CHARACTERIZATION OF AXENIC IMMUNE DEFICIENCY
IN ARABIDOPSIS THALIANA
By
James Michael Kremer

A DISSERTATION
Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of
Microbiology and Molecular Genetics – Doctor of Philosophy
2017

ABSTRACT
CHARACTERIZATION OF AXENIC IMMUNE DEFICIENCY
IN ARABIDOPSIS THALIANA
By
James Michael Kremer
Evolution of land plants began and has since occurred, in concert with complex
communities of microorganisms, giving rise to a vast spectrum of plant-microbe
relationships. Over the past decade, plant molecular biologists and microbial
ecologists have worked together to identify drivers of microbiome composition that
inspire hypotheses about microbiome functional potential, but many fall short of
offering empirical evidence of microbiome-mediated influence on host phenotypes.
Herein, I introduce a new suite of tools to explore microbiome function and report that
many facets of plant immunocompetence are microbiome-dependent.
Chapter One summarizes the current understanding of plant innate immunity and
notable progress of plant microbiome research, including: (1) detection and response
to microbe-associated molecular patterns, (2) hormone signaling during biotic
interactions, (3) technology for exploration of plant microbiome ecology, (4) factors that
influence microbiome community structure, and (5) a review of relevant model systems
and gnotobiotic growth platforms. Chapter Two describes the development of a novel
“FlowPot” growth system: a peat-based platform conducive to axenic (microbe-free),
gnotobiotic (defined microbiota), and holoxenic (undefined, complex microbiota)
Arabidopsis thaliana growth. This system provides the ability to maintain control of
abiotic parameters and exogenous microbiota, thus providing a valuable platform for

discovery for plant microbiome research. The FlowPot system and offers a substantial
improvement over alternative growth systems regarding plant health, tractability to
greenhouse conditions, and maintenance of bacterial alpha diversity upon inoculation
with soil-derived microbiota. An implementation of the growth system is detailed in
Chapter Three, featuring a comparative analysis of the axenic vs. holoxenic
Arabidopsis transcriptome, metabolome, and immunocompetence. Axenic Arabidopsis
has a reduced level of defense- and immunity-associated gene expression and the
defense hormone salicylic acid (SA). We report that axenic Arabidopsis is
compromised in defense against the foliar pathogen Pseudomonas syringae pv. tomato
DC3000 (Pst). Immune elicitation experiments revealed that axenic Arabidopsis is also
compromised in the ability to recognize and/or mount normal defense responses to the
microbe-associated molecular pattern flg22. Axenic susceptibility to Pst is partially
explained by defective innate immunity. Finally, we report the identity of differentially
abundant metabolites and transcripts in axenic and holoxenic Arabidopsis that may be
involved in microbiome-influenced host phenotypes. Collectively, research described in
this dissertation provides new tools and a discovery platform to empirically
characterize function of plant microbiota, as well as detailed characterization of axenic
phenotypes and axenic immune deficiency.

Copyright by
JAMES MICHAEL KREMER
2017

This dissertation is dedicated to you, the reader. I hope that this will inspire new
hypotheses and future experiments.

v

ACKNOWLEDGEMENTS
First and foremost, I must acknowledge my mentors, Dr. Sheng Yang He and Dr. Jim
Tiedje. Without your support and guidance, none of this work would have been
possible. I was originally drawn to Michigan State University because of your
respective expertise in microbial suppression of plant innate immunity and the ecology
of root-associated microbiota. Over time, as I rotated in your respective laboratories
and eventually began to focus my dissertation research, I realized how unique and
incredible an opportunity it was to be at a place like Michigan State University;
nowhere else in the world could I have such brilliant advisors with different but
complementary perspectives. Both of you identify and cultivate passion in your
students towards the pursuit of areas of research that inspire us individually. When the
inevitable roadblocks and experimental setbacks arise, you inspire tenacity. Sheng
Yang, you were incredibly patient when I was resistant to attempt certain experiments,
and likewise, you trusted me to conduct many experiments that did not necessarily
coincide with the primary objectives of this dissertation. By doing so, you have
provided me with the incredible opportunity to make mistakes and retrospectively
evaluate depth versus breadth of research, and that has taught me more than you can
imagine.
Jim, you inspire me to think globally, and taught me the importance of creativity and
communication in science. Your academic legacy and success of past students are
inspirational, and I cannot understate how honored and fortunate I am to receive your
mentorship. Both of you have provided me with numerous opportunities to present at

vi

academic conferences and build a network of collaborators, and for this, I thank you. I
also must acknowledge Dr. Paul Schulze-Lefert. You extended an invitation to visit
your group in Cologne, and immediately embraced me as one of your own students. I
learned a tremendous amount from your pragmatic approach to tackle highly complex
scientific questions. Drs. Amine Hassani, Sergio de los Santos, JP Jerome and Brian
Kvitko, you have all been incredible scientific collaborators and lifelong friends. To
Kevin Nehil, Alec Bonifer, Kristin Drumheller, Alan Mundakkal, David Rhodes, Caleigh
Griffin, and all of the other undergraduate researchers I have had the pleasure of
advising, you have taught me more than you can realize and I cannot wait to see where
life takes you. I also want to acknowledge Brad Paasch for being a great colleague and
friend in our mutual pursuit to characterize microbiome function. To the many members
of the He and Tiedje lab groups, you have been tremendous colleagues and thank you
for all the discussions and input. It has been a pleasure to work with many of you, and I
look forward to crossing paths again in the future.
Finally, I must acknowledge my family. Without the tremendous unwavering love and
support from my parents, I never would have finished this dissertation. To my aunt Dr.
Margaret McLaughlin, your help and wisdom during the “final push” made this all
possible. And to my grandmother Bernice Neuberg, your creativity and love for plants
no doubt inspired me to pursue this Ph.D.

vii

TABLE OF CONTENTS
LIST OF TABLES

x

LIST OF FIGURES

xi

Chapter 1
History and review of plant microbiome research
Introduction
Early plant pathology
Microbial ecology of plant and soil environments
The Arabidopsis thaliana model
Arabidopsis innate immunity
Defense hormone signaling
Systemic acquired resistance and induced systemic resistance

1
2
5
9
12
14
18
21

Chapter 2
Construction and optimization of a soil-based growth system conducive to
axenic, gnotobiotic, and holoxenic Arabidopsis growth
Abstract
Introduction
Materials and Methods
Sterilization of the FlowPot peat-based substrate
FlowPot growth system assembly
Calcined clay system assembly
Soil collection
Preparation of input community microbiota
Growth system inoculation
Arabidopsis growth conditions
Protein extraction and quantification of photosynthesis-associated
proteins
Sample collection and DNA extraction
16S rRNA gene fragment amplification and MiSeq Library preparation
Preprocessing of 16S rRNA gene fragment amplicons
Results
Arabidopsis growth in axenic FlowPots
Beta-diversity
Alpha-diversity
Bacterial community differentiation at the Phylum level
Bacterial community differentiation at the OTU level
Discussion

viii

23
24
26
31
31
31
34
34
35
35
36
36
37
38
39
40
40
43
45
48
54
62

Chapter 3
Characterization of axenic Arabidopsis thaliana
Abstract
Introduction
Materials and Methods
Soil collection and biochemical analysis
Arabidopsis seeds and growth conditions
DNA extractions
ITS and 16S rRNA gene targeted amplification
Bioinformatic analysis of ITS1 and 16S rRNA gene fragment amplicons
RNA extraction and quantitative PCR
RNA seq library preparation and analysis
Tissue collection flg22 elicitation experiments
Protein extraction
Phytohormone quantification by UPLC/MS
Metabolite quantification by GC/MS
Bacterial growth conditions, sources, and strain construction
Infiltration of rosettes with bacterial suspensions
Bacterial multiplication in planta protocol
Oxidative burst assay
Callose deposition assay
Results
Bacterial and fungal compositions of input soil communities are distinct
Fungal and bacterial community differentiation
Axenic plants lack normal basal expression of immune-associated genes
Basal: metabolite analysis (GC/MS)
Basal: Salicylic acid quantification (UPLC/MS)
Transcriptional response to immune elicitation is defective in axenic
plants
Compromised axenic posttranslational immune response
Oxidative burst and callose deposition are compromised in axenic plants
Defective priming and defense against pathogens in axenic plants
Discussion
Chapter 4
Discussion and future directions for plant microbiota research
Development of model host-microbiome systems with associated microbial
culture collections and reference genomes
Define core microbiomes and metagenomes
Rules of synthetic, functionally programmable microbiome assembly
Proposed improvements to FlowPot design and assembly
Strategies for characterization of axenic immunodeficiency
REFERENCES

76
77
78
84
84
84
85
86
87
88
89
90
91
91
93
93
94
95
96
98
99
99
107
120
138
144
146
148
150
152
156
177
180
182
185
188
193
200

ix

LIST OF TABLES
Table 2.1. Enrichment of phylum by sample type relative to input and to bulk.

53

Table 2.2. Absolute OTU abundance by abundance qualifiers.

56

Table 2.3. Classification of root-enriched OTUs and magnitude of enrichment.

57

Table 2.4. Abundance of reads by phylum.

70

Table 2.5. Phylum level classifications of most abundantly ranked OTUs
that account for 90% of reads

71

Table 2.6. Natural soil, clay system. Top 10 OTUs by rank abundance

72

Table 2.7. Natural soil, FlowPot system. Top 10 OTUs by rank abundance

73

Table 2.8. At-SPHERE, clay system. Top 10 OTUs by rank abundance.

74

Table 2.9. At-SPHERE, FlowPot system. Top 10 OTUs by rank abundance.

75

Table 3.1. Attributes of soils used in this study

101

Table 3.2. Classifications of core 16S OTU classifications from input soil

104

Table 3.3. Classifications of core ITS1 OTU classifications from input soil

105

Table 3.4. Rosette enriched OTUs relative to abundance in bulk substrate

112

Table 3.5. Differentially expressed genes (DEG) with stringent criteria

124

Table 3.6. Differentially expressed genes (DEG) with moderate criteria

125

Table 3.7. Differentially expressed genes (DEG) with sensitive criteria

126

Table 3.8. Core differential expression meeting significance criteria at
moderate stringency

127

Table 3.9. Expressed genes (EG)

164

Table 3.10. Pan- differential expression at moderate stringency

165

Table 3.11. Metabolite quantification by GC/MS

172

x

LIST OF FIGURES
Figure 1.1. Plant disease triangle

8

Figure 1.2. Immune responses at the plasma membrane in Arabidopsis.

15

Figure 1.3. MAMP-triggered immune response.

16

Figure 2.1. Schematics of FlowPot construction and irrigation.

33

Figure 2.2. Arabidopsis thaliana grown in FlowPots with axenic or holoxenic
substrate, and photosynthetic protein detection from leaf tissue.

42

Figure 2.3. Principle coordinate analysis (PCoA) of weighted UniFrac distances 44
Figure 2.4. Alpha diversity indices for each sample type

47

Figure 2.5. Mean percent abundance of OTU-classified reads by phylum

52

Figure 2.6. Root-associated OTU enrichment for samples with natural
soil microbiota

58

Figure 2.7. Root-associated OTU enrichment relative to abundance in
bulk substrate for samples with At-SPHERE microbiota

59

Figure 2.8. Phylum-level resolution of root-associated OTU enrichment relative
to abundance in bulk substrate for samples with natural soil microbiota

60

Figure 2.9. Phylum-level resolution of root-associated OTU enrichment relative
to abundance in bulk substrate for samples with At-SPHERE microbiota

61

Figure 2.10. Changes in substrate chemical properties after treatments

68

Figure 2.11. Instruments for the aseptic manipulation of FlowPots, and
illustration of FlowPots on a stand and within the Eco2 microbox

69

Figure 3.1. Map of input community soil collection locations

100

Figure 3.2. Venn diagram of classified bacterial 16S rRNA OTUs from
input communities

106

Figure 3.3. Experimental design for 3-community FlowPot inoculation

108

Figure 3.4. Mean percent abundance of 16S OTU classified reads by phylum

113

xi

Figure 3.5. UniFrac distances between samples based on 16S OTU clustering.

114

Figure 3.6. Alpha diversity indices for each sample type

115

Figure 3.7. Abundance of 16S OTUs for plant (whole rosette) communities with
genus classifications
116
Figure 3.8. Rosette microbiome 16S OTU enrichment relative to abundance
in bulk substrate for Agricultural microbiota samples.
Figure 3.9. Rosette microbiome 16S OTU enrichment relative to abundance
in bulk substrate for Prairie microbiota samples.

117
118

Figure 3.10. Rosette microbiome 16S OTU enrichment relative to abundance
in bulk substrate for Arid microbiota samples.

119

Figure 3.11. RNA extraction experimental design and pipeline.

121

Figure 3.12. Total differentially expressed genes considering input community

129

Figure 3.13. Volcano plot of pan differential gene expression
(all axenic/holoxenic).

130

Figure 3.14. Principal component analysis for transcriptome profiles

131

Figure 3.15. Gene set enrichment analysis of microbiome composition-dependent
changes in host transcriptome, relative to corresponding axenic samples.
132
Figure 3.16. Community specific gene set enrichment analysis, contrasting
holoxenic samples by input community.

133

Figure 3.17. Differential and absolute expression of PTI recognition
and response genes

134

Figure 3.18. Differential and absolute expression of jasmonic acid associated
(JA) and pathogenesis related (PR) marker genes.

135

Figure 3.19. Differential and absolute expression of salicylic acid associated
genes.

136

Figure 3.20. Differential and absolute expression of WRKY transcription factors. 137
Figure 3.21. Relative abundance of metabolites extracted from whole rosette
tissue and quantified by GC/MS

xii

140

Figure 3.22. Defense hormone salicylic acid quantification

145

Figure 3.23. Expression of immunity and defense associated genes

147

Figure 3.24. MAPK protein detection and quantification

149

Figure 3.25. Response to immune elicitation by flg22

152

Figure 3.26. Pathogen susceptibility of axenic and holoxenic Arabidopsis

154

Figure 3.27. Flow cytometry scatter plot of suspended Arabidopsis leaf
disc samples from bacterial counts protocol

176

Figure 4.1. Schematic of FlowPot modification

191

Figure 4.2. Revised disease triangle model

199

xiii

Chapter 1
History and review of plant microbiome research

1

Introduction
Agriculture is arguably the single-most important human innovation. It is an essential
element of civilization. Crop diseases pose an enormous threat to agricultural
productivity and stability, and epidemics have resulted in economic devastation and
countless deaths due to famine. Effective strategies to reduce the incidence and
severity of crop disease are greatly needed. Thus, the majority of plant-microbe
research focuses on model plant-pathogen interactions. One must not discount that
multicellular life is inevitably intertwined with a microbial world, and a vast spectrum of
host-microbe and microbe-microbe relationships exist within a plant microbiome. In
nature, a plant provides numerous distinct habitats colonized by diverse microbiota
throughout each phase of development. Our understanding of plant-microbe
interactions is biased towards a disproportionate amount of in-depth research on
model pathosystems in a laboratory setting. While such studies provide a fantastic
foundation for the molecular underpinnings of important pathogenic interactions, there
are billions of microscopic “elephants” in the room that are often ignored.
Like mammals, plants have an innate immune system to recognize and respond to
microbial pathogens, thereby defending against pathogenic invasion and disease.
Pattern recognition receptors (PRRs) offer surveillance for certain microbe-associated
molecular patterns (MAMPs), and upon ligand binding, an immune response ensues.
Interestingly, certain symbionts rely on immune responses by the plant host to
successfully colonize. Many, if not all, microorganisms, have MAMPs that a plant can
recognize. What are the phenotypic impacts of perpetual exposure to MAMPs? While

2

there are some reports of PRR-triggered immunity (PTI) sculpting plant microbiome
community composition, environmental and ecological complexity, make such
questions difficult to address. Axenic (germfree) animal models have proven valuable to
determine that endogenous microbiota are necessary for the development of a fully
functional immune system and nutrition, but due to lack of infrastructure, analogous
conclusions cannot be drawn for plants.
The plant microbiome, defined here as the collective host-associated environments
naturally colonized by microbiota, is dynamic and compartment-specific. The major
plant microbiomes are the root surface (rhizoplane) and immediately surrounding soil
(rhizosphere), the surface of leaves (phyllosphere), and within tissue beneath the
epidermis (endosphere). Microbiota perpetually colonize these habitats in nature, but
the benefit or detriment of a colonized microbiome is largely unknown. Most microbiota
do not confer obvious phenotypic impacts on their respective plant hosts and
historically have long been regarded as commensal.
With new and accessible ‘-omics’ technologies, along with the foresight to develop
effective agricultural strategies to meet future food demands of a growing global
population, a multidisciplinary plant microbiome research field is emerging. Over the
past 20 years, innovations in DNA sequencing technologies have allowed us to explore
new horizons with culture-independent methodology to characterize microbial
environments. We now know from the Human Microbiome Project that the cells of a
typical human are outnumbered by associated microbial cells ten-fold, and the number
of genes from any given human microbiome outnumbers human genes by
approximately 100:1 (Goodman et al., 2011). Studies over the past decade provide us

3

with insight into plant-associated communities as well, revealing that astounding
microbial diversity is present on (and within) the roots, vascular tissue, and foliage in
nature (Buée et al., 2009; Delmotte et al., 2009; Ikeda et al., 2009). Historically, plant
microbiome research has been approached from two different angles: a
mechanistic/reductionist approach (plant pathology and plant molecular biology) and
an ecological/systems approach (microbial ecology). In this chapter, I will highlight
important technological breakthroughs and discoveries in the fields of molecular plant
pathology/biology and microbial ecology that establish the foundation of plant
microbiome research. This chapter also provides a review of key recent publications to
provide the reader with a current status of the field. The gradual convergence of
molecular plant biology and microbial ecology marks the beginning of a plant
microbiome renaissance. A nearly infinite reservoir of genetic functional potential within
the plant microbiome conceptually extends the phenotypic capacity of the plant. A
better understanding of microbiome structure and function will bestow the power to
intelligently manipulate plant-associated microbial communities to control disease,
increase crop yield, and achieve optimal phenotypes.

4

Early plant pathology
Crop diseases have had a profound impact on civilizations over the course of history.
Global potential yield loss due to disease is estimated to be approximately 40%, and
regional epidemics pose a significant threat to food security and economic stability
(Savary et al., 2012). Plant pathogenic microorganisms, including certain bacteria,
fungi, oomycetes, and nematodes, are responsible for many crop diseases. The Great
Potato Famine of the 1840s, caused by the oomycete pathogen Phytophthora
infestans, scourged the country of Ireland and resulted in a 25% population drop
(Agrios, 2005). A fungal pathogen responsible for brown spot disease in rice, Bipolaris
oryzae, contributed to the Bengal Famine of 1943, resulting in loss of an estimated 4
million lives (Dasgupta, 1984). In the 1970s, the Southern Corn Leaf Blight of North
America caused by the fungal pathogen Cochliobolus heterostrophus resulted in over
$1 billion USD in crop losses (Tatum, 1971). Citrus greening (‘Ca. Liberibacter spp.'),
the most devastating disease of citrus, cost the Florida citrus industry $3.63 billion and
6,611 jobs from 2006-2012(Alvarez et al., 2016). A number of bacterial pathogens,
particularly of the genera Pseudomonas, Burkholderia, Xanthomonas, Ralstonia and
Erwinia, are of significant economic importance (Mansfield et al., 2012; Pfeilmeier et al.,
2016). Worldwide, potential yield losses caused by microbial phytopathogens, insects,
and invasive weeds, are estimated to account for up to $500 billion USD (Agrios, 2005).
Early plant-microbe interaction research and the beginnings of plant pathology were
driven by the desire to understand and combat crop disease. Plant pathology was a
field driven by necessity, and some would argue, the need to control crop diseases set
it apart from basic science (Sequeira, 1988). The enigma of the potato late blight was

5

solved in 1853 when Anton de Bary, widely regarded as the father of plant pathology,
discovered the disease was caused a microorganism (later classified as the oomycete
Phytopthora infestans) (Agrios, 2005). It is worthwhile noting that in addition to his
research into plant-pathogen interactions, de Bary was the first to coin the term
“symbiosis”, which he used to describe the symbiotic relationship of fungi and algae
that comprise lichen (de Bary, 1879). One of many noteworthy alumni of de Bary’s
laboratory is Pierre-Marie-Alexis Millardet. As an early pioneer of translating plant
pathology research to the field, Millardet is accredited with the first successful
chemical intervention to control downy mildew, by use of the “Bordeaux Mixture”
fungicide, consisting of copper (II) sulfate and slaked lime (Swingle, 1896). Another
early pioneer of plant pathology was T.J. Burrill, who discovered that the bacterium
Erwinia amylovora is the causative pathogen of fire blight, a destructive disease to
many fruit trees (Agrios, 2005). Analogous to medical pathology, plant pathology was
driven by the paradigm that a plant disease was the result of infection of a susceptible
host by a compatible pathogen, and the objective of a plant pathologist was to identify
said pathogen and characterize the binary interaction to elucidate the basis of the
disease.
Eventually, the concept of a disease-conducive environment came to light. This
expanded the historical notion that the mere presence of a pathogen on susceptible
host tissue was sufficient to cause disease. Certain environmental variables, such as
temperature and humidity, were important factors for the establishment of diseaseconducive or disease-suppressive environments. A conceptual “plant disease triangle”
was proposed to emphasize the importance of the environment in disease, stating that

6

the three prerequisites for plant disease epidemics are 1) a susceptible host, 2) a
compatible pathogen, and 3) a conducive environment (Figure 1; Stevens, 1960). Yet
still to this day, plant-microbe interaction research is overwhelmingly dominated by a
handful binary pathogenic interactions.

7

Figure 1.1. Plant disease triangle (adapted from Stevens 1960), depicting the factors
necessary for a crop disease to occur and develop to an epidemic. A host plant must be
developmentally and genetically susceptible to the pathogen. The environment must be
conducive to pathogen proliferation, dispersal, and must confer disease-conducive host
physiology to accommodate pathogen proliferation. The pathogen must have appropriate
virulence factors to replicate to high levels within the host plant.

8

Microbial ecology of plant and soil environments
Diverse microorganisms colonize soil in great abundance and provide ecosystem
services upon which all life depends. An early pioneer of microbial ecology was
Beijerinck. Among his many important discoveries in the field of microbiology,
Beijerinck was the first to document biological nitrogen fixation and sulfate reduction,
thus demonstrating that microorganisms play a critical role in soil fertility. Beijerinck
further reported that N-fixing bacteria were capable of colonizing and forming nodules
in the roots of leguminous plants (Chung and Ferris, 1996). Winogradsky, another
pioneering soil biologist and microbial ecologist, made major contributions to the field
by isolating the first nitrifying bacteria and demonstrating critical roles for
microorganisms in the sulfur and nitrogen cycles. Beijerinck and Winogradsky were
among the first to promote the ecological concept of microorganisms. The ecology of
microorganisms was not considered by most of their contemporaries in the latter half
of the 19th century, who instead focused on individual pathogenic microorganisms.
Interestingly, Winogradsky began his career as a plant biologist. He later conducted
research under the supervision of Anton de Bary, the aforementioned father of plant
pathology, but focused on sulfur cycling in soil and pioneered microbial isolation by
enrichment and dilution (Dworkin, 2012). Winogradsky was also an influential
proponent of environmental stabilization and reduction of complexity for reproducible
experimentation. As recounted by Selman Waksman, “...experiments carried out by the
youthful Winogradsky may well be considered as among the first careful investigations
ever made on the influence of controlled environment on the growth of microorganisms
in pure culture, under well-defined experimental conditions” (Dworkin, 2012). Needless

9

to say, although he did not study the plant microbiome in the context of microbial
communities interacting with a host, a strong case can be made that plant microbiome
research was first conceived in the laboratory of de Bary.
Winogradsky’s careful technique to establish axenic cultures, free of contamination,
was also exemplified in the work of the French scientist and close colleague of Louis
Pasteur, Émile Duclaux. In addition to his many seminal contributions to medicine,
microbiology and the Germ Theory of Disease, a lesser-known contribution by Pasteur
that helped to shape the future of microbiome research came from a challenge he
posed to the French Academy of Sciences. Perhaps inspired by the concept of
pasteurization, Duclaux reported that peas and beans could not thrive in thoroughly
sterilized soil and postulated that plants rely upon microbiota for adequate nutrition
(Fairman, 1887). Pasteur addressed the French Academy to recognize Duclaux’s work,
and during his speech, digressed to express doubt that complex multicellular life was
capable of survival under axenic, or germ-free, conditions (Pasteur, 1885). Maybe this
was true skepticism, or perhaps it was a call to challenge dogma, but nonetheless,
Pasteur's skepticism catalyzed the emergence of a new field – Gnotobiology. Animal
researchers were not far behind the plant researchers, and systems to rear various
animals under effectively germfree conditions were developed by the 1960s (Luckey,
1965). In the words of Duclaux, indeed, “the most fruitful periods of Science are those
in which dogmas are shaken” (Duclaux, 1920).
If plant microbiome research was conceived in Anton de Bary’s laboratory in
Strasbourg, surely it was born in the laboratory of Lorenz Hiltner. Based on culturebased methodology and microscopic investigation, Hiltner postulated that a microbial

10

community distinct from soil colonizes the roots of plants, likely feeding off root
exudate material (Hiltner, 1904). He used the now-colloquial term “rhizosphere” to
describe the root-associated microbiome. He further postulated that rhizosphere
composition substantially influences plant nutrition (Hartmann et al., 2008). Perhaps
Hiltner’s most prophetic insight was the idea that the taxonomic composition of
rhizosphere microbiota directly influences plant resistance to pathogens, a concept
that is still an active topic of research more than a century later (Hartmann et al., 2008).
Interestingly, the entrepreneurial Hiltner shares the first patent for agricultural usage of
a microorganism on the use of microbiota for Rhizobium inoculants (Hartmann et al.,
2008). To put this in perspective, in 2016, the agricultural biostimulants industry (not
including biofertilizers and bioinsecticides) is valued at approximately $3 billion USD
with an annual projected growth of greater than 10% through 2022 (Raj Khosla,
personal communication).
It took nearly 50 years for the microbial communities associated with the above-ground
portion of plants to gain attention by the scientific community. In terms of surface area,
the plant leaf surface is reported to be the largest biological interface (Delmotte et al.,
2009). This environment, including the air-filled apoplastic space of leaf interiors, is
termed the “phyllosphere” (Ruinen, 1961). The early history of phyllosphere research is
not as rich as that of the rhizosphere, and most literature on leaf-associated microbiota
prior to 1960 is almost exclusively focused on individual pathogenic microorganisms.
However, by 1930, reports had been published about nutrient exudation and water
exchange via the leaf surface, and that leaf-associated microbiota colonized
heterogeneously. Microscopic and culture-based evaluations of phyllosphere

11

microbiota can be found, claiming significantly different fungal and bacterial
composition than root-associated communities (Ruinen, 1961). As early as the 1960s,
N-fixing bacteria had been isolated from the leaves of tropical plants, but whether or
not they contributed to plant nutrition or phenotypes was unknown (Ruinen, 1961).
Ruinen further reported that precipitation and moisture content of leaves profoundly
impacted microbial abundance and diversity (Ruinen, 1961). Among collections from
diverse plants in tropical forests of Indonesia, Beijerinckia spp were the most
ubiquitous N-fixing bacteria, but other unidentified bacteria, fungi, algae, and protists
could be spotted via microscopic evaluation (Ruinen, 1961). Notably, in Rubin’s 1961
report, he claims that under normal circumstances, nutrients on the water droplets on a
leaf surface are only sufficient for a limited number of cellular divisions, and that
vigorous microbial growth only occurs during a fully mature leaf robust photosynthetic
activity, and only within pools of water on the leaf; the waxy cuticle is not conducive to
much microbial replication.

The Arabidopsis thaliana model
Domesticated crops and their pathogens are often difficult to research for various
reasons, prompting the adoption of model systems with more powerful tools and fewer
logistical constraints. It is important to have multiple diverse models to account for
genetic variation across the plant kingdom, but the clear poster child of plant model
systems is mouse-ear cress (Arabidopsis thaliana, hereafter Arabidopsis). Arabidopsis
has proven to be an outstanding model organism for plant research, and it has been
used to make seminal discoveries in hormone signaling, stress tolerance, growth, and

12

development, and disease resistance (Provart et al., 2016). Arabidopsis is an
herbaceous flowering dicot of the Brassicaceae family (mustard family), with a relatively
short generation time and a fully sequenced genome of 135 Mbp (Arabidopsis Genome
Initiative, 2000). By the 1980s, Arabidopsis began to rapidly gain popularity for plant
researchers around the world (reviewed by Meyerowitz, 2001). After the initial discovery
that Arabidopsis was conducive to transformation by Agrobacterium tumerfacians,
early methods began to emerge to generate selectable Arabidopsis mutants via T-DNA
insertion of antibiotic resistance genes (Lloyd et al., 1986). Subsequently, the genome
of Arabidopsis has been sequenced, and several large and well-characterized T-DNA
mutant libraries are now available to the public. Arabidopsis is physically small relative
to most domesticated crops, has a short generation time, grows under relatively simple
environmental conditions, can be crossed, and has high fecundity. Translation of
research is important for model systems, and indeed, key discoveries applicable to
other plants have been made in Arabidopsis (Chang et al., 2016). Perhaps most
importantly, there is a massive community of Arabidopsis researchers, thus allowing for
experimental replication across different labs and a plethora of tools. Arabidopsis is a
well-established organism for molecular host-microbe interaction research. One such
example is the interaction between Arabidopsis and the pathogen Pseudomonas
syringae pv tomato DC3000 (Pst), the causative agent of bacterial speck disease (Xin
and He, 2013). In the early 1990s, it was discovered that Pst had the capacity to cause
disease in Arabidopsis, and the Pst-Arabidopsis pathosystem quickly gained popularity
as a platform to dissect bacterial virulence and plant defense (Whalen et al., 1991). A
number of plant-microbe model systems are now established that utilize Arabidopsis

13

as the host, many of which have been used to report fundamental mechanisms of
pathogenesis and eukaryotic innate immunity.

Arabidopsis innate immunity
Beyond physical barriers such as the waxy leaf cuticle, the first layer of defense against
subversion by microorganisms in Arabidopsis is the innate immune response. The cost
of pathogenic relationships and benefits of symbiosis provide adequate selective
pressure for the evolution of defense adaptations and intricate signaling networks
collectively termed innate immunity. Successful organisms allocate resources towards
growth and defense in a niche- and stress-dependent fashion. Defense overspending
can be detrimental. Thus, mechanisms to control immune response sensitivity and
specificity are important. Despite the fact that host organisms are perpetually colonized
in a microbial milieu, very little is known about host-microbiome dynamics or the
impact of innate immunity on associated microbiota. Recognition and response to nonself molecular patterns by pattern recognition receptors (PRRs) is collectively termed
innate immunity. To initiate a pattern-triggered immunity (PTI) response, a non-selfligand is first bound by the extracellular domain of a PRR, resulting in
autophosphorylation of the cytosolic ectodomain and heterodimerization with other.
One such example is recognition of a 22AA motif of the flagellin monomer of certain
bacteria (flg22) by the PRR FLAGELLIN SENSITIVE 2 (FLS2) (Gómez-Gómez and
Boller, 2000) (Figure 2).

14

Figure 1.2. Immune responses at the plasma membrane in Arabidopsis. The extracellular
leucine-rich repeat domain of a receptor-like kinase (LRR-RLK) pattern recognition receptor
(PRR) protein (for example, EFR or FLS2) bind the foreign ligand (EFR binds 18AA ‘elf18’ from
certain bacterial EF-tu; FLS2 binds 22AA ‘flg22’ from certain bacterial flagellin), resulting in
autophosphorylation of the ectodomain and the ectodomain of the co-receptor BAK1.
Activated BAK1 directly phosphorylates the NADPH oxidase RBOHD, and also phosphorylates
BIK1, which also phosphorylates RBOHD but at a higher affinity. RBOHD reduces dioxygen to
superoxide, and in the apoplastic intercellular space between cells, superoxide dismutases
(SOD) convert the free superoxide to hydrogen peroxide. This process is called the reactive
oxygen species (ROS) burst. Increased concentrations of apoplastic H2O2 results in positive
feedback and influx of calcium into the cytosol. CPK enzymes interact with RBOHD to increase
its activity. (reproduced from Kadota et al., 2015).

15

Figure 1.3. MAMP-triggered immune response. PTI activation results in activation of the
mitogen-associated protein kinase (MAP kinase) cascades. Activated MAP kinase proteins
(MEKK1, MKK5, MPK3/6, and/or MEKK1, MKK1/MKK2, MPK4, MKS1) subsequently localize
to the nucleus (Couto and Zipfel, 2016; Xin and He, 2013). Activated MAP kinase proteins bind
defense-associated transcription factors, for example, certain WRKY transcription factors),
resulting in defense associated gene expression (for example Frk1 and Pr1) (Boudsocq et al.,
2010). Additional responses include biosynthesis of Pathogenesis Related proteins (PR
proteins); changes in vesicle trafficking, change in redox homeostasis, and increased levels of
cytosolic free salicylic acid (SA) (Chisholm et al., 2006). A later response is cell wall
reinforcement with callose.

16

Plant PRRs are the convergently-evolved counterparts to Toll-like receptors in animals
(Gómez-Gómez and Boller, 2000). For example, some plants perceive a particular motif
of bacterial flagellin through the PRR FLS2 (FLAGELLIN SENSITIVE 2). Similarly,
mammals use the Toll-like receptor TLR5 to perceive bacterial flagellin and mount a
downstream immune response (Hayashi et al., 2001), albeit a different motif.
During PTI, plants undergo multiple physical and chemical changes. For example,
papillae (callose- containing cell-wall appositions) form at infection loci. Callose is a
glucan polymer that serves as a matrix in which plant-produced antimicrobials are
localized and suspended (Figure 3; Gómez-Gómez et al., 1999). Activated PRRs also
initiate downstream defense responses, such as anion fluxes, phosphorylation
cascades (including the MAP-kinase cascade), reactive oxygen species accumulation,
and defense gene activation (reviewed by (Luna et al., 2011). As demonstrated by
Gomez-Gomez and Boller in 2000, exogenous application of MAMP is sufficient to
activate MTI. The three best characterized MAMPs/PRRs for Arabidopsis thaliana Col-0
are: the semi-conserved N-terminus of bacterial flagellin by FLS2, an 18-AA motif of
bacterial EF-Tu by EFR (Kunze et al., 2004), and various forms of fungal chitin and
chitosan by CERK1 (reviewed by Chisholm et al., 2006). Peptidoglycan is also known
to be a MAMP. A number of other plasma membrane-bound LRR-RLKs also exist in
Arabidopsis, to which no respective MAMPs have been identified. It is known that
MAMP-containing microorganisms regularly inhabit plant tissue, but basal activation of
MTI has not been reported. Studies have failed to directly address the effect of
exogenous MAMP application on plant-associated microbial communities. Although
putative genes from some plant-associated bacteria suggest the potential for reactive

17

oxygen species tolerance and alkalization tolerance, it is unknown whether some
organisms are more tolerant to MTI than others, and likewise, it is not known whether
the plant has differential responses to different MAMPs perceived by the same
coreceptor complex, or whether this is a generalized response.

Defense hormone signaling
Salicylic acid and jasmonic acid are the two major plant defense hormones and key
regulators of plant innate immunity. Jasmonic acid signaling pathway has a significant
amount of overlap with the ethylene hormone signaling pathway. The jasmonic acid
and salicylic acid hormone signaling pathways are regarded as antagonistic (Bari and
Jones, 2009). Salicylic acid signaling is implicated in defense against (hemi-) biotrophic
pathogens, while defense against necrotrophic pathogens, including many fungal
pathogens, oomycetes, and insects, are dependent on jasmonic acid signaling.
Pathogen-induced SA biosynthesis occurs in the chloroplast and is dependent upon
isochorismate synthase enzymes, most prominent of which is ICS1 (ISOCHORISMATE
SYNTHASE 1). Although salicylic acid can be synthesized in planta via at least two
independent pathways, the isochorismate (IC) and the phenylalanine ammonia-lyase
(PAL) pathways, the IC pathway is reported to be the most prominent for pathogeninduced production.
The concentration of salicylic acid increases subsequent to an interaction with a
pathogenic microorganism. Despite considerable interest, all the steps of salicylic acid
biosynthesis have not been elucidated fully. Isochorismate synthase is a critical
enzyme in the IC pathway and mutants impaired in isochorismate synthase are more

18

susceptible to pathogen invasion (Dempsey et al., 2011; Wildermuth et al., 2001).
Various forms and modifications of salicylic acid exist in plants. In Arabidopsis, salicylic
acid can be methylated, sulfonated, hydroxylated, conjugated with amino acids, and
glucosylated. Potentially salicylic acid can be modified to other forms, and further
investigation is necessary to make such claims. SA increases the production of
reactive oxygen species such as hydrogen peroxide and in general creates a more
oxidative environment in the cytosol (Tata, 2005). This will increase the concentration
superoxide dismutases. Directly relevant to defense against pathogens, SA prompts
the expression of Pathogenesis related (PR) proteins, some of which are reported to
have antimicrobial activities. Many SA-inducible genes perpetuate a positive feedback
loop of SA-associated defense responses. For example, the SA-inducible protein PAD4
promotes SA biosynthesis in coordination with a similar lipase-like protein EDS1 (Zhou
et al., 1998). The master regulator of SA- and JA mediated transcriptional
reprogramming during immunity is NPR1. Sweeping REDOX changes upon SA
accumulation leads oligomerization of cytosolic NPR1 oligomers, and monomers are
released into the nucleus. Numerous post-translational modifications occur to defenseassociated proteins after pathogen encounter and in response to elevated SA levels
(Withers and Dong, 2017).
Jasmonic Acid and Salicylic acid have been implicated in influencing microbiome
community composition in Arabidopsis. In a field study, JA/ET and SA defensecompromised mutants (in the same A. thaliana Col-0 background), as well as
Arabidopsis with exogenous MeJA or SA application were planted among wildtype
controls, and culturable leaf-epiphytic and leaf-endophytic bacterial populations were

19

monitored (Kniskern et al., 2007). Major findings from this study are 1) foliageassociated epiphytic diversity is significantly greater than endophytic diversity in WT
plants; 2) induction of SA-mediated defenses reduced endophytic bacterial community
diversity; 3) JA-mediated defenses deficiencies confer greater epiphytic bacterial
diversity. The small sample size, use of semi-selective culture media, decision to ignore
fungi, failure to examine non-culturable populations, single sample site without controls
for microhabitat variation, and failure to simultaneously examine root-associated
communities leaves us with many unanswered questions regarding the impact of plant
defense hormones on associated communities. Another study focused on fitness
effects of activated defense signaling (Traw et al., 2007). The Traw et al. study reports
that induction of SAR, but not JA/ET related defenses, increases the fitness of
Arabidopsis in situ, despite the (untested) costs that accompany SAR induction.
Fitness in the Traw et al. study is defined as seed production (seed viability not
addressed), and there was a lack of controls to determine the effects of SA and/or JA
application have on a germ- free plant, and therefore the fitness benefit conferred by
exogenous SA application cannot be attributed to the correlated loss of bacterial
diversity. Another study utilized culture-dependent techniques, as well the cultureindependent community fingerprinting technique denaturing gradient– gel
electrophoresis (DGGE) targeted to the 16S rRNA gene, to examine the effects of
JA/ET and SA on the Arabidopsis rhizosphere (Doornbos et al., 2011). The Doornbos
study was able to identify significantly different rhizosphere community fingerprints for
each of the different Arabidopsis genotypes and control plants of other species, only
three biological replicates were used for each genotype, multiple knock-out lines were

20

not used to determine if indeed the “fingerprint” can be attributed to the mutant allele.
More recently, a reverse genetics approach was published by Lebeis et al. reported
that SA-signaling is required for the homeostasis of root-associated microbiota and
that the taxonomic composition of the root microbiome is sculpted in a SA-dependent
manner. Lebeis et al hypothesize that SA-mediated modulation may occur by a
conceptual “gating” of particular bacterial taxa as a in a manner of tolerance to SAmediated defense, but this raises questions regarding whether previously
uncharacterized branches of immune regulatory networks play a substantial role in
microbiome governance (Lebeis et al. 2015).

