soil and Water Sciences Department (2016), the

American Society for Microbiology Student Travel Award (2017), the UF IFAS Travel

Grant (2017), the Dr. James Davidson Graduate Student Travel Scholarship (2016), and second place in the graduate student poster competition at the American Society for Microbiology: Southeastern Branch Meeting (2015).

Upon completion of his Ph.D., Ryan will join the Department of Civil and

Environmental Engineering at Northwestern University as a Postdoctoral Fellow.

Working with Dr. Erica Hartmann, he will focus on microbiomes of the built environment, conducting studies that integrate “-omics” and environmental chemistry.

185