Detection of Cellulose Synthase Antisense Transcripts Involved in Regulating Cell Wall

Biosynthesis in Barley, Brachypodium and Arabidopsis

A dissertation presented to

the faculty of

the College of Arts and Sciences of Ohio University

In partial fulfillment

of the requirements for the degree

Doctor of Philosophy

Daniel B. Nething

August 2017

This dissertation titled

Detection of Cellulose Synthase Antisense Transcripts Involved in Regulating Cell Wall

Biosynthesis in Barley, Brachypodium and Arabidopsis

DANIEL B. NETHING

has been approved for

the Department of Chemistry and Biochemistry

and the College of Arts and Sciences by

Michael A. Held

Associate Professor of Chemistry and Biochemistry

Robert Frank

Dean, College of Arts and Sciences

ABSTRACT

NETHING, DANIEL B., Ph.D., August 2017, Chemistry

Detection of Cellulose Synthase Antisense Transcripts Involved in Regulating Cell Wall

Biosynthesis in Barley, Brachypodium and Arabidopsis

Director of Dissertation: Michael A. Held

Current understanding of the regulatory mechanisms governing plant cell wall development is incomplete. Hormonal and transcriptional mechanisms that activate wall biosynthesis are well understood, but mechanisms that control the fine attenuation and repression of wall growth remain elusive. My efforts are to evaluate a potential mechanism that utilizes cis and/or trans-acting small RNAs to broadly control the available transcript levels in the cellulose synthase (CESA) gene superfamily. Previous work (Held et al., 2008) has shown that endogenously expressed natural antisense transcripts (NATs) derived from the HvCESA6 (Hordeum vulgare) gene are degraded to yield small RNAs capable of influencing wall biosynthesis. Expanding on these studies, I have now identified antisense transcripts from additional members of the CESA families in barley, brachypodium, and arabidopsis, and identified dynamic expression of

HvCESA1 antisense transcripts and small RNAs during leaf development similar to

HvCESA6. This complements results from a custom microarray indicating artificial siRNA production targeting the HvCESAs shows concurrent silencing of related cell wall biosynthesis genes. My data indicate that CESA-derived antisense transcripts and siRNAs could act via an RNA-induced silencing mechanism to degrade cis transcripts, which could also trigger trans-acting silencing of related genes with similar sequences. Such a 4 mechanism may be important for selectively attenuating CESA genes, as observed in the transition from primary to secondary wall biosynthesis.

DEDICATION

I dedicate this work to my family and friends, as I did not walk this path alone. To those

of you who walked it with me, it wasn’t possible without your support. Thank you.

ACKNOWLEDGMENTS

To start, I’d like to thank my advisor Dr. Michael Held. You gave me opportunity when I was drifting, patient instruction when I was lost, and were a guiding example as how to be both a good scientist and a good educator. Thank you for the wisdom and conviction, and putting up with me when I probably was making you pull your hair out.

This was not possible without you.

Thank you to my committee members, Dr. Kieliszewski, Dr. Showalter, Dr. Chen, and Dr. Cimatu, for your insight and rigor. A scientist is only as good as those who push and teach them, and you have fulfilled that role well.

Thank you to my lab mates, Wen, Yadi, and John, for being there with me, and always being willing to troubleshoot and learn from each other. I feel fortunate to have worked with you all, and I’ll always consider you family.

Thank you to the members of the Plant Cell Wall Journal Club, past and present, for being a constant source of knowledge and constructive criticism, and for the chance to hone my presentation skills. I would not know nearly as much as I do now without participating in the extended discussions and arguments we all had about plant science over the years.

I’d also like to thank several people who were not involved in my research, but were essential to keeping me going. Dr. Barlag, thank you for your emotional support and sanity checks. You gave me strength when I needed it. Dr. Nyasulu, thank you for the conversations and the trust. You often lifted my mood on a dark day. Members of Alpha

Chi Sigma, Gamma Nu chapter, thank you for the brotherhood. You gave me joy and 7 strong friendships. Amber, Ally, and Sara, thank you for being there, and not letting me get too serious for my own good.

Finally, I’d like to thank my parents and brother for being with me from the very beginning of this trip. Dad, you made it possible for me to even consider graduate school, and were always there when things were literally falling down around my ears. Mom, you listened when I needed an ear, and talked to me when I needed advice, and made sure I didn’t want for anything. Ben, you showed up when I needed you, and always were there to talk it out. Without you all, this never could have been.

TABLE OF CONTENTS

Page

Abstract ...... 3 Dedication ...... 5 Acknowledgments...... 6 List of Tables ...... 12 List of Figures ...... 13 List of Abbreviaitons (Alphabetized) ...... 15 Chapter 1: Introduction ...... 19 Chapter 2: Background ...... 24 The Plant Cell Wall ...... 24 Differences between Primary and Secondary Cell Wall ...... 27 Primary Cell Wall ...... 27 Secondary Cell Wall ...... 28 Differences between Type I and Type II Primary Cell Wall ...... 29 Type I Primary Cell Wall ...... 29 Type II Primary Cell Wall ...... 30 Cellulose and Hemicellulose Biosynthesis ...... 32 Regulatory Control Mechanisms for Cell Wall Biosynthesis ...... 35 Transcriptional Control Network ...... 35 Post Transcriptional, Small RNA Control Networks ...... 40 Antisense RNA and siRNA Regulation of CESA Genes ...... 53 Experimental Aims ...... 58 Chapter 3: Detection of CESA Antisense Transcripts and Small RNAs in the Grass Species Hordeum vulgare ...... 61 Introduction ...... 61 Materials and Methods ...... 63 Construction and Use of BSMV Vectors ...... 63 Viral Inoculation of Barley Plants ...... 63 Harvest of VIGS Tissue and Preparation of RNA ...... 64 qPCR Verification of HvCESA Silencing ...... 64 Construction of Custom Microarray ...... 65 9