Systemic acquired resistance and induced systemic resistance
Systemic acquired resistance (SAR) and induced systemic resistance (ISR) are two
important microbe-induced phenomena with long-term systemic effects on plant
immune response and physiology. After infection, endogenous salicylic acid levels rise
throughout the plant, resulting in systemic “priming” against infection by some, but not
all, microbial pathogens (Durrant and Dong, 2004; Jung et al., 2009). On one hand,
plants with activated SAR have higher basal levels of PR proteins (for example, PR-1)
and other not fully characterized proteins with hypothetical roles in defense. On the
other hand, ISR activated by non-pathogenic rhizobacteria is jasmonate-signaling
dependent, and functions independent of SA signaling (Van Wees and De Swart, 2000).
Both ISR and SAR are dependent on the function of NPR-1, a master defense
response regulator protein (Dong, 2004). Although many intricacies of plant defense
signaling and priming have yet to be unraveled, it appears that SAR is most effective

21

against biographic (derive nutrients from living cells) pathogens and ISR is most
effective against necrotrophic pathogens (Durrant and Dong, 2004).

22

Chapter 2

Construction and optimization of a soil-based growth system conducive to
axenic, gnotobiotic, and holoxenic Arabidopsis growth

Acknowledgements: Western blot contributed by Dr. John Froehlich. Some Plant
images contributed by Cait Thireault. Experiments in the calcined clay system
conducted in collaboration with Dr. Amine Hassani.

23

Abstract
The presence of endogenous microbiota on and inside plants is hypothesized to
influence many phenotypic attributes of the host. Likewise, host factors and microbemicrobe interactions are believed to influence microbiome community assembly.
Rigorous testing of these hypotheses necessitates the ability to grow plants in the
absence or presence of endogenous microbiota. We introduce the FlowPot axenic
growth platform and report vigorous Arabidopsis growth in the presence or absence of
microbiota in the system’s peat-based substrate. Mechanically, the FlowPot system is
unique in that it allows for total-saturation “flushing” of sterile substrate with water
and/or nutrient solution, contributing to the healthy appearance of axenic plants in situ.
We compare bacterial community establishment in two easily-constructed axenic
growth systems on the model plant Arabidopsis thaliana: an established system with
calcined clay substrate and a novel system with peat substrate. We report bacterial
community establishment trends in both systems after inoculation with directlyextracted soil-derived microbiota and the established “At-SPHERE” community that
collectively represents Arabidopsis-associated bacteria cultured from nature. We
report system-specific community shifts, as well as commonalities and discrepancies
in communities by natural and defined input communities. Relative to the calcined clay
growth system, the peat-based FlowPot system retains greater alpha diversity of rootassociated communities, and in particular, allows for more proteobacterial growth. In
contrast, the calcined clay system is conducive to an actinobacterial-dominated
community and increased granularity when comparing rhizosphere versus bulk
substrate bacterial communities. This standardized suite of tools and colonization

24

protocols empowers the plant microbiome research community to conduct harmonized
experiments to reveal mechanistic bases of microbial community succession and the
impact of microbiota on host phenotypes.

25

Introduction
In an 1885 address to the French Academy of Sciences, Louis Pasteur expressed
doubts that axenic (i.e., germ-free) animals were capable of survival (Pasteur, 1885).
Pasteur went on to credit the work of Duclaux who reported that peas and beans could
not thrive in thoroughly sterilized soil and that plants likely require nutritive assistance
from bacteria (Fairman, 1887). Pasteur's skepticism catalyzed the emergence of a new
field – Gnotobiology. Animal researchers were not far behind the plant researchers, and
systems to raise mice, chickens and guinea pigs devoid of conventionally culturable
microorganisms were established before the first half of the twentieth century (Luckey,
1965). The axenic mouse model, in particular, has proven to be a tremendously
valuable tool for microbiome research. Seminal studies demonstrate that enteric
microbiota is essential for proper nutrition, enteric physiology, cardiovascular function,
and immune function (Clemente et al., 2012; Gensollen et al., 2016; Smith et al., 2007).
Two major motivations for establishing axenic experimental systems that also
accommodate gnotobiotic (defined microbiota) or holoxenic (undefined microbiota)
hosts are 1) to elucidate effects of host-associated microbiota and 2) to identify factors
that influence microbiome structure and function. Careful consideration must be taken
when designing an axenic growth system to minimize artifacts and sample variability,
but not at the expense of versatility.
Environmental factors are the major drivers of differential taxonomic composition and
diversity among host-associated microbiomes (Adair and Douglas, 2017; Peiffer et al.,
2013; Spor et al., 2011). Thus, abiotic factors must be accounted for when performing
a microbial colonization experiments for reproducibility of host phenotypes and

26

community dynamics. Variation in the food source from commercial suppliers of nearisogenic mice can result in different microbiome composition, and as a result, altered
host immune phenotypes (Celaj et al., 2014). In the model plant Arabidopsis, humidity
influences microbiome composition and the plant’s ability to defend against foliar
pathogens (Xin et al., 2016). Ecological drivers of soil microbiomes are associated with
physical and chemical factors, including, but not limited to porosity and water
retention, C/N/P availability, pH, percent organic content, and cation exchange
capacity (Andrew et al., 2012; Fierer and Jackson, 2006). When conducting a
microbiome colonization experiment, and comparing gnotobiotic or holoxenic plants to
axenic plants, it is essential to consider abiotic factors of the substrate as well as the
growth environment, particularly if the intention is to recapitulate a microbial
community reminiscent of a “natural” soil microbiome.
Relative to the mechanical, technical, and financial requirements to establish and
maintain an axenic mouse colony, an axenic system for a model plant such as
Arabidopsis is straightforward (Nicklas et al., 2015). Axenic Arabidopsis growth can be
accomplished using routine tissue culture methodology, which in brief entails aseptic
placement of a decontaminated seed on a phytonutrient agar substrate (or similar)
contained within a sterile, gas-permeable, transparent vessel. Tissue culture systems
do not provide a soil-like scaffold for microbial colonization. Furthermore, agar-based
systems are notorious for non-uniform nutrient and O2 delivery over time (Gunning and
Cahill, 2009). Hydroponic and aeroponic systems can alleviate the uniformity of nutrient
and O2-delivery by agitation and media replenishment, but do not provide a soil-like
scaffold for microbial colonization (Jackson et al., 1991). Furthermore, it can be

27

challenging to maintain axenic conditions or prevent cross-contamination in commonreservoir hydroponic systems.
Non-soil substrates such as sand, quartz, vermiculite and calcined clay are frequently
used in gnotobiotic systems (Bai et al., 2015; Henry et al., 2006; Kloepper et al., 1981;
Lebeis et al., 2015). These substrates are porous, thus providing surface area for
microbial colonization and root penetration. However, batch-to-batch variation of
ceramic substrates can result in a wide range of labile ions (Henry et al., 2006).
Calcined clay, for example, has sorptive properties that can result in reduced labile
concentrations of P, Fe, Cu and Zn, and desorptive properties that can result in excess
labile B, Mg, Ca, S, K, and sometimes to toxic levels, e.g. Mn (Adams et al., 2014).
While thorough washing or soaking of the non-soil substrate can reduce the initial
excess of labile ions, flow and drainage are important to reduce significant changes in
chemistry over time. Furthermore, the aforementioned non-soil substrates lack
significant organic carbon typical of soil, unless supplemented.
Soil has also been used as substrate in axenic systems, although it can present
challenges due to contamination from insufficient sterilization and hindered plant
growth due to suboptimal sterilization protocols. Numerous sterilization methods have
been used with soil, including: autoclaving, dry heat, irradiation, microwave, fumigation
by gaseous chemicals, saturation with various sterilants (i.e., mercuric chloride, sodium
azide, formaldehyde) (Trevors, 1996). All methods of soil sterilization alter, to some
extent, physical and/or chemical properties of soil. Chemical sterilization methods are
not appropriate for plant growth systems due to phytotoxic effects. Autoclaving soil
has been shown to increase levels of water-soluble carbon and reduce pH (Shaw et al.,

28

1999), but not significantly alter ion exchange capacity. Gamma-irradiation has been
reported to minimally disrupt the physical attributes of soils but can result in the
generation of reactive oxygen species, capable of depolymerizing C-C of
polysaccharides (Bank et al., 2008). Both autoclaving and gamma irradiation can result
in changes of the physical structure of the soil, exposing more surface area and thus
altering sorptive properties. Complete sterilization of some soils can be achieved with
minimal chemical alterations by autoclaving a thin layer of soil for three short (<45 min)
autoclave cycles with day intervals (Berns et al., 2008; Lotrario et al., 1995; Wolf et al.,
1989). Subsequent flushing of the sterile substrate can increase plant performance,
presumably by rinsing away soluble phytotoxic byproducts.
Here, we compare bacterial community establishment in two easily-constructed axenic
Arabidopsis growth systems: an established system with calcined clay substrate
(Bulgarelli et al., 2012; Lundberg et al., 2012) and a novel system with peat substrate.
The peat-based FlowPot system features an inoculation port on each individual vessel
that allows for substrate flushing to remove soluble byproducts of sterilization,
provides drainage, and accommodates homogenous inoculation with microbiota
and/or nutrients. We inoculate both systems with natural soil-derived microbiota from a
field abundant with Arabidopsis, and the established “At-SPHERE” synthetic
community that collectively represents Arabidopsis-associated bacteria (Bai et al.,
2015). This set of experiments seeks to empirically determine whether growth system
influences microbial community succession in synthetic and natural communities. This
standardized suite of tools and colonization protocols empowers the plant microbiome

29

research community to conduct experiments to reveal mechanistic bases of microbial
community succession and the impact of microbiota on host phenotypes.

30

Materials and Methods
Sterilization of the FlowPot peat-based substrate
Peat potting mix (Sunshine Mix (LG3), SunGro Horticulture, USA) was blended with
horticulture-grade medium vermiculite (1:1). The substrate mixture was sprayed with
distilled water to achieve ~20% moisture content. Substrate was evenly spread at a
depth of 4 cm onto polypropylene or glass trays and covered with aluminum foil in
such a way to prevent condensate from pooling and unevenly saturating the substrate,
and autoclaved for 30 min on liquid cycle (121.1°C, 15 PSI, slow exhaust with forced
liquid cooling). After autoclaving, the substrate was stored for 24 to 48 h at room
temperature. The autoclave cycle was repeated, and the substrate was stored for an
additional 24 to 48h. The substrate was then aseptically homogenized and ready for
use. The substrate was prepared immediately before FlowPot construction. In our
experience, depending on the initial moisture content of the substrate, relative
humidity, and the calibration of the autoclave, autoclave parameters may need to be
optimized empirically.
FlowPot growth system assembly
FlowPots were constructed and assembled using easily-obtained laboratory materials
(for supplier details, see Table 1). Luer taper syringes (50 ml, Polypropylene, nonlubricated) were sectioned using a mitre saw with a reversed fine-tooth blade at the
"20 ml" mark, retaining only the portion with the Luer connector. Residual debris was
removed, and syringes were soaked for 20 min in Multi-Terge ionic detergent (2% v/v),
followed by thorough rinsing with distilled water and sterilization by autoclave.
Fiberglass screen (Phiferglass, Pfifer Inc., USA) was sectioned into 5 cm squares and

31

autoclaved. To create the FlowPot platforms, autoclave-compatible plastic
(polypropylene or polycarbonate, typical for pipet tip boxes) was sectioned into 12 cm
× 8 cm × 1 cm blocks and four holes (8.8 mm diameter) were drilled. Each stand was
fastened to the inside bottom of Eco2 micro box (model TPD1600 (XXL), Combiness,
USA), secured with filament tape.
Sterile 3 mm soda-glass beads (n=10) were aseptically added to each syringe top and
placed on a culture tube rack. The syringe tips were gently packed with the twiceautoclaved substrate mixture until slightly heaping (Figure 1). In our experience,
consistency of substrate packing density and avoidance of overly-compressing the
substrate were important for optimal plant growth results. The FlowPot construction
was completed by covering the substrate-filled syringe tips with a fiberglass screen
(one per unit), and securing the assembly with a polycarbonate cable tie (Figure 1).
Completed FlowPots were placed on a culture tube racks, put into covered
polypropylene bins, covered with autoclave paper or foil, and autoclaved for 45 min on
liquid cycle (121.1°C, 15 PSI, slow exhaust with forced liquid cooling). Immediately
after autoclaving, the FlowPots were placed in a sterile laminar flow hood or biosafety
cabinet to a sterile laminar flow hood. Constructed microboxes were placed in
polypropylene bins and covered with autoclave paper. Micro boxes were not stacked,
and lids were not secured to allow for thorough autoclave sterilization.

32

Figure 2.1. Schematics of FlowPot construction and irrigation. Each FlowPot was prepared
by fastening a (i) zip-tie to a (ii) mesh retainer to secure (iii) sterilize peat/vermiculite substrate
and (iv) glass drainage beads within the (v) 20 ml portion of a 50 ml Luer-lock syringe barrel.
Assembled FlowPots were autoclaved, (vi) aseptically rinsed with sterile water, and (vii)
irrigated with nutrient solution and (optionally) microbiota.

33

Calcined clay system assembly
The calcined clay growth system was assembled according to a previously published
protocol (Bai et al., 2015; Lebeis et al., 2015). In brief, the substrate (Diamond Pro
Calcined Clay Drying 570 Agent, Diamond Pro, Arlington, TX, USA) was thoroughly
rinsed with sterile ddH2O, then autoclaved for 45 min and incubated at 65°C until
leather-dry. Calcined clay was aseptically added to Magenta boxes at 100 g per box
(approximately half-filled).
Soil collection
German Soil (CAS10) was collected during Spring 2014 from an agricultural field in
Cologne, Germany (50.958°N, 6.865°E), at the Max Planck Institute for Plant Breeding
Research. The soil has not been subjected to cultivation or chemical treatments for a
minimum of 10 years. Large particulate debris was removed from the soil using a steel
sieve (3 mm). This identical soil has been used in previously a published study (Zgadzaj
et al. 2016).
Michigan Soil (MS15MSU) was collected during Fall 2015 from a sandy loam
agricultural field at Michigan State University, East Lansing, Michigan, USA (42.709°N,
-84.466°E). The field was being used for Miscanthus cultivation and had not been
subjected to tilling, fertilization, or any intervention for a minimum of 10 years. The soil
was collected from 5 to 15 cm below the surface. Upon collection, the soil was spread
out on tables and allowed to sit for one week at room temperature with ~50% relative
humidity. Large debris was removed using a 3 mm galvanized metal screen, and 50 g
aliquots were prepared in Whirl-Pak bags and stored at 4°C in the dark.

34

Preparation of input community microbiota
The soil was suspended in sterile distilled water (50 g∙L−1) in a covered 2L Erlenmeyer
flask and agitated on a rotary shaker for 20 min at 22°C. Afterwards, the soil slurry was
allowed to settle for 5 min and filtered through a 40 μm cell strainer into a sterile
medium bottle. When required for axenic growth system preparation, a portion of the
soil slurry was autoclaved for 45 min. Immediately prior to growth system inoculation,
Murashige Skoog nutrient solution (pH 5.7, MES-buffered, Gamborg vitamins) was
added to the slurry, bringing the final concentration to either 1/2X or 1/4X, as specified
(Murashige and Skoog, 1962). The culture-derived input community At-SPHERE was
prepared as published previously and resuspended in Murashige Skoog nutrient
solution (pH 5.7, MES-buffered, Gamborg vitamins) at 106 CFU∙ml−1 concentration.
Growth system inoculation
Using a sterile test tube clamp or equivalent holder, each FlowPot was aseptically
inverted over a sterile funnel placed atop a waste flask. While inverted, each FlowPot
was aseptically irrigated by upward flow from the bottom over the course of 30 s with
sterile distilled water (50 ml, 65C) using a female-female Luer coupler on a 50 ml
syringe (Figure 1). To reduce the risk of contamination, infiltrations were performed in a
biosafety cabinet cleaned with Spor-Klenz (STERIS, USA), instruments (Supplemental
Figure 1) were flame-sterilized between each unit, and infiltrated FlowPots were placed
on a sterile culture tube rack between infiltrations. After irrigation with water, FlowPots
were allowed to sit for 30 min. Subsequently, FlowPots were irrigated with the
prepared input microbiota. Inoculated FlowPots were then aseptically placed in sterile
micro boxes, totaling four per micro box.

35

Arabidopsis growth conditions
Arabidopsis seeds (250-300 µm) were decontaminated using a chlorine gas sterilization
protocol for 6-8 h (Clough and Bent, 1998). Seed-borne contamination and germination
efficiency were evaluated by incubating an aliquot of seeds on R2A agar for one week
at 22°C. Prior to sowing, seeds were suspended in sterile distilled water at 4°C in the
dark for 48-72 h. Per FlowPot or Magenta Box, eight seeds were sown. Plants were
grown in growth cabinets set to the following parameters: 23°C with 12/12 h day/night
light cycle at an intensity of ~80 µE∙m−2∙s−1. Approximately two weeks after
germination, seedlings were thinned from each FlowPot and Magenta Box, leaving
three plants per FlowPot. For an additional contamination check, removed seedlings
were incubated on R2A agar for seven days.
Protein extraction and quantification of photosynthesis-associated proteins
Briefly, single Arabidopsis leaves (3 weeks old) from FlowPot-grown axenic or
holoxenic plants (MS13MSU community) were collected and homogenized using
Minute-Chloroplast Isolation Kit (Invent Biotechnology, Inc. Plymouth, MN, USA)
according to manufacturer protocol. Chlorophyll content of total leaf homogenate was
determined by the method of Arnon (Arnon, 1949) to give a final concentration of 1mg
chlorophyll/ml for all samples. All total leaf homogenate (5 mg chlorophyll total)
samples were solubilized in Laemmli buffer (Laemmli, 1970) and resolved by SDSPAGE and either stained with Coomassie Blue or transferred onto polyvinylidene
difluoride (PVDF) membrane (Invitrogen) and probed with the following antibodies: antiD1 Protein (Agrisera, AS01 016), OEC33 (Agrisera, AS06 142-33), OEC23 (Agrisera,
AS06 167) or antibodies produced in-house: Anti-Toc159, Anti-Toc75, anti-Tic110,

36

anti-ClpC. All primary antibodies were incubated using a 1:4,000 dilution. The
detection method employed used a secondary anti-rabbit conjugated to alkaline
phosphatase (KLP, Inc. Gaithersburg, MD) at 1:5,000 dilution for 1 hour in 5%
DM/TBST. The blots were developed using a standard Alkaline Phosphatase (AP)
detection system with BCIP/NBT as substrates (Sigma-Aldrich, St. Louis, MO, USA).
Quantification of photosynthesis-associated proteins was performed by Dr. John
Froehlich, Michigan State University.
Sample collection and DNA extraction
At-SPHERE input microbiota aliquots (3 x 15 ml) were collected at the time of each
inoculation, concentrated by centrifugation (15 min at 4,000 x g). Bacterial pellets were
resuspended in nuclease-free water (Qiagen) stored until Lysing Matrix E tubes at -80C
until further processing. German soil input microbiota aliquots (3 x 500 mg) were
collected in Lysing Matrix E (MP Biomedicals), and stored at -80C until further
processing.
After six weeks of co-incubation in a light cabinet, roots and substrate samples were
collected separately from each growth vessel. Roots from each magenta box or micro
box were pooled per container and washed twice in sodium phosphate buffer
containing Silwet L-77 (0.02% v/v) to remove adherent debris. Within 2 hours, roots
were placed in a Lysing Matrix E tube (MP Biomedicals), immediately frozen in liquid
nitrogen, and stored at -80°C until further sample processing.
Microbiota from clay substrate were suspended by vigorously shaking in sodium
phosphate containing Silwet L-77 (0.02% v/v) and concentrated by centrifugation (10
min at 4,000 x g). Bacterial pellets were resuspended in nuclease-free water and

37

transferred to Lysing Matrix E to be stored at -80°C until further sample processing.
For peat substrate, a 500 mg aliquot was collected per FlowPot in a Lysing Matrix E
tube and stored at -80C until further processing.
For DNA extractions, samples were homogenized twice by a Precellys 24 tissue lyser
(Bertin Technologies), and DNA was extracted according to the manufacturer's
protocol (FastDNA SPIN kit for Soil; MP Biomedicals). DNA was eluted in water and
stored at -20°C.
16S rRNA gene fragment amplification and MiSeq Library preparation
The 16S rRNA gene fragment DNA libraries for Illumina sequencing were prepared
using a previously published two-step amplification protocol (Bai et al., 2015). In brief,
DNA concentrations of gel-purified amplicon DNA were adjusted to approximately 3.5
ng∙µl-1. We targeted the V5-V7 region of the 16S rRNA gene using PCR primers 799F
(5'-AACMGGATTAGATACCCKG) and 1193R (5'-ACGTCATCCCCACCTTCC).
Thermocycling conditions and reagent concentrations were performed according to a
previously published protocol (Bai et al., 2015). All reactions were performed in
triplicate, and products subsequently pooled. Enzymatic cleanup of the PCR reactions
to remove residual nucleotides was performed using 20U Exonuclease I (20U) and
antarctic phosphatase (5U) for a 15 min incubation (37°C), followed by a 15 min heatinactivation (85°C) (New England, BioLabs). The enzymatically processed DNA
amplicons were then PCR-barcoded with 12 bp GoLay tags (Caporaso et al., 2012)
and Illumina sequencing adapters according to the manufacturer’s protocol. Barcoded
amplicons were gel-purified according to the manufacturer’s protocol (QIAquick Gel
Extraction Kit; Qiagen). DNA samples were quantified, pooled, and equilibrated to

38

~7 ng∙µl-1. The amplicon libraries were then subjected to AMPure XP bead purification,
according to the manufacturer’s protocol (Agencourt, Beckman Coulter). Sequencing
was performed using the Illumina MiSeq platform with the MiSeq Reagent kit (version 3
chemistry) and 2 x 300 bp paired-end sequencing protocol.
Preprocessing of 16S rRNA gene fragment amplicons
Quality-filtered reads were demultiplexed by barcode, allowing for one mismatch per
barcode. Paired-end reads were merged using PANDAseq (version 2.11; (Masella et
al., 2012) using the default parameters for the Ribosomal Database Project pipeline
and a minimum PHRED score of 25 (Cole et al., 2014). The PANDAseq parameters
were used: “-A rdp_mle -C min_phred:25 -d rbfkms”. Reads were further processed
using the previously published unsupervised UPARSE method, using the USEARCH
software package (version v10.0.240; Edgar, 2013). OTUs were clustered at 97%
identity into operational taxonomic units (OTUs), and were classified using the RDP
reference database to the highest taxonomic level with >70% confidence (release 11.5;
(Cole et al., 2014). Bray-Curtis and weighted UniFrac distances were computed using
the R package “phyloseq”. Weighted UniFrac distances were calculated between each
sample’s OTU assemblages (Lozupone et al., 2011). For alpha-diversity analysis of the
bacterial communities, read counts were rarefied to an even sequencing depth of
5,000 reads per sample. For beta-diversity analysis of the bacterial communities, read
counts were normalized by cumulative sum scaling normalization factors.

39

Results
Arabidopsis growth in axenic FlowPots
For all experiments presented herein, axenic plants were grown in tandem. A common
problem with axenic growth systems that use soil substrate is the release of phytotoxic
byproducts during sterilization, resulting in stunted growth and chlorosis or
anthocyanin accumulation (Jackson et al., 1991; Trevors, 1996). To assess whether
Arabidopsis grown under axenic conditions in the FlowPot system, which involves a
critical flushing step to remove possibly phytotoxic byproducts, exhibited any
symptoms of phytotoxicity, we carefully examined whole rosettes up to 9 weeks post
germination in parallel with Arabidopsis grown in holoxenic FlowPots inoculated with
agricultural soil collected from Michigan. Over the course of 9 weeks, we failed to
identify any chlorotic tissue or visible anthocyanin accumulation (Figure 2). There were
no visible signs of stress in Arabidopsis in the calcined clay system. Microbial
contamination was checked using R2A medium at the time of sowing and at the time of
collection, and any micro boxes that contained contaminated samples were discarded.
To further assess the health of axenic Arabidopsis from the FlowPot system, we
quantified seven different photosynthesis-associated proteins from total protein
extracted from whole rosette tissue. The abundance of axenic Arabidopsis
photosynthesis-associated proteins was quantified relative to simultaneously-grown
Arabidopsis in holoxenic substrate (inoculated with MS15MSU Michigan soil
microbiota) via Western Blot using native antibodies (Figure 2). The proteins quantified
were two outer envelope chloroplast translocases (Toc159, Toc75), an inner envelope
chloroplast translocase (Tic110), ATP-Dependent Chloroplast Protease (ClpC), and

40

three photosystem II complex proteins (D1 protein, OEC33, OEC23). All seven proteins
were equivalently abundant in holoxenic and axenic rosettes.

41

B

A

Axenic

Holoxenic

Axenic Holo
Toc / Tic Complex PSII Complex

Toc159
Toc 75
Tic 110
ClpC
D1 Protein
OEC33
OEC23

RuBiSCO LSU

C

2.5 w

4.5 w

6.5 w

Figure 2.2. Arabidopsis thaliana grown in FlowPots with axenic or holoxenic substrate,
and photosynthetic protein detection from leaf tissue. (A) Photosynthesis-associated
protein quantification from total protein extracts of rosette tissue 3 w post germination. (B)
Arabidopsis in FlowPots 4 weeks post germination. Holoxenic substrate was inoculated with a
natural soil from Michigan (MS15MSU). (C) Axenic Arabidopsis growth, photographed at 2.5
weeks, 4.5 weeks and 6.5 weeks post germination. Rosette images and the protein gel is
representative of at least three replicated experiments. Western blot contributed by Dr. John
Froehlich (Michigan State University) and plant photographs contributed by Cait Thireault
(Michigan State University).

42

Beta-diversity
We hypothesized that substrate from different plant growth platforms will influence the
microbial community differentiation, resulting in distinct root-associated microbiomes.
To test this hypothesis, we inoculated the Magenta box calcined clay system and the
FlowPot peat system with established At-SPHERE bacteria (Bai et al., 2015), or with
complex microbiota directly extracted from German soil (Supplemental Table). We
sequenced and performed community analysis on 16S rRNA gene fragment amplicons
from DNA collected from (1) At-SPHERE and soil input communities, (2) bulk substrate
six weeks after inoculation and (3) from whole Arabidopsis roots at six weeks. This
resulted in a total of 1,830 OTUs across 90 samples. Samples clustered together by
growth system (calcined clay or peat substrate), and by compartment (root or bulk
substrate) based on principle coordinate analysis (PCoA) of weighted UniFrac
distances (Figure 3). This observation was consistent for samples derived from systems
inoculated with German soil microbiota (Figure 3A) and the At-SPHERE defined
bacterial collection (Figure 3B). For soil-inoculated samples, sample type (root or bulk
substrate) was explained by axis PCoA2 (16.0% of variation), and growth system (clay
versus peat FlowPot) was explained by axis PCoA2 (60.0% of variation). Trends across
axes were less clear for the defined At-SPHERE inoculated samples. Irrespective of
input community, clusters from clay root and substrate samples had greater distinct
separation than FlowPot samples.

43

Figure 2.3. Principle coordinates analysis (PCoA) on weighted UniFrac distances of OTU
assemblages by growth system (clay or peat FlowPot) and by sample type (root or bulk). Analysis was
performed for samples collected from growth systems inoculated with (A) microbiota from German soil
or (B) the defined At-SPHERE bacterial culture collection. This figure was created in collaboration with
Dr. Amine Hassani, Max Planck Institute for Plant Breeding.

44

Alpha-diversity
We hypothesized that alpha diversity would be highest for the input communities,
followed by corresponding bulk substrates, and the selective root environment would
produce samples with the lowest alpha diversity. Furthermore, we wanted to determine
whether the growth systems accommodated equivalent alpha diversity in their
respective substrates. Alpha diversity was measured for mean OTU values from each
sample using the Inverse Simpson and Shannon indices, both of which are frequently
used for 16S community analysis (McMurdie and Holmes, 2013). Community profiling
was performed on fewer than four of each input community (as opposed to n=8 or
more for all other samples), and therefore significance testing was not performed on
input community data, however, mean alpha diversity measurements with a standard
error of the mean were calculated and plotted (Figure 4).
Consistent with our hypothesis, the bulk substrate from both systems inoculated with
natural soil microbiota had relatively higher alpha diversity than root samples from the
corresponding system. Bulk substrate from the peat FlowPot system with natural soil
microbiota had significantly higher alpha diversity than bulk substrate from the clay
system treated with the same soil extract inocula (p<0.0005 for both indices). Root
samples from the FlowPot system had greater alpha diversity than the bulk substrate
of the calcined clay system. However, root samples collected from the FlowPot system
inoculated with defined At-SPHERE bacteria had equivalent alpha diversity to the
corresponding substrate. In contrast, a reduction of alpha diversity was detected in
when comparing root and bulk samples in the calcined clay system inoculated with AtSPHERE bacteria. The Shannon diversity index (but not Inverse Simpson) indicates

45

that At-SPHERE inoculated FlowPots support greater bacterial diversity in the
substrate than the calcined clay system (Figure 4).

46

Figure 2.4. Alpha diversity indices for each sample type: (A) inverse Simpson and (B) Shannon (base
2). Significance determined by two-tailed pairwise t-test. Error bars represent standard error of the
mean. Alpha diversity indices are calculated from rarefied count table to 5000 reads per sample. All bulk
and root samples have a minimum of n=8 samples and consist of all data points from three independent
experiments. Input samples are n=2 or n=3 and were not included in significance testing. * p<0.05, **
p<0.005, *** p<0.0005.

47

Bacterial community differentiation at the Phylum level
Based on principal coordinate and alpha diversity analyses, we have determined that
both the peat-based FlowPot system and the calcined clay system support axenic
Arabidopsis growth, and with either complex or defined community inocula.
Furthermore, we observed a rhizosphere effect in both systems, characterized by a
reduction in root-associated bacterial diversity relative to the bulk substrate and
distinct clustering of root and substrate samples upon principal coordinate analysis.
We hypothesized that phylogenetic signals could explain community differences
between the inoculated peat and calcined clay substrates. To address this hypothesis,
we calculated average read distribution at the phylum/subphylum level for the German
soil and At-SPHERE input communities and corresponding bulk substrate and root
samples from the FlowPot and calcined clay systems (Figure 5, Supplemental table).
Reads from the soil input community were predominantly distributed across seven
phyla and subphyla: Actinobacteria, Bacteroidetes, Firmicutes, Acidobacteria,
Alphaproteobacteria, Betaproteobacteria, and Gammaproteobacteria. Actinobacteria
reads were most abundant (33.35 ± 6.36%), and reads classified to rare phyla
accounted for 7.72 ± 2.8% of soil input (Figure 5, Supplemental Table). Actinobacteria
was enriched 0.96-fold in calcined clay bulk substrate and depleted 2.76-fold in peat
bulk substrate relative to the soil input (Figure 5, Table 1). Although Actinobacteria was
the most abundant phylum in calcined clay bulk samples (65.64 ± 4.74% of reads) and
only the fourth-most abundant phylum in FlowPot bulk samples (10.52 ± 0.83% of
reads), Actinobacteria abundance in root samples from both systems were similar
(FlowPot: 7.00 ± 1.25%, calcined clay: 6.08 ± 1.20%). Betaproteobacteria was the

48

most abundant phylum for root samples from the both FlowPot and calcined clay
systems (FlowPot: 50.58 ± 3.83%, calcined clay: 88.75 ± 1.69%). Root and bulk
samples from the soil-inoculated FlowPot system had similar Alphaproteobacteria
abundances (bulk substrate: 16.17 ± 1.12% and root: 11.00 ± 1.32%). However,
Alphaproteobacteria in samples from the soil-inoculated calcined clay system were 20fold less abundant than the soil input (bulk: 0.38 ± 0.12% and root: 0.29 ± 0.05%).
Strikingly, Alphaproteobacteria abundance did not exceed 0.5% in any sample
collected from the soil-inoculated calcined clay groups. Gammaproteobacteria was
equivalently abundant (± 2%) in FlowPot bulk substrate and root samples, but more
than 30-fold depleted in root samples from the soil-inoculated calcined clay system
relative to abundance in respective bulk substrate (bulk: 16.7 ± 3.28%, root: 0.36 ±
0.03%). Bacteroidetes-classified reads were enriched in the FlowPot system relative to
the input (input: 2.08 ± 0.84%, bulk: 7.1 ± 1.01%, root: 20.37 ± 2.72%), accounting for
a 3.1-fold enrichment in bulk substrate relative to soil input, and a 2.3-fold root
enrichment relative to bulk substrate. In contrast, Bacteroidetes represented less than
0.2% of reads from soil-inoculated calcined clay bulk substrate and root samples.
Firmicutes were enriched in the FlowPot bulk substrate relative to the soil input
community, and very rare in the calcined clay bulk substrate (input: 7.25 ± 2.23%,
FlowPot bulk: 22.23 ± 2.25%, calcined clay bulk: 0.13 ± 0.08%). Acidobacteria was
found a very low abundance or undetected in all systems (<0.1%), in contrast to 2.66 ±
1.11% of reads from the soil input community.
At phylum-level resolution, bulk substrate of FlowPot samples inoculated with the AtSPHERE bacteria were similar to the input (Figure 5, Table 1). Collectively, four phyla

49

accounted for 95.79% of input community reads Actinobacteria, Alphaproteobacteria,
Betaproteobacteria, and Gammaproteobacteria. Individually, none of the four dominant
phyla had greater than 0.11-fold enrichment or depletion in FlowPot substrate relative
to the input. The calcined clay substrate, however, had much larger differences in the
read distribution amongst the four dominant phyla relative to the At-SPHERE input:
4.5-fold enrichment of Actinobacteria, 2.8-fold depletion of Betaproteobacteria, 22.8fold depletion of Alphaproteobacteria, and 0.6-fold depletion of Gammaproteobacteria.
Trends observed for Actinobacteria read distribution from soil-inoculated systems were
strikingly similar to trends for At-SPHERE inoculated samples. For both calcined clay
and FlowPot systems, Actinobacteria preferentially colonized the substrate relative to
the root. Relative to Actinobacteria abundance in bulk substrate, FlowPot root samples
had 1.6-fold less (bulk: 17.07 ± 1.26%; root: 7.04 ± 0.86%), and calcined clay root
samples had 10.6-fold less (68.22 ± 6.74% versus 7.1 ± 1.76%). Notably, the total
abundance of Actinobacteria in root samples from both systems was approximately
7%. Betaproteobacteria were root-enriched 14.9-fold in calcined clay relative to bulk,
but less than 0.1-fold differentially abundant in FlowPot samples. A subtle enrichment
of Gammaproteobacteria in FlowPot root samples was observed relative to the bulk
substrate, and collectively Gammaproteobacteria abundance in At-SPHERE FlowPot
roots was very high (49.17 ± 3.09%), in contrast to At-SPHERE calcined clay root
samples (5.83 ± 2.52%). High variability of Gammaproteobacteria abundance could
possibly be explained by opportunistic colonization of one or several OTUs, or a
founder’s effect. Firmicutes accounted for 2.78 ± 1.63% of At-SPHERE input reads, in

50

contrast to soil-inoculated FlowPots, the highest average abundance of Firmicutes in
At-SPHERE inoculated bulk or root samples was 0.53 ± 0.17%.

51

Figure 2.5. Mean percent abundance of OTU-classified reads by phylum. The right bound
maximum across the x-axis represents 100%. Error bars represent standard error of the mean
for each sample type.

52

Table 2.1. Enrichment of phylum by sample type relative to input and to bulk.
natural soil

At-SPHERE

enrichment over input

over bulk

enrichment over input

over bulk

FlowPot

clay

FP

clay

FlowPot

clay

FP

clay

bulk

root

bulk

root

root

root

bulk

root

bulk

root

root

root

Act

-1.66

-2.25

0.98

-2.46

-0.59

-3.43

0.13

-1.15

2.12

-1.14

-1.28

-3.26

BP

0.71

1.3

-0.58

2.11

0.59

2.69

-0.02

0.22

-1.66

2.2

0.24

3.86

AP

0.59

0.04

-4.82

-5.21

-0.56

-0.39

-0.33

-0.27

-4.77

-3.83

0.06

0.93

GP

-1.25

-1.67

0.09

-5.44

-0.41

-5.54

0.17

0.36

-0.75

-2.72

0.19

-1.97

Bct

1.77

3.29

-inf

-3.89

1.52

n/a

-inf

-4.29

-inf

-0.96

n/a

n/a

Frm

1.62

-1.05

-5.8

-2.51

-2.66

3.29

-2.83

-2.53

-3.95

-2.39

0.3

1.56

Acd

-5.73

-5.73

-inf

-6.47

0

n/a

n/a

n/a

n/a

n/a

n/a

n/a

Values expressed as log2(x/y). Act= Actinobacteria, BP= Betaproteobacteria,
AP= Alphaproteobacteria, GP= Gammaproteobacteria, Bct= Bacteroidetes,
Frm= Firmicutes, Acd= Acidobacteria

53

Bacterial community differentiation at the OTU level
We hypothesize that there are root-enriched OTUs shared among the growth systems,
as well as growth system-specific root microbiota. Furthermore, we hypothesize that
some root-enriched OTUs will be shared among At-SPHERE and German soil
communities.
Root-enriched OTUs were identified according to three criteria (1) minimum average
abundance on the root sample of 5/5,000 (0.1%) reads, (2) minimum fold-enrichment of
the OTU abundance over the corresponding bulk substrate sample of
log2(root/bulk)>0.5, and (3) significance of root-associated enrichment of p < 0.1,
determined by pairwise two-tailed t-test. The number of OTUs with greater than five
reads for each root sample type is 22 for clay with natural soil, 80 peat FlowPots with
natural soil, 26 for clay with At-SPHERE, and 39 for peat FlowPots with At-SPHERE
(Table 2). For each sample, OTUs were ranked in order of abundance, and the number
of OTUs to reach 90% of the total reads was assessed. The clay system had very few
OTUs account for 90% of total reads, relative to the peat FlowPot system (Table 2).
In the clay system with natural soil microbiota, 9 OTUs classified as seven genera met
the root enrichment criteria: Burkholderia (n=2), Ralstonia, Herbiconiux,
Staphylococcus, Streptococcus, Propionibacterium, unclassified Oxalobacteraceae,
unclassified Proteobacteria (Figure 6, Figure 8, and Table 3). The Ralstonia OTU1 is
18.49-fold root enriched (p < 10-13). The peat FlowPot system with natural soil
microbiota had 11 OTUs root-enriched: Hoeflea, Ensifer, Rhizobium (n=2), unclassified
Rhizobiaceae, Flavobacterium, Ralstonia, unclassified Oxalobactereaceae (n=2)
Massilia, and unclassified Sinobacteraceae. Of the root-enriched OTUs from the

54

calcined clay and FlowPot systems with natural soil microbiota, only two are shared:
Ralstonia (OTU1) and an unclassified member of Oxalobacteraceae (OTU10). In the
FlowPots with natural soil microbiota, Ralstonia (OTU1) was root-enriched 3.61-fold
(p<0.032), accounting for 5.3% of reads, while in the clay samples with natural soil
microbiota, OTU1 is 18.49-fold root enriched (p < 10-13) and accounted for >87% of
reads.
Samples inoculated with the At-SPHERE bacteria were subjected to the same rootenrichment criteria as the holoxenic samples (Figure 7, Figure 9, and Table 3). AtSPHERE calcined clay samples yielded eight root-enriched OTUs, classified as the
following genera: Burkholderia (n=3), Ralstonia, Staphylococcus, Propionibacterium,
Flavobacterium, and unclassified Proteobacteria. Six of the eight root-enriched OTUs
are shared with the holoxenic calcined clay samples (Table 3).

55

Table 2.2. Absolute OTU abundance by abundance qualifiers.
natural soil

At-SPHERE

FlowPot

clay

FlowPot

clay

qualifier

input

bulk

root

bulk

root

input

bulk

root

bulk

root

#>5

138

104

80

47

22

48

39

39

25

26

#>1

482

325

279

149

162

187

133

134

114

145

top 90%

167

88

58

20

4

23

20

19

10

9

56

Table 2.3. Classification of root-enriched OTUs and magnitude of enrichment.
natural soil
FlowPot

At-SPHERE
clay

FlowPot

clay

OTU

Classification

%R

R/B

%R

R/B

%R

R/B

%R

R/B

100

Act

Propionibacterium

-

-

0.6

4.4

-

-

0.5

21.2

320

Act

Herbiconiux

-

-

0.4

18.5

-

-

-

-

39

AP

Devosia

-

-

-

-

2.6

5.8

-

-

47

AP

Microvirga

-

-

-

-

0.9

0.5

-

-

55

AP

Hoeflea

0.1

7.8

-

-

0.8

0.6

-

-

51

AP

Ensifer

0.1

0.6

-

-

0.6

1.0

-

-

63

AP

Rhizobium

0.6

5.3

-

-

0.5

0.6

-

-

122

AP

(Rhizobiaceae)

0.4

1.2

-

-

-

-

-

-

352

AP

Rhizobium

0.3

0.6

-

-

-

-

-

-

19

Bct

Flavobacterium

16.8

2.3

-

-

-

-

0.2

10.9

1

BP

Ralstonia

5.3

3.6

87.3

18.5

10.4

1.4

73.1

19.4

88

BP

Burkholderia

-

-

0.7

10.2

-

-

0.7

26.0

1175

BP

Burkholderia

-

-

0.2

12.3

-

-

0.2

13.0

46

BP

Burkholderia

-

-

-

-

-

-

0.1

7.8

37

BP

Methylophilus

-

-

-

-

2.6

15.2

-

-

10

BP

(Oxalobacteraceae)

8.0

7.3

0.1

4.0

-

-

-

-

208

BP

Massilia

7.6

11.6

-

-

-

-

-

-

382

BP

(Oxalobacteraceae)

1.6

3.2

-

-

-

-

-

-

179

Frm

Staphylococcus

-

-

0.2

10.9

-

-

0.3

16.0

181

Frm

Streptococcus

-

-

0.2

13.0

-

-

-

-

23

GP

Rhizobacter

-

-

-

-

2.4

7.8

-

-

48

GP

Lysobacter

-

-

-

-

1.7

2.3

-

-

226

GP

(Sinobacteraceae)

0.2

7.3

-

-

-

-

-

-

%R= mean percent of root sample reads of the sample group, R/B= root enrichment over bulk;
Act= Actinobacteria, BP= Betaproteobacteria,
AP= Alphaproteobacteria, GP= Gammaproteobacteria, Bct= Bacteroidetes,
Frm= Firmicutes

57

A 15

substrate

significance -log10(p)

13

root

clay
Ralstonia

11
9
7
5
3
1
-1

significance -log10(p)

B

-8
7

-6

-4

-2

0

substrate

2

4

6

8

root

FlowPot

Paenibacillus

5
Paenibacillus

Phenylobacterium
Flavobacterium

3

Ralstonia

Massilia

1
7

7

-1
-4

-2

7

5

5

5 0

2

differential abundance3 log2(root/bulk)
3

OTU abundance:
1

3

> 5%
1

1-5%

0.1-1%

1
Figure 2.6. Root-associated OTU enrichment for
samples
with natural soil microbiota
-1
(German CAS10) from the (A) calcined
(B)
FlowPot
Statistical
-1 clay system and the
-6
-3
0 system.
3
6
-6 comparison
-3 -1 0 by pairwise
3
6 t-test and plotted on the yconfidence was calculated for each OTU
-6
-3
0
3
6
axis as -log10 (p).
Natural	soil

58

4

A

7

substrate

Ralstonia

Terrabacter

5

significance -log10(p)

root

clay
Arthrobacter

3

1

-1
-8

B

-6

-4

-2

0

4

6

8

7

substrate
significance -log10(p)

2

root

FlowPot

5

Methylophilus
Devosia
Rhizobacter

3

1

7

7

-1
-6

-4

7

5-2

5

0

5

2

4

differential abundance3 log2(root/bulk)
3

OTU abundance:
1

3
1

> 5%
1

1-5%

0.1-1%

Figure 2.7. Root-associated OTU enrichment relative
-1 to abundance in bulk substrate for
-1
samples with At-SPHERE microbiota from the (A) calcined
clay
and
-6
-3 system
0
3 the6(B) FlowPot
-6
-3for-1
0 OTU
3 comparison
6
system. Statistical confidence was calculated
each
by pairwise t-test and
-6
-3
0
3
6
plotted on the y-axis as -log10At-SPHERE
(p).

59

6

Calcined	Clay, German soil
A

Alphaproteobacteria

Actinobacteria

Firmicutes

14

14

14

9

9

9

4

4

4

-1

-1

-1

-8 -6 -4 -2 0 2 4 6 8

-8 -6 -4 -2 0 2 4 6 8

-8 -6 -4 -2 0 2 4 6 8

Betaproteobacteria

Gammaproteobacteria

Bacteroidetes

14

14

14

9

9

9

4

4

4

-1

-1
-8 -6 -4 -2 0 2 4 6 8

B

Actinobacteria

7

-1
-8 -6 -4 -2 0 2 4 6 8

-8 -6 -4 -2 0 2 4 6 8

Alphaproteobacteria

7

Firmicutes
7

5

5

5

3

3

3

1

1

1

-1

-1
-4

-2

0

2

4

-4

Betaproteobacteria

7

7

5

5

3

3

-2

-4

-2

0

2

4

3

-1
-4

-2

0

2

4

-2

0

2

4

Bacteroidetes

5
3

7

7
-4

-2

OTU abundance:
1

4

7

5-1

-1

2

Gammaproteobacteria

71

1

0

5
3

1

5
-1

0 3

2

> 5%
1

4

1-5%

-4

0.1-1%

1
Figure 2.8. Phylum-level resolution of root-associated
OTU enrichment relative to
-1
abundance in bulk substrate-1for samples with
natural
soil
microbiota
from the (A) calcined
-6
-3
0
3
6
-1 0
-6
-3
3
6
clay system and the (B) FlowPot system. Statistical confidence was calculated for each OTU
-6
-3
0
3
6
comparison by pairwise t-test and plotted on the y-axis as -log10 (p).

60

At-SPHERE.		German soil

A

Actinobacteria

7

5

5

3

3

3

1

1

1

-1
-8

-4

0

4

-1
-8

8

Betaproteobacteria

7

-4

0

4

8

Gammaproteobacteria

7

-8

5

5

3

3

3

1

1

1

-1
-8

-4

0

4

Actinobacteria

7

-4

0

4

8

Alphaproteobacteria

7

-8

5

5

3

3

3

1

1

1

-1

-1
-3

0

3

7

7

-3

0

3

6

Gammaproteobacteria

-6

5

5

3

3

3

1

1

1

-1
-6

-3

0

3

6

-4

0

4

8

-6

-3

0

3

6

-3

0

3

6

Bacteroidetes

7

5

-1

8

-1
-6

6

Betaproteobacteria

4

Firmicutes

7

5

-6

0

-1
-8

8

-4

Bacteroidetes

7

5

-1

Firmicutes

7

5

-1

B

Alphaproteobacteria

7

-1
-6

-3

0

3

6

Figure 2.9. Phylum-level resolution of root-associated OTU enrichment relative to
abundance in bulk substrate for samples with At-SPHERE microbiota from the (A) calcined
clay system and the (B) FlowPot system. Statistical confidence was calculated for each OTU
comparison by pairwise t-test and plotted on the y-axis as -log10 (p).

61

Discussion
Over the past decade, great progress has been towards identifying the core microbiota
of Arabidopsis using culture-independent methods (Lundberg et al., 2012; Bulgarelli et
al., 2012; Delmotte et al., 2009; Kniskern et al., 2007). More recently, microbiome
colonization experiments on Arabidopsis under gnotobiotic conditions with synthetic
communities have been used to identify host factors that influence the microbiome
(Vogel et al., 2016; Lebeis et al., 2015; Bai et al., 2015; Bodenhausen et al., 2014). The
trend is shifting from hypothesis-generating descriptive studies to hypothesis-testing
mechanistic laboratory studies (Busby et al., 2017). Just as gnotobiotic research in
mice has successfully led to discoveries of medical importance (Ley et al., 2006; Kau et
al., 2011; Hooper et al., 2012), a major goal of plant microbiome recolonization
research in a laboratory setting is to discern plant-microbiome interactions that occur
in nature or in an agricultural setting with reduced complexity. It is crucial to know if a
growth system can capture and support the microbiota from a natural setting. Based
on many studies aiming to identify host factors that influence microbiome assembly,
environmental factors have been noted to exert profound impact (McCoy et al., 2017;
Goodrich et al., 2014). Ideally, multiple experimental systems with different
characteristics should be used to capture different aspects of natural plant-microbiome
interactions, resulting in a more robust and realistic general principle.
Here, we introduce the FlowPot: a peat-based growth system that supports vigorous
Arabidopsis growth under the following conditions: axenic (total absence of detectable
microbiota), gnotobiotic (defined or synthetic microbiota), and holoxenic (direct transfer
of a soil microbial community). In this study, we performed a soil microbiome

62

transplant and characterized microbial community feedback after this perturbation, as
well as inoculation with members of the defined At-SPHERE Arabidopsis bacterial
microbiota, many of which were isolated from plants growing in the same soil (Bai et
al., 2015). Relative to the natural soil input community, the bulk substrate of a calcined
clay system, and to a lesser extent the FlowPot system, has a reduction in detectable
OTUs, consistent with previous reports (Howard et al., 2017; Lau and Lennon, 2012).
FlowPot system maintains a similar bacterial phylogenetic profile to a natural soil
collected from an agricultural field at the phylum level. Relative to the established the
calcined clay growth system, the FlowPot system maintains higher alpha diversity,
more closely simulating an agriculturally relevant microbial community. This is probably
because calcined clay lacks robust, complex carbon, nitrogen and/or certain
micronutrient sources for microorganisms. Although in this study both systems were
supplemented with a plant nutrient solution, the carbon source (other than CO2) in the
calcined clay growth system likely comes from photosynthates exuded by the plant.
This is in contrast to the FlowPot system, in which the substrate is mostly organic
material. We speculate that after inoculation, some populations of microorganisms in
the calcined clay system may be in a state of metabolic dormancy, some may not even
survive. It would be interesting to test the recoverability of viable taxa over time in
calcined clay and FlowPots after inoculation. The strong enrichment of Actinobacteria
in the calcined clay substrate relative to the root is interesting and suggests that
populations of certain taxa may be more capable of surviving the calcined clay
environment than other taxa.

63

The “rhizosphere effect” is the term to describe a root-associated microbiome
taxonomically and functionally distinct from surrounding roots (Hiltner, 1904).
Previously, the rhizosphere effect in Arabidopsis has been observed in Magenta boxes
with calcined clay substrate inoculated with mixtures of defined cultures of
Arabidopsis-associated bacteria (Bai et al., 2015; Lebeis et al., 2015). Relative to the
FlowPot system the calcined clay system showed a stronger rhizosphere effect as
evident from the beta-diversity analysis. We speculate that multiple nutrient sources
exist in the FlowPot system, and that metabolically flexible microorganisms can
establish in the bulk substrate or the plant root environment. In the calcined clay
system, many microorganisms may be forced to compete for the root environment,
become/remain metabolically dormant, or die. Some OTUs were present in the soil
input community, but not detected at significant abundance the substrate or root. For
example, six Acidobacteria OTUs are detected in the soil input at 0.1-0.62%
abundance, but not in the calcined clay or FlowPot systems. One hypothesis is that the
Acidobacteria are slow-growing in this system, and although they may be alive in the
new substrate, they fall below the detection limit due to other faster-growing bacteria.
Another possibility is that the substrate lacks physical or chemical properties to sustain
the survival of these particular Acidobacteria, although Acidobacteria has been
reported to be a dominant phylum in natural Sphagnum peat habitats (Pankratov et al.,
2008). A third possibility to consider is that some bacteria maybe be non-viable dead at
the time of inoculation. A shortcoming of culture-independent (shotgun and amplicon)
community profiling from environmental DNA is the inability to discern live versus dead
microorganisms. It is also possible that the perturbation event of inoculation prompted

64

populations of microorganisms to shift from a dormant to an active lifestyle, and
outcompete other microorganisms. In this study, we classified a microorganism as
“root-enriched” if it is more relatively abundant in a root sample than the corresponding
soil substrate. Some organisms, however, are highly abundant in both the substrate
and the root. Just because an organism is not more relatively abundant in a root as
compared to the bulk substrate does not exclude the fact that it may have a functional
interaction with the plant.
Some OTUs are present in the substrate of one system, but not the other. For example,
Pseudonocardia (Actinobacteria, Otu17 and Otu22) is >2.5% abundant in soilinoculated clay, but not detected in FlowPots. Perhaps there are circumstances where
Pseudonocardia is capable of survival in the FlowPot substrate as well, but the
FlowPot substrate also sustains populations of microorganisms that outcompete
Pseudonocardia. It is critical to consider the dynamics of microbe-microbe interactions
and their role in the ecology of microbiomes. The lack of proliferation of a
microorganism in a growth system may be due to microbial antagonism and antibiosis
(i.e., siderophores, type-VI secretion, bacteriocin production, antibiotics, etc.) rather
than abiotic factors (Ma et al., 2014; Loper et al., 1991; Jones et al., 2007). An OTU is
root-enriched in one system, and although it is detectable in the other system, it is not
root-enriched. Herbiconiux (Actinobacteria, Otu320) is root-enriched in the soilinoculated calcined clay system, and detected in the FlowPot system but at equal
equivalent abundance in the root and substrate. This OTU is very low abundance, and
increased sequencing depth could potentially solve this problem. Otu320 is very low
abundance in both soil-inoculated systems: ~0.4% in clay roots, 0% in clay substrate,

65

~0.18% in FlowPot roots, ~0.1% in FlowPot substrate. Notably, all root-enriched OTUs
specific to a system were less than 0.2% abundant in the other system (substrate and
root samples). Six Acidobacteria OTUs are detected in the German soil input at 0.10.62% abundance, but not in the calcined clay or FlowPot systems.
One must consider that in nature, a plant microbiome is populated by many nonbacterial microorganisms, including fungi, oomycetes, and protists. Furthermore,
macroscopic organisms in the soil such as nematodes and springtails may affect a
plant as well. The 16S rRNA gene amplicon approach to microbiome analysis only
reveals a portion of the microbiome. Non-bacterial microorganisms may influence the
bacterial constituency of a plant’s microbiome via competition for nutrients, predation,
influencing host metabolism, changing soil features, and providing a new repertoire of
metabolites in the ecosystem. The German soil inoculum likely contains viable fungi;
traces of fungal and protist mitochondrial DNA were detected in the 16S amplicon
reads. The extent to which non-bacterial microorganisms influenced the bacterial
interaction in this set of experiments is unknown. Follow-up experimentation is
required to address these questions.
The complexity of the plant microbiome is not well understood. In the agricultural
biotechnology industry, microbial products have been applied to the field –to stimulate
plant growth (biostimulants), providing additional nutrient sources (biofertilizers), and
controlling pests (biopesticides). The agricultural biologicals industry is rapidly
emerging, but a better understanding of plant-microbe associations in a community
context would most certainly lead to innovations (Ciancio et al., 2016; Busby et al.,
2017; Mueller and Sachs, 2015). A better grasp of the complexities of the plant

66

microbiome could lead to new and effective technologies that can enhance crop
growth in what is considered marginal land, improve plant nutrient use efficiency, and
confer tolerance to both biotic and abiotic stress tolerance. A future direction for the
FlowPot system and other axenic technologies is to develop its use for agriculturally
relevant model systems such as maize and soybean. Furthermore, expansion of
reference culture collections such as At-SPHERE to include isolates from other
environments, or to build collections based on function rather than 16S, will increase
the power of recolonization experiments. One application of an axenic colonization
system is to determine the most successful plant colonizers because presumably, they
form a stronger association with the plant and/or effectively out-compete other
microorganisms. Coupled with the power of fully-sequenced culture collections, one
can hypothesize which genomic features are necessary for successful colonization of a
plant host. We can also introduce microorganisms into established microbiomes on a
large-scale and determine which isolates can invade and persist. As we move forward
in the plant microbiome field an important question is now possible to address: how
does the microbiome influence host phenotype? Healthy and robust plant growth in a
nutrient-replete axenic FlowPot system allows us to study a plant in an axenic state,
and directly compare this to colonized plants. Is there anything wrong with an axenic
plant? Although we could not detect or identify any signs of stress in our axenic
Arabidopsis, a further in-depth study of the physiology of axenic Arabidopsis relative to
Arabidopsis colonized by conventional soil community is of interest.

67

Figure 2.10. Changes in substrate chemical properties after treatments. Chemical analysis
of peat substrate after four treatments: as prepared by the manufacturer (no treatment), 7 d
after autoclave regime (no flushing), 7 d after autoclave regime and flushing regime with ddH2O
followed by Michigan soil microbiota suspended in 1/4X MS media (liquid), 7 d after autoclave
regime and flushing regime with ddH2O followed by autoclaved Michigan soil microbiota
suspended in 1/4X MS media (liquid).

68

Figure 2.11. Instruments for the aseptic manipulation of FlowPots, and illustration of
FlowPots on a stand and within the Eco2 microbox.

69

Table 2.4. Abundance of reads by phylum.
natural soil
FlowPot
in
bulk
root

clay
bulk

root

At-SPHERE
FlowPot
in
bulk
root

clay
bulk

root

Act

33.4
± 6.4

10.5
± 0.8

7.0
± 1.3

65.6
± 4.7

6.1
± 1.2

15.7
± 7.8

17.1
± 1.3

7.0
± 0.9

68.2
± 6.7

7.1
± 1.8

BP

20.6
± 3.5

33.7
± 2.3

50.6
± 3.8

13.8
± 2.9

88.8
± 1.7

17.9
± 2.3

17.7
± 1.3

20.9
± 2.9

5.7
± 1.4

82.4
± 5.5

AP

10.7 ±
0.9

16.2 ±
1.2

11.0 ±
1.3

0.4 ±
0.1

0.3 ±
0.1

24.0 ±
4.7

19.1 ±
0.4

19.9 ±
1.1

0.9 ±
0.3

1.7 ±
1.4

GP

15.7 ±
8.3

6.6 ±
1.2

4.9 ±
0.9

16.7 ±
3.3

0.4 ±
0.0

38.3 ±
11.7

43.0 ±
1.1

49.2 ±
3.1

22.8 ±
5.1

5.8 ±
2.5

Bct

2.1 ±
0.8

7.1 ±
1.0

20.4 ±
2.7

< 0.1

0.1 ±
0.0

0.4 ±
0.2

< 0.1

< 0.1

< 0.1

0.2 ±
0.1

Frm

7.3 ±
2.2

22.2 ±
2.5

3.5 ±
0.6

0.1 ±
0.1

1.3 ±
0.6

2.8 ±
1.6

0.4 ±
0.1

0.5 ±
0.1

0.2 ±
0.1

0.5 ±
0.2

Values represent mean of samples, ± standard error of the mean.
Act= Actinobacteria, BP= Betaproteobacteria,
AP= Alphaproteobacteria, GP= Gammaproteobacteria, Bct= Bacteroidetes,
Frm= Firmicutes

70

Table 2.5. Phylum level classifications of most abundantly ranked OTUs that account for 90%
of reads
natural soil

At-SPHERE

FlowPot

clay

FlowPot

clay

in

bulk

root

bulk

root

in

bulk

root

bulk

root

Act

51

21

12

16

2

8

8

3

6

4

BP

29

16

15

3

2

3

3

4

2

2

AP

30

26

17

-

-

6

6

8

-

1

GP

15

6

4

1

-

5

3

4

2

2

Bct

4

3

3

-

-

-

-

-

-

-

Frm

12

16

7

-

-

1

-

-

-

-

Acd

10

-

-

-

-

-

-

-

-

-

Act= Actinobacteria, BP= Betaproteobacteria,
AP= Alphaproteobacteria, GP= Gammaproteobacteria, Bct= Bacteroidetes,
Frm= Firmicutes, Acd= Acidobacteria

71

Table 2.6. Natural soil, clay system. Top 10 OTUs by rank abundance.
Sample type
bulk

root

OTU

reads/5000

Phylum

Genus

3

1028 ± 404

Actinobacteria

Arthrobacter

2

809 ± 160

Gammaproteobacteria

Pseudomonas

7

527 ± 172

Actinobacteria

Patulibacter

8

371 ± 110

Betaproteobacteria

Burkholderia

5

369 ± 111

Actinobacteria

Nocardioides

14

237 ± 55

Actinobacteria

Solirubrobacter

1

217 ± 148

Betaproteobacteria

Ralstonia

11

204 ± 45

Actinobacteria

17

190 ± 47

Actinobacteria

Pseudonocardia

22

110 ± 55

Actinobacteria

Pseudonocardia

1

4367 ± 85

Betaproteobacteria

Ralstonia

38

117 ± 52

Actinobacteria

Streptomyces

88

35 ± 3

Betaproteobacteria

Burkholderia

86

30 ± 14

Actinobacteria

Streptomyces

100

29 ± 7

Actinobacteria

Propionibacterium

309

27 ± 21

Firmicutes

Streptococcus

17

22 ± 9

Actinobacteria

Pseudonocardia

320

19 ± 2

Actinobacteria

Herbiconiux

7

16 ± 16

Actinobacteria

Patulibacter

181

11 ± 7

Firmicutes

Streptococcus

72

Table 2.7. Natural soil, FlowPot system. Top 10 OTUs by rank abundance.
Sample type
bulk

root

OTU

reads/5000

Phylum

Genus

599

524 ± 72

Betaproteobacteria

Massilia

42

372 ± 67

Firmicutes

Paenibacillus

40

356 ± 95

Betaproteobacteria

Massilia

19

291 ± 47

Bacteroidetes

Flavobacterium

34

258 ± 53

Firmicutes

Bacillus

2

246 ± 58

Gammaproteobacteria

Pseudomonas

16

183 ± 29

Betaproteobacteria

Massilia

28

179 ± 13

Alphaproteobacteria

Phenylobacterium

81

146 ± 29

Betaproteobacteria

Massilia

53

110 ± 42

Alphaproteobacteria

Brevundimonas

19

842 ± 158

Bacteroidetes

Flavobacterium

40

447 ± 107

Betaproteobacteria

Massilia

10

402 ± 129

Betaproteobacteria

208

382 ± 102

Betaproteobacteria

Massilia

1

263 ± 82

Betaproteobacteria

Ralstonia

4

194 ± 185

Actinobacteria

Arthrobacter

12

193 ± 178

Gammaproteobacteria

Dyella

89

159 ± 128

Betaproteobacteria

Herminiimonas

599

157 ± 49

Betaproteobacteria

Massilia

2

142 ± 28

Gammaproteobacteria

Pseudomonas

73

Table 2.8. At-SPHERE, clay system. Top 10 OTUs by rank abundance.
Sample type
bulk

root

OTU

reads/5000

Phylum

Genus

3

1168 ± 183

Actinobacteria

Arthrobacter

2

974 ± 259

Gammaproteobacteria

Pseudomonas

5

857 ± 185

Actinobacteria

Nocardioides

6

394 ± 58

Actinobacteria

Terrabacter

4

344 ± 72

Actinobacteria

Arthrobacter

93

259 ± 53

Actinobacteria

1

177 ± 55

Betaproteobacteria

Ralstonia

15

170 ± 45

Actinobacteria

Nocardioides

12

135 ± 18

Gammaproteobacteria

Dyella

9

103 ± 32

Betaproteobacteria

Variovorax

1

3657 ± 453

Betaproteobacteria

Ralstonia

9

382 ± 226

Betaproteobacteria

Variovorax

2

116 ± 53

Gammaproteobacteria

Pseudomonas

12

100 ± 40

Gammaproteobacteria

Dyella

3

73 ± 25

Actinobacteria

Arthrobacter

25

66 ± 63

Alphaproteobacteria

Rhizobium

4

53 ± 22

Actinobacteria

Arthrobacter

5

43 ± 10

Actinobacteria

Nocardioides

15

37 ± 14

Actinobacteria

Nocardioides

320

36 ± 21

Actinobacteria

Herbiconiux

74

Table 2.9. At-SPHERE, FlowPot system. Top 10 OTUs by rank abundance.
Sample type
bulk

root

OTU

reads/5000

Phylum

Genus

2

1990 ± 74

Gammaproteobacteria

Pseudomonas

10

441 ± 57

Betaproteobacteria

13

320 ± 15

Alphaproteobacteria

Caulobacter

1

228 ± 14

Betaproteobacteria

Ralstonia

27

219 ± 67

Actinobacteria

Oerskovia

4

186 ± 52

Actinobacteria

Arthrobacter

9

181 ± 23

Betaproteobacteria

Variovorax

32

157 ± 18

Alphaproteobacteria

Sphingopyxis

115

117 ± 19

Alphaproteobacteria

Caulobacter

12

104 ± 28

Gammaproteobacteria

Dyella

2

2065 ± 214

Gammaproteobacteria

Pseudomonas

1

521 ± 151

Betaproteobacteria

Ralstonia

13

313 ± 41

Alphaproteobacteria

Caulobacter

10

238 ± 43

Betaproteobacteria

12

177 ± 67

Gammaproteobacteria

Dyella

9

132 ± 21

Betaproteobacteria

Variovorax

37

129 ± 21

Betaproteobacteria

Methylophilus

27

128 ± 46

Actinobacteria

Oerskovia

39

128 ± 29

Alphaproteobacteria

Devosia

23

120 ± 32

Gammaproteobacteria

Rhizobacter

75

Chapter 3
Characterization of axenic Arabidopsis thaliana

Acknowledgements: MAPK and UPLC/MS experiments were contributed by Brad
Paasch. Non-targeted GC/MS was performed by Dr. Tim Tschliplinsky (Oak Ridge
National Lab). Assistance with plant growth and FlowPot assembly from David Rhodes
and Caleigh Griffin were greatly appreciated. Thoughtful discussions and guidance
from Dr. JP Jerome and Dr. Brian Kvitko were instrumental.

76

Abstract
Plants and animals are naturally colonized by diverse microbiota, and although
mechanistic studies on select host-microbe model systems have revealed important
molecular underpinnings of symbiosis and pathogenesis, little is known about the
dynamic relationship between a host and its associated microbiota in a community
context. Utilizing the FlowPot growth system described in Chapter 2, here we
characterize differential immune-associated phenotypes in holoxenic and axenic
Arabidopsis, revealing functional attributes of plant microbiota by proxy of host
phenotypic characterization. Despite similar appearances and growth rates in
nutrient-replete substrate, normal basal abundance of immunity-associated transcripts
are lacking in axenic plants. Likewise, metabolic profiling and phytohormone
quantification reveal a deficiency of defense and immune-associated metabolites in
axenic plants. We further report that axenic plants are compromised in transcriptional,
translational, and post-translational responses to the immune elicitation and priming
capacity. Axenic plants are compromised in defense against the foliar bacterial
pathogen Pseudomonas syringae pv. tomato DC3000. Diseases assays using a type-III
secretion system deficient virulence-compromised DC3000 mutant strain and an
immune-deficient Arabidopsis polymutant reveal that axenic susceptibility is partially
explained by a compromised innate immune system. The identity of differentially
abundant transcripts and metabolites from axenic and holoxenic Arabidopsis are
reported herein, which may contribute to microbiome-influenced host phenotypes.
These results provide the first direct evidence that endogenous microbiota are required
for the development of a normal level of immunocompetence in soil-grown plants.

77

Introduction
Despite the vast microbial diversity of the plant microbiome, only a select few
pathogenic and symbiotic interactions have been studied in depth. The majority of
plant-microbe interactions and their phenotypic repercussions to the host are poorly
understood, particularly in a community context. Dysbiosis of a plant’s microbiome can
occur when a virulent pathogenic microorganism infects a plant host, resulting in
disease manifestation. The dogmatic Plant Disease Triangle has been the most popular
model to explain prerequisites for plant disease to occur: (1) a disease-conducive
environment, (2) a virulent pathogen, and (3) a susceptible host (Scholthof, 2007).
Indeed, these elements are necessary for disease to occur, but a potentially neglected
factor that influences disease susceptibility is the microbiome.
Previous research indicates that the plant microbiome is intimately linked to plant
nutrition. For example, phosphorus solubilization, nitrogen fixation, and carbon
conversion are all ecosystem services provided by soil and plant-associated
microbiota, for which the health of plants is dependent upon (Berendsen et al., 2012;
DeAngelis et al., 2008; Jackson et al., 2008; Turner and Haygarth, 2001). Given the
presumed importance of the plant microbiome towards plant health, numerous studies
attempt to identify host factors that influence microbiome composition (Bodenhausen
et al., 2014; Horton et al., 2014; Lebeis et al., 2015; Peiffer et al., 2013). The function of
the plant microbiome, regarding influence over host phenotype, has been studied from
various approaches. A traditional approach, albeit removed from understanding plantmicrobe interactions in a community context, is to isolate and exogenously apply
microorganisms to a plant and compare the host phenotype to a mock-inoculated

78

control, either in the lab or the field. Correlation studies have been performed to
generate hypotheses about microbiome function, whereupon taxonomic-level
microbiome community profiling is correlated with corresponding plant phenotypic
measurements (Kembel et al., 2014; Wagner et al., 2014). Another approach is to
survey the microbiome in depth using meta-’omics approaches (i.e., metagenomics,
metatranscriptomics, meta-proteomics, etc.), and based on database annotations of
what is identified, hypothesize what effect the microbiome may have on the host
(Delmotte et al., 2009; Garoutte, 2016; Ryffel et al., 2016; Tringe et al., 2005). Another
approach is to inoculate a plant or soil with a microbial consortium, either via soil or a
mix of microbial cultures and assess the host phenotype (Mueller and Sachs, 2015;
Panke-Buisse et al., 2015). Here, we propose an alternative approach-- characterize
phenotypic attributes of an axenic (germ-free) plant and contrast them directly with
holoxenic (colonized by natural microbiota) plants grown under identical conditions.
Correlation studies that link plant phenotypes with associated taxonomic microbiome
composition, in both diseased and non-diseased states, are the basis for many
hypotheses of microbiome-influenced plant phenotypes. One such example links
microbiome composition with flowering time, whereupon researchers identified a
correlation between flowering time and reproductive fitness of an Arabidopsis relative
to soil microbiome composition by studying natural populations (Wagner et al., 2014).
Correlations between flowering phenology and relative abundance of particular taxa
enabled Wagner et al. to formulate hypotheses regarding which taxa play a role in
flowering phenology. In another study, researchers surveyed phyllosphere bacterial
communities from over 50 species of trees in a neotropical forest in Panama and

79

identified correlations between host attributes (i.e., plant species, nutrient
concentrations, leaf mass, host phenology, etc.) (Kembel et al., 2014). Both Wagner et
al. and Kembal et al. were able to formulate testable hypotheses regarding which
microorganisms may play a role in host phenology.
Numerous in-depth descriptive studies that utilize meta-’omics approaches (i.e.,
metagenomics, metatranscriptomics, metaproteomics, etc.) report attributes of plant
microbiota, likewise providing a foundation for the hypothesis of functional plantmicrobiome interactions. With the advent of high throughput sequencing technology,
researchers are able to extract environmental DNA and perform shotgun metagenome
sequencing and hypothesize about community functional potential based on the
database annotations of identified reads. This approach has been used on rhizosphere
and soil microbiomes. Comparative metagenomics across different environments
allows one to hypothesize about microbial traits necessary for colonization, and
identification of genes whose annotations (based on previous plant-microbe interaction
mechanistic studies) suggest may play a role in the microbial manipulation of host
phenology, one can hypothesize about community functional potential (Bai et al., 2015;
Tringe et al., 2005). A major challenge with shotgun metagenomics is the inability to
discern which genes are expressed, and under what circumstances. To mitigate this
and get a step closer to function, metatranscriptomic and hybrid
metagenomic/proteomic approaches have been used to identify the activity of
microbial populations in situ (Delmotte et al., 2009; Garoutte, 2016). The hybrid and
hybrid metagenomic/proteomic approach, in essence, captures mass spectroscopy
signatures on peptide fragments from an environment, and subsequently mapped to

80

curated databases as well as to the corresponding metagenome (Delmotte et al., 2009;
Knief et al., 2012). The community proteogenomics approach may be difficult in the
rhizosphere and endophytic environments, due to artifacts from the soil as well as
“contaminating” plant proteins.

Reducing environmental complexity is often necessary to identify strong
correlations between microbial taxonomic composition and plant phenology,
due to stochastic and uncontrollable abiotic parameters. Thus, some
researchers have, in a sense, brought the ecology to the lab by stabilizing
environmental variables and working with synthetic microbiota or complex
microbial consortia. Inoculations of plant material with simplified microbial
populations have been performed by many groups, but few have taken the
analysis to the next level to assess corresponding host phenology. One group
has used pure cultures of abundant phyllosphere bacteria to inoculate leaf
material and utilize novel mass spectroscopy approaches to identify and
quantify carbohydrate signatures on the leaf surface in the presence/absence of
select phyllosphere microbiota (Ryffel et al., 2016). Likewise, host transcriptome
profiling has been performed on Arabidopsis inoculated with a few select
bacterial cultures isolated from the phyllosphere, but in a highly artificial tissue
culture environment (Vogel et al., 2016).
An alternative “top-down” approach with additional complexity is to inoculate a plant
or soil with microbial consortia and select for the desired host phenotype (i.e. heat

81

tolerance, salt tolerance, etc.) over multiple cycles of artificial selection. This has been
performed by both industrial (for example, Bioconsortia) and academic groups, and
has allowed researchers to identify consortia that directly correlate with a host
phenotype (Mueller and Sachs, 2015; Panke-Buisse et al., 2015). There is a lack,
however, of high-resolution profiling of host phenology tied to microbiota.
An alternative approach to identifying functional roles for host-associated microbiota is
to establish an axenic, or germ-free, system and assess relative phenology in the
presence and absence of microbiota. These axenic model systems have been used to
form testable hypotheses about microbiome function, by proxy of identifying axenic
abnormalities relative to conventionally colonized controls. As previously described in
Chapter 2, the germ-free mouse model has been widely used in mammalian research.
Here we characterize phenotypic attributes of an axenic (germ-free) plant and contrast
them directly with holoxenic (colonized by natural microbiota) plants grown under
identical conditions. Although conceptually similar studies have been performed (Badri
et al., 2013; Carvalhais et al., 2013), we rigorously optimized our axenic system to
isolate the microbiome variable, and normalize abiotic factors between axenic and
holoxenic plants. For the purposes of this study, we chose to characterize axenic and
holoxenic Arabidopsis thaliana (hereafter, Arabidopsis) to identify microbiomedependent phenotypes. This study was performed under nutrient-replete conditions
because pleiotropic effects of nutrient deprivation could obscure the ability to identify
additional and potentially more subtle phenotypes, particularly in a non-diseased state.
We took a top-down approach and broadly characterized axenic attributes using
transcriptomics, metabolomic and phytohormone quantification to generate

82

hypotheses of microbiome function based upon axenic/holoxenic discrepancies.
Multiple indications of immunity-related abnormalities prompted us to focus our efforts
and perform, to our knowledge, the first in-depth and focused characterization of
innate immunity of axenic Arabidopsis.

83

Materials and Methods
Soil collection and biochemical analysis
The soils used in this study were chosen from a collection of soils with a presumed
high degree of dissimilarity in community composition: sandy loam agricultural soil
(hereafter Ag soil) was chosen from a Miscanthus plot on Michigan State University
campus (42.7089°N, -84.4656°E; altitude 261 M; East Lansing, Michigan, USA), fine
loam soil from an arid region of California (hereafter Arid soil) was collected from a
backslope near the Cleveland National Forest (33.3037°N, -116.8511°E; altitude 1,604
M; California, USA), and fine silt soil from an undisturbed grassland (hereafter Prairie
soil) in Iowa (41.8361°N, -93.0078°E; altitude 271 M; Iowa, USA) (Figure 2). Soil
collections were performed by removing the top ~5 cm of soil and sampling ~1 kg of
soil in the 5-10 cm depth zone. A three-point linear transect separated by 1 m intervals
was implemented for each location. After collection, the soil was homogenized, passed
through a 3 mm test sieve, and dried for one week at room temperature under ambient
humidity (~40%). Soil was stored in 50 or 100 g aliquots in Whirl-Pak bags (Nasco) at
4°C in the dark. Chemical analyses were performed on soil samples by the Michigan
State University Soil and Plant Nutrient Laboratory.
Arabidopsis seeds and growth conditions
Arabidopsis thaliana accession Col-0 was obtained from ABRC (www.arabidopsis.org).
The following mutants in Col-0 background were obtained from various laboratories:
fls2-1 (Thomas Boller), npr1-1 (Xinnian Dong), coi1-16 (John Turner) and sid2 (Mary
Wildermuth). The bak1 bkk1-2 cerk1 and efr fls2 cerk1 mutants were a gift from Cyril
Zipfel. All seeds were size-selected (250-300 µm) using stacked test sieves.

84

Seed-associated microbiota were killed using a vapor-phase sterilization protocol,
exposing aliquots of ~400 seeds to 6-8 h of chlorine gas (Clough and Bent, 1998). After
vapor-phase sterilization, seed aliquots in 1.5 ml microfuge tubes were placed under
sterile laminar airflow for ~10 min to remove any residual gas, and promptly closed and
put into a desiccator at 4°C in the dark for storage. Seeds were suspended in 1 ml of
sterile ddH2O for 48 - 72 h at 4°C in the dark to imbibe H2O and synchronize
germination prior to sowing. FlowPot assembly and plant growth protocol were
performed as previously described in Chapter 2 of this dissertation.
DNA extractions
Whole rosettes were aseptically collected three weeks after germination and placed in
2.0 ml high-density polypropylene tubes with three 3 mm zirconium beads. Samples
were frozen in liquid nitrogen and stored at -80°C until further processing. Three
rosettes from the same FlowPot were pooled per sample collection tube. Sample tubes
were refrozen in liquid nitrogen and contents homogenized using a Tissuelyser II
(Qiagen, 30 Hz for 2 x 1 min). DNA collections were performed using a Powersoil kit
(MoBio). Lysis solution from the PowerSoil kit was added to the homogenized leaf
sample, resuspended by pipetting, and transferred back to the PowerSoil
homogenization tube. Samples were then heated to 70C for 5 min, and remaining
steps were performed according to the manufacturer's instructions. Matrix and soil
DNA extractions were performed according to manufacturer's instructions with one
modification: with the additional 5 min 70C incubation step during initial lysis. DNA was
quantified and checked for quality using a ND-1000 UV Nanodrop spectrophotometer
(Thermo Scientific).

85

ITS and 16S rRNA gene-targeted amplification
Targeted amplification of 16S rRNA gene fragments (regions V5-V7) was performed
using primers 799F and 1192R, designed to exclude Arabidopsis mtDNA and pDNA
(Bulgarelli et al., 2012). To accommodate compatibility with the Fluidigm Access Array
tagging workflow, primers were modified with 5' CS-adapter sequences for tagging:
CS1-799F [5'-ACACTGACGACATGGTTCTACA-AACMGGATTAGATACCCKG-3'] and
CS2-1192R [5'-TACGGTAGCAGAGACTTGGTCT-ACGTCATCCCCACCTTCC-3']. PCR
reactions (25 µl) were performed in triplicate and pooled after gel separation:
AccuPrime HiFi Polymerase (0.15 µl, Thermo Fisher), DMSO (1.00 µl), forward and
reverse primers (200 nM each), 10X AccuPrime Buffer II (2.50 µl, Thermo Fisher),
template DNA (~20 ng), DNA-free water. Thermocycling conditions were as follows: hot
start 94°C (1 min); 30 cycles of [94°C (20 s), 53°C (30 s), 68°C (45 s)], and a final
extension at 68°C (2 min). No-template control reactions were performed as a
contamination check, and entire PCR runs were repeated in the event of
contamination. Amplicon DNA was gel purified, pooled by sample, checked for quality
using an ND-1000 NanoDrop spectrophotometer (Thermo Scientific). Illumina adapters
and barcodes were added during a second PCR according to the Access Array
workflow (Fluidigm), priming from the CS1 and CS2 tags. Secondary PCR was
performed using CS-16S fusion primers to increase fidelity and improve library quality.
Amplicon products were normalized for concentration with SequalPrep DNA
Normalization Plates (Invitrogen), and tagged amplicons were then pooled. DNA
concentrations were measured by Qubit dsDNA and Kapa qPCR after all the tagged
libraries were pooled. The pool was loaded on a MiSeq flowcell (Illumina, version 2

86

chemistry) according to the manufacturer's specified loading concentrations,
supplemented with PhiX to reduce highly similar cluster density, and sequenced in a 2
x 250 bp format using a 500-cycle reagent cartridge. Base calling was done by Real
Time Analysis software (Illumina, RTA v1.8.54), and sequencing was performed at the
Michigan State University Research Technology Support Facility.
Template DNA from all samples was also processed by the Michigan State University
Research Technology Support Facility to amplify ITS1 for fungal community profiling
using primers ITS1F1 and ITS2R (Schmidt et al., 2013), modified with 5' Fluidigm CS1
and CS2 oligo sequences: CS1-ITS1F1
[5'-ACACTGACGACATGGTTCTACA-GAACCWGCGGARGGATCA-3'] and CS2-ITS2R
[5'-TACGGTAGCAGAGACTTGGTCT-TGTGTTCTTCATCATG-3'].
Bioinformatic analysis of ITS1 and 16S rRNA gene fragment amplicons
Raw Illumina output files were checked for quality using FastQC and preprocessed
using the Ribosomal Database Project pipeline (Cole et al., 2014). In brief, sequences
were demultiplexed and trimmed for length, adapter removal, and quality (Q-score
cutoff of 26). Paired-end reads from the 16S amplicons were assembled using the
RDP-modified version of PANDAseq (Masella et al., 2012), and all remaining singleton
sequences were discarded. ITS amplicons were quality trimmed, but not assembled.
BBDuk was used to remove Arabidopsis reads from both 16S and ITS datasets,
retaining summary statistics for the abundance of mitochondrial and chloroplast reads
per sample (Bushnell et al., 2016). All sequences were rarified and stripped of
singletons using USEARCH and UPARSE, respectively (Edgar et al., 2011). Clustering
(97% identity cutoff) was independently performed on 16S and ITS sequences using

87

UPARSE (Edgar, 2013). Chimeric sequences were removed from the 16S and ITS
datasets using UCHIME (Edgar, 2013), and ITS/16S OTU counts were determined by
mapping reads back to respective ‘.uc' files. Representative sequences for 16S OTUs
were extracted, aligned using the RDP Infernal 1.1 alignment model, and assigned
taxonomy using the RDP classifier classification from the RDP database (Cole et al.,
2014). Representative sequences for ITS OTUs were assigned taxonomy using the
RDP classifier trained to the UNITE fungal database (ITS trainset 07-04-2014). ITS
OTUs were defined using the USEARCH pipeline (version 9.2) and refined using a 97%
identity cutoff of denoised sequences. Each pooled sample was rarefied to 100,000
reads. ITS OTUs were classified using the UNITE ITS database (version 7.1) trimmed to
ITS1 and 50 flanking bases using ITSx (version 1.0.11) and HMMER (version 3.1b2).
RNA extraction and quantitative PCR
Total RNA was collected and purified from whole rosette tissue using the RNeasy Plant
Mini Kit (Qiagen) according to manufacturer's protocol with optional on-column DNase
digestion. Purified RNA was eluted in TE buffer (Tris-HCl 10 mM, pH 7.5, EDTA 1 mM).
RNA concentrations were determined using an ND-1000 NanoDrop spectrophotometer
(Thermo Scientific) or by Qubit RNA HS fluorometric assay (Thermo Fisher). Total RNA
samples were collected in 2.0 ml nucleic acid LoBind tubes (Eppendorf) and stored
at -80°C.
For quantitative PCR analysis, total RNA collections were used for cDNA biosynthesis.
In brief, first strand cDNA libraries were generated using M-MLV reverse transcriptase
according to previously published protocols (Withers and Yao, 2013). Quantitative PCR
was performed in 10 µl reaction volumes with ~0.5 ng·μl−1 cDNA template using SYBR

88

Green reagents (Applied Biosystems) and primers specific to each experiment. All
qPCR reactions were performed on an ABI 7500 Fast qPCR instrument (Applied
Biosystems) with three technical replicates and a minimum of three biological
replicates. A dissociation curve to confirm the presence of a single peak for each
primer set was generated after the final cycle of each reaction using standard
parameters (15 s at 95°C, 60 s ramp from 60°C to 95°C in 1°C increments, and 15 s at
95°C). All qPCR experiments were repeated a minimum of three times with consistent
results.
RNA-seq library preparation and analysis
Total RNA was extracted from whole rosettes of FlowPot grown Arabidopsis with Ag,
Prairie or Arid input microbiota. A biological replicate is defined as a pool of eight
rosettes collected from four FlowPots within the same micro box. Three biological
replicates per condition were collected, totaling nine holoxenic and nine axenic
replicates. Total RNA was checked for quality using a Bioanalyzer 2100 (Agilent), and
all samples were determined to have an RNA integrity (RIN) score of six or greater.
Stranded sequencing libraries were prepared using the NuGEN Ovation RNA-SEQ
System for Model Organisms (Arabidopsis) according to manufacturer's protocol
(NuGEN). Library preparation and sequencing was performed by the Michigan State
University RTSF. Sequencing was performed on the HiSeq 2500 (Illumina) with a 1 x 50
bp single read stranded format using Illumina HiSeq SBS reagents (version 4). Base
calling was done by Illumina Real Time Analysis (RTA version 1.18.64). FastQ files were
checked for quality using FastQC (version 0.11.5; Andrews, 2010), then processed
using the Joint Genome Institute BBMap tool suite (version 36.86; (Bushnell, 2016).

89

Preprocessing was performed using BBDuk to remove low-quality regions, PhiX reads,
Illumina adapters, and poly-A tails. The following parameters were implemented for
adapter and poly-A removal: k = 23, mink = 11, hdist = 1, ktrim = r, trimq = 12. BBMap
was used to map reads to the Arabidopsis thaliana Col-0 reference genome file in SAM
format (ARAPORT11 assembly; Krishnakumar et al., 2015). A minimum of 30 million
quality-passed reads were used for each sample. Splice variants were collapsed so
that a single expression value could be determined for each gene locus. Structural
RNAs were excluded from expression value calculations. Ambiguous read mappings
were assigned to the “best” match according to default parameters. Expression values
were determined from read counts normalized as fragments per kilobase per million
reads mapped (FPKM). The significance of differential expression was determined for p
values using Benjamini-Hochberg correction.
Tissue collection flg22 elicitation experiments
Plants were grown for three weeks in FlowPots under axenic conditions or with
specified microbiota. Whole rosettes were severed at root-shoot junctions and
carefully transferred to sterile distilled H2O (20 ml, 22°C) in 50 ml conical tubes or Petri
dishes. Suspended rosettes were gently rocked for 2 h, then bottom-loaded with
aqueous flg22 (20 ml, 22°C, 250 nM final concentration, 0.1% DMSO). Samples were
gently rocked at a 45-degree angle until the specified post-elicitation time points. For
MAPK phosphorylation assays, the final concentration of flg22 was adjusted to 1 µm,
rocking was omitted, and individual leaves were collected rather than whole rosettes.
Mock-treated samples were subjected to the equivalent volume of 0.1% DMSO. At the
time of collection, plant tissue was quickly blotted dry, immediately frozen in liquid

90

nitrogen, and stored at -80°C until homogenization for RNA or protein extraction. A
biological sample is defined as three rosettes pooled from the same FlowPot or 3-4
leaves pooled from the same FlowPot for RNA and protein extraction, respectively.
Protein extraction
Membrane proteins were extracted as previously described. Briefly, tissues were
mechanically disrupted and taken up in extraction buffer (50 mM Tris HCl pH 8.0, 150
mM NaCl, 10% glycerol, 1% Igepal CA 630 [Sigma Aldrich] and 0.5% sodium
deoxycholate), supplemented with protease and phosphatase inhibitor cocktails
(Roche) as needed). Soluble proteins were taken up in a buffer consisting of 50 mM
Tris HCl, pH 8.0, and 150 mM NaCl, Samples were cleared of debris, normalized to
total protein concentration. The relative abundance of each protein was visualized by
western blot using anti AtMPK3 (Sigma Aldrich), anti AtMPK6 (Sigma Aldrich), anti
BAK1 (Agrisera) or anti FLS2 (Agrisera) antibodies. Phosphorylation of AtMPK3/6 was
measured using an anti-phosphor-p44/42 MAPK.
Phytohormone quantification by UPLC/MS
Plants were grown under the previously described FlowPot regime in the presence or
absence of Michigan (Ag), Iowa (Prairie), or California (Arid) soil microbiota for three
weeks. Extraction was performed according to previously published methodology, with
some modifications (Zeng et al., 2011). In brief, rosette tissue (~50-100 mg) was
weighed, harvested, and frozen in liquid nitrogen. Tissue and cells were homogenized
using three 3 mm Zirconium beads (Glen Mills) and Tissuelyser II (Qiagen, 30 Hz for 2 x
1 min). Homogenates were vortexed and spun down at 12,000 × g for 10 min at 4°C.

91

Extraction was performed with 1 ml methanol	water and taken up in phytohormone
extraction buffer (methanol: water (80:20 v/v), 0.1% v/v formic acid, 0.1 mg·ml−1
butylated hydroxytoluene (BHT) supplemented with 100 nM deuterated-ABA
(ABA-2H6) as an internal standard. Upon overnight incubation at 4°C with gentle
agitation, samples were cleared by centrifugation, filtered through a 0.2 µm PTFE
membrane (Millipore) and transferred to autosampler vials. Calibration standards (1.56
nM -100 nM) were prepared using SA, IAA, ABA, JA, MeJA (Sigma-Aldrich) and JA-Ile
(Paul Staswick) supplemented with 100 nM ABA-2H6. Separation, quantification, and
analysis were performed as described (Zeng et al., 2011). Transitions from
deprotonated molecules to characteristic product ions were monitored for SA (m/z
137>93), JA (m/z, 209.1>59), ABA (m/z 263.1>153.1), JA-Ile (m/z 322.2>130.2, and
ABA-2H6 (m/z 269.1>159.1) in negative ion mode and IAA (m/z 176.1>130.1) and
MeJA (m/z 225.1>151.1 in positive ion mode. For SAG analysis, a gradient of 100 mM
ammonium formate acidified with 0.05% formic acid (solvent A) and methanol (solvent
B) was applied for 5 min over the same column at 0.4 ml·min−1. The gradient increased
from 98:2 A:B to 50:50 A:B for 180 s, then immediately increased to 100% B for 60 s
and returned to 98:2 A:B for the remainder of the run. Transitions from deprotonated
molecules to characteristic product ions were monitored for SAG (m/z 299>137) using
a cone voltage and collision energy of 22 V and 11 eV, respectively. UPLC/MS work
was performed by Brad Paasch with assistance from Dr. Tony Schilmiller and Dr. Dan
Jones at Michigan State University RTSF Mass Spectrometry Facility.

92

Metabolite quantification by GC/MS
Plants were grown under the previously described FlowPot regime in the presence or
absence of Ag, Prairie, or Arid soil microbiota for three weeks. Extraction was
performed according to previously published methodology (Zeng et al., 2011). In brief,
~200 mg (fresh weight) of rosette tissue were frozen in liquid nitrogen, homogenized
upon the addition of three 3 mm Zirconium beads (Glen Mills) using a Tissuelyser II (30
Hz for 2 x 1 min, Qiagen), and extracted with 1 ml methanol∶water (1∶1 v/v) containing
formic acid [0.1% (v/v)] and butylated hydroxytoluene (BHT, 0.1 mg·ml−1) at 4°C for 24
h. Homogenates were vortexed and spun down at 12,000 × g for 10 min at 4°C, and
supernatants were collected, filtered through 0.2 µm PTFE membrane (EMD Millipore),
and transferred to autosampler vials. Due to ion suppression of detectable signal,
sorbitol was added as an internal standard to each of the extracts. Extracts were dried
under nitrogen, derivatized, and processed according to previously published
methodology (Li et al., 2012; Tschaplinski et al., 2012). Metabolite quantities are
reported as µg·g−1 fresh weight (sorbitol equivalent), derived from the mean of at least
three biological replicates from three experimental replicates (nine samples per
condition). Statistical confidence is reported as P-values calculated using a student's
t-test, and error is reported as standard error of mean values. Sample processing and
GC/MS was performed in collaboration with Oak Ridge National Laboratory.
Bacterial growth conditions, sources, and strain construction
All bacterial strains were obtained from the He lab culture collection, with the exception
of Pseudomonas syringae pv. tomato DC3000 Tn7-GFP, which was constructed for

93

this study by triparental mating according to previously published methods (Teal et al.,
2006). The insertion locus of the mini-Tn7-nptII-gfp cassette was confirmed by PCR to
be in the predicted attTn7 site downstream of glmS, and donor plasmid was removed
by supplementing growth media with 10% sucrose for sacB-mediated
counter-selection. All Pseudomonas strains were grown at 28°C, and E. coli strains
grown at 37°C in LM or LB media with appropriate antibiotics, respectively.
Infiltration of rosettes with bacterial suspensions
Overnight cultures of the specified Pseudomonas strains were prepared in LM broth
supplemented with rifampicin and incubated at 28°C. The overnight cultures were
diluted to OD600 = 0.002 (~5 x 106 CFU·ml−1) in liquid LM media, spread onto LM agar
plates, and incubated for 24 h at 28°C to form lawns. The working infection titre was
prepared by gently rinsing the lawns in sterile MgCl2 [1 mM] and diluting to OD600 =
0.0002 (~5 x 105 CFU·ml−1) in MgCl2 [1 mM] unless specified otherwise. Bacterial
concentrations were confirmed by plating a serial dilution of the suspension in triplicate
on LM agar supplemented with appropriate antibiotics, and CFUs were counted after
48 h of incubation at 28°C. Silwet L-77 [0.0025% v/v] (Lehle Seeds, Round Rock, TX,
USA) was added 10 min prior to infiltrations, and homogenized with a magnetic stir rod
at low speed to prevent foaming. Sterile 80 ml beakers were each filled with 60 ml of
bacterial suspension. Immediately prior to vacuum infiltration, each FlowPot was
immersed, stabilized by a sterile paper clamp fastened to the Luer end. The immersed
FlowPots (six at a time) were placed into a glass bell vacuum desiccator (Wheaton).
Vacuum was applied until the pressure within the glass bell desiccator reached 70 kPa,
whereupon the vacuum was maintained for 1 min and slowly released. The vacuum

94

step was repeated for a total of two iterations. FlowPots were removed from
suspensions and placed on KimWipes. Residual liquid was gently blotted from
rosettes, and the FlowPots were placed back into the Eco2 boxes. Miracloth (EMD
Millipore) was placed over the opening of the Eco2 box and fastened using the
cut-away lip portion of an Eco2 lid. Each covered Eco2 box was placed in a BSL2
hood for ~1 h, or until the rosettes no longer appeared water-soaked. Microbox lid
covers were then replaced, and boxes were put back in the growth chamber at 23°C
[100 μmol·m-2·s−1] for 2, 3 or 4 d as specified. Mock inoculations were performed in
parallel.
Bacterial multiplication in planta protocol
Whole rosettes of representative plants were collected after vacuum infiltration at the
time point specified. Leaf discs (3 mm2) were punched from representative leaves at
approximately the same developmental stage and were placed in 2.0 ml high-density
screw-cap microtubes containing three zirconium beads (3 mm). Per biological
replicate, a total of four leaf discs were pooled from plants sharing the same FlowPot.
To each tube, MgCl2 buffer [1 mM] or phosphate buffer (PB, 10 mM, pH 7) was added
(500 µl), and leaf discs were mechanically homogenized using a Tissuelyser II (Qiagen,
30 Hz for 2 x 1 min). After homogenization, an additional 500 µl of MgCl2 or PBS buffer
was added to each tube, vortexed, and a 3-step 100-fold serial dilution was performed
in a round-bottom 96-well microplate with 500 µl well capacity (Nunc, #267334). For
assays in which the infiltrated bacteria constitutively express gfp (Pst-GFP), cell counts
were enumerated using an Accuri C6 flow cytometer (BD Biosciences) equipped with a
robotic sampler arm (BD Biosciences). In brief, a 10X dilution of a Pst

95

DC3000-infiltrated leaf homogenate was analyzed, followed by a wash step and a 100X
dilution of Pst-GFP infiltrated leaf homogenate 3 dpi with estimated bacterial counts of
~1x108 viable cells·ml−1. A volume of 100 µl was analyzed from each sample at medium
fluidics flow rate. The Accuri C6 was set to the following parameters: 6,000 SSC
threshold, FL1 filter (510/15), FL2 filter (530/40), FL3 (670LP). A GFP-specific gate was
established by removing background events from the Pst DC3000 infiltrated leaf
homogenate sample viewed on log10 scale scatter plots of FL1 by FSC and FL2 by FSC
axes (supplemental figure). A gate for the positive control Pst DC3000-GFP sample
was established by simultaneous visual inspection of the Pst DC3000-GFP and Pst
DC3000 leaf homogenate scatter plots, and events within the GFP-specific gate were
determined to be analogous to CFU·cm-2 when compared to plate counts of serial
dilutions. Experimental cells·cm2 data was recorded for samples with 100-6,000 GFP
events·s-1 according to manufacturer's suggestions. Multiple dilutions within the
100-6,000 GFP events·s-1 range for the same biological sample are considered
technical replicates and data points are reported as the mean of technical replicate
values. A minimum of six biological replicates were analyzed for each condition per
experiment, and all multiplication assays were repeated a minimum of three times with
consistent results. Multiplication assays for non-GFP bacteria were performed
according to a previously published protocol (Katagiri et al., 2002).
Oxidative burst assay
Reactive oxygen species generated upon flg22 elicitation were quantified using a
luminol-based kinetic assay (Leslie and Heese, 2014). In brief, micro boxes containing
3-4 weeks old Arabidopsis were partially opened to allow relative humidity equilibration

96

for ~30 m. Leaf discs (3 mm2) from representative plants were transferred to a white
flat-bottom 96-well microplate (Berthold Technologies), and suspended in ddH2O (100
µl per well) with abaxial side facing down. Leaf discs were floated for 12-14 h to
attenuate wounding responses. Horseradish peroxidase (HRP, 10 mg·ml−1) was
dissolved in ddH2O, and luminol [17 mg·ml−1] was dissolved in DMSO. From stock
solutions, a combined aqueous working solution of HRP [80 µg·ml−1] and luminol [13.6
µg·ml−1] was prepared. A separate elicitor solution was prepared by diluting flg22 in
ddH2O [800 nM]. The HRP/luminol working solution was added to each well containing
a leaf disc (50 µl per well), and the plate was then incubated in the dark for 15 min. The
elicitor solution was added to each well (50 µl per well), and the suspended leaf discs
were immediately placed in a SpectraMax L Microplate Reader (Molecular Devices) to
collect luminescence data. Final flg22 concentration for oxidative burst assays was 200
nm. The SpectraMax L parameters were set to the following: kinetic read mode,
photon counting mode, integration time 1 s, target wavelength 470 nm, and assay time
90 min. Leaf discs from multiple plants sharing a common FlowPot were considered
technical replicates, and the mean of 4-6 technical replicates is considered a biological
replicate. Each experiment used a minimum of three biological replicates from different
micro boxes, and the flagellin-insensitive mutant fls2-1 or efr fls2-1 was used as a
negative control. Data is reported as mean photons per square centimeter of leaf tissue
per second, and error bars represent standard deviation of the mean. Experiments
were repeated a minimum of three times with consistent results.

97

Callose deposition assay
Staining for callose depositions was performed according to a previously published
protocol with some modification (Nomura et al., 2011). In brief, representative 2-3 w
rosettes were vacuum-infiltrated with flg22 (500 nM + 0.0025% Silwet L-77) and placed
in micro boxes. After a 24 h incubation period under standard growth conditions, whole
rosettes were collected and immersed in a clearing solution (1:3 acetic acid: ethanol)
for 1 h at 65°C, then incubated overnight at room temperature with gentle rocking.
Destaining solution was replaced if saturated. Clearing solution was aspirated and
replaced with rinse buffer (K2HPO4 [150 mM]) for ~30 min, and replaced with aniline
blue staining solution (K2HPO4 [150 mM] pH 8.5 and aniline blue [1 mg·ml−1]) for 2 h.
Representative leaves were mounted in glycerol [50% v/v] and observed under UV
excitation to detect and photograph callose depositions. Depositions were enumerated
using ImageJ software and are reported as mean depositions per square centimeter for
three leaves (leaf two, three, and four), among eight biological replicates. Error bars
represent the standard deviation of depositions per square centimeter. Callose
deposition assays were repeated a minimum of three times with similar deposition
enumeration and identical conclusions. Flg22-insensitive mutant fls2-1 was used as a
negative control.

98

Results
Bacterial and fungal compositions of input soil communities are distinct
To minimize the community-specific effects of a microbiome on Arabidopsis
phenotypic attribute, I elected to use three microbial input communities from distinct
soil types and geographic locations. Chemical analysis of the soils revealed that the
Prairie soil was closest to neutral pH (6.6), and that the Ag and Arid soils are more
acidic (pH 5.6 and 5.4, respectively), and that the Ag and Arid soils have low buffering
capacity (Table 1). Other notable differences between the soils are relatively low
concentrations of P (9 µg∙g-1), K (77 µg∙g-1), Ca (186 µg∙g-1), Mg (52 µg∙g-1) and Mn (7.3
µg∙g-1) in Ag soil, a relatively high concentration of P (200 µg∙g-1) and organic content
(6.3%) in Prairie soil, and relatively high concentrations of K (379 µg∙g-1), Fe (190
µg∙g-1), Na (28 µg∙g-1), S (357 µg∙g-1), Cl (1,000 µg∙g-1) and soluble salts (3.52 µg∙g-1) in
Arid soil (Table 1). Precipitation and seasonal temperature fluctuations also contribute
to different environmental factors that likely contribute to composition and diversity of
each respective soil microbiome.

99

Figure 3.1. Map of input community soil collection locations and photographs taken at the
time of collection. From Left to Right: a fine loam “Arid” soil from a backslope in southern
California (33.3037°N, -116.8511°E; altitude 1,604 M), a sandy loam “Ag” agricultural soil from
Michigan from an unamended Miscanthus plot (42.7089°N, -84.4656°E; altitude 261 M) and a
fine silt “Prairie” soil from an undisturbed grassland in Iowa (41.8361°N, -93.0078°E; altitude
271 M).

100

Table 3.1. Attributes of soils used in this study
Agricultural

Prairie

Arid

%Organic

2.1

6.3

2.3

pH

5.6

6.6

5.4

BpH

6.7

6.9

6.9

P

9

200

72

K

77

170

379

Ca

186

1499

1958

Mg

52

296

284

Mn

7.3

49.8

68

Cu

1.1

6.7

6.2

Fe

35

46

190

NO3-

22

7

317.9

NH4+

2.1

8.8

5.6

S

14

12

357

Na

1

2

28

Soluble Salts

0.12

0.23

3.52

Chloride

151

178

1000

All measurements are in µg∙g-1

101

Bacterial and fungal communities of the Ag, Arid, and Prairie soils were analyzed for
microbial community composition by targeted amplicon sequencing of the 16S rRNA
SSU gene fragment and the ribosomal intergenic spacer region (ITS1), respectively.
Among the three input soil samples, 367 ITS1 OTUs accounted for a minimum of 0.1%
of the reads for at least one sample. Although 14 OTUs could not be classified to the
phylum level, the majority of OTUs were classified to four known phyla (n=353):
Ascomycota (n=257), Basidiomycota (n=64), Zygomycota (n=30), and Glomeromycota
(n=2). OTU classifications fell into sixteen different classes, of which 182 were
classified to the genus level (>0.5 confidence from RDP classifier). Only four core OTUs
were detected, classified as the following genera: Fusarium (n=2) and Trichoderma
(n=2 (Figure 2C). Collectively, these four OTUs accounted for <10% of the total
classified ITS reads for each sample (Arid=2.7%, Prairie=3.9%, Ag=9.6%) (Table 3).
Two OTUs were exclusively shared among Arid and Prairie, both of which are classified
as Alternaria. Seven OTUs are exclusively shared among Arid and Ag, five of which are
classified as fungal genera (within Ascomycota): Humicola (n=2), unidentified
Chaetothyriales (n=2), and Drechslera (n=1). Notably, Zygomycota was far less
abundant in Arid soil relative to Prairie and Ag (Zygomycota reads as percent of total:
Arid=0.17%, Prairie=31.53%, Ag=4.94%).
Bacterial community composition analysis of the input soils yielded 298 16S rRNA SSU
OTUs (97% cutoff), each accounting for a minimum of 0.1% average of the reads for at
least one sample group. Relative to ITS1 OTUs, there is a greater number of core 16S
OTUs are shared among all input communities (n=13; Figure 2A), also accounting for a
relatively higher portion of shared reads (Arid=18.7%, Prairie=22.9%, Ag=36.6%).

102

Notably, Prairie soil contains over 10-fold more Firmicutes and Bacteroidetes than Arid
or Ag soils (Figure 2B; Figure 5). Agricultural soil had the greatest relative abundance of
Acidobacteria, positively correlating with the acidic pH of the soil. The Arid soil had the
highest relative abundance of Actinobacteria (Figure 2).

103

Table 3.2. Classifications of core 16S OTU classifications from input soil
Phylum

Family

Genus

Actinobacteria

Mycobacteriaceae

Mycobacterium

Firmicutes

Bacillaceae

Bacillus

Alphaproteobacteria

Bradyrhizobiaceae

Bradyrhizobium

Alphaproteobacteria

Bradyrhizobiaceae

unclassified

Alphaproteobacteria

Rhizomicrobium

Rhizomicrobium

Alphaproteobacteria

Sphingomonadaceae

Sphingomonas

Alphaproteobacteria

Sphingomonadaceae

Sphingomonas

Alphaproteobacteria

Xanthobacteraceae

Pseudolabrys

Alphaproteobacteria

Hyphomicrobiaceae

Rhodoplanes

Betaproteobacteria

Burkholderiaceae

Burkholderia

Betaproteobacteria

Comamonadaceae

Variovorax

Betaproteobacteria

unclassified

unclassified

Betaproteobacteria

Comamonadaceae

unclassified

Classifications based on RDP taxonomy with at least 80% confidence.

104

Table 3.3. Classifications of core ITS1 OTU classifications from input soil
Phylum

Family

Genus

Ascomycota

Nectriaceae

Fusarium

Ascomycota

Nectriaceae

Fusarium

Ascomycota

Hypocreaceae

Trichoderma

Ascomycota

Hypocreaceae

Trichoderma

Classifications based on RDP taxonomy with at least 80% confidence.

105

Figure 3.2. Venn diagram of classified bacterial 16S rRNA OTUs from input communities,
displaying number of shared OTUs among each sample, along with (B) relative abundance of
OTU (>0.1%) classifications at the Phylum level. (C) Venn diagram of classified fungal ITS1
OTUs, displaying number of shared OTUs among each sample, (D) relative abundance of OTU
(>0.1%) classifications at the Phylum level.

106

Fungal and bacterial community differentiation
We have established that the input soil communities are distinct, and in Chapter 2, we
determined that both synthetic and naturally-extracted soil communities differentiate to
form distinct root-associated and substrate-inhabiting communities six weeks after
inoculation in the FlowPot system. As a basis for downstream phenotypic analysis of
the host, we further characterized differentiation of the Ag, Arid, and Prairie input
communities at week three, comparing within-pot differentiation of rosette-associated
microbiota from substrate-associated, as well as comparisons of community
composition as the result of different input community inoculum. This experiment was
conducted in the FlowPot system according to the schema depicted in Figure 3. Total
rosette tissue and bulk substrate was collected three weeks after germination (occurs
2-3 days after sowing).

107

Figure 3.3. Experimental design for 3-community FlowPot inoculation: Agricultural soil,
Prairie soil, and Arid soil. (A) Three aliquots of each soil input community are collected at t=0.
For each soil input community, 6 micro boxes with 4 FlowPots each are (B) inoculated. At week
3, (B) whole rosettes are collected, along with substrate, and samples are pooled by box. Thus,
each micro box containing 4 individual FlowPots is considered a single sample.

108

In addition to the steps taken for input ITS1 fungal community analysis, for substrate
and rosette samples we took an additional step to remove artefactual OTUs resulting
from amplification of Arabidopsis DNA, with kmer-based decontamination tool BBDuk
(version 36.92, kmer length=150). A total of 13.26% of unique sequences prior to
denoising and OTU clustering. Additionally, a BLAST search was performed using the
TAIR10 assembly of Arabidopsis thaliana Col-0 to scan for OTUs with high similarity to
Arabidopsis DNA. One OTU (OTU1) with 99.1% identity across 100% of the query to a
region on A. thaliana chromosome 3, annotated as the ITS1 region, was removed.
Thus, despite USEARCH classification of OTS1 as Xylaria with 98.3% confidence,
OTU1 was treated as plant contamination. The reference sequence to which OTU1 has
the highest identity is Xylaria sp. B101 (genbank accession KJ512146.1). The ITS
sequence of Xylaria sp. B101 is an unpublished direct submission to NCBI described
as a fungal endophyte of Cleisostoma williiamsonii, and no additional information is
provided regarding whether steps were taken to eliminate plant genomic DNA during
template preparation. Among samples with >50,000 high quality read pairs,
Arabidopsis (OTU1) accounted for an average of 0.001%, 11.762% and 99.419% of
classified paired reads for respective soil input, bulk, and rosette DNA samples.
Among the detected ITS1 OTUs across all samples, 1413 are classified as Fungi, 15
are classified as Plantae, 72 are classified as Protista, and 20 as unidentified. OTUs are
classified at the phylum level for 988/1413 OTUs. Plantae OTUs are unclassified at
higher resolution than the domain level. All Protista OTUs are classified as Cercozoa,
and 44/72 of the Cercozoa classifications have greater than 90% confidence of
classification. Fungal-classified OTUs accounted for an average of 98.644%, 85.413%

109

and 0.580% of paired reads for respective soil input, matrix, and rosette samples.
Although rosette DNA with Ag microbiota has <0.1% read pairs classified as fungal
OTUs, rosettes with Arid-derived and Prairie-derived microbiota have >1.1% and
>0.6% of read pairs classified as fungal OTUs, respectively. Read pairs classified as
Protista account for <1.0% of all samples with only two exceptions: Prairie input
(2.05%) and Ag substrate (5.30%). A single core OTU (OTU4) identified as Fusarium sp.
(74.85% confidence) was detected among rosettes with Arid microbiota. Three core
OTUs identified as Penicillium sp. (OTU10, 85.29% confidence), Mucor sp. (OTU39,
86.78% confidence), and Umbelopsis sp (OTU16, 99.16% confidence) were detected
among rosettes with Iowa-derived microbiota. Rosettes with Ag microbiota have a
single core ITS1 OTU classified as a member of Chaetomiaceae (OTU2, 80.3%
confidence).
The bulk substrate bacterial community composition for all three samples was mostly
(>80%) composed of Proteobacteria (by decreasing abundance: Betaproteobacteria,
Alphaproteobacteria, and Gammaproteobacteria) (Figure 4). While Arid and Prairie
groups respectively had 3.7% and 4.1% Bacteroidetes reads, Ag had >0.1%
Bacteroidetes. None of the bulk substrate samples had >0.5% Actinobacteria. Beta
diversity analysis using UniFrac distances revealed strong clustering by sample type
(Figure 5). Despite similar taxonomic profiles at the phylum-level resolution, bulk
substrate samples only 16 core OTUs (Figure 6C), and rosettes share only 6 (Figure
6D). Among soil input communities, Ag and Arid had the greatest alpha diversity based
on the inverse Simpson index (Figure 6A), but bulk substrate and rosettes with Ag and
Arid microbiota had less alpha diversity than samples with Prairie microbiota. Input

110

communities all had greater alpha diversity than bulk substrate and rosettes. For each
corresponding sample, the alpha diversity of bulk substrate was not significantly
different from rosette microbiomes (Figure 6AB). The most highly abundant OTUs in
rosette samples with Arid microbiota were classified as: Massilia Otu2 (30.9±5.6% of
reads), Chryseobacterium Otu5 (21.5±8.7% of reads), Achromombacter Otu8
(17.9±4.3% of reads), and Pseudomonas Otu6 (9.8±1.7% of reads). Dominant Prairie
rosette microbiota were: Novosphingobium Otu4 (20.1±4.3%), Pseudomonas Otu6
(15.1±3.8% of reads), Rhizobium Otu13 (9.6±4.0%), and Cupriavidus/Ralstonia Otu11
(5.3±1.6%). For rosettes with Agricultural microbiota were predominantly composed of:
Paraburkholderia Otu1 (55.7±1.6%), Rhizobium Otu3 (16.4±1.4%), Paraburkholderia
Otu14 (7/0±1.9%). OTUs with greater relative abundance in the rosette that the
corresponding bulk substrate were considered rosette-enriched (also >0.1% abundant,
pairwise t-test p<0.05). Rosettes with Arid microbiota were enriched in 9 OTUs (Figure
8), Prairie with 21 OTUs (Figure 9), and Agricultural with 11 OTUs (Figure 10; Table 4).

111

Table 3.4. Rosette-enriched OTUs relative to abundance in bulk substrate.

OTU
42
90
64
114
48
5
40
10
941
1429
352
4
13
160
27
3
59
8
1
166
7
1530
21
2
43
61
11
20
24
592
6
99

Classification
Ac
Gordonia
Ac
Gordonia
Ac
Leifsonia
Ac
Nocardioides
Ac
Rhodococcus
Bc
Chryseobacterium
Fr
Paenibacillus
AP Caulobacter
AP
AP
AP
AP Novosphingobium
AP Rhizobium
AP Rhizobium
AP Rhizobium
AP Rhizobium
AP Sphingomonas
BP Achromobacter
BP Paraburkholderia
BP Paraburkholderia
BP Paraburkholderia
BP Herbaspirillum
BP Herbaspirillum
BP Massilia
BP Massilia
BP
BP Ralstonia
GP Erwinia
GP Pantoea
GP Pseudomonas
GP Pseudomonas
GP Stenotrophomonas

Arid
%Rs
1.1
0.3

21.5
0.6
0.4

Rs/B
17.2
8.5

6.5
24.1
23.5

17.9
0.4

22
0.1

0.2

1.7

1
31
0.2

8.4
0.3
9.9

0.8
3.1

0.4
5.9

9.8
0.2

5.7
0.9

p
*
*

*
*

**

**
**

*
**

Prairie
%Rs Rs/B
0.1
6.7

sig

0.5
0.1
0.6
1.5
0.1
0.1
0.2

16
5.4
3
16.2
10
1
19.3

*
*
*
**
*
**
*

0.2
20.1
9.6
0.4
2.2
0.4

8.3
16
27.7
8.6
3.1
2.4

***
***
*
*
*

2.4
0.5

0.7
1.5

1.6
2.7
0.1

0.6
4.8
4.9

5.3
1.3
0.1
0.6
15.1
0.8

2.9
3.2
0.7
1.5
1.2
5.6

**
**

Agricultural
%Rs
Rs/B

p

0.1

6

*

0.6

35.7

0.2

12.9

0.5

32.2

16.4
0.1
0.6
55.7
0.4
1.7
0.1

47.2
3.1
0
0.4
1.1
10.2
1.4

***
***

2.3
0.3
0.4

2.1
29.8
28.7

*
***
***

1.6

1

***

**
*
***
*

*
*
*
*
*

%Rs= mean percent of rosette sample reads of the sample group, R/B= root enrichment over
bulk. Only OTUs with >0.1% abundance in one or more rosette samples were included. Ac=
Actinobacteria, Bc=Bacteroidetes, Fr= Firmicutes, AP= Alphaproteobacteria, BP=
Betaproteobacteria, GP= Gammaproteobacteria, p-values from two-tailed t-test: p < 0.05 *, p
< 0.005 **, p < 0.0005 ***.

112

Figure 3.4. Mean percent abundance of 16S OTU-classified reads by phylum. The right
bound maximum across the x-axis represents 100%. Error bars represent standard error of the
mean for each sample type. Ar= Arid soil; P=Prairie soil; Ag=Agricultural soil.

113

Figure 3.5. UniFrac distances between samples based on 16S OTU clustering.

114

Figure 3.6. Alpha diversity indices for each sample type: (A) inverse Simpson and (B)
Shannon (base 2). Error bars represent standard error of the mean. Alpha diversity indices are
calculated from rarefied count table to 10,000 reads per sample. All bulk and root samples
have a minimum of n=5 samples. The number of shared OTUs with a minimum of 0.1% mean
read abundance in at least one sample group was determined for (C) bulk substrate and (D)
plant rosettes. Soil input microbiota abbreviations: Ar=Arid, P=Prairie, Ag=Agricultural.

115

Figure 3.7. Abundance of 16S OTUs for plant (whole rosette) communities with genus
classifications. Each of the 36 OTUs individually accounted for >0.5% of the average reads in
at least one rosette sample group. Each column represents a sample replicate. Reads are sum
scale normalized.

116

Figure 3.8. Rosette microbiome 16S OTU enrichment relative to abundance in bulk
substrate for Agricultural microbiota samples. (A) All 16S OTUs plotted and OTUs filtered by
phylum (B-G). Log2(rosette/bulk) abundance plotted across the x-axis. Statistical confidence
was calculated for each OTU comparison by pairwise t-test and plotted on the y-axis
as -log10(p). OTU data point symbols are size scaled to reflect abundance in rosette
microbiome (if x>0) or bulk substrate microbiome (if x<0).

117

Figure 3.9. Rosette microbiome 16S OTU enrichment relative to abundance in bulk
substrate for Prairie microbiota samples. (A) All 16S OTUs plotted and OTUs filtered by
phylum (B-G). Log2(rosette/bulk) abundance plotted across the x-axis. Statistical confidence
was calculated for each OTU comparison by pairwise t-test and plotted on the y-axis
as -log10(p). OTU data point symbols are size scaled to reflect abundance in rosette
microbiome (if x>0) or bulk substrate microbiome (if x<0).

118

Figure 3.10. Rosette microbiome 16S OTU enrichment relative to abundance in bulk
substrate for Arid microbiota samples. (A) All 16S OTUs plotted and OTUs filtered by phylum
(B-G). Log2(rosette/bulk) abundance plotted across the x-axis. Statistical confidence was
calculated for each OTU comparison by pairwise t-test and plotted on the y-axis as -log10(p).
OTU data point symbols are size scaled to reflect abundance in rosette microbiome (if x>0) or
bulk substrate microbiome (if x<0).

119

Axenic plants lack normal basal expression of immune-associated genes
To achieve a global snapshot of basal transcriptional differences between axenic and
holoxenic Arabidopsis, we used the FlowPot system to grow axenic plants in parallel
with plants in substrate inoculated with Ag (MI), Arid (CA) or Prairie (IA) soil microbiota
(Figure 8), collected whole rosette tissue at week 3, and performed RNAseq on total
RNA. Axenic plants were confirmed to be truly axenic, or devoid of culturable
microbiota, by incubating sterilized inoculum at t=0 on R2A medium, as well as
incubating plant material on R2A medium at the time of tissue collection for RNA
extraction. There was no detectable microbial growth from material from any axenic
micro boxes used in this experiment.

120

Figure 3.11. RNA extraction experimental design and pipeline. (B) Images of plants at the
time of collection for total RNA extraction.

121

RNA-seq data was analyzed using two separate group definitions: all holoxenic (n=9)
vs all axenic (n=9), and input-specific holoxenic (3 groups of n=3) vs. input-specific
axenic (3 groups of n=3). In the all vs. all (axenic vs holoxenic) analysis, biological
replicates do not account for the input community source (holoxenic n=9, axenic n=9).
This allowed for greater statistical power and noise reduction to determine core
differentially expressed genes. A gene’s expression value is defined as mean reads per
million (RPM), averaged across all nine replicates. The input-specific analysis provides
granularity to determine whether certain genes are differentially expressed when
colonized by certain taxa, but at the expense of statistical power. This allows for the
determination of pan differentially expressed genes.
Using stringent criteria to define an expressed gene (RPM>5), we detected the pan
expression of 15,050 genes, accounting for ~54% of annotated protein-coding
Arabidopsis genes (Table 5). Differential expression of genes was analyzed at three
levels of stringency: |log2(RPM/RPM)|>2.0 (Table 5B), |log2(RPM/RPM)|>1.0 (Table 5C),
and |log2(RPM/RPM)|>0.5 (Table 5D). Collectively, there are only three ‘core’
differentially expressed genes at the |log2(RPM/RPM)|>2.0 stringency, all of which are
less abundant in axenic: a peroxidase superfamily protein-coding gene (AT5G64120), a
MLO-family protein-coding gene (MLO12), and a chitinase family protein-coding gene.
Notably, all three of these genes are defense-associated based upon gene ontology.
The peroxidase superfamily protein is involved in lignification of cell walls and induced
during fungal defense, the MLO12 protein is a calmodulin-binding and is annotated as
a mildew resistance locus, and the chitinase family protein is also involved in fungal
defense (Supplemental Table).

122

Of the 30 core differentially expressed genes (i.e., differentially expressed axenic vs. all
holoxenics) using moderate stringency criteria, 29 are less abundant in axenics (Table
6). The only gene expressed at relatively higher levels in axenic is a putative
sugar-phosphate exchange protein (AT4G32480) (mean log2(ax/holo)=1.5). Gene
ontology analysis (GO:BP), based on direct experimental evidence, indicates that 11 of
the axenic-depleted genes are categorized as “defense response to other organism
[GO:0098542]”, (p<0.0008, Benjamini-Hochberg corrected). Notably, this list includes
two defense-associated WRKY transcription factors (WRKY48, WRKY53),
REDOX-associated glutathione S-transferase (GSTF6), genes involved in glucosinolate
metabolism (CYP81F2, IGMT2), a cell wall-associated peroxidase (AT5G64120),
receptor-like kinases with leucine-rich repeat domains (IOS1, MLO12, CRK14), and a
chitinase family protein (AT2G43620).

123

Table 3.5. Differentially expressed genes (DEG) with stringent criteria.

Agricultural (Ag)
Ag+Pr
Prairie (Pr)
Arid+Pr
Arid
Arid+Ag
Arid+Ag+Pr (core)
Pan (holo pooled, ax pooled)

ax enriched
genes
%exp
1
0.01
1
0.01
5
0.03
0
0
0
0
0
0
0
0
0
0

Criteria: log2(RPM/RPM)| ≥ 2.0 p < 0.05, RPM ≥ 5

124

ax depleted
genes
%exp
46
0.3
23
0.16
31
0.2
3
0.02
15
0.1
4
0.03
3
0.02
22
0.15

differential (DEG)
genes
%exp
47
0.31
24
0.16
36
0.24
3
0.02
15
0.1
4
0.03
3
0.02
22
0.15

Table 3.6. Differentially expressed genes (DEG) with moderate criteria.

Agricultural (Ag)
Ag+Pr
Prairie (Pr)
Arid+Pr
Arid
Arid+Ag
Arid+Ag+Pr (core)
Pan (holo pooled, ax pooled)

ax enriched
genes
%exp
22
0.15
5
0.03
36
0.24
1
0.01
28
0.18
2
0.01
1
0.01
9
0.06

Criteria: log2(RPM/RPM)| ≥ 1.0 p < 0.05, RPM ≥ 5

125

ax depleted
genes
%exp
188
1.24
97
0.66
164
1.08
33
0.22
173
1.14
55
0.37
29
0.2
136
0.9

differential (DEG)
genes
%exp
210
1.39
102
0.69
200
1.32
36
0.24
201
1.32
57
0.39
30
0.21
145
0.96

Table 3.7. Differentially expressed genes (DEG) with sensitive criteria.

Agricultural (Ag)
Ag+Pr
Prairie (Pr)
Arid+Pr
Arid
Arid+Ag
Arid+Ag+Pr (core)
Pan (holo pooled, ax pooled)

ax enriched
genes
%exp
248
1.64
37
0.25
281
1.85
21
0.14
246
1.62
41
0.28
10
0.07
104
0.69

Criteria: log2(RPM/RPM)| ≥ 0.5 p < 0.05, RPM ≥ 5

126

ax depleted
genes
%exp
634
4.18
207
1.40
535
3.53
97
0.66
605
3.98
174
1.18
70
0.48
501
3.33

differential (DEG)
genes
%exp
882
5.82
246
1.66
816
5.39
140
0.95
851
5.59
219
1.48
81
0.55
605
4.02

Table 3.8. Core differential expression meeting significance criteria at moderate stringency.
Name

Locus

Arid

Ag

Prairie

Description

AT5G64120

AT5G64120

-2.69

-3.98

-4.45

Peroxidase superfamily protein

MLO12

AT2G39200

-2.25

-3.93

-3.78

Seven transmembrane MLO family

AT2G43620

AT2G43620

-2.00

-3.56

-3.70

Chitinase family protein

IOS1

AT1G51800

-1.55

-4.08

-2.81

Leucine-rich repeat RLK

IGMT2

AT1G21120

-2.34

-2.07

-3.12

O-methyltransferase family protein

AT1G30720

AT1G30720

-1.60

-2.82

-2.51

FAD-binding Berberine family

GSTF6

AT1G02930

-1.15

-2.34

-2.95

glutathione S-transferase 6

AT1G30730

AT1G30730

-1.35

-2.71

-2.43

FAD-binding Berberine family

AT2G27660

AT2G27660

-1.52

-2.96

-1.74

Cysteine/Histidine-rich C1 domain

AT1G26420

AT1G26420

-1.44

-2.24

-2.45

FAD-binding Berberine family protein

WRKY48

AT5G49520

-2.42

-1.97

-1.68

WRKY DNA-binding protein 48

BCB

AT5G20230

-2.17

-1.78

-1.72

blue-copper-binding protein

AT1G72060

AT1G72060

-1.63

-1.92

-1.91

serine-type endopeptidase inhibitor

AT2G18690

AT2G18690

-1.73

-1.87

-1.91

transmembrane protein

FRB1

AT5G01100

-2.25

-1.80

-1.45

O-fucosyltransferase family protein

CRK14

AT4G23220

-1.69

-1.80

-1.46

cysteine-rich receptor kinase

CYP81F2

AT5G57220

-1.15

-1.47

-2.19

cytochrome P450

AT5G67340

AT5G67340

-1.02

-1.99

-1.73

ARM repeat superfamily protein

AT5G35735

AT5G35735

-1.39

-1.77

-1.63

Auxin-responsive family protein

WRKY53

AT4G23810

-1.46

-1.59

-1.43

WRKY family transcription factor

AT1G49000

AT1G49000

-1.37

-1.54

-1.36

transmembrane protein

AT1G65845

AT1G65845

-1.33

-1.51

-1.34

transmembrane protein

EXO70B2

AT1G07000

-1.44

-1.59

-1.15

exocyst subunit exo70 family

CNI1

AT5G27420

-1.07

-1.25

-1.73

carbon/nitrogen insensitive 1

PMT6

AT4G36670

-1.08

-1.59

-1.15

Major facilitator superfamily protein

AT5G52750

AT5G52750

-1.23

-1.47

-1.23

Heavy metal transport/detoxification

AT1G61360

AT1G61360

-1.28

-1.16

-1.53

S-locus lectin protein kinase family

AT4G39830

AT4G39830

-1.20

-1.39

-1.37

Cupredoxin superfamily protein

DIC2

AT4G24570

-1.36

-1.02

-1.16

dicarboxylate carrier 2

AT4G32480

AT4G32480

+1.75

+1.51

+1.10

sugar phosphate exchanger

Criteria (for all sample groups): |log2(RPM/RPM)| > 1.0, RPM > 5, p < 0.05 (BenjaminiHochberg corrected)

127

At high-sensitivity differential gene expression criteria, there are 605
differentially-expressed genes (4.02% of all protein-coding genes): 501
axenic-depleted and 104 axenic-enriched. To simultaneously visualize significance and
magnitude of differential expression, a volcano plot was generated, with sensitive
parameters for differential gene expression indicated by solid black lines (Figure 11).
Gene ontology enrichment analysis (GO:BP biological process) was performed on
these significantly enriched and significantly depleted regions. Overwhelmingly, the
most strongly enriched category with the highest confidence is GO:BP involved in
defense, with a p-value (Benjamini-Hochberg corrected) of p < 3.62 x 10-42. Of the 501
axenic-depleted transcripts, 137 are annotated as defense-associated (Figure 12).
Collectively, at stringent and moderate differential expression criteria, there are more
differentially-expressed genes in common among holoxenic plants grown in Ag
(Michigan) and Prairie (Iowa) inoculated substrate. A distance matrix was generated
based on gene expression values (RPM), and principal component analysis was
performed and plotted to visualize whether community-specific gene expression
profiles cluster together (Figure 11). Indeed, while all axenic samples cluster together,
Ag (Michigan) and Prairie (Iowa) samples cluster away from the axenic, and gene
expression profile differences are explained entirely by the x-axis (PC1). The Arid
(California) holoxenic plants had expression profiles that clustered together, but apart
from the Arid and Ag holoxenic plants, forming a third discrete cluster.

128

Figure 3.12. Total differentially expressed genes considering input community (sample
size = 3 for each), displayed with at three stringency levels: (A) strict, (B) moderate, and (C)
sensitive.

129

Figure 3.13. Volcano plot of pan differential gene expression (all axenic/holoxenic).
Positive values across the x-axis indicate the gene is more highly expressed in axenic relative
to holoxenic (log2(RPMAX/RPMHO). The boxed-out regions on the left and right correspond to
the most sensitive criteria for differential expression (log2(RPMAX/RPMHO)|>0.5). Gene ontology
analysis (GO:BP biological process) was performed on these significantly enriched and
significantly depleted regions. Annotations that indicate the gene is involved in biotic defense
are highlighted in pink. Of the 501 axenic-depleted transcripts, 137 are annotated as
defense-associated. The p-value (Benjamini-Hochberg corrected) of this enrichment analysis is
p < 3.62 x 10-42. Gene names or corresponding brief gene descriptions are annotated as labels
next to the symbol representing the corresponding gene.

130

Figure 3.14. Principal component analysis for transcriptome profiles in this experiment.
Profiles for axenic plants all cluster together and separate from holoxenic samples. Iowa and
Michigan holoxenic cluster together, while holoxenic California clusters away from the other
two groups. Michigan=Ag; Iowa=Prairie; California=Arid.

131

Figure 3.15. Gene set enrichment analysis of microbiome composition-dependent
changes in host transcriptome, relative to corresponding axenic samples. Confidence is
expressed as –log10 of a p-value (Benjamini-Hochberg corrected).

132

Figure 3.16. Community-specific gene set enrichment analysis, contrasting holoxenic
samples by input community. (A-C) Ternary plots where each gene is represented by a
symbol, and deviation from the mean RPM expression value is represented by skewing
towards the input community for which the expression is relatively higher. The dotted line
represents one standard deviation from the mean; the dark gray symbols represent two
standard deviations from the mean, and the pink symbols represent statistical significance by
2-way ANOVA. Light gray symbols indicate the p-value for enrichment is p>0.05 and/or that
the mean RPM for that sample group is within 0.5 standard deviations of total mean RPM
among all holoxenic groups. CA=Cali=Arid, Iowa=IA=Prairie, MI=Mich=Ag.

133

BAK1

log2(ax/ho)

log2RPM

Gene

Ar Ag Pr all p

ax ho

IOS1
BIR1

**
**

impaired oomycete susceptibility; LRR-RLK
BAK1-interacting receptor-like kinase

FLS2
RBOHD

**
***

immune receptor for bacterial flagellin
NADPH/respiratory burst oxidase protein

EFR
BAK1
PEPR1

*
**
*

immune receptor for bacterial EF-TU
BRI1-associated receptor kinase
amplifies immune response to pathogens

BRI1

*

brassinosteroid insensitive

**

BRI1-associated kinase

***
**

chitin elicitor receptor kinase
involved in chitin perception
peptidoglycan sensing and response

CERK1 BKK
CERK1
LIK1
LYM1
LYM3
PTI

Description

peptidoglycan-binding

FRK1
FOX1

**
*

MAPK-specific flg22-induced RLK
FAD-linked oxidoreductase

AT5G39580
PER4

*
*

haem peroxidase
peroxidase

YLS9
WAK2

*
**

similar to NDR1; senescence-associated
wall-associated kinase; SA-inducible

Absolute: Log2(RPM)
0

10
Enrichment: Log2(ax/ho)

-2

2

Figure 3.17. Differential and absolute expression of PTI recognition and response genes,
coding for proteins of the BAK1 pattern recognition complex, CERK1 pattern recognition
complex, and pattern-triggered immune (PTI) response marker genes. Differential expression is
defined as log2(RPMAX/RPMHO); absolute expression is log2(RPM). Significance determined by
t-test, p-values: * p<0.5, ** p<0.005, *** p<0.0005, **** p<0.00005 (Benjamini-Hochberg
corrected). Soil input microbiota abbreviations: Ar=Arid, P=Prairie, Ag=Agricultural.

134

Gene
JA

log2(ax/ho)
Ar Ag Pr all p

JAZ1
OPR3

*

jasmonate-zim-domain protein
required for JA biosynthesis

JAZ5

*

jasmonate-zim-domain protein

MYC2
MYC4
JAM2
JAZ8
AOS
LOX2
JAR1
MYC3
MES16
PR

log2RPM
ax ho
Description

flavonoid biosynth and REDOX response
bHLH transcription factor
JA associated bHLH transcription factor
jasmonate-zim-domain protein
Involved in JA biosynthesis
lipoxygenase required for JA accumulation
catalyzes the formation of jasmonyl-Ile
bHLH transcription factor
methyl esterase

***
*

*
***

PR4
PR2
PR1
PR3
PR5

*
*
*

pathogenesis-related; chitin-binding
beta 1,3-glucanase
pathogenesis-related gene; SA-responsive
basic chitinase
pathogenesis-related gene

*

Absolute: Log2(RPM)
0

10
Enrichment: Log2(ax/ho)

-2

2

Figure 3.18. Differential and absolute expression of jasmonic acid-associated (JA) and
pathogenesis-related (PR) marker genes. Differential expression is defined as
log2(RPMAX/RPMHO); absolute expression is log2(RPM). Significance determined by t-test,
p-values: * p<0.5, ** p<0.005, *** p<0.0005, **** p<0.00005 (Benjamini-Hochberg corrected).
Soil input microbiota abbreviations: Ar=Arid, P=Prairie, Ag=Agricultural.

135

Gene
SA

log2(ax/ho)
Ar Ag Pr all p

AZI1
ANK
BCS1
SYP122
CBP60G
PEN3
WAK1
SARD1
WAK2
PAD4
NPR1
TGA5
EDS1
BSMT1
PBS3
ICS1
CPR1
UGT75B1
EPS1
PAL3

log2RPM
ax ho
Description

**
*
***
***
**
***
*

azelaic acid induced; involved in SAR
ankryn protein; SA-inducible
amplifies SA signaling
syntaxin, neg regulation of SA/JA signaling
MAMP-inducible; involved in SA
SA-dependent non-host resistance
wall-associated kinase
SAR deficient; SA biosynthesis regulator
wall-associated kinase; SA-inducible
phytoalexin deficient, lipase; SA signaling
key regulator of SA-dep SAR
NPR1-interacting B-ZIP protein
enhanced disease susceptibility
methylates salicylic acid and benzoic acid
AvrPphB susceptible; auxin-responsive
isochorismate synthase; SA accumulation
constitutive PR genes, neg reg of defense
UDP-glucosyltransferase; callose
pathogen-induced SA biosynthesis
phenyl alanine ammonia-lyase

**
*
***
*

*

Absolute: Log2(RPM)
0

10
Enrichment: Log2(ax/ho)

-2

2

Figure 3.19. Differential and absolute expression of salicylic acid-associated genes.
Differential expression is defined as log2(RPMMAX/RPMHO); absolute expression is log2(RPM).
Significance determined by t-test, p-values: * p<0.5, ** p<0.005, *** p<0.0005, **** p<0.00005
(Benjamini-Hochberg corrected). Soil input microbiota abbreviations: Ar=Arid, P=Prairie,
Ag=Agricultural.

136

Gene
WRKY
TFs

log2(ax/ho)
Ar Ag Pr all p

WRKY48
WRKY18
WRKY30
WRKY60
WRKY53
WRKY47
WRKY46
WRKY6
WRKY8
WRKY33
WRKY58
WRKY25
WRKY15
WRKY40
WRKY54
WRKY17
WRKY3
WRKY4
WRKY7
WRKY22
WRKY45
WRKY69
WRKY29
WRKY51
WRKY28
WRKY21
WRKY70

log2RPM
ax ho
Name

***
**
***
**
***
***
***
***
**
**
*
**
***
*

+ reg of defense against Pst
+ reg of ETI; complex with WRKY40/60

*
*
**

- reg of Pst resistance

pathogen-responsive
+ reg of senescence and defense
WRKY53-mediated defense against Pst

- reg of SA signaling

+ reg of ETI; confers Pst defense

- reg of Pst resistance
mediates dark-induced leaf senescence

*

JA responsive
interacts with TCP20 to regulate SA
- reg of defense
Absolute: Log2(RPM)
0

10
Enrichment: Log2(ax/ho)

-2

2

Figure 3.20. Differential and absolute expression of WRKY transcription factors.
Differential expression is defined as log2(RPMAX/RPMHO); absolute expression is log2(RPM).
Significance determined by two-tailed t-test, p-values: * p<0.5, ** p<0.005, *** p<0.0005, ****
p<0.00005 (Benjamini-Hochberg corrected).

137

Basal metabolite analysis (GC/MS)
To further characterize the basal state of the axenic plants and contrast with the basal
state of holoxenic plants grown in holoxenic substrate (with the Ag, Arid or Prairie
microbiota), we performed metabolic profiling in collaboration with Oak Ridge National
Lab. Relative to all holoxenic communities, phenyl-B-D-glucopyranoside was ~100-fold
more abundant in axenic plants (Figure 21, Supplemental Table). Four additional
metabolites were significantly more abundant in axenic plants, two of which were
identified as sucrose and the polyamine putrescine, and the other two were not fully
identified beyond classification as a feruloyl glycoside and an N-containing metabolite,
respectively (Figure 21). Based on the GC/MS spectral peak library of Oak Ridge
National Lab, 93 metabolites were above the lower threshold of detection. Each
metabolite was quantified relative to a sorbitol standard, and differential abundance of
each metabolite was assessed by comparing community-specific holoxenic
abundance to the corresponding axenic sample in triplicate and tested for significance
using a t-test. The number of differentially abundant metabolites is community-specific:
18/93 for Arid (n=14 axenic-enriched), 25/93 for Ag (n=20 axenic-enriched), and 39/93
for Prairie (n=8 axenic-enriched). Ascorbic acid is significantly enriched in Ag and
Prairie axenic, but not in Arid axenic. The compound “13.79 204 231 glycoside” is
significantly depleted in axenic plants relative to holoxenic-Arid and holoxenic Prairie
holoxenic, but not for holoxenic-Ag. Only two metabolites were consistently lacking in
axenic plants relative to all holoxenic plants: citric acid and 2-ethylhexanoic acid.
Based on our transcriptome data, we hypothesized that azelaic acid and salicylic acid
would be reduced basal levels in axenic plants. However, azelaic acid differential

138

abundance was either not significant (Arid, Prairie), or slightly more abundant in axenic
(Ag: axenic 9.4±1.0, holoxenic 5.6±0.5 µg·g-1 FW).

139

Arid
phenyl-B-D-glucopyranoside
13.27 feruloylglycoside
11.07 174 329 314 256 213 157 N-metabolite
sucrose
12.74 320 192 467
ornithine
citrulline
putrescine
glycine
asparagine
glutamine
9.52 155 229 214 187 172 N-metabolite
ascorbic acid
myo-inositol
azelaic acid
lysine
allantoin
serine
glucose 6-phosphate
B-cyano-alanine
glucose
uric acid
14.18 299 387 370 204 217 sugar-phosphate
12.22 232 347 449 N-metabolite
11.47 359 269 174 N-metabolite

p

Ag

***
***
***
*
*

***
*

p

Prairie p

***
***
**
***

***
***
***
**

*
*
**
**
*
*
*
***
*
*

*
*
*

***

*
*

*

Enrichment: Log2(ax/ho)
-2

2

Figure 3.21. Relative abundance of metabolites extracted from whole rosette tissue and
quantified by GC/MS (against a sorbitol standard). Differential metabolite abundance is
defined as log2(Axenic/Holoxenic) in µg·g-1 FW; corresponding raw concentrations are in
Table S3-2. Significance determined by t-test, p-values: * p<0.5, ** p<0.005, *** p<0.0005, ****
p<0.00005.(continued on next page [1/4])

140

Figure 3.21 (cont’d)

Arid

p

Ag

p

Prairie p

threonine
cysteine
tyrosine

*

tryptophan
threonic acid
B-alanine

*

phenylalanine

***

spermidine
10.86 306 288 N-metabolite

*
*

galactose
glyceric acid
glutamic acid
11.25 217 450 dehydrosugar
sinapic acid-4-O-glucoside

*
***

12.64 320 479 464
10.65 295 310 267 phenolic

**

12.52 507 103 189 285 249 261

*

12.98 320 307 217
isoleucine
leucine
valine
4-hydroxybenzoic acid
phosphate

*

gamma-aminobutyric acid
3,4-dihydroxybenzoic acid

*

Enrichment: log2(ax/ho)
-2

2

141

Figure 3.21 (cont’d)

Arid

p

Ag

ethanolamine
5-oxo-proline
salicylic acid
methylphosphate
erythronic acid
phloroglucinol
12.69 395 249 410 519 504
caffeic acid
monopalmitin
12.03 219 235
aspartic acid
succinic acid
alanine
1,2,4-benzenetriol
glycerol
3-hydroxybenzoic acid
13.35 409
9.99 98 288 390 N-metabolite
sinapoyl malate
13.47 342 299 315 P-metabolite
stearic acid
malic acid
fumaric acid
8.24 273 258 231 216
maleic acid

*
*
*

*
*
*
*
*
*
*
*
*
*
**
*
*

2

142

Prairie p

*

Enrichment: log2(ax/ho)
-2

p

Figure 3.21 (cont’d)

Arid
palmitic acid
13.79 204 231 glycoside
dihydrouracil
sinapic acid
9.79 346 174 N-metabolite
phytol
8.06 249 379 115
oleic acid
stigmasterol
linoleic acid
monostearin
B-sitosterol
12-hydroxyoctadecanoic acid
glycerol-1/3-phosphate
14.39 320 217 361 119
citric acid
12-hydroxyoctadecanoic acid methyl ester
2-ethylhexanoic acid

p

Ag

*

*
*
*
*

*
*
*

*
***

***

*
*
**
*
***

***

***

***

2

143

Prairie p

***

Enrichment: log2(ax/ho)
-2

p

Basal: Salicylic acid quantification (UPLC/MS)
To further characterize the basal status of axenic and holoxenic Arabidopsis, we
followed the clues that there may be defense associated phenotypic differences.
Although GC/MS did not reveal a significant different in salicylic acid accumulation, we
elected to quantify using a more sensitive UPLC approach with a salicylic acid
standard. Additionally, we quantified the glucosylated form of salicylic acid
(SA-glucosyl). Relative to Arabidopsis grown in any holoxenic substrate tested,
corresponding axenic plants were deficient in basal SA levels (Figure 22). Although
rosettes from plants grown in holoxenic Prairie substrate had similar levels of
SA-glucosyl to plants grown in corresponding axenic substrate, holoxenic Prairie
rosettes had the highest relative and absolute level of free SA (Figure 22).

144

Figure 3.22. Defense hormone salicylic acid quantification as (A) free SA and (B)
SA-glucosyl. Quantifications were performed with eight biological replicates per condition, and
with three experimental replicates all showing the same trends. Significance determined by
two-tailed t-tests; (Michigan=Ag, Iowa=Prairie, California=Arid);
* p<0.05, ** p<0.005;

145

Transcriptional response to immune elicitation is defective in axenic plants
Based on abnormally low levels of defense-associated gene expression and reduced
basal salicylic acid levels, we hypothesized that axenic plants may be immunodeficient.
To further interrogate immune-associated phenotypes of axenic Arabidopsis, we
performed a series of immune elicitation experiments using a fragment of bacterial
flagellin protein “flg22” (Felix et al. 1999). Initially, comparatively analyzed holoxenic
and axenic flg22-inducible expression of the following immune-associated genes were
quantified by qRT-PCR: ICS1 (ISOCHORISMATE SYNTHASE 1), FLS2
(FLAGELLIN-SENSITIVE 2), and FRK1 (FLG22-induced receptor-like kinase 1).
Although the induction of the expression of all three genes was detected in holoxenic
and axenic plants, the absolute expression levels of all three genes was higher in
holoxenic, except for FLS2 in Prairie(Iowa) holoxenic, which was not significantly
different from the corresponding axenic.

146

Figure 3.23. Expression of immunity and defense associated genes 2 hours after treatment
with flg22. (Michigan=Ag, Iowa=Prairie, California=Arid). Significance testing by one-way
ANOVA with post-hoc Tukey-Kramer test.

147

Compromised axenic posttranslational immune response
An early event that occurs during PTI after immune elicitation by flg22 is
phosphorylation of mitogen activated protein kinase (MAPK) proteins (Ligterink et al.
1997). First, we interrogated whether axenic and holoxenic plants had equivalent basal
levels of MPK3/6 and then whether flg22-induced phosphorylation of MAPK3/6
differentially occurs in axenic plants. The Axenic abundance of both MPK3 and MPK6
are equivalent to Ag-holoxenic in a basal state (Figure 24B). However, upon flg22
elicitation, axenic plants are partially compromised in phosphorylation of MPK3/6
relative to Ag-holoxenic (Figure 24A). At a high concentration of flg22 [1000 nM], both
axenic and holoxenic plants underwent MPK3/6 phosphorylation, but holoxenic plants
had a greater relative abundance of p-MPK3/6 (Figure 24A). At lower concentrations of
flg22 [100nM], phosphorylated MPK3/6 was not consistently detected in axenic plants
(Figure 24A).

148

Figure 3.24. MAPK protein detection and quantification by Western blot of (A)
Mitogen-Activated Protein Kinase 3/6 (MPK3/6), and (B) phosphorylated Mitogen-Activated
Protein Kinase 3/6 (p-MPK3/6) after treatment with 1000nM or 100nM flg22. The visible band in
the Ponceau stained gels are from the RuBiSCO LSU. Holoxenic refers to Ag soil inocula.

149

Oxidative burst and callose deposition are compromised in axenic plants
Foliar treatment of Arabidopsis with flg22 elicits an immune response also
characterized by reinforcement of cell walls with callose and a rapid apoplastic “burst”
of reactive oxygen species (ROS) (Dixon et al. 1994). Although flg22-induced callose
depositions were observed in axenic leaves, they tended to be concentrated in
clusters, in contrast to homogenous deposition patterns in holoxenic leaves (Figure
25A). Collectively, axenic Arabidopsis had fewer callose depositions per unit area than
holoxenic. Furthermore, a luminol ROS quantification assay revealed that axenic plants
are partially compromised in oxidative burst after elicitation (Figure 25D).

150

Figure 3.25. Response to immune elicitation by flg22 [500 nM] quantified by (A-B) callose
deposition counts, and oxidative burst of reactive oxygen species (C-D). Significance testing
by one-way ANOVA with post-hoc Tukey-Kramer test or t-test, * 0.05 > p.

151

Defective priming and defense against pathogens in axenic plants
Based on numerous indications of defective innate immunity, we hypothesized that
axenic Arabidopsis would have decreased capacity to defend pathogen invasion.
Indeed, the foliar pathogen Pseudomonas syringae pv. tomato DC3000 (Pst) was able
to grow to a significantly higher population density in axenic plants two days after
vacuum infiltration (with 5 x 105 CFUmg-2) relative to holoxenic plants with Ag, Arid or
Prairie microbiota (Figure 26A). Notably, among the three holoxenic plants, Pst had a
significantly lower population in Arabidopsis with Arid microbiota relative to the Ag or
Prairie microbiota. Foliar disease symptoms, particularly chlorosis and tissue collapse,
were more pronounced in axenic infected plants (Figure 26C). We further tested
whether the salicylic acid analog benzothiadiazole (BTH) would prove to be equally
efficacious at priming the defenses of axenic and holoxenic plants (Görlach et al.

1996). BTH was efficacious in both axenic and holoxenic plants in regard to reducing
the pathogenicity of Pst, but the priming was more effective in holoxenic plants (Figure
26B). Notably, BTH pretreatment of axenic plants and mock pretreatment of holoxenic
plants resulted in equivalent susceptibility to Pst. To interrogate whether indeed innate
immunity could explain axenic susceptibility, we performed diseases assays using Pst
∆hrcC, which has a defective type III secretion system and is unable to deliver
immune-suppressing effectors to inhibit host innate immunity. Likewise, we performed
vacuum infiltration disease assays using used a PTI-compromised Arabidopsis triple
mutant bak1 bkk cerk1 (bbc hereinafter) with wild type Pst and Pst ∆hrcC (Figure 26C).
Consistent with wild type Arabidopsis (Col-0), the bbc mutant grown in holoxenic

152

condition was ~10-fold resistant to Pst, compared to that grown in axenic condition,
and there was not a significant difference between WT and bbc mutant Arabidopsis
based on this vacuum infiltration assay. However, the magnitude of axenic
susceptibility to Pst ∆hrcC in the bbc mutant background was significantly greater
(Figure 26C).

153

Figure 3.26. Pathogen susceptibility of axenic and holoxenic Arabidopsis to (A)
Pseudomonas syringae pv tomato DC3000 measured as colony forming units per area at two
days post inoculation. (B) BTH protection assay using Pst DC3000 in wild-type Arabidopsis
(Col-0), two days post infection and three days post BTH pretreatment. (C) Disease symptoms

154

Figure 3.26. (cont’d)
in five-week-old Arabidopsis, three days after infection with Pst DC3000. (D) Susceptibility of
WT Arabidopsis (Col-0) and the bak1 bkk cerk1 (bbc) mutant to Pst and Pst ∆hrcC.
Significance testing by one-way ANOVA with post-hoc Tukey-Kramer test. (Michigan=Ag,
Iowa=Prairie, California=Arid);

155

Discussion
Here, we report that soil-grown axenic Arabidopsis has numerous dysfunctions in
innate immunity. In this study, our initial objective was to perform an in-depth basal
characterization of axenic Arabidopsis relative to holoxenic Arabidopsis with three
distinct microbiomes, and our findings inspired us to further characterize the
microbiome effect on plant innate immunity. All experiments herein were conducted in
the homogenous environment of the FlowPot system under nutrient-replete conditions,
enabling us to minimize abiotic stress and look beyond the nutritional services
provided by microbiota to identify subtle but significant differences in the basal
transcriptome and metabolome. In our hands, the presence/absence of soil microbiota
in a non-diseased state results in significant differential expression of merely ~0.2% of
Arabidopsis protein-coding genes, based on standard criteria (log2 fold change >2, pvalue < 0.05, minimum expression > 5 RPM). However, at higher sensitivity, we
discovered axenic plants are defective in the basal expression of a significantly
disproportionate amount of defense and immunity-associated genes, including genes
associated with the defense hormone Specifically, genes associated with the defense
hormone salicylic acid (SA), pattern-triggered immunity (PTI) and systemic acquired
resistance (SAR). Targeted phytohormone quantification validated that indeed, axenic
plants have reduced levels of SA. Three microbiomes from geographically, physically
and chemically distinct soils were used as holoxenic input microbiota, enabling us to
identify and report a repertoire of “core” microbiome-modulated plant genes, as well as
community-specific influence over the host transcriptome. In-depth basal
characterization of axenic and holoxenic Arabidopsis also revealed numerous

156

differentially abundant transcripts and metabolites that have not been previously
reported to be related to biotic interactions. Empirical investigation into axenic
immunodeficiency revealed that recognition and/or response to elicitation by flg22, a
common bacterial molecular pattern, is less robust. We also report that axenic
Arabidopsis is significantly compromised in the ability to defend against a bacterial
foliar pathogen. Likewise, protection against infection via pre-activation of SA signaling
is less efficacious in axenic plants. Foliar disease assays with a PTI-compromised
Arabidopsis bbc mutant and virulence-compromised Pst ∆hrc mutant definitively
revealed that immunodeficiency is not sufficient to explain increased axenic
susceptibility to infection, but other mechanisms such as microbe-microbe antagonism
and niche exclusion are also likely involved. The absence or drastic reduction in basal
expression of multiple PTI marker genes in axenic plants relative to all three holoxenic
plants include such markers as: FRK1, FOX1, AT5G39580, PER4, YLS9, and WAK2.
This is not unexpected, because an axenic plant is naive to MAMPs, whereas
holoxenic plants constitutively encounter low-level amounts of MAMPs. A low-level
activation of PTI may coincide with other defense phenotypes observed in axenic
plants. For example, the lower expression of SA-responsive gene expression, including
AZI1, PEN3, PAD4, ACD6 and NPR1, are indicative of a compromised SA signaling in
axenic plants. There is also a subtle down-regulation of Pathogenesis Related 1 (PR1),
a very widely used SA marker (Tsuda et al., 2013), in axenic plants. This was further
validated by the finding of a reduced basal concentration of SA in axenic plants.
Interestingly, some well-characterized pattern recognition receptors (PRRs) or coreceptors (IOS1, BIR1, FLS2, EFR, BAK1, PEPR1, BRI1, BKK, CERK1, LIK1) as well as

157

several uncharacterized LRR-RLK proteins (AT1G51890, AT1G51790, AT1G51850),
and AT4G14370 (TIR-NBS-LRR) are also expressed at a lower basal level than in
holoxenic plants. The protein BAK1 is of particular significance because it is a
coreceptor of multiple PRRs (e.g., EFR and FLS2). Also interesting is the finding of
numerous axenic-downregulated genes are specifically implicated as important for
SAR, including ACD6 and NPR1. Based on gene ontology (GO:BP) analysis, “Defense
Response” to “Biotic Stimulus”, “Response to Bacterium”, “Response to Fungi”, and
“Response to Stress” comprise the core down-regulated categories in axenic plants.
Another striking example of depleted gene expression in axenic plants is the family of
WRKY transcription factor genes. WRKY transcription factors are often associated
with biotic interactions and stress response in Arabidopsis (Birkenbihl et al., 2017).
Consistent with a previous report that WRKY70 is more abundant in axenic plants
(Zhou et al., 2017), this gene was not found in our list of genes with repressed
expression in axenic plants.
Three distinct microbiomes, intentionally selected from geographically and
physiochemically distinct soils, were chosen for this study to represent diverse
holoxenic reconstitutions. As a result, we were able to identify community-specific
effects. For example the Arid (California) microbiota conferred a significant relative
enrichment of gene expression associated with abscisic acid (ABA). Interestingly, ABA
response is important for drought tolerance and desiccation stress, and given the
climate in which “Arid” microbiota were collected, it is an attractive hypothesis that
microbiota from arid regions may confer drought tolerance. Additionally, JA response
was significantly greater in the California holoxenics relative to the Prairie and

158

Agricultural holoxenics. Furthermore, Arid holoxenic plants had a greater relative
enrichment of glucosinolate biosynthesis genes expressed. Glucosinolates are required
for innate immune response in Arabidopsis (Clay et al., 2009). Although Ag and Prairie
microbiota conferred enrichment in JA subcategories (GO:BP ‘defense response to
fungus’ and GO:BP ‘response to fungus’), the overarching JA ontology category was
only significantly enriched for in Arid-holoxenic. The Agricultural soil conferred higher
relative expression of secondary metabolism associated genes, and the Prairie
microbiota conferred the greatest relative enrichment of ‘defense response to bacteria’
genes. In this study, we also identified microbiome community-dependent effects on
the Arabidopsis metabolome. Relative to axenic plants, as well as plants with
agricultural (Ag) or arid microbiota, Arabidopsis with prairie (Pr) microbiota have
significantly higher levels of stearic acid, malic acid, fumaric acid, maleic acid, palmitic
acid, sinapic acid, phytol, oleic acid, and linoleic acid. Prairie microbiota plants also
have significantly higher levels of glycerol 1/3-phosphate (G3P), which has been
reported as a defense-priming compound (Conrath et al., 2015; Kachroo and Robin,
2013). The community composition of Arabidopsis with Prairie, Ag, and Arid microbiota
is likely responsible for the differential phenotypes of holoxenic plants, but specifically
which microbiota are responsible is unknown. The Arabidopsis with Prairie microbiota
is dominated by the following bacteria, for some of which there are examples of
isolates that elicit biocontrol and/or prime defenses: Chryseobacterium (Domenech et
al., 2006), Pseudomonas (Pieterse et al., 2014), Achromobacter, and Massilia. Less
abundant, but consistently present microbiota in Arabidopsis with Prairie microbiota
are Enterobacter, Erwinia, Pantoea and Herbaspirillum. Compared to other holoxenics,

159

Arabidopsis with Prairie microbiota also had the greatest abundance of rosetteassociated fungal reads relative to chloroplast reads, dominated by four OTUs
classified as two genera: Mucor and Penicillium. Potentially, one or more members of
these genera have the ability to further prime defenses in Arabidopsis (Marques et al.,
2006; Thuerig et al., 2005). Likewise, it could be rhizosphere-associated microbiota or
substrate-colonizing microbiota that emit volatiles that can prime the defenses of the
plant (Lee et al., 2012). It is also possible that the increased resistance of Arid
holoxenic is entirely attributed to microbe-microbe antagonism.
This study also provides a reference data for the identification of genes relevant to
plant-microbiome interactions. For example, the expression of multiple genes encoding
FAD-binding berberines are depleted in axenic Arabidopsis. FAD-binding berberines
are involved in cell wall modification and REDOX reactions, but not biotic interactions
per se (Daniel et al., 2015). Likewise, the expression of multiple protease inhibitor/seed
storage/lipid transfer protein-coding genes are depleted in axenic plants without any
experimental evidence to suggest a function. It would be interesting to further
investigate the role of these genes in plant/microbiome interactions.

160

Arabidopsis has can recognize a 22 AA peptide (flg22) of bacterial flagellin via ligand
binding by the transmembrane leucine-rich repeat receptor-like kinase protein FLS2.
Upon flg22 binding, FLS2 autophosphorylation and activation of several branches of
PTI responses ensue. An early detectable response (<15 mins) is apoplastic burst of
reactive oxygen species (ROS burst), mediated by the association of the NADPH
oxidase RboHD with the FLS2 receptor complex (Couto and Zipfel, 2016). Later
responses for which axenic Arabidopsis is impaired include transcriptional responses
(SA and PTI marker genes), post-translational response (MAPK phosphorylation), early
REDOX response (apoplastic oxidative burst), and later cell wall reinforcement (callose
deposition). We speculate that axenic Arabidopsis may be compromised in PTI
responses not due to lack receptors or recognition problems, as justified by an inability
to detect differential abundance of FLS2 (Brad Paasch, personal communication).
Rather, we speculate there could be a basal REDOX homeostasis issue or simply
deficient levels of basal SA could explain axenic supe-susceptibility of axenic plants.
REDOX homeostasis has been attributed as an important element for stress tolerance
in Arabidopsis, and differential expression of sulfur-containing protein-coding genes
and glutathione-related genes could induce a higher redox status in holoxenic plants
(Dutilleul et al., 2003), a possibility that needs further investigation.

161

It is possible that factors in addition to compromised immunity contribute to increased
axenic susceptibility. These factors could include but not limited to, defects in PTI, lack
of microbe-microbe antagonism, and the increased availability of nutrient sources in
the axenic phyllosphere such as sucrose. Metabolic profiling revealed that axenic
plants have significantly higher levels of phenyl-B-D-glucopyranoside, sucrose,
putrescine (and trending increased abundances of other polyamines), feruloylglycoside.
and ascorbic acid. The underlying basis for this is unknown, but warrants further
investigation. Likewise, axenic plants a significant reduction in basal levels of citric acid
and 2-ethylhexanoic acid, and trending reduced abundance of linoleic acid and
glycerol-1/3/phosphate. Citric acid is potentially a REDOX sink, and G3P has been
previously implicated in defense, but 2-ethylhexanoic acid is puzzling and the
biosynthetic pathway for this compound in Arabidopsis is unclear, although there are
reports of its priming potential.

162

Complex multicellular life is inevitably intertwined with the microbial world. While some
host-microbe associations are mutualistic, many microorganisms have evolved
mechanisms to exploit eukaryotes for nutrient resources at the expense of the host
(i.e., parasitism). Presumably, the cost of parasitism and benefits of mutualism prompt
the selection for defense adaptations, ultimately giving rise to the evolution of intricate
signaling networks collectively termed innate immunity. But despite the fact that all
plants and animals have evolved in concert with microbiota, very little is known about
host- microbiome dynamics and the functional impact the microbiome has on host
phenotypes, particularly in plants. The innate immune system of plants is essential for
defense against pathogens; a sessile lifestyle rooted in soil bestows an inability to
escape stress in a milieu of potential pathogens. The molecular underpinnings of plant
innate immunity have traditionally been studied using model pathosystems (i.e., a wellcharacterized plant such as Arabidopsis thaliana treated with an artificially high
concentration of microorganism). Although this approach has built a fantastic
framework for understanding plant defense signaling, technological limitations have
hindered our ability to study plant immune response to the indigenous microbiome,
thus leaving a fundamental gap in our understanding of the evolution and dynamic
interplays between plants, indigenous microbiota and pathogens. For future directions,
I believe that the FlowPot gnotobiotic system and the dataset obtained in this study will
provide the foundation for future inquiry into a better mechanistic understanding of
functional impact of the plant microbiome on plant-microbe interactions and other
phenotypic aspects of plant biology.

163

Table 3.9. Expressed genes (EG)
Sample grouping
Agricultural (Ag)
Ag+Pr
Prairie (Pr)
Arid+Pr
Arid
Arid+Ag
Arid+Ag+Pr (core)
Pan (holo pooled, ax pooled)

n
15162
14794
15151
14794
15214
14758
14597
15050

%total
55.25
53.90
55.20
53.90
55.43
53.77
53.19
54.84

Criteria: mean RPM ≥ 5, 27445 loci

164

Table 3.10. Pan- differential expression at moderate stringency

Name
AT5G64120
MLO12
AT2G43620
IOS1
FRK1
XBAT34
AZI1
IGMT2
AT1G30720
AT1G51890
AT3G46280
AT4G22470
GSTF6
AT4G12490
AT1G30730
AT1G51790
GSTF7
AT2G27660
AT1G26380
AT1G26420
WRKY48
ACS7
FLOT1
ERD7
EARLI1

Description
Peroxidase superfamily protein
Seven transmembrane MLO family
protein
Chitinase family protein
Leucine-rich repeat protein kinase
flg22-induced receptor-like kinase 1
hypothetical protein
azelaic acid induced 1;
SAR component
O-methyltransferase
FAD-binding Berberine
Leucine-rich repeat protein kinase
kinase-like protein
Protease inhibitor/seed storage/LTP
glutathione S-transferase 6
Protease inhibitor/seed storage/LTP
FAD-binding Berberine
Leucine-rich repeat protein kinase
glutathione S-transferase 7
Cysteine/Histidine-rich C1 domain
FAD-binding Berberine
FAD-binding Berberine
WRKY transcription factor
Enzyme involved in ethylene
biosynthesis
SPFH/Band 7/PHB domain
membrane-associated
Senescence/dehydration-associated
protein
Protease inhibitor/seed storage/LTP

ax
3.4
1.8

expression
ho
p
56.9
***
28.2
**

all
-3.7
-3.4

Log2(axenic/holo)
Ar
Ag
Pr
-2.7 -4.0 -4.5
-2.3 -3.9 -3.8

1.4
1.9
1.1
2.3
1.6

21.4
23.0
15.7
19.4
14.3

**
**
**
***
**

-3.2
-3.1
-3.0
-2.6
-2.6

-2.0
-1.6
-1.4
-2.2
-1.5

-3.6
-4.1
-3.4
-3.4
-3.2

-3.7
-2.8
-3.3
-2.2
-2.6

2.9
2.8
1.2
1.2
1.8
3.0
1.1
2.1
1.4
2.4
2.1
1.1
1.4
3.9
1.4

21.1
19.4
10.4
10.1
12.7
18.4
9.2
13.9
10.5
14.7
13.1
8.4
9.4
19.2
8.6

***
***
**
*
*
**
**
***
*
**
*
*
***
****
*

-2.5
-2.4
-2.4
-2.3
-2.3
-2.3
-2.3
-2.3
-2.3
-2.2
-2.2
-2.2
-2.1
-2.0
-2.0

-2.3
-1.6
-1.2
-1.2
-0.8
-1.2
-1.0
-1.4
-0.2
-1.4
-1.5
-0.7
-1.4
-2.4
-0.8

-2.1
-2.8
-2.8
-3.2
-3.2
-2.3
-2.8
-2.7
-3.2
-2.2
-3.0
-2.1
-2.2
-2.0
-2.7

-3.1
-2.5
-2.8
-2.0
-2.1
-3.0
-2.6
-2.4
-2.4
-2.9
-1.7
-3.0
-2.5
-1.7
-2.2

1.8

10.1

**

-2.0

-0.9

-2.5

-2.3

19.0

76.8

**

-2.0

-1.8

-3.0

-0.8

1.2

7.4

*

-1.9

-0.7

-2.6

-1.9

Criteria: log2(RPM/RPM)| ≥ 1.0, p < 0.05, RPM ≥ 5. Sorted by ascending average enrichment.
Statistical confidence was determined by pairwise t-test (all holoxenic vs all axenic), with
Benjamini-Hochberg correction. * p < 0.05, ** p < 0.005, *** p < 0.0005, **** p < 0.00005

165

Table 3.10. (cont’d)
Name
PLA2A
AT5G48657
BCB
AT1G72060
AT1G51850
AT2G18690
FRB1
CNGC3
AT1G13470
AT4G14370
CRK5
AT2G36690
CYP71B23
CRK14
WRKY18
AT1G05675
AT3G18250
AT1G56660
CYP81F2
STZ
AT5G67340
CRK6
AT1G10340
AT5G35735
ARCK1

Description
phospholipase A 2A
defense protein-like protein
blue-copper-binding protein;
oxidative stress resp.
serine-type endopeptidase
inhibitor
Leucine-rich repeat protein kinase
transmembrane protein
O-fucosyltransferase family protein
cyclic nucleotide gated channel 3
hypothetical protein (DUF1262)
Disease resistance protein
(TIR-NBS-LRR)
cysteine-rich receptor-like kinase
2-oxoglutarate and
Fe(II)-dependent oxygenase
cytochrome P450
cysteine-rich receptor-like kinase
WRKY transcription factor
UDP-Glycosyltransferase
Putative membrane lipoprotein
MAEBL domain protein
cytochrome P450
salt tolerance zinc finger
ARM repeat superfamily protein
cysteine-rich receptor-like kinase
Ankyrin repeat family protein
Auxin-responsive family protein
receptor-like cytosolic kinase; ABA
neg regulator

expression
ax
ho
5.4
23.2
1.9
9.9
12.9
49.7

p
**
*
****

Log2(axenic/holo)
all
Ar
Ag
-1.9
-1.0
-2.4
-1.9
-1.0
-2.7
-1.9
-2.2
-1.8

Pr
-2.0
-1.7
-1.7

10.6

40.7

***

-1.9

-1.6

-1.9

-1.9

1.4
2.0
3.9
4.3
1.4
2.3

7.5
9.6
16.3
16.6
6.9
9.8

*
***
****
**
*
**

-1.8
-1.8
-1.8
-1.7
-1.7
-1.7

-0.3
-1.7
-2.3
-1.5
-0.1
-0.9

-2.4
-1.9
-1.8
-2.2
-2.7
-2.1

-2.2
-1.9
-1.5
-1.3
-1.5
-2.1

2.8
1.3

11.4
6.3

**
**

-1.7
-1.7

-0.4
-0.5

-2.1
-2.0

-2.3
-2.0

3.0
3.4
10.7
2.9
1.3
31.0
9.6
4.3
4.5
1.8
1.7
8.3
1.5

11.7
13.0
36.3
11.2
6.2
98.2
31.7
15.3
15.9
7.5
7.1
26.9
6.4

**
****
**
**
*
**
***
**
**
*
**
****
**

-1.7
-1.7
-1.7
-1.7
-1.7
-1.6
-1.6
-1.6
-1.6
-1.6
-1.6
-1.6
-1.6

-0.8
-1.7
-2.6
-1.4
-0.5
-1.2
-1.2
-2.0
-1.0
-0.8
-0.9
-1.4
-1.2

-2.0
-1.8
-2.0
-1.3
-1.8
-2.4
-1.5
-1.0
-2.0
-2.3
-1.9
-1.8
-2.4

-1.7
-1.5
-0.5
-2.2
-2.2
-1.1
-2.2
-1.7
-1.7
-1.3
-1.6
-1.6
-1.1

Criteria: log2(RPM/RPM)| ≥ 1.0, p < 0.05, RPM ≥ 5. Sorted by ascending average enrichment.
Statistical confidence was determined by pairwise t-test (all holoxenic vs all axenic), with
Benjamini-Hochberg correction. * p < 0.05, ** p < 0.005, *** p < 0.0005, **** p < 0.00005

166

Table 3.10. (cont’d)
Name
AT1G27020
AT5G10760
AT5G37600
LOX4
WRKY30
AT2G39210
ABCG42
JAZ1
WRKY60
RLP21
PCC1
WRKY53
ERD10
AT1G49000
PP2-A5
WAKL10
UCP5
PYL6
ATPMEPCRB
AT1G65845
ACD6
EXT4

EXO70B2
PP2-B13
GLR2.8

Description
plant/protein
Eukaryotic aspartyl protease
family protein
cytosolic glutamine synthetase
with NH3 affinity
PLAT/LH2 domain
lipoxygenase
WRKY transcription factor
Major facilitator superfamily
ABC-2 and Plant PDR
ABC-type transporter
jasmonate-zim-domain 1
WRKY transcription factor
receptor like protein 21
pathogen and circadian
controlled 1
WRKY transcription factor
Dehydrin; bacteriostatic
transmembrane protein
phloem protein 2 A5
Leucine-rich repeat protein
kinase, wall-associated
mitochondrial dicarboxylate
carriers
PYR1-like 6; ABA-responsive
pectin methylesterase
inhibitor; BR regulator
transmembrane protein
ankyrin repeat family protein;
Pst resistance
Extensin: cell-wall
hydroxyproline-rich
glycoproteins
exocyst subunit exo70 family
protein B2
phloem protein 2-B13
glutamate receptor 2.8

expression
ax
ho
3.3
11.7
1.2
5.5

p
*
*

Log2(axenic/holo)
all
Ar
Ag
-1.6
-0.9
-2.1
-1.6
-0.6
-2.0

Pr
-1.2
-1.8

10.8

33.6

**

-1.6

-0.8

-2.1

-1.5

19.4

58.3

**

-1.5

-2.1

-1.4

-0.9

1.2
16.1
7.6

5.4
47.9
23.7

***
**
**

-1.5
-1.5
-1.5

-1.2
-0.9
-0.9

-1.6
-2.0
-2.0

-1.8
-1.5
-1.3

14.6
5.6
1.4
3.3

43.4
17.7
5.8
11.2

*
**
*
*

-1.5
-1.5
-1.5
-1.5

-2.2
-0.7
-0.8
0.0

-1.1
-2.1
-2.2
-2.5

-0.8
-1.6
-1.2
-1.4

14.1
93.5
1.9
6.6
1.3

41.8
257
6.9
19.8
5.3

****
**
***
**
**

-1.5
-1.5
-1.5
-1.5
-1.5

-1.5
-2.3
-1.4
-1.5
-0.9

-1.6
-1.7
-1.5
-1.6
-1.4

-1.4
-0.2
-1.4
-1.1
-2.0

8.3

23.8

***

-1.4

-1.7

-1.6

-1.0

2.3
11.2

7.8
31.6

*
**

-1.4
-1.4

0.4
-1.1

-2.2
-2.1

-1.7
-1.0

22.6
35.6

61.8
96.1

***
**

-1.4
-1.4

-1.3
-0.7

-1.5
-2.0

-1.3
-1.3

106

279

**

-1.4

-0.9

-1.6

-1.6

12.1

33.4

****

-1.4

-1.4

-1.6

-1.2

2.1
1.6

7.1
5.8

**
****

-1.4
-1.4

-0.9
-1.1

-1.4
-1.8

-1.6
-1.3

Criteria: log2(RPM/RPM)| ≥ 1.0, p < 0.05, RPM ≥ 5. Sorted by ascending average enrichment.
Statistical confidence was determined by pairwise t-test (all holoxenic vs all axenic), with
Benjamini-Hochberg correction. * p < 0.05, ** p < 0.005, *** p < 0.0005, **** p < 0.00005

167

Table 3.10. (cont’d)
Name
AT3G16530
ANK
AT3G22235
AT3G61280
RLP23
RLK1
CNI1
FBS1
PMT6
AT5G52750
AT1G11960
AT1G61360
AT4G39830
CHX16
BCS1
CYP707A3
VSR7
AT3G28540
EXT3
MC2
AT1G51820
AT1G52200
AT1G30700
PTR3
DIC2

Description
Legume lectin; chitin-inducible
ankyrin
cysteine-rich TM module stress
tolerance protein
O-glucosyltransferase rumi-like
receptor like protein 23
receptor-like protein kinase 1,
mannose binding
carbon/nitrogen insensitive 1
F-box family protein
Major facilitator superfamily
protein
Heavy metal
transport/detoxification protein
ERD (early-responsive to
dehydration stress)
S-locus lectin protein kinase
Cupredoxin superfamily protein
cation/H exchanger 16
cytochrome BC1 synth; amplifies
SA signal
cytochrome P450; ABA
catabolism
Vacuolar sorting receptor 7
P-loop with nucleoside
triphosphate hydrolases
extensin 3
metacaspase 2
LRR repeat protein kinase
PLAC8 family protein
FAD-binding Berberine
peptide transporter 3; pathogen
response
dicarboxylate carrier 2

expression
ax
ho
12.7
35.0
1.6
5.8
4.9
14.4

p
**
*
*

Log2(axenic/holoxenic)
all
Ar
Ag
Pr
-1.4
-0.7
-1.0
-2.3
-1.4
-0.4
-1.8
-1.5
-1.4
0.1
-2.4
-1.4

2.8
1.9
2.3

8.8
6.4
7.4

*
*
**

-1.4
-1.4
-1.4

-0.7
0.3
-0.3

-2.1
-1.9
-2.0

-1.0
-2.0
-1.6

4.2
2.5
12.7

12.2
7.8
33.5

****
*
**

-1.3
-1.3
-1.3

-1.1
-2.3
-1.1

-1.3
-0.7
-1.6

-1.7
-0.5
-1.2

2.5

7.8

***

-1.3

-1.2

-1.5

-1.2

16.4

42.3

**

-1.3

-1.1

-2.0

-0.8

6.0
3.2
3.2
3.8

16.5
9.5
9.3
10.7

****
***
**
***

-1.3
-1.3
-1.3
-1.3

-1.3
-1.2
-0.5
-1.8

-1.2
-1.4
-1.7
-0.7

-1.5
-1.4
-1.4
-1.3

9.1

23.7

**

-1.3

-1.7

-1.5

-0.7

2.0
9.9

6.3
25.3

*
*

-1.3
-1.3

-0.2
-0.2

-1.8
-1.8

-1.5
-1.5

25.5
2.1
2.7
3.6
2.0
1.7

62.3
6.4
7.8
9.8
6.0
5.3

*
*
*
**
*
*

-1.3
-1.3
-1.3
-1.2
-1.2
-1.2

0.4
0.2
-0.3
-0.9
-0.2
-0.8

-2.2
-2.0
-1.7
-1.5
-1.5
-1.8

-1.8
-1.4
-1.6
-1.2
-1.8
-1.0

39.3

91.9

***

-1.2

-1.4

-1.0

-1.2

Criteria: log2(RPM/RPM)| ≥ 1.0, p < 0.05, RPM ≥ 5. Sorted by ascending average enrichment.
Statistical confidence was determined by pairwise t-test (all holoxenic vs all axenic), with
Benjamini-Hochberg correction. * p < 0.05, ** p < 0.005, *** p < 0.0005, **** p < 0.00005

168

Table 3.10. (cont’d)
Name
AT5G41750
AT4G38560
PR4
AT5G28630
ORP1A
AT3G07195
RLP52
AT1G18390
AT2G37130
PIRL2
AT5G25930
AT2G26440
WRKY47
RLP33
AT4G29780
ACA.l
NAC036
MSS1
IGMT3
SYP122

Description
Disease resistance protein
(TIR-NBS-LRR)
phospholipase-like protein PEARLI
4
pathogenesis-related 4
glycine-rich protein
OSBP (oxysterol binding protein)
RPM1-interacting protein 4 (RIN4)
family
disease resistance protein induced
by chitin
Serine/Threonine kinase family
catalytic domain
Peroxidase superfamily protein
plant intracellular ras group-related
LRR 2
LRR protein kinase
Plant invertase/pectin
methylesterase inhibitor
WRKY transcription factor
receptor like protein 33
nuclease
autoinhibited Ca2/ATPase II
NAC domain containing protein
high affinity, hexose-specific/H
symporter
O-methyltransferase
Syntaxin; regulator of JA/SA and
apotosis

expression
ax
ho
9.5
23.1

p

Log2(axenic/holoxenic)
all
Ar
Ag
Pr
**
-1.2 -0.9 -1.4 -1.2

2.6

7.2

*

-1.2

-0.2

-1.6

-1.5

12.8
10.1
2.5
1.9

30.3
24.2
6.8
5.5

*
**
**
**

-1.2
-1.2
-1.2
-1.2

-0.2
-1.2
-0.7
-0.6

-1.3
-1.5
-1.7
-1.4

-1.7
-0.8
-0.9
-1.3

2.4

6.6

***

-1.2

-0.8

-1.3

-1.4

6.5

15.6

****

-1.2

-0.9

-1.1

-1.5

9.6
5.2

22.5
12.8

**
**

-1.2
-1.2

-0.7
-0.6

-1.2
-1.0

-1.5
-1.9

10.9
11.8

25.2
26.9

***
*

-1.1
-1.1

-0.7
-0.2

-1.2
-1.6

-1.5
-1.2

34.8
3.6
16.5
57.7
5.2
5.3

77.0
8.9
36.7
125.0
12.3
12.5

***
**
**
***
**
**

-1.1
-1.1
-1.1
-1.1
-1.1
-1.1

-1.0
-0.5
-1.5
-1.0
-1.4
-1.4

-1.4
-1.3
-0.6
-1.1
-1.4
-0.9

-0.9
-1.6
-1.0
-1.2
-0.5
-1.0

7.1
18.3

16.3
40.2

**
***

-1.1
-1.1

-0.7
-1.6

-0.8
-1.2

-1.7
-0.7

Criteria: log2(RPM/RPM)| ≥ 1.0, p < 0.05, RPM ≥ 5. Sorted by ascending average enrichment.
Statistical confidence was determined by pairwise t-test (all holoxenic vs all axenic), with
Benjamini-Hochberg correction. * p < 0.05, ** p < 0.005, *** p < 0.0005, **** p < 0.00005

169

Table 3.10. (cont’d)
Name
AT1G35710
AR781
UGT72E1
AT5G08760
AT3G47090
GSTF2
AT4G35985
BIR1
AT1G36060
AT5G05600
ERF5
RGXT2
LHT1
AT4G38550

Description
Leucine-rich repeat protein
kinase
pheromone receptor-like protein
(DUF1645)
UDP-glucosyl transferase for
lignin metabolism
transmembrane protein
LRR protein kinase
glutathione S-transferase PHI 2
Senescence/dehydrationassociated
BAK1-interacting leucine-rich
repeat protein kinase
Integrase-type DNA-binding
2-oxoglutarate and Fe(II)dependent oxygenase
ethylene responsive element
binding factor 5
rhamnogalacturonan
xylosyltransferase 2
lysine histidine transporter 1
phospholipase-like protein
(PEARLI 4)

expression
ax
ho
38.6
82.6

p
*

Log2(axenic/holoxenic)
all
Ar
Ag
Pr
-1.1 -0.1 -1.6 -1.1

8.3

18.7

**

-1.1

-1.6

-0.9

-0.8

30.2

65.0

**

-1.1

-0.4

-1.4

-1.3

8.6
3.2
41.0
5.5

19.3
7.8
87.3
12.6

**
**
**
*

-1.1
-1.1
-1.1
-1.1

-0.9
-0.5
-0.5
-1.4

-1.6
-1.4
-1.1
-1.7

-0.7
-1.4
-1.4
-0.1

16.3

34.7

**

-1.1

-1.0

-1.5

-0.6

2.5
13.5

6.2
28.9

*
*

-1.0
-1.0

0.0
-1.8

-1.3
-1.3

-1.4
0.4

6.9

15.2

**

-1.0

-1.2

-0.6

-1.4

2.5

6.1

*

-1.0

-0.1

-1.6

-1.1

3.1
256.6

7.3
517.4

*
****

-1.0
-1.0

-0.3
-1.0

-1.7
-1.1

-0.9
-0.9

Criteria: log2(RPM/RPM)| ≥ 1.0, p < 0.05, RPM ≥ 5. Sorted by ascending average enrichment.
Statistical confidence was determined by pairwise t-test (all holoxenic vs all axenic), with
Benjamini-Hochberg correction. * p < 0.05, ** p < 0.005, *** p < 0.0005, **** p < 0.00005

170

Table 3.10. (cont’d)

Name
CYP71A13
AT5G05250
AT4G22495
UMAMIT33
AT4G22505
AT4G22475
AT4G22485
CYP81D11
AT4G32480

Description
cytochrome P450 family 71
polypeptide
hypothetical protein
hypothetical protein
nodulin MtN21 /EamA-like
transporter
Protease inhibitor/seed
storage/LTP
transmembrane protein
Protease inhibitor/seed
storage/LTP
Cytochrome P450;
cis-JA responsive
sugar phosphate exchanger,
putative (DUF506)

expression
ax
ho
5.5
1.9

p
**

Log2(axenic/holoxenic)
all
Ar
Ag
Pr
1.2
1.4
1.2
0.9

19.5
297.4
9.8

8.1
124.6
3.4

*
*
**

1.2
1.3
1.3

1.0
-0.2
1.4

1.5
2.2
1.5

0.9
3.1
1.0

372.2

149.7

*

1.3

-0.2

2.3

3.2

475.0
452.4

189.9
179.8

*
*

1.3
1.3

0.0
0.0

2.1
2.1

2.9
2.9

83.9

30.2

***

1.4

0.6

2.7

2.8

136.0

49.2

****

1.5

1.8

1.5

1.1

Criteria: log2(RPM/RPM)| ≥ 1.0, p < 0.05, RPM ≥ 5. Sorted by ascending average enrichment.
Statistical confidence was determined by pairwise t-test (all holoxenic vs all axenic), with
Benjamini-Hochberg correction. * p < 0.05, ** p < 0.005, *** p < 0.0005, **** p < 0.00005

171

Table 3.11. Metabolite quantification by GC/MS

phenyl-B-D
glucopyranoside
13.27
feruloylglycoside
unk.
N-metabolite
sucrose
12.74 320 192
467
ornithine
citrulline
putrescine
glycine
asparagine
glutamine
unk.
N-metabolite
ascorbic acid
myo-inositol
azelaic acid
lysine
allantoin
serine
glucose
6-phosphate
B-cyano-alanine
glucose
uric acid
unk.
sugar-phosphate
unk.
N-metabolite
unk.
N-metabolite

Arid
axenic
47.1±4.6

holo
1.6±0.4

Ag
axenic
50.2±8.6

holo
0.7±0.4

Prairie
axenic
57.3±3.6

holo
0.5±0.3

6.0±0.5

0.5±0.2

7.5±1.0

1.2±0.2

6.8±0.5

1.4±0.4

49.7±4.1

19.8±3.3

46.5±6.1

16.6±1.3

53.2±4.1

26.2±3.2

118.0±17.5
3.9±0.3

67.1±7.3
2.4±0.3

336.1±25.9
4.3±1.3

133.5±15.9
1.7±0.2

233.7±25.5
4.2±0.5

115.2±11.1
2.7±0.5

173.6±20.5
121.0±12.9
48.3±3.9
38.2±4.2
316.8±33.2
951.8±101.3
177.7±14.7

123.2±34.6
81.4±23.8
21.4±3.0
22.1±3.6
229.3±54.1
844.4±200.4
149.2±40.2

152.2±27.1
102.4±18.9
40.7±5.1
49.4±7.0
266.8±45.6
667.5±124.7
257.0±40.8

70.0±7.2
50.2±5.6
18.0±1.4
22.5±2.5
132.6±14.3
319.6±39.5
115.1±10.5

194.2±19.1
135.2±13.3
46.4±3.9
40.3±3.8
328.9±37.1
908.2±107.3
179.7±26.1

119.2±17.6
79.6±12.9
31.2±3.4
30.6±4.8
245.6±28.2
747.9±78.8
162.5±19.9

206.9±12.9
18.6±2.5
12.3±1.6
27.2±2.4
28.3±3.1
44.3±5.9
5.5±0.5

176.5±10.8
12.5±1.7
10.0±2.3
24.0±4.6
28.7±5.6
30.1±5.7
4.2±0.7

319.4±14.9
26.3±3.6
9.4±1.0
25.1±3.9
24.6±3.1
33.3±4.1
6.3±0.4

201.1±8.0
15.9±1.7
5.6±0.5
16.8±1.2
15.5±1.1
23.0±2.9
4.9±0.3

274.6±9.3
22.6±2.6
15.4±1.5
26.1±2.4
28.4±2.2
48.8±11.4
6.9±0.6

185.0±15.7
17.1±0.9
13.5±1.7
22.2±3.6
26.3±5.6
42.5±11.9
5.7±0.6

11.5±3.3
24.1±3.0
17.2±2.2
2.6±0.2

5.9±1.9
18.2±2.3
11.7±0.9
2.6±0.4

6.2±2.2
27.4±4.5
19.0±1.8
3.0±0.2

5.5±1.7
17.2±1.1
13.7±2.2
2.5±0.2

9.6±4.5
27.6±2.7
15.8±2.0
3.3±0.2

7.0±1.9
28.8±3.9
14.3±2.7
2.7±0.3

51.8±6.1

43.7±9.2

32.3±6.8

23.7±2.8

49.4±4.5

46.4±5.7

4.3±0.5

3.6±0.8

2.7±0.4

2.1±0.2

4.0±0.4

3.7±0.5

RT (min) key m/z; Values are µg·g-1 FW (sorbitol equivalent) ± standard error of the mean.

172

Table 3.11. (cont’d)

threonine
cysteine
tyrosine
tryptophan
threonic acid
B-alanine
phenylalanine
spermidine
10.86 306 288
N-metabolite
galactose
glyceric acid
glutamic acid
11.25 217 450
dehydrosugar
sinapic
acid-4-O-glucoside
12.64 320 479 464
unk. phenolic
12.52 507 103 189
285 249 261
12.98 320 307 217
isoleucine
leucine
valine
4-hydroxybenzoic
acid
phosphate
gamma-aminobutyric
acid
3,4-dihydroxybenzoic
acid

Arid
axenic
56.5±8.7
4.8±0.3
10.4±0.8
3.4±0.2
2.9±0.3
9.4±0.8
32.1±2.8
52.0±2.8
4.8±0.3

holo
39.4±4.6
3.7±0.4
10.2±1.5
2.9±0.6
2.5±0.3
7.5±1.1
28.1±4.1
37.0±4.0
4.0±0.4

Ag
axenic
39.8±4.8
4.6±0.4
10.0±1.4
3.4±0.5
3.3±0.2
7.7±1.1
32.0±4.0
41.2±4.3
5.9±0.1

holo
33.8±3.3
3.3±0.3
8.4±0.8
2.9±0.3
2.7±0.2
5.8±0.7
23.7±1.6
40.0±2.9
4.5±0.3

Prairie
axenic
50.6±7.7
4.3±0.3
10.9±1.2
3.6±0.3
3.1±0.3
8.2±0.9
29.8±3.0
46.1±3.7
5.5±0.3

holo
46.1±6.8
4.7±0.5
10.7±1.9
3.5±0.4
3.1±0.3
9.2±1.7
35.4±5.5
42.2±3.2
6.7±0.9

5.2±0.8
5.2±0.3
230.7±28.2
51.7±3.5

4.8±0.8
3.9±0.5
215.5±39.2
57.6±6.0

6.2±0.5
4.2±0.2
247.4±32.6
74.2±5.7

4.7±0.4
4.0±0.2
172.8±18.1
65.3±5.2

6.4±0.6
4.8±0.3
238.4±25.3
63.6±3.2

8.0±1.0
4.8±0.5
338.1±55.9
73.5±5.2

1.5±0.2

1.6±0.3

1.5±0.3

1.3±0.3

1.4±0.1

1.7±0.3

3.1±0.4
8.9±1.5
28.2±3.2

1.9±0.2
1.0±0.3
31.5±3.8

1.7±0.2
3.7±0.9
19.9±1.9

1.4±0.1
4.2±0.5
14.9±2.2

2.2±0.3
5.3±0.5
19.7±2.4

2.7±0.5
5.2±1.1
30.1±3.9

6.2±0.6
17.9±2.0
13.5±1.7
41.2±4.5
10.8±1.0

3.6±0.5
14.6±1.7
11.9±1.3
36.7±4.7
9.4±1.1

3.4±0.2
11.8±1.8
8.8±1.1
31.1±4.1
12.2±1.6

3.5±0.4
12.0±1.6
9.1±1.3
30.4±2.8
11.6±1.5

5.1±0.5
14.7±1.5
10.8±1.3
35.4±3.4
11.4±0.7

5.9±0.9
18.6±4.0
13.6±2.7
47.6±9.6
16.9±3.1

689.6±47.4
28.9±2.9

771.0±75.2
22.1±1.8

544.9±51.6
19.5±3.0

539.0±28.6
17.1±1.1

579.4±34.8
21.7±2.4

847.8±91.7
37.2±7.6

13.2±1.8

12.9±1.5

16.3±2.2

14.7±2.1

14.4±1.3

24.4±4.3

RT (min) key m/z; Values are µg·g-1 FW (sorbitol equivalent) ± standard error of the mean.

173

Table 3.11. (cont’d)

ethanolamine

Arid
axenic
235.4±21.1

holo
181.3±17.0

5 oxo proline

896.7±61.9

895.1±115.9

salicylic acid
methylphosphat
e
erythronic acid
phloroglucinol
12.69 395 249
410 519 504
caffeic acid
monopalmitin
12.03 219 235

20.6±2.6
36.4±3.0

19.6±2.3
29.6±3.7

Ag
axenic
118.9±12.
5
647.9±81.
3
29.2±4.0
25.2±3.3

28.2±3.2
36.9±1.7

Prairie
axenic
176.7±19.
4
687.6±62.
5
23.6±2.3
29.3±3.2

7.5±0.8
2.3±0.6
42.9±5.9

9.1±1.2
1.9±0.3
42.4±6.1

6.4±0.7
3.0±0.4
17.6±2.1

6.8±0.6
2.5±0.3
22.6±3.4

6.3±0.8
2.1±0.2
35.1±3.7

10.6±1.4
4.5±0.9
49.2±8.8

1.1±0.3
121.7±29.6
154.7±21.8

0.8±0.1
119.3±13.8
148.5±28.9

1.2±0.1
68.9±10.9
95.6±10.1

1.4±0.1
81.9±16.5
121.8±16.3

1.9±0.4
87.5±19.3
250.3±62.3

476.0±37.9

593.8±66.1

503.8±43.1

succinic acid
alanine

14.8±2.2
196.9±20.8

13.3±1.8
177.8±39.2

1,2,4
benzenetriol
glycerol
3
hydroxybenzoic
acid
13.35 409
9.99 98 288 390
N metabolite
sinapoyl malate
13.47 342 299
315 P
metabolite
stearic acid
malic acid
fumaric acid

4.9±0.6

5.2±0.6

431.0±51.
1
11.0±1.2
213.4±27.
4
4.5±0.6

4.8±0.6

1.2±0.1
57.1±11.6
174.0±19.
6
418.0±48.
9
13.2±1.5
184.5±20.
1
4.8±0.5

aspartic acid

60.3±7.6
2.8±0.3

67.5±7.6
3.4±0.4

57.1±6.1
3.7±0.5

56.2±7.1
3.9±0.6

32.0±3.1
3.1±0.3

67.1±11.9
6.2±1.2

2.3±0.4
814.5±148.
9
8.4±1.5
22.4±2.8

2.1±0.3
1046.5±179.
3
8.4±1.8
25.3±3.4

1.4±0.2
437.8±59.
0
6.5±1.1
18.6±3.3

1.5±0.2
542.5±122.
9
5.4±1.1
20.9±3.5

1.6±0.2
658.2±78.
4
6.8±1.2
20.7±2.0

3.2±0.7
1152.3±188.
9
18.2±3.3
40.9±7.8

82.1±11.0
71.1±5.8
639.0±30.6

94.5±12.0
68.9±9.0
541.2±96.5

76.9±12.3
70.4±5.2
618.8±36.8

264.2±43.6

280.0±28.7

75.4±7.5
60.9±4.5
512.3±64.
4
261.5±30.
0

150.7±28.8
105.6±12.3
919.8±119.4

8.24 273 258
231 216

66.9±10.6
52.9±6.8
471.1±76.
3
191.4±25.
7

holo
150.5±13.5
750.5±55.2

12.8±1.5
238.7±23.6

226.7±30.1

holo
227.4±25.0
967.2±115.1
41.1±7.4
35.5±5.5

668.9±82.9
21.6±3.5
318.3±60.7
9.3±1.8

526.3±123.5

RT (min) key m/z; Values are µg·g-1 FW (sorbitol equivalent) ± standard error of the mean.

174

Table 3.11. (cont’d)

maleic acid
palmitic acid
13.79 204 231
glycoside
dihydrouracil
sinapic acid
9.79 346 174
N metabolite
phytol
8.06 249 379
115
oleic acid
stigmasterol
linoleic acid
monostearin
B sitosterol
12
hydroxyoctade
canoic acid
glycerol 1/3
phosphate
14.39 320 217
361 119
citric acid
12
hydroxyoctade
canoic acid
methyl ester
2
ethylhexanoic
acid

Arid
axenic
holo
27.7±3.3
28.6±3.7
116.3±16.9
121.7±16.6
25.4±1.8
59.4±5.5

Ag
axenic
22.4±3.5
98.0±19.0
141.2±41.1

holo
29.2±5.1
96.1±15.6
62.7±8.1

Prairie
axenic
23.0±2.8
94.0±10.2
24.0±1.5

holo
42.6±7.7
231.6±48.6
135.2±42.8

13.2±1.5
654.5±39.5
31.2±5.1

12.3±1.3
805.9±79.3
28.8±7.5

10.1±1.3
511.1±61.4
13.1±1.9

12.3±1.7
748.5±93.0
18.0±1.9

10.4±1.3
542.3±55.3
22.7±2.0

21.5±4.3
983.0±141.5
44.4±14.0

424.8±55.4
613.6±106.2

506.0±52.0
519.1±144.0

307.5±19.1
308.8±89.8

490.2±80.6
375.6±94.4

478.7±54.0
344.6±65.7

836.5±210.5
869.5±262.5

6.1±1.0
10.5±3.9
13.6±2.1
332.4±80.2
33.7±6.2
9.6±3.2

6.1±1.0
6.3±0.9
17.7±2.6
313.7±48.4
35.3±4.5
8.9±2.6

3.3±0.4
2.9±0.2
9.0±1.4
100.5±14.1
13.5±1.7
5.2±1.1

4.5±0.6
4.3±1.0
12.1±3.4
176.6±46.9
21.6±4.1
5.6±2.2

5.4±0.5
5.9±0.8
17.3±2.3
128.4±28.7
28.6±3.5
5.5±0.9

13.1±2.8
13.9±3.8
44.9±11.3
283.0±73.0
69.2±16.7
19.9±4.9

9.4±1.2

10.9±1.6

6.4±0.7

7.6±1.1

8.7±0.8

29.0±6.0

2.0±0.4

3.8±0.4

5.3±1.5

3.9±0.7

1.5±0.2

8.7±2.2

56.0±2.3
56.8±23.7

139.8±6.9
56.0±20.3

59.4±3.3
15.9±4.2

142.4±8.7
19.9±7.9

52.6±4.3
20.3±3.9

133.8±10.7
112.4±49.2

16.3±2.9

82.7±8.7

25.3±6.1

78.7±6.7

13.1±1.2

188.5±35.8

RT (min) key m/z; Values are µg·g-1 FW (sorbitol equivalent) ± standard error of the mean.

175

Figure 3.27. Flow cytometry scatter plot of suspended Arabidopsis leaf disc samples
from bacterial counts protocol. (A) mock-inoculation control and (B) infected with Pst-GFP
5x106 CFUml-1. Events were enumerated from the gate (in orange) as a function of volume.
Forward scatted (FSC) is on the Y-axis and green fluorescence intensity (530/40 [488]) is on the
X-axis.

176

Chapter 4

Discussion and future directions for plant microbiota research

177

Plant microbiome research is the product of a multidisciplinary alliance among plant
biologists and, enabled by –omics technology, inspired by coordinated human
microbiome initiatives, and motivated by food demands of a growing global population.
Great strides have been made in the past decade towards a better understanding of
which microorganisms inhabit plant microbiomes. Well-designed plant microbiome
survey experiments have revealed factors likely to drive microbiome community
composition, and cutting- edge ‑omics technologies (genomics, targeted/shotgun
metagenomics, transcriptomics, metabolomics, proteomics, etc.) are powerful for tools
for hypothesis generation. Fundamentally, hypothesis-driven empirical research that
allows us to interpret these enormous datasets beyond correlations of patterns. The
annotation for each gene/transcript/peptide/metabolite and taxonomic classification of
each microorganism in a modern plant microbiome survey is based on homology or
shared attributes to a biologically characterized example. Perhaps impeded by
technological or logistical hurdles, all too often, modern microbiome studies will report
correlations and postulate functional repercussions, but fall short of hypothesis testing.
Hypothesis testing is essential to translate plant microbiome research into real-world
agricultural practice.
A recent summit of leading plant microbiome researchers in Asilomar, California,
gathered to discuss the status of the field and to establish research priorities for
harnessing plant microbiomes in sustainable agriculture (Busby et al., 2017): (1) define
core microbiomes and metagenomes in these model systems, (2) elucidate the rules of
synthetic, functionally programmable microbiome assembly, (3) determine functional

178

mechanisms of plant-microbiome interactions, (4) characterize and refine plant
genotype-by-environment-by-microbiome-by-management interactions. In this
section, I offer my perspective on and propose experiments and/or endeavors towards
achieving some of these objectives, with emphasis on potential applications for the
FlowPot system in this pursuit. As is true for most technologies, the FlowPot system
design can be refined and improved. Where appropriate, I offer specific design and
protocol improvement suggestions for future models. I offer my perspective towards
achieving a better mechanistic understanding of axenic immune deficiency.

179

Development of model host-microbiome systems with associated microbial
culture collections and reference genomes
Development of culture collections for plant microbiomes across different model
species is no doubt a research community task. Currently, large culture collections
exist for Bacteria, fungi, oomycetes, and protists exist for Arabidopsis thaliana in
various research laboratories, some of which have been fully sequenced, including:
Paul Schulze Lefert (Max Planck, Cologne, Germany), Erick Kemen (Max Planck,
Tubingenden, Germany), Detlef Weigel (Max Planck, Tubingenden, Germany), Jeff
Dangl (University of North Carolina, Chapel Hill, USA), Julia Vorholt (ETH Zurich,
Switzerland) and Joy Bergelson (University of Chicago, USA). I also established a 400member bacterial culture collection of Arabidopsis phyllosphere-associated bacteria,
along with full-length 16S rRNA gene sequences, currently in the laboratory of Sheng
Yang He. Ideally, a central repository should be established with each isolate, along
with comprehensive associated metadata, following mutually agreed-upon standards
and guidelines (Bowers et al., 2017; Chain et al., 2009). It is the opinion of this author
that full genome sequencing information will be required for organism identification in
the near future, and thus the genomes should be sequenced and annotated for all
isolates included in the collection. Along the same framework as the Earth Microbiome
Project, the collection should include a diversity of sampling locations to capture
increased diversity (Thompson et al., 2017). Associated metadata should be included
in a publically accessible database (Arabidopsis accession, isolation location, tissue
type, environmental details, growth media, culture conditions, etc.). Such an effort
would require substantial funding, organization, and a willingness among participating

180

laboratories to contribute fairly. Perhaps it would be the highest probability of success
would be to coordinate such an effort with separate funding agencies in the EU, United
States, Asia, and Australia/New Zealand. An essential element of this endeavor is strict
adherence to laboratory and database standards, and even if the strains are not
located in a single internationally-accessible location, the database containing
genomes and associated metadata should be globally accessible.
Once established for Arabidopsis, this will provide a framework to build additional
databases for alternative model systems, expanding upon already initiated efforts for
barley (Bulgarelli et al., 2015), wheat (Donn et al., 2015), corn (Aira et al., 2010), rice
(Edwards et al., 2015), soybean (Mendes et al., 2014), medicago (Stanton-Geddes et
al., 2013), Populus (Hacquard and Schadt, 2015), and tomato (Tian et al., 2015).
Current culture collections exist for many of the additional proposed models. An
international summit among laboratory directors and governors of each respective
culture collection would greatly facilitate coordination and centralize what resources
are currently available. Participation by industry would also be greatly valuable, as
biotechnology companies may have large culture collections.

181

Define core microbiomes and metagenomes
The concept of an absolute core microbiome for a plant species is controversial, and
there are several schools of thought that fundamentally come down to the species
concept of microbiota and where thresholds are set for gene sequence clustering. In
communities of bacteria, there are reports of very distantly-related microorganisms
having functionally redundancy. Thus, a global absolute core community of microbiota
inhabiting the Arabidopsis microbiome may not exist at conventional OTU identity
cutoffs. Indeed, at low-resolution classification levels, such as phylum-level
classification, core microbiomes exist. However, among a single genus such as
Pseudomonas, there are reports of approximately 70% of gene conservation, despite
identical 16S gene sequences (Silby et al., 2009). One of the earliest (now disproven)
concepts in microbial ecology is “everything is everywhere, and the environment
selects” (de Wit and Bouvier, 2006). Certain organisms have obligate symbioses and
vertically-transmitted microbiota. This is a common phenomenon for insects (Baumann
et al., 1997; Hansen et al., 2012; Moran, 2001). Seed‑borne vertical transmission of
microbiota may occur in some plants, as has been reported in maize (Johnston-Monje
and Raizada, 2011). However, as demonstrated in this dissertation, axenic Arabidopsis
can acquire microbiota from extremely different input communities with minimal
overlap of OTUs and grow vigorously. It is the opinion of this author that core microbial
species likely exist within a confined geographic area, but biogeography of
microorganisms and environmental variability suggest there is not a core microbiome,
as defined by 16S rRNA gene sequences. In other words, core is an appropriate term
to use in a relative context, but not an absolute context. It is conceivable that global

182

core microbiome functionality exists, although functions can be convergently evolved
and non-orthologous genes can code for very different proteins with similar
functionality. Thus, in a shotgun sequencing experiment among many plant
microbiome samples, fewer core gene sequences might exist than core functional
elements. Based on gene presence/absence, some genes that are common to all
bacteria will presumably be “core” metagenomic components, such as gyrB, rpoE,
rpoN, tufAB, (Parks et al., 2015). It is also possible that core sequence motifs will be
identified in such an experiment for which no reliable annotation exists. Thus,
extrapolation of biologically interpretable data to formulate testable hypotheses may be
limited.
Core taxonomic overlap was identified for input, substrate, and plant-associated
communities in FlowPot experiments for both Chapter 2 and Chapter 3, as were core
differentially expressed genes. The plant-enriched taxa (relative to substrate
abundance) in both the calcined clay system and the FlowPot system may have
attributes to establish and maintain a more robust association with the host plant than
other input microbiota. This does not necessarily indicate that microbiota with a more
robust host association are beneficial to the host, pathogens and parasites have
evolved adaptations for close host associations as well. However, it is of interest to
identify microbial traits that bestow the ability to form robust host associations in the
presence of a diverse microbial community. Such traits could include one or more of
the following abilities: to suppress/evade/tolerate host immune responses, to
manipulate exudation of nutrient sources to maintain a metabolic relationship with the
host, to outcompete other microorganisms for the nutrient-rich host-associated

183

environment, to form close adhesion structures to maintain a close physical
association with the host. From the taxa identified as plant-enriched, it would be
interesting to pursue a forward or reverse genetic approach to identify microbial genes
of importance to maintain the close association. One candidate for this is OTU1
Ralstonia from Chapter 2, which was root-enriched from the natural German soil, as
well as the synthetic community. Another commonly-enriched genus with very limited
research published on it is Massilia. Knowledge of successful plant colonization
strategies may help us to predict the success of candidate biostimulants,
biopesticides, and biofertilzers in the field based on their genomic features.

184

Rules of synthetic, functionally programmable microbiome assembly
A fundamental question in plant microbiome research, and an essential prerequisite to
translate microbiome research into effective real-world agricultural practice is “how
does the microbiome influence host phenotypes of interest?” There are far too many
combinations of microbiomes, hosts, and phenotypes to assess by just one research
group, and a research community is necessary to address this question. Experimental
reproducibility of microbiome experiments is often confounded by environmental
complexity and stochasticity, making hypothesis testing and reproducibility major
challenges in plant microbiome research. Recognizing a lack of adequate growth
systems for microbiome recolonization experiments that simulate agriculturally relevant
conditions, I focused on the development of the FlowPot system-- an inexpensive
growth platform constructed with common materials and standard greenhouse peat
substrate, for highly-controlled plant growth. This FlowPot system was designed to
maximize reproducibility by normalization of environmental variables. Microbial
communities can be inoculated directly into the soil, allowing for true isolation of the
“microbiome” variable. Given that there is no “standard” soil microbiome, I developed
the FlowPot system to accommodate axenic (germfree) plant growth, so that the
system can be inoculated with microbiota of choice, including but not limited to
defined synthetic microbiota, natural soil microbiota, or mock-inoculated with no
microbiota. Additional variables can be manipulated, for example: nutrient
supplementation, chemical addition, climate simulation, and pathogen introduction.
The overall objective and purpose of the FlowPot system is to provide a standardized
growth platform for the plant microbiome research community. Data harmonization is

185

key to drawing accurate comparisons of experimental microbiome results, and to truly
leverage the power of a research community, model systems and standards are
essential.
Of course, the FlowPot system is not only possible growth platform for plant
microbiome research, nor is it the first. One such alternative platform used in several
previous high-profile studies, the calcined clay growth system, was compared with the
FlowPot system in Chapter 2. The calcined clay system was previously used to grow
Arabidopsis in the presence of defined bacterial microbiota to address whether
isolation environment (leaf/soil/root) defined where bacterial would preferentially
colonize and whether modulation of salicylic acid signaling in the host would
predictably influence microbiome composition, respectively(Bai et al., 2015; Lebeis et
al., 2015). Using the same synthetic bacterial community from(Bai et al., 2015), we
performed recolonization experiments in tandem with the calcined clay system, and
report that in fact the FlowPot system accommodates more bacterial diversity more
closely recapitulating the microbiome of a greenhouse or agricultural setting.
Furthermore, we determined that the enormous microbiota community profile
differences between the bulk substrate and the root communities in the calcined clay
system we more subtle in FlowPots, perhaps more accurately recapitulating or
simulating what would occur in an agricultural setting. It is ideal to have multiple
growth platforms for microbiome research, especially since our experiments
corroborate the many previous reports that environmental factors are major drivers of
community differentiation (Zhou and Ning, 2017).

186

The breadth of colonization experiments conducted in Chapter 2 could be expanded in
a number of interesting ways, depending on the hypothesis to be tested. The objective
of the study described in Chapter 2 was primarily to introduce the FlowPot system and
validate the health of axenic Arabidopsis. A secondary object was to determine
whether bacteria differentially colonize FlowPots and calcined clay, and in what
manner. With those objectives accomplished, follow-up on this study could go in
various directions

187

Proposed improvements to FlowPot design and assembly
Current conditions have been optimized for Arabidopsis growth. Arabidopsis is
attractive because before bolting, it grows relatively flat. Plants with greater vertical
requirements would need an alternative growth vessel. The FlowPots themselves are
modular, and will easily fit into containers other than the Eco2 micro boxes. The
fiberglass mesh may not accommodate plants with thicker stems than Arabidopsis, so
these may need to be adapted with either wider-gage mesh, or removable mesh.
Plants with a more elaborate and larger root system than Arabidopsis may require
additional growth substrate contained by a larger cylinder than the 50 mL syringe
barrel. Companies such as Wilburn Medical USA make a 100 mL syringe that could
potentially accommodate for this. As soil volume increases, one must take into
consideration the effect this will have on humidity within the chamber and water
retention in the substrate. Ideally, empirical experimentation should be conducted on a
single lot of sphagnum peat with titrated vermiculite and perlite to determine the water
retention. A major bottleneck is the manual labor associated with assembly, along with
the risk for contamination. The assembly facility here at Michigan State would
immensely benefit from a collaboration with an industrial engineering group and
experts from the packaging school.
Despite extensive efforts to optimize batch-to-batch consistency of FlowPots, for
unknown reasons, batches occasionally fail, and unpredictably entire boxes of plants
will be unhealthy or not survive. In my experience, sometimes this is input community
dependent, and other times there is no correlation to microbiome colonization status.
The Michigan State University greenhouse staff performs chemical testing on each lot

188

of Sunshine® Redi-Earth Plug & Seedling mix they receive, and indeed there are
variations in nutrient levels. Many factors, however, are not routinely tested. For
example, Redi-Earth is blended with an unspecified amount of dolomite limestone,
vermiculite, proprietary wetting agent, gypsum, and a proprietary silicon additive. This
substrate was used for FlowPots in all experiments conducted at Michigan State
because this is the standard greenhouse substrate, but ideally, the substrate should
not include proprietary additives. Furthermore, Redi-Earth is pasteurized or steamed by
an unspecified protocol. Since there is lot-to-lot variation of nutrient levels, presumably
there is lot-to-lot variation of the proprietary additives. It is unknown how the additives
may affect microbiota. Greater alpha diversity was present in the Michigan (ag) soil
inoculated FlowPots from chapter 2 than the chapter 3 FlowPots inoculated with
identical input microbiota. The FlowPots used in chapter 2 were assembled with
German peat substrate, while the chapter 3 FlowPots were assembled with Redi-Earth.
The extent to which this influences microbial community composition is unknown
because the experiment was not conducted simultaneously and time points were not
identical (six weeks in Chapter 2, three weeks in Chapter 3).
For future FlowPot optimization, I propose that rather than using the Redi-Earth
product, a substrate consisting of pure sphagnum peat blended with medium
vermiculite be used. Additional nutrient supplementation and/or wetting agent addition
may be necessary to achieve optimal Arabidopsis growth, but this needs to be
empirically validated. It is likely that peat from different locations around the world may
differ in chemical and physical attributes, so it is important that the peat source is

189

reported for each experiment. It would be ideal to conduct an experiment testing
whether peat from different sources influences the microbial composition.
The Eco2 micro boxes in this set of experiments contain a gas-permeable filter that
allows for passive diffusion of air, thus allowing for transpiration, and to some extent,
moisture loss. After several autoclave cycles, some filters become discolored and
appear partially melted, while others do not. Gas exchange rate and accumulation of
volatile organic compounds can affect many aspects of plant physiology, and
presumably, the partially melted filters are compromised in full gas exchange capacity.
To improve consistency, I suggest that the Eco2 boxes be replaced with similar
polypropylene containers, but with an opening port on the side of the box, to which
disposable adhesive air filter can be adhered. The side is a better suited place for the
filter because if one desires to adapt the system to an overhead optical phenotyping
platform, a lid filter may obscure the light and image path. Packing the substrate into
the FlowPots by hand is prone to inconsistency. Similar to tamping an espresso,
variations in packing intensity greatly influence compaction, and thus may influence
consistency.

190

Figure 4.1. Schematic of FlowPot modification to allow it to fit in a DEPI chamber without
light reflection interference. Further improvements are illustrated, including filter placement and
septum addition for VOC studies and headspace measurements.

191

The FlowPot system is conducive to high-throughput experimentation given its small
form factor, and thus is an attractive platform to perform larger screening experiments.
In Chapter 3, we only examined three input communities to determine core microbiome
function and core colonization of the plant and substrate. To gain greater statistical
power to draw more robust correlations, one could assemble a panel of diverse soil
types from a spectrum or gradient of an environmental parameter of choice, and
determine to what extend the perturbation of a “transplant” event influences
community structure via targeted amplicon sequencing. Presumably, input soil
microbiota from peat-like environments would be more amenable to high diversity
transfer after colonization in the peat substrate, but this remains to be tested. The
FlowPot system, in general, was used as a proof of concept with the Arabidopsis
model, and now that this platform has been validated, the research community is
encouraged to conduct diverse experiments, cross-check the results from other labs,
and establish additional tools to further enhance FlowPots, as well as the field.

192

Strategies for characterization of axenic immunodeficiency
Based on the experiments conducted in Chapters 3, as well as the dual-RNAseq
experiment in axenic and holoxenic Arabidopsis infected with Pseudomonas syringae,
we can conclude that axenic Arabidopsis is compromised in the ability to defend
against Pst. The relative magnitude of susceptibility to Pst is greater in the Pst hrcCmutant, which is unable to suppress innate immunity. It would be interesting to
determine whether the epigenome of axenic plants differs from that of conventionally
colonized. There are numerous reports of epigenetic variations in chromatin structure
and methylation patterns in DNA that could perhaps be playing a role in host
phenotypes. Defense priming and systemic acquired resistance are reported to be, in
part, dependent on epigenetic effects on the genome (Conrath et al., 2015; Dowen et
al., 2012). It would be interesting to determine if axenic plants have hot spots of relative
epigenetic differences as opposed to conventionally colonized plants. An epigeneticlevel dissection of microbiota influence over host plants has not been examined yet,
and such a study would be a technology-enabled dive into the unknown, perhaps
revealing previously undetected layers of plant-microbiome relationships. One could
approach this experiment by manipulating the “microbiome variable”, using ~3
reference microbial communities, and perform a time course bs-seq and methyl-seq
experiment (Cokus et al., 2008; Hing et al., 2015).
Certain microorganisms will likely have competitive advantages over others, whether
they be by antibiosis, niche exclusion, toxin/antitoxin systems, type-VI secretion, direct
competition for resources, or other antagonistic characteristics. Rather than starting
with a 240 member synthetic community, as is used in Chapter 2, a more practical

193

approach would be to use a 30-member synthetic community, such as the SynCom
used in (Lebeis et al., 2015), but with a high degree of replication to account for
stochastic colonization events. Both synCom and At-SPHERE collections are based
strictly on taxonomic identification using the 16S rRNA subunit sequences at standard
97% identity cut-offs. One must consider whether these taxa are functionally
representative of the Arabidopsis microbiome. Furthermore, the synthetic communities
lack fungi, which is appropriate in some experiments, but for prediction of invasion and
persistence with robust confidence for success even in a greenhouse, one might
consider the addition of fungi or perform the recolonization in an endogenous
microbiota background.
An important goal for the plant microbiome research community is to identify
characteristics that contribute to or enable a microorganism to invade and persist in a
pre-established microbiome. Towards this objective, one could take a microbial
genome-wide associated survey approach (Horton et al., 2014) based on many
replicates of closely related taxa. The close relatedness of a taxonomic group in
question could better equip the scientist to bioinformatically subtract shared among
successful and non-successful colonizers, moving towards the objective of identifying
functionally relevant genes. It would be prudent to perform these experiments in
multiple growth platforms and with multiple background microbiota, because as
demonstrated in Chapter 2, substrate and environmental variables influence
colonization by certain taxa, and we hypothesize that certain taxa are precluded based
on inadequate physical or chemical attributes. The genetic properties of a successful
competitive microorganism are valuable information, regardless of whether or not the

194

microorganism positively influences plant health. Perhaps one has at their disposal a
large database of annotated genomes for many microorganisms; multiple selection
criteria could be implemented based on HMMs of desired gene sequences. The
FlowPot system is conducive to high throughput experimentation with multiple
replicates, and would be well suited for a GWAS study to identify functionally relevant
genes.
The form factor of the FlowPot system also offers the unique retrofit boxes with septa
for headspace measurements, as well as visual growth and photosynthesis
measurements. For example, if one were to measure PS-II efficiency fluorometrically as
well as growth, the FlowPot system can be used and placed directly in an imaging
chamber. The modularity of FlowPots and their containment vessels should maintain
sterility or prevent post-inoculation microbiome immigration. The highly controlled and
normalized environment of FlowPots and their boxes could potentially enable one to
catch more subtle phenotypes.
A major focus of this dissertation was on the characterization of innate immunity
deficits and compromised ability to defend against pathogen invasion in axenic
conditions, versus colonized by three distinct microbiomes. Although we were able to
confirm that all three holoxenic Arabidopsis groups were significantly more
immunocompetent than axenic groups, and likewise, inept at defense against Pst, it is
unknown whether single or multiple members of the microbiome confer this
phenotype. For a follow-up experiment, to identify organisms or groups of organisms
that confer a desired phenotype, the three-sample panel could be expanded across a
gradient or group of similar soils, and introduce only one stress variable. Such an

195

experiment may provide enough granularity to discern differential effects on pathogen
fitness in the host and susceptibility to infection. Through gene expression profiling and
metabolic profiling, we were able to identify subtle differences between the colonized
plant based upon input community. However since only three input communities were
used, the size of this data set is insufficient to draw significant correlations between
small taxonomic groups and phenotypic attributes.
In this study, we defined a plant’s susceptibility by enumerating precise pathogen
population density over time. Pathogen replication in situ was greatly expedited by
using the novel flow cytometry enumeration protocol and a th7-GFP-tagged strain.
This is an invasive experiment because the leaves we macerated for community
extraction. For future experimentation, Pst with constitutive lux expression, along with
a retrofitted IVIS imager or appropriate CCD camera, one could non-invasively monitor
pathogen proliferation by correlating luminescent intensity.
To examine the spectrum of axenic susceptibility, it would be informative to perform
disease assays using more pathogens/pests than Pst. In particular, it would be
informative to perform an insect herbivory assay and disease assays using one or more
necrotrophic pathogens. If indeed, axenic are more susceptible to both biotrophs and
necrotrophs, this suggests a distinct SA/JA-independent arm of defense signaling is
microbiome-influenced. The increased relative abundance of hexanoic acid is
interesting in the holoxenic plants. Many of the attributes of axenic susceptibility, or
holoxenic resistance, are parallel to literature published about the priming capacity of
hexanoic acid (Scalschi et al., 2013). Scalschi et al. demonstrated that hexanoic acid
could be used as a biostimulant to prime plants for resistance to necrotrophs and

196

biotrophs, but not at the expense of growth (Campos et al., 2016). One could further
investigate the basis for axenic susceptibility by performing disease assays in JAinsensitive and SA-insensitive backgrounds. Rather than using the wildtype Pst
DC3000 strain, it would be a cleaner experiment to use a coronatine-deficient hrcCstrain. This way the pathogen cannot manipulate hormone signaling directly other than
through PTI. I attempted 3-4 disease assay trials in coi1-16, npr1-1 and, sid2 mutants,
but the results were inconclusive. As expected, coi1-16 was resistant and npr1-1 and,
sid2 were more susceptible to Pst, but the relative magnitude of susceptibility was
unhindered (Katsir et al., 2008; Tsuda, 2008; Yu et al., 2001).
One puzzling result that remains unresolved comes from protection assays. BTH
protection assays worked very reliable and consistently, always conferring at least 10fold protection against Pst, but flg22 protection assays would not work with the same
batch of plants, and a wide range of titrated flg22 concentrations. Finally, after two
years of attempts at the hands of 4 different scientists, Dr. Kinya Nomura was able to
achieve consistent results in four week old plants, but only with a reduced surfactant
concentration (personal communication). This seems to suggest that the original higher
concentration of surfactant was masking the effects of PTI priming in the growth
system. A previous report ties PTI to apoplastic water availability, and perhaps SAmediated defense activation (BTH is an SA analog) does not change moisture
availability in the apoplast, but PTI does (Xin et al., 2016).
Prior to the development of the FlowPot system, an earlier version of a soil-based
axenic growth platform was used to grow and collect 2wo Arabidopsis rosettes for
RNA extraction (data not included). We performed mRNA-seq, as well as smRNA-seq,

197

and identified miR398 as highly abundant in axenic Arabidopsis. This miRNA targets
copper superoxide dismutases 2/3, as well as their chaperone. The gene expression
from the FlowPot mRNA-seq suggests that indeed, miR398 may be enriched in axenic
plants, but this warrants follow up experiments with 398 mutant T-DNA lines (knockout
and overexpressor), and qPCR validation of miRNA abundance tracking. miR398 has
previously been implicated in innate immunity and defense (Li et al., 2011; Zhu et al.,
2011).
A coordinated effort amongst plant microbiome community, as well as adequate
financial support from funding agencies and adherence to standards and standard
operating protocols will lead to major strides and steps forward and microbiome
research. The global population is growing at an alarming rate and our food resources
and current agricultural productivity cannot keep pace, especially in areas of the world
that need massive technological improvements the most. It is the belief of this author
that innovations in microbiome technologies and intelligent application of
biostimulants, biopesticides and biofertilizers can offer sustainable efficacious
alternatives or supplements to current agricultural practices. With the advent of
accessible and inexpensive next-gen sequencing technology platforms at our fingertips
and a growing interest in plant microbiome research, perhaps motivated by human
microbiome initiatives, we are embarking upon a renaissance of plant microbiome
discoveries and innovations.

198

Figure 4.2. Revised disease triangle model, incorporating more vertices that integrate
‘management’ and ‘microbiome’. Plant disease triangle (adapted from Stevens 1960), depicting
the factors necessary for a crop disease to occur and develop to an epidemic. A host plant
must be developmentally and genetically susceptible to the pathogen. The environment must
be conducive to pathogen proliferation, dispersal, and must confer disease-conducive host
physiology to accommodate pathogen proliferation. The pathogen must have appropriate
virulence factors to replicate to high levels within the host plant. An extension of this model
highlights the opportunity for inappropriate management regimes to influence the risk for
epidemic, such as import/export of diseased plant material, high-density monoculture cropping
systems, and use of one or very few chemicals/traits for pest resistance, thus facilitating the
evolution of resistance.

199

REFERENCES

200

REFERENCES
Adair KL, Douglas AE. (2017). Making a microbiome: the many determinants of hostassociated microbial community composition. Curr Opin Microbiol 35: 23–29.
Adams C, Jacobson A, Bugbee B. (2014). Ceramic Aggregate Sorption and Desorption
Chemistry: Implications for Use as a Component of Soilless Media. J Plant Nutr 37:
1345–1357.
Agrios, G.N. (2005). Plant Pathology. 5th eds. Department of Plant Pathology.
University of Florida.
Aira, M., Gómez-Brandón, M., Lazcano, C., Bååth, E., and Domínguez, J. (2010). Plant
genotype strongly modifies the structure and growth of maize rhizosphere microbial
communities. Soil Biol. Biochem. 42, 2276–2281.
Alvarez, S., Rohrig, E., Solís, D., and Thomas, M.H. (2016). Citrus Greening Disease
(Huanglongbing) in Florida: Economic Impact, Management and the Potential for
Biological Control. Agric. Res. 5, 109–118.
Andrew DR, Fitak RR, Munguia-Vega A, Racolta A, Martinson VG, Dontsova K. (2012).
Abiotic factors shape microbial diversity in Sonoran Desert soils. Appl Environ
Microbiol 78: 7527–7537.
Andrews, S. (2010). FastQC: a quality control tool for high throughput sequence data.
Arabidopsis Genome Initiative (2000). Analysis of the genome sequence of the
flowering plant Arabidopsis thaliana. Nature 408, 796–815.
Arnon DI. (1949). Copper enzymes in isolated chloroplasts. Polyphenoloxidase in beta
vulgaris. Plant Physiol 24: 1–15.
Badri, D.V., Zolla, G., Bakker, M.G., Manter, D.K., and Vivanco, J.M. (2013). Potential
impact of soil microbiomes on the leaf metabolome and on herbivore feeding behavior.
New Phytol. 198, 264–273.
Bai Y, Müller DB, Srinivas G, Garrido-Oter R, Potthoff E, Rott M, et al. (2015).
Functional overlap of the Arabidopsis leaf and root microbiota. Nature 528: 364–369.
Bai, Y., Müller, D.B., Srinivas, G., Garrido-Oter, R., Potthoff, E., Rott, M., Dombrowski,
N., Münch, P.C., Spaepen, S., Remus-Emsermann, M., et al. (2015). Functional overlap
of the Arabidopsis leaf and root microbiota. Nature 528, 364–369.

201

Bank TL, Kukkadapu RK, Madden AS, Ginder-Vogel MA, Baldwin ME, Jardine PM.
(2008). Effects of gamma-sterilization on the physico-chemical properties of natural
sediments. Chem Geol 251: 1–7.
Baumann, P., Moran, N.A., and Baumann, L. (1997). The Evolution and Genetics of
Aphid Endosymbionts. Bioscience 47, 12–20.
Berendsen, R.L., Pieterse, C.M.J., and Bakker, P.A.H.M. (2012). The rhizosphere
microbiome and plant health. Trends Plant Sci. 17, 478–486.
Berns AE, Philipp H, Narres H-D, Burauel P, Vereecken H, Tappe W. (2008). Effect of
gamma-sterilization and autoclaving on soil organic matter structure as studied by
solid state NMR, UV and fluorescence spectroscopy. Eur J Soil Sci 59: 540–550.
Birkenbihl, R.P., Liu, S., and Somssich, I.E. (2017). Transcriptional events defining plant
immune responses. Curr. Opin. Plant Biol. 38, 1–9.
Bodenhausen N, Bortfeld-Miller M, Ackermann M, Vorholt JA. (2014). A synthetic
community approach reveals plant genotypes affecting the phyllosphere microbiota.
PLoS Genet 10: e1004283.
Bodenhausen, N., Bortfeld-Miller, M., Ackermann, M., and Vorholt, J.A. (2014). A
synthetic community approach reveals plant genotypes affecting the phyllosphere
microbiota. PLoS Genet. 10, e1004283.
Boudsocq, M., Willmann, M.R., McCormack, M., Lee, H., Shan, L., He, P., Bush, J.,
Cheng, S.-H., and Sheen, J. (2010). Differential innate immune signaling via Ca(2+)
sensor protein kinases. Nature 464, 418–422.
Bowers, R.M., Kyrpides, N.C., Stepanauskas, R., Harmon-Smith, M., Doud, D., Reddy,
T.B.K., Schulz, F., Jarett, J., Rivers, A.R., Eloe-Fadrosh, E.A., et al. (2017). Minimum
information about a single amplified genome (MISAG) and a metagenome-assembled
genome (MIMAG) of bacteria and archaea. Nat. Biotechnol. 35, 725–731.
Buée, M., Reich, M., Murat, C., Morin, E., Nilsson, R.H., Uroz, S., and Martin, F. (2009).
454 Pyrosequencing analyses of forest soils reveal an unexpectedly high fungal
diversity. New Phytol. 184, 449–456.
Bulgarelli D, Rott M, Schlaeppi K, Ver Loren van Themaat E, Ahmadinejad N, Assenza
F, et al. (2012). Revealing structure and assembly cues for Arabidopsis root-inhabiting
bacterial microbiota. Nature 488: 91–95.
Bulgarelli, D., Garrido-Oter, R., Münch, P.C., Weiman, A., Dröge, J., Pan, Y., McHardy,
A.C., and Schulze-Lefert, P. (2015). Structure and function of the bacterial root
microbiota in wild and domesticated barley. Cell Host Microbe 17, 392–403.

202

Bulgarelli, D., Rott, M., Schlaeppi, K., Ver Loren van Themaat, E., Ahmadinejad, N.,
Assenza, F., Rauf, P., Huettel, B., Reinhardt, R., Schmelzer, E., et al. (2012). Revealing
structure and assembly cues for Arabidopsis root-inhabiting bacterial microbiota.
Nature 488, 91–95.
Busby PE, Soman C, Wagner MR, Friesen ML, Kremer J, Bennett A, et al. (2017).
Research priorities for harnessing plant microbiomes in sustainable agriculture. PLoS
Biol 15: e2001793.
Bushnell, B. (2016). BBMap short read aligner. http://sourceforge.net/projects/bbmap
Campos, M.L., Yoshida, Y., Major, I.T., de Oliveira Ferreira, D., Weraduwage, S.M.,
Froehlich, J.E., Johnson, B.F., Kramer, D.M., Jander, G., Sharkey, T.D., et al. (2016).
Rewiring of jasmonate and phytochrome B signalling uncouples plant growth-defense
tradeoffs. Nat. Commun. 7, 12570.
Caporaso JG, Lauber CL, Walters WA, Berg-Lyons D, Huntley J, Fierer N, et al. (2012).
Ultra-high-throughput microbial community analysis on the Illumina HiSeq and MiSeq
platforms. ISME J 6: 1621–1624.
Carvalhais, L.C., Muzzi, F., Tan, C.-H., Hsien-Choo, J., and Schenk, P.M. (2013). Plant
growth in Arabidopsis is assisted by compost soil-derived microbial communities.
Front. Plant Sci. 4, 235.
Celaj S, Gleeson MW, Deng J, O’Toole GA, Hampton TH, Toft MF, et al. (2014). The
microbiota regulates susceptibility to Fas-mediated acute hepatic injury. Lab Invest 94:
938–949.
Chain, P.S.G., Grafham, D.V., Fulton, R.S., Fitzgerald, M.G., Hostetler, J., Muzny, D.,
Ali, J., Birren, B., Bruce, D.C., Buhay, C., et al. (2009). Genomics. Genome project
standards in a new era of sequencing. Science 326, 236–237.
Chang, C., Bowman, J.L., and Meyerowitz, E.M. (2016). Field Guide to Plant Model
Systems. Cell 167, 325–339.
Chisholm, S.T., Coaker, G., Day, B., and Staskawicz, B.J. (2006). Host-microbe
interactions: shaping the evolution of the plant immune response. Cell 124, 803–814.
Chung, K., and Ferris, D.H. (1996). Martinus Willem Beijerinck (1851–1931). J. Am.
Med. Assoc.
Ciancio A, Pieterse CMJ, Mercado-Blanco J. (2016). Editorial: Harnessing Useful
Rhizosphere Microorganisms for Pathogen and Pest Biocontrol. Front Microbiol 7:
1620.

203

Clay, N.K., Adio, A.M., Denoux, C., Jander, G., and Ausubel, F.M. (2009). Glucosinolate
metabolites required for an Arabidopsis innate immune response. Science 323, 95–
101.
Clemente JC, Ursell LK, Parfrey LW, Knight R. (2012). The impact of the gut microbiota
on human health: an integrative view. Cell 148: 1258–1270.
Clough SJ, Bent AF. (1998). Transformation of Arabidopsis thaliana. Plant J 16: 735–
743.
Cokus, S.J., Feng, S., Zhang, X., Chen, Z., Merriman, B., Haudenschild, C.D., Pradhan,
S., Nelson, S.F., Pellegrini, M., and Jacobsen, S.E. (2008). Shotgun bisulphite
sequencing of the Arabidopsis genome reveals DNA methylation patterning. Nature
452, 215–219.
Cole JR, Wang Q, Fish JA, Chai B, McGarrell DM, Sun Y, et al. (2014). Ribosomal
Database Project: data and tools for high throughput rRNA analysis. Nucleic Acids Res
42: D633–42.
Cole, J.R., Wang, Q., Fish, J.A., Chai, B., McGarrell, D.M., Sun, Y., Brown, C.T.,
Porras-Alfaro, A., Kuske, C.R., and Tiedje, J.M. (2014). Ribosomal Database Project:
data and tools for high throughput rRNA analysis. Nucleic Acids Res. 42, D633–D642.
Conrath, U., Beckers, G.J.M., Langenbach, C.J.G., and Jaskiewicz, M.R. (2015).
Priming for enhanced defense. Annu. Rev. Phytopathol. 53, 97–119.
Couto, D., and Zipfel, C. (2016). Regulation of pattern recognition receptor signalling in
plants. Nat. Rev. Immunol. 16, 537–552.
Daniel, B., Pavkov-Keller, T., Steiner, B., Dordic, A., Gutmann, A., Nidetzky, B., Sensen,
C.W., van der Graaff, E., Wallner, S., Gruber, K., et al. (2015). Oxidation of Monolignols
by Members of the Berberine Bridge Enzyme Family Suggests a Role in Plant Cell Wall
Metabolism. J. Biol. Chem. 290, 18770–18781.
Dasgupta, M.K. (1984). The Bengal famine, 1943 and the brown spot of rice--an inquiry
into their relations. Hist. Agric. 2, 1–18.
De Bary, A. (1879). Die erscheinung der symbiose.
de Wit, R., and Bouvier, T. (2006). “Everything is everywhere, but, the environment
selects”; what did Baas Becking and Beijerinck really say? Environ. Microbiol. 8, 755–
758.
DeAngelis, K.M., Lindow, S.E., and Firestone, M.K. (2008). Bacterial quorum sensing
and nitrogen cycling in rhizosphere soil. FEMS Microbiol. Ecol. 66, 197–207.

204

Delmotte, N., Knief, C., Chaffron, S., Innerebner, G., Roschitzki, B., Schlapbach, R.,
von Mering, C., and Vorholt, J.A. (2009). Community proteogenomics reveals insights
into the physiology of phyllosphere bacteria. Proc. Natl. Acad. Sci. U. S. A. 106,
16428–16433.
Delmotte, N., Knief, C., Chaffron, S., Innerebner, G., Roschitzki, B., Schlapbach, R.,
von Mering, C., and Vorholt, J.A. (2009). Community proteogenomics reveals insights
into the physiology of phyllosphere bacteria. Proc. Natl. Acad. Sci. U. S. A. 106,
16428–16433.
Dempsey, D.A., Vlot, A.C., Wildermuth, M.C., and Klessig, D.F. (2011). Salicylic Acid
biosynthesis and metabolism. Arabidopsis Book 9, e0156.
Dixon, R.A., Harrison, M.J., and Lamb, C.J. (1994). Early Events in the Activation of
Plant Defense Responses. Annu. Rev. Phytopathol. 32, 479–501.
Domenech, J., Reddy, M.S., Kloepper, J.W., Ramos, B., and Gutierrez-Mañero, J.
(2006). Combined Application of the Biological Product LS213 with Bacillus,
Pseudomonas or Chryseobacterium for Growth Promotion and Biological Control of
Soil-Borne Diseases in Pepper and Tomato. Biocontrol 51, 245.
Dong, X. (2004). NPR1, all things considered. Curr. Opin. Plant Biol. 7, 547–552.
Donn, S., Kirkegaard, J.A., Perera, G., Richardson, A.E., and Watt, M. (2015). Evolution
of bacterial communities in the wheat crop rhizosphere. Environ. Microbiol. 17, 610–
621.
Doornbos, R.F., Geraats, B.P.J., Kuramae, E.E., Van Loon, L.C., and Bakker, P.A.H.M.
(2011). Effects of jasmonic acid, ethylene, and salicylic acid signaling on the
rhizosphere bacterial community of Arabidopsis thaliana. Mol. Plant. Microbe. Interact.
24, 395–407.
Dowen, R.H., Pelizzola, M., Schmitz, R.J., Lister, R., Dowen, J.M., Nery, J.R., Dixon,
J.E., and Ecker, J.R. (2012). Widespread dynamic DNA methylation in response to
biotic stress. Proc. Natl. Acad. Sci. U. S. A. 109, E2183–E2191.
Duclaux, E. (1920). Pasteur: The History of a Mind (W.B. Saunders Company).
Durrant, W.E., and Dong, X. (2004). Systemic acquired resistance. Annu. Rev.
Phytopathol. 42, 185–209.

205

Dutilleul, C., Garmier, M., Noctor, G., Mathieu, C., Chétrit, P., Foyer, C.H., and de
Paepe, R. (2003). Leaf mitochondria modulate whole cell redox homeostasis, set
antioxidant capacity, and determine stress resistance through altered signaling and
diurnal regulation. Plant Cell 15, 1212–1226.
Dworkin, M. (2012). Sergei Winogradsky: a founder of modern microbiology and the
first microbial ecologist. FEMS Microbiol. Rev. 36, 364–379.
Edgar RC. (2013). UPARSE: highly accurate OTU sequences from microbial amplicon
reads. Nat Methods 10: 996–998.
Edgar, R.C., Haas, B.J., Clemente, J.C., Quince, C., and Knight, R. (2011). UCHIME
improves sensitivity and speed of chimera detection. Bioinformatics 27, 2194–2200.
Edwards, J., Johnson, C., Santos-Medellín, C., Lurie, E., Podishetty, N.K., Bhatnagar,
S., Eisen, J.A., and Sundaresan, V. (2015). Structure, variation, and assembly of the
root-associated microbiomes of rice. Proc. Natl. Acad. Sci. U. S. A. 112, E911–E920.
Fairman CE. (1887). Chemical Relations of Bacteria.
Felix, G., Duran, J.D., Volko, S., and Boller, T. (1999). Plants have a sensitive
perception system for the most conserved domain of bacterial flagellin. Plant J. 18,
265–276.
Fierer N, Jackson RB. (2006). The diversity and biogeography of soil bacterial
communities. Proc Natl Acad Sci U S A 103: 626–631.
Garoutte, A. (2016). Identifying the activities of rhizosphere microbial communities
using metatranscriptomics.
Gensollen T, Iyer SS, Kasper DL, Blumberg RS. (2016). How colonization by microbiota
in early life shapes the immune system. Science 352: 539–544.
Gómez-Gómez, L., and Boller, T. (2000). FLS2: An LRR Receptor–like Kinase Involved
in the Perception of the Bacterial Elicitor Flagellin in Arabidopsis. Mol. Cell 5, 1003–
1011.
Gómez-Gómez, L., Felix, G., and Boller, T. (1999). A single locus determines sensitivity
to bacterial flagellin in Arabidopsis thaliana. Plant J. 18, 277–284.
Goodman, A.L., Kallstrom, G., Faith, J.J., Reyes, A., Moore, A., Dantas, G., and
Gordon, J.I. (2011). Extensive personal human gut microbiota culture collections
characterized and manipulated in gnotobiotic mice. Proc. Natl. Acad. Sci. U. S. A. 108,
6252–6257.

206

Goodrich JK, Di Rienzi SC, Poole AC, Koren O, Walters WA, Caporaso JG, et al.
(2014). Conducting a microbiome study. Cell 158: 250–262.
Görlach, J., Volrath, S., Knauf-Beiter, G., Hengy, G., Beckhove, U., Kogel, K.H.,
Oostendorp, M., Staub, T., Ward, E., Kessmann, H., et al. (1996). Benzothiadiazole, a
novel class of inducers of systemic acquired resistance, activates gene expression and
disease resistance in wheat. Plant Cell 8, 629–643.
Gunning T, Cahill DM. (2009). A Soil-free Plant Growth System to Facilitate Analysis of
Plant Pathogen Interactions in Roots. Journal of Phytopathology 157: 497–501.
Hacquard, S., and Schadt, C.W. (2015). Towards a holistic understanding of the
beneficial interactions across the Populus microbiome. New Phytol. 205, 1424–1430.
Hansen, A.K., Vorburger, C., and Moran, N.A. (2012). Genomic basis of endosymbiontconferred protection against an insect parasitoid. Genome Res. 22, 106–114.
Hartmann, A., Rothballer, M., and Schmid, M. (2008). Lorenz Hiltner, a pioneer in
rhizosphere microbial ecology and soil bacteriology research. Plant Soil 312, 7–14.
Hayashi, F., Smith, K.D., Ozinsky, A., Hawn, T.R., Yi, E.C., Goodlett, D.R., Eng, J.K.,
Akira, S., Underhill, D.M., and Aderem, A. (2001). The innate immune response to
bacterial flagellin is mediated by Toll-like receptor 5. Nature 410, 1099–1103.
Henry A, Doucette W, Norton J, Jones S, Chard J, Bugbee B. (2006). An axenic plant
culture system for optimal growth in long-term studies. J Environ Qual 35: 590–598.
Hiltner L. (1904). Uber neuere Erfahrungen und Probleme auf dem Gebiet der
Bodenbakteriologie und unter besonderer Berucksichtigung der Grundungung and
Brache. Arbeiten der Deutschen Landwirtschaftlichen Gesellschaft 98: 59–78.
Hing, B., Ramos, E., Braun, P., McKane, M., Jancic, D., Tamashiro, K.L.K., Lee, R.S.,
Michaelson, J.J., Druley, T.E., and Potash, J.B. (2015). Adaptation of the targeted
capture Methyl-Seq platform for the mouse genome identifies novel tissue-specific
DNA methylation patterns of genes involved in neurodevelopment. Epigenetics 10,
581–596.
Hooper LV, Littman DR, Macpherson AJ. (2012). Interactions between the microbiota
and the immune system. Science 336: 1268–1273.
Horton, M.W., Bodenhausen, N., Beilsmith, K., Meng, D., Muegge, B.D., Subramanian,
S., Vetter, M.M., Vilhjálmsson, B.J., Nordborg, M., Gordon, J.I., et al. (2014). Genomewide association study of Arabidopsis thaliana leaf microbial community. Nat.
Commun. 5, 5320.

207

Howard MM, Bell TH, Kao-Kniffin J. (2017). Soil microbiome transfer method affects
microbiome composition, including dominant microorganisms, in a novel environment.
FEMS Microbiol Lett 364. e-pub ahead of print, doi: 10.1093/femsle/fnx092.
Ikeda, S., Kaneko, T., Okubo, T., Rallos, L.E.E., Eda, S., Mitsui, H., Sato, S., Nakamura,
Y., Tabata, S., and Minamisawa, K. (2009). Development of a bacterial cell enrichment
method and its application to the community analysis in soybean stems. Microb. Ecol.
58, 703–714.
Jackson MB, Abbott AJ, Belcher AR, Hall KC, Butler R, Cameron, J. (AFRC Institute of
Arable Crops Research, Long Ashton Research Station, University of Bristol, Bristol
BS18 9AF (United Kingdom). (1991). Ventilation in plant tissue cultures and effects of
poor aeration on ethylene and carbon dioxide accumulation, oxygen depletion and
explant development. Ann Bot 67. http://agris.fao.org/agrissearch/search.do?recordID=GB9129134.
Jackson, L.E., Burger, M., and Cavagnaro, T.R. (2008). Roots, nitrogen
transformations, and ecosystem services. Annu. Rev. Plant Biol. 59, 341–363.
Johnston-Monje, D., and Raizada, M.N. (2011). Conservation and diversity of seed
associated endophytes in Zea across boundaries of evolution, ethnography and
ecology. PLoS One 6, e20396.
Jones AM, Lindow SE, Wildermuth MC. (2007). Salicylic acid, yersiniabactin, and
pyoverdin production by the model phytopathogen Pseudomonas syringae pv. tomato
DC3000: synthesis, regulation, and impact on tomato and Arabidopsis host plants. J
Bacteriol 189: 6773–6786.
Jung, H.W., Tschaplinski, T.J., Wang, L., Glazebrook, J., and Greenberg, J.T. (2009).
Priming in systemic plant immunity. Science 324, 89–91.
Kachroo, A., and Robin, G.P. (2013). Systemic signaling during plant defense. Curr.
Opin. Plant Biol. 16, 527–533.
Kadota, Y., Shirasu, K., and Zipfel, C. (2015). Regulation of the NADPH Oxidase
RBOHD During Plant Immunity. Plant Cell Physiol. 56, 1472–1480.
Katagiri, F., Thilmony, R., and He, S.Y. (2002). The Arabidopsis thaliana-pseudomonas
syringae interaction. Arabidopsis Book 1, e0039.
Katsir, L., Schilmiller, A.L., Staswick, P.E., He, S.Y., and Howe, G.A. (2008). COI1 is a
critical component of a receptor for jasmonate and the bacterial virulence factor
coronatine. Proc. Natl. Acad. Sci. U. S. A. 105, 7100–7105.

208

Kau AL, Ahern PP, Griffin NW, Goodman AL, Gordon JI. (2011). Human nutrition, the
gut microbiome and the immune system. Nature 474: 327–336.
Kembel, S.W., O’Connor, T.K., Arnold, H.K., Hubbell, S.P., Wright, S.J., and Green, J.L.
(2014). Relationships between phyllosphere bacterial communities and plant functional
traits in a neotropical forest. Proc. Natl. Acad. Sci. U. S. A. 111, 13715–13720.
Kloepper JW, Schroth - Phytopathology MN, 1981. (1981). Plant growth-promoting
rhizobacteria and plant growth under gnotobiotic conditions. apsnet.org.
https://www.apsnet.org/publications/phytopathology/backissues/Documents/1981Arti
cles/Phyto71n06_642.pdf.
Knief, C., Delmotte, N., Chaffron, S., Stark, M., Innerebner, G., Wassmann, R., von
Mering, C., and Vorholt, J.A. (2012). Metaproteogenomic analysis of microbial
communities in the phyllosphere and rhizosphere of rice. ISME J. 6, 1378–1390.
Kniskern JM, Traw MB, Bergelson J. (2007). Salicylic acid and jasmonic acid signaling
defense pathways reduce natural bacterial diversity on Arabidopsis thaliana. Mol Plant
Microbe Interact 20: 1512–1522.
Krishnakumar, V., Hanlon, M.R., Contrino, S., Ferlanti, E.S., Karamycheva, S., Kim, M.,
Rosen, B.D., Cheng, C.-Y., Moreira, W., Mock, S.A., et al. (2015). Araport: the
Arabidopsis information portal. Nucleic Acids Res. 43, D1003–D1009.
Kunze, G., Zipfel, C., Robatzek, S., Niehaus, K., Boller, T., and Felix, G. (2004). The N
terminus of bacterial elongation factor Tu elicits innate immunity in Arabidopsis plants.
Plant Cell 16, 3496–3507.
Laemmli. (1970). Cleavage of structural proteins during the assembly of the head of
bacteriophage T4. Springer. http://link.springer.com/article/10.1038/227680a0.
Lau JA, Lennon JT. (2012). Rapid responses of soil microorganisms improve plant
fitness in novel environments. Proc Natl Acad Sci U S A 109: 14058–14062.
Lebeis, S.L., Paredes, S.H., Lundberg, D.S., Breakfield, N., Gehring, J., McDonald, M.,
Malfatti, S., del Rio, T.G., Jones, C.D., Tringe, S.G., et al. (2015). Salicylic acid
modulates colonization of the root microbiome by specific bacterial taxa. Science 349,
860–864.
Lee, B., Farag, M.A., Park, H.B., Kloepper, J.W., Lee, S.H., and Ryu, C.-M. (2012).
Induced resistance by a long-chain bacterial volatile: elicitation of plant systemic
defense by a C13 volatile produced by Paenibacillus polymyxa. PLoS One 7, e48744.

209

Leslie, M.E., and Heese, A. (2014). A re-elicitation assay to correlate flg22-signaling
competency with ligand-induced endocytic degradation of the FLS2 receptor. Methods
Mol. Biol. 1209, 149–162.
Ley RE, Turnbaugh PJ, Klein S, Gordon JI. (2006). Microbial ecology: human gut
microbes associated with obesity. Nature 444: 1022–1023.
Li, Y., Tschaplinski, T.J., Engle, N.L., Hamilton, C.Y., Rodriguez, M., Jr, Liao, J.C.,
Schadt, C.W., Guss, A.M., Yang, Y., and Graham, D.E. (2012). Combined inactivation
of the Clostridium cellulolyticum lactate and malate dehydrogenase genes substantially
increases ethanol yield from cellulose and switchgrass fermentations. Biotechnol.
Biofuels 5, 2.
Li, Y., Wang, W., and Zhou, J.-M. (2011). Role of small RNAs in the interaction between
Arabidopsis and Pseudomonas syringae. Front. Biol. 6, 462–467.
Ligterink, W., Kroj, T., zur Nieden, U., Hirt, H., and Scheel, D. (1997). Receptormediated activation of a MAP kinase in pathogen defense of plants. Science 276,
2054–2057.
Lloyd, A.M., Barnason, A.R., Rogers, S.G., Byrne, M.C., Fraley, R.T., and Horsch, R.B.
(1986). Transformation of Arabidopsis thaliana with Agrobacterium tumefaciens.
Science 234, 464–466.
Loper JE, Buyer - Mol. Plant-Microbe Interact JS, 1991. (1991). Siderophores in
microbial interactions on plant surfaces. apsnet.org.
https://www.apsnet.org/publications/mpmi/BackIssues/Documents/1991Articles/Micro
be04_005.pdf.
Lotrario JB, Stuart BJ, Lam T, Arands RR, O’Connor OA, Kosson DS. (1995). Effects of
sterilization methods on the physical characteristics of soil: implications for sorption
isotherm analyses. Bull Environ Contam Toxicol 54: 668–675.
Lozupone C, Lladser ME, Knights D, Stombaugh J, Knight R. (2011). UniFrac: an
effective distance metric for microbial community comparison. ISME J 5: 169–172.
Luckey TD. (1965). Effects of Microbes on Germfree Animals11Presented in modified
form as the principal talk at the International Symposium on Microecology, Berlin,
September, 1964. In: Umbreit WW (ed) Vol. 7. Advances in Applied Microbiology.
Academic Press, pp 169–223.
Luna, E., Pastor, V., Robert, J., Flors, V., Mauch-Mani, B., and Ton, J. (2011). Callose
deposition: a multifaceted plant defense response. Mol. Plant. Microbe. Interact. 24,
183–193.

210

Lundberg DS, Lebeis SL, Paredes SH, Yourstone S, Gehring J, Malfatti S, et al. (2012).
Defining the core Arabidopsis thaliana root microbiome. Nature 488: 86–90.
Ma L-S, Hachani A, Lin J-S, Filloux A, Lai E-M. (2014). Agrobacterium tumefaciens
deploys a superfamily of type VI secretion DNase effectors as weapons for
interbacterial competition in planta. Cell Host Microbe 16: 94–104.
Mansfield, J., Genin, S., Magori, S., Citovsky, V., Sriariyanum, M., Ronald, P., Dow, M.,
Verdier, V., Beer, S.V., Machado, M.A., et al. (2012). Top 10 plant pathogenic bacteria
in molecular plant pathology. Mol. Plant Pathol. 13, 614–629.
Marques, M.R., Buckeridge, M.S., Braga, M.R., and Dietrich, S.M.C. (2006).
Characterization of an extracellular endopolygalacturonase from the saprobe Mucor
ramosissimus Samutsevitsch and its action as trigger of defensive response in tropical
plants. Mycopathologia 162, 337–346.
Masella, A.P., Bartram, A.K., Truszkowski, J.M., Brown, D.G., and Neufeld, J.D. (2012).
PANDAseq: paired-end assembler for illumina sequences. BMC Bioinformatics 13, 31.
McCoy KD, Geuking MB, Ronchi F. (2017). Gut Microbiome Standardization in Control
and Experimental Mice. Curr Protoc Immunol 117: 23.1.1–23.1.13.
McMurdie PJ, Holmes S. (2013). phyloseq: an R package for reproducible interactive
analysis and graphics of microbiome census data. PLoS One 8: e61217.
Mendes, L.W., Kuramae, E.E., Navarrete, A.A., van Veen, J.A., and Tsai, S.M. (2014).
Taxonomical and functional microbial community selection in soybean rhizosphere.
ISME J. 8, 1577–1587.
Meyerowitz, E.M. (2001). Prehistory and history of Arabidopsis research. Plant Physiol.
125, 15–19.
Moran, N.A. (2001). The Coevolution of Bacterial Endosymbionts and Phloem-Feeding
Insects. Ann. Mo. Bot. Gard. 88, 35–44.
Mueller UG, Sachs JL. (2015). Engineering Microbiomes to Improve Plant and Animal
Health. Trends Microbiol 23: 606–617.
Nicklas W, Keubler L, Bleich A. (2015). Maintaining and Monitoring the Defined
Microbiota Status of Gnotobiotic Rodents. ILAR J 56: 241–249.
Nomura, K., Mecey, C., Lee, Y.-N., Imboden, L.A., Chang, J.H., and He, S.Y. (2011).
Effector-triggered immunity blocks pathogen degradation of an immunity-associated
vesicle traffic regulator in Arabidopsis. Proc. Natl. Acad. Sci. U. S. A. 108, 10774–
10779.

211

Panke-Buisse, K., Poole, A.C., Goodrich, J.K., Ley, R.E., and Kao-Kniffin, J. (2015).
Selection on soil microbiomes reveals reproducible impacts on plant function. ISME J.
9, 980–989.
Pankratov TA, Serkebaeva YM, Kulichevskaya IS, Liesack W, Dedysh SN. (2008).
Substrate-induced growth and isolation of Acidobacteria from acidic Sphagnum peat.
ISME J 2: 551–560.
Parks, D.H., Imelfort, M., Skennerton, C.T., Hugenholtz, P., and Tyson, G.W. (2015).
CheckM: assessing the quality of microbial genomes recovered from isolates, single
cells, and metagenomes. Genome Res. 25, 1043–1055.
Pasteur, L. (1885). Observations relatives à la note précédente de M. Duclaux. C.R.
Acad. Sci. 100, 68.
Peiffer, J.A., Spor, A., Koren, O., Jin, Z., Tringe, S.G., Dangl, J.L., Buckler, E.S., and
Ley, R.E. (2013). Diversity and heritability of the maize rhizosphere microbiome under
field conditions. Proc. Natl. Acad. Sci. U. S. A. 110, 6548–6553.
Pfeilmeier, S., Caly, D.L., and Malone, J.G. (2016). Bacterial pathogenesis of plants:
future challenges from a microbial perspective: Challenges in Bacterial Molecular Plant
Pathology. Mol. Plant Pathol. 17, 1298–1313.
Pieterse, C.M.J., Zamioudis, C., Berendsen, R.L., Weller, D.M., Van Wees, S.C.M., and
Bakker, P.A.H.M. (2014). Induced systemic resistance by beneficial microbes. Annu.
Rev. Phytopathol. 52, 347–375.
Preston, G., Deng, W.L., Huang, H.C., and Collmer, A. (1998). Negative regulation of
hrp genes in Pseudomonas syringae by HrpV. J. Bacteriol. 180, 4532–4537.
Provart, N.J., Alonso, J., Assmann, S.M., Bergmann, D., Brady, S.M., Brkljacic, J.,
Browse, J., Chapple, C., Colot, V., Cutler, S., et al. (2016). 50 years of Arabidopsis
research: highlights and future directions. New Phytol. 209, 921–944.
Ruinen, J. (1961). The phyllosphere. Plant Soil 15, 81–109.
Ryffel, F., Helfrich, E.J.N., Kiefer, P., Peyriga, L., Portais, J.-C., Piel, J., and Vorholt,
J.A. (2016). Metabolic footprint of epiphytic bacteria on Arabidopsis thaliana leaves.
ISME J. 10, 632–643.
Savary, S., Ficke, A., Aubertot, J.-N., and Hollier, C. (2012). Crop losses due to
diseases and their implications for global food production losses and food security.
Food Sec. 4, 519–537.

212

Scalschi, L., Vicedo, B., Camañes, G., Fernandez-Crespo, E., Lapeña, L., GonzálezBosch, C., and García-Agustín, P. (2013). Hexanoic acid is a resistance inducer that
protects tomato plants against Pseudomonas syringae by priming the jasmonic acid
and salicylic acid pathways. Mol. Plant Pathol. 14, 342–355.
Schmidt, P.-A., Bálint, M., Greshake, B., Bandow, C., Römbke, J., and Schmitt, I.
(2013). Illumina metabarcoding of a soil fungal community. Soil Biol. Biochem. 65, 128–
132.
Scholthof, K.-B.G. (2007). The disease triangle: pathogens, the environment and
society. Nat. Rev. Microbiol. 5, 152–156.
Sequeira, L. (1988). On becoming a plant pathologist: the changing scene. Annu. Rev.
Phytopathol. 26, 1–14.
Shaw LJ, Beaton Y, Glover LA, Killham K, Meharg AA. (1999). Re-inoculation of
autoclaved soil as a non-sterile treatment for xenobiotic sorption and biodegradation
studies. Appl Soil Ecol 11: 217–226.
Silby, M.W., Cerdeño-Tárraga, A.M., Vernikos, G.S., Giddens, S.R., Jackson, R.W.,
Preston, G.M., Zhang, X.-X., Moon, C.D., Gehrig, S.M., Godfrey, S.A.C., et al. (2009).
Genomic and genetic analyses of diversity and plant interactions of Pseudomonas
fluorescens. Genome Biol. 10, R51.
Smith K, McCoy KD, Macpherson AJ. (2007). Use of axenic animals in studying the
adaptation of mammals to their commensal intestinal microbiota. Semin Immunol 19:
59–69.
Spor A, Koren O, Ley R. (2011). Unravelling the effects of the environment and host
genotype on the gut microbiome. Nat Rev Microbiol 9: 279–290.
Stanton-Geddes, J., Paape, T., Epstein, B., Briskine, R., Yoder, J., Mudge, J., Bharti,
A.K., Farmer, A.D., Zhou, P., Denny, R., et al. (2013). Candidate genes and genetic
architecture of symbiotic and agronomic traits revealed by whole-genome, sequencebased association genetics in Medicago truncatula. PLoS One 8, e65688.
Stevens, R.B. (1960). Cultural practices in disease control. Plant Pathol.
Swingle, W.T. (1896). Bordeaux Mixture: Its Chemistry, Physical Properties, and Toxic
Effects on Fungi and Algae, by Walter T. Swingle.
Tata, J.R. (2005). From oxidative phosphorylation to transcription--a postdoctoral
adventure. Trends Biochem. Sci. 30, 529–534.
Tatum, L.A. (1971). The southern corn leaf blight epidemic. Science 171, 1113–1116.

213

Teal, T.K., Lies, D.P., Wold, B.J., and Newman, D.K. (2006). Spatiometabolic
stratification of Shewanella oneidensis biofilms. Appl. Environ. Microbiol. 72, 7324–
7330.
Thompson, L.R., Sanders, J.G., McDonald, D., Amir, A., Ladau, J., Locey, K.J., Prill,
R.J., Tripathi, A., Gibbons, S.M., Ackermann, G., et al. (2017). A communal catalogue
reveals Earth’s multiscale microbial diversity. Nature 551, 457–463.
Thuerig, B., Felix, G., Binder, A., Boller, T., and Tamm, L. (2005). An extract of
Penicillium chrysogenum elicits early defense-related responses and induces
resistance in Arabidopsis thaliana independently of known signalling pathways.
Physiol. Mol. Plant Pathol. 67, 180–193.
Tian, B.-Y., Cao, Y., and 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.
Traw, M.B., Kniskern, J.M., and Bergelson, J. (2007). SAR increases fitness of
Arabidopsis thaliana in the presence of natural bacterial pathogens. Evolution 61,
2444–2449.
Trevors JT. (1996). Sterilization and inhibition of microbial activity in soil. J Microbiol
Methods 26: 53–59.
Tringe, S.G., von Mering, C., Kobayashi, A., Salamov, A.A., Chen, K., Chang, H.W.,
Podar, M., Short, J.M., Mathur, E.J., Detter, J.C., et al. (2005). Comparative
metagenomics of microbial communities. Science 308, 554–557.
Tschaplinski, T.J., Standaert, R.F., Engle, N.L., Martin, M.Z., Sangha, A.K., Parks, J.M.,
Smith, J.C., Samuel, R., Jiang, N., Pu, Y., et al. (2012). Down-regulation of the caffeic
acid O-methyltransferase gene in switchgrass reveals a novel monolignol analog.
Biotechnol. Biofuels 5, 71.

214

Tsuda, K., Mine, A., Bethke, G., Igarashi, D., Botanga, C.J., Tsuda, Y., Glazebrook, J.,
Sato, M., and Katagiri, F. (2013). Dual regulation of gene expression mediated by
extended MAPK activation and salicylic acid contributes to robust innate immunity in
Arabidopsis thaliana. PLoS Genet. 9, e1004015.
Turner, B.L., and Haygarth, P.M. (2001). Biogeochemistry. Phosphorus solubilization in
rewetted soils. Nature 411, 258.
Van Wees, S., and De Swart, E. (2000). Enhancement of induced disease resistance by
simultaneous activation of salicylate-and jasmonate-dependent defense pathways in
Arabidopsis thaliana. Proceedings of the.
Vogei C., Bodenhausen N., Gruissem W., Vorholt J.A. (2016). The Arabidopsis leaf
transcriptome reveals distinct but also overlapping responses to colonization by
phyllosphere commensals and pathogen infection with impact on plant health. New
Phytol 212: 192–207.
Wagner, M.R., Lundberg, D.S., Coleman-Derr, D., Tringe, S.G., Dangl, J.L., and
Mitchell-Olds, T. (2014). Natural soil microbes alter flowering phenology and the
intensity of selection on flowering time in a wild Arabidopsis relative. Ecol. Lett. 17,
717–726.
Whalen, M.C., Innes, R.W., Bent, A.F., and Staskawicz, B.J. (1991). Identification of
Pseudomonas syringae pathogens of Arabidopsis and a bacterial locus determining
avirulence on both Arabidopsis and soybean. Plant Cell 3, 49–59.
Wildermuth, M.C., Dewdney, J., Wu, G., and Ausubel, F.M. (2001). Isochorismate
synthase is required to synthesize salicylic acid for plant defense. Nature 414, 562–
565.
Withers, J., and Dong, X. (2017). Post-translational regulation of plant immunity. Curr.
Opin. Plant Biol. 38, 124–132.
Wolf DC, Dao TH, Scott HD, Lavy TL. (1989). Influence of Sterilization Methods on
Selected Soil Microbiological, Physical, and Chemical Properties. J Environ Qual 18:
39–44.
Xin, X.-F., and He, S.Y. (2013). Pseudomonas syringae pv. tomato DC3000: a model
pathogen for probing disease susceptibility and hormone signaling in plants. Annu.
Rev. Phytopathol. 51, 473–498.
Xin, X.-F., Nomura, K., Aung, K., Velásquez, A.C., Yao, J., Boutrot, F., Chang, J.H.,
Zipfel, C., and He, S.Y. (2016). Bacteria establish an aqueous living space in plants
crucial for virulence. Nature 539, 524–529.

215

Yu, D., Chen, C., and Chen, Z. (2001). Evidence for an important role of WRKY DNA
binding proteins in the regulation of NPR1 gene expression. Plant Cell 13, 1527–1540.
Zeng, W., Brutus, A., Kremer, J.M., Withers, J.C., Gao, X., Jones, A.D., and He, S.Y.
(2011). A genetic screen reveals Arabidopsis stomatal and/or apoplastic defenses
against Pseudomonas syringae pv. tomato DC3000. PLoS Pathog. 7, e1002291.
Zgadzaj R, Garrido-Oter R, Jensen DB, Koprivova A, Schulze-Lefert P, Radutoiu S.
(2016). Root nodule symbiosis in Lotus japonicus drives the establishment of
distinctive rhizosphere, root, and nodule bacterial communities. Proc Natl Acad Sci U S
A 113: E7996–E8005.
Zhou, J., and Ning, D. (2017). Stochastic Community Assembly: Does It Matter in
Microbial Ecology? Microbiol. Mol. Biol. Rev. 81.
Zhou, M., Lu, Y., Bethke, G., Harrison, B.T., Hatsugai, N., Katagiri, F., and Glazebrook,
J. (2017). WRKY70 prevents axenic activation of plant immunity by direct repression of
SARD1. New Phytol.
Zhou, N., Tootle, T.L., Tsui, F., Klessig, D.F., and Glazebrook, J. (1998). PAD4 functions
upstream from salicylic acid to control defense responses in Arabidopsis. Plant Cell 10,
1021–1030.
Zhu, C., Ding, Y., and Liu, H. (2011). MiR398 and plant stress responses. Physiol.
Plant. 143, 1–9.

216