Microarray Hybridization and Data Extraction ...... 65 Processing of Microarray Data ...... 65 Design of Gene-Specific Primers for Tagged-SS-RT-PCR of HvCESAs ...... 66 Preparation of Barley Tissue ...... 66 Preparation of Barley RNA ...... 67 Preparation of HvCESA Antisense cDNA ...... 67 Amplification of HvCESA Antisense cDNA ...... 68 Characterization of HvCESA Amplicons ...... 68 RNA Loading Control for HvCESA1 Antisense Time Course ...... 69 Preparation of HvCESA1 Antisense Time Course cDNA ...... 69 Amplification of HvCESA1 Antisense Time Course cDNA ...... 69 Characterization of HvCESA1 Time Course Amplicons ...... 70 Design of HvCESA1 RPA Probes ...... 70 HvCESA1 Time Course RPA ...... 71 Characterization of HvCESA1 Time Course RPA Gels ...... 71 Results ...... 71 Antisense Transcripts Corresponding to Multiple Cellulose Synthase Genes Are Detectible in Developing Barley Leaf ...... 71 Expression of HvCESA1 Antisense Transcripts Are Dynamic during Leaf Growth ...... 75 Dynamic Expression of 21 and 24 Nucleotide Antisense smRNAs ...... 78 Microarray Analysis of Specific HvCESA Silenced Tissues Shows Changes in Other Cell Wall Related Genes ...... 79 Discussion ...... 85 Conclusion ...... 91 Chapter 4: Detection of CESA Antisense Transcripts in the Grass Species Brachypodium Distachyon ...... 93 Introduction ...... 93 Materials and Methods ...... 94 Preparation of Brachypodium Tissue...... 94 Preparation of Brachypodium RNA...... 94 Design of Gene-Specific Primers for Tagged SS-RT-PCR of BdCESAs ...... 95 Preparation of BdCESA cDNA ...... 95 Amplification of Brachypodium cDNA...... 95 10

Characterization of BdCESA Survey Amplicons ...... 96 RNA Loading Control for BdCESA Antisense Time Courses ...... 96 Preparation of BdCESA Antisense Time Course cDNA ...... 97 Amplification of BdCESA Antisense Time Course cDNA ...... 97 Characterization of BdCESA Time Course Amplicons...... 97 qPCR Analysis of BdCESA mRNA Expression ...... 97 CESA Phylogenetic Tree ...... 98 Data Mining for BdCESA smRNAs ...... 98 Results ...... 98 Antisense Transcripts and smRNAs of CESAs Are Present in the Model Grass Brachypodium distachyon ...... 98 Expression of BdCESA Antisense Transcripts Decrease Over Time ...... 103 Expression of BdCESA mRNAs Show Correlation with Antisense Transcripts 106 Discussion ...... 111 Conclusion ...... 115 Chapter 5: Detection of CESA Antisense Trancripts in Arabidopsis and Attempts at Identifying Antisense Source ...... 116 Introduction ...... 116 Materials and Methods ...... 116 CESA Phylogenetic Tree ...... 116 Arabidopsis thaliana Liquid Culture ...... 117 Preparation of Arabidopsis RNA ...... 117 Preparation of AtCESA Antisense cDNA ...... 118 Amplification of AtCESA Antisense cDNA ...... 118 Characterization of AtCESA Antisense Amplicons ...... 118 Growing Arabidopsis thaliana Inflorescences ...... 119 Results ...... 119 Targeted Detection of Antisense AtCESA Transcripts ...... 119 Efforts Towards Identifying Antisense Transcript Source ...... 122 Discussion ...... 124 Conclusion ...... 125 Chapter 6: Lessons Learned; Refinement of a Strand-Specific Reverse-Transcription Assay to Detect Antisense Transcripts ...... 126 Introduction ...... 126 11

Materials and Methods ...... 127 Design of SS-RT-PCR Tag1 Primer ...... 127 Total RNA Isolation and Purification ...... 127 cDNA Synthesis and Dilution ...... 127 Spin Column cDNA Cleanup Method ...... 128 Generic Method for cDNA Amplification ...... 128 Results ...... 128 Identification of Strand Specificity Problems with SS-RT-PCR ...... 128 Improving Antisense Assay Signal and Reproducibility ...... 129 Discussion ...... 132 Conclusion ...... 139 Chapter 7: Future Directions ...... 140 Specific Aim 1. Identify More Specific and Granular Tissues for Examining Primary to Secondary Wall Transition ...... 143 Experimental Design ...... 144 Specific Aim 2. Determine a Better Way to Get a Global Picture of Antisense Transcripts and Their Relationship with Antisense Transcripts and Small RNAs ... 146 Experimental Design ...... 146 Specific Aim 3. Examine the Known RDR Pathway Mutants for mRNA, Antisense Transcript, and siRNA Behavior...... 149 Experimental Design ...... 149 References ...... 151

LIST OF TABLES

Page

Table 1 Summary table of hormone effects on cell wall biosynthesis ...... 36 Table 2 Primer sequences, melting temperatures, and extension times for HvCESA antisense transcript survey and ribonuclease protection assay ...... 73 Table 3 Groupings of significantly up and downregulated genes following VIGS of the HvCESA gene family ...... 83 Table 4 Statistically significant microarray results...... 84 Table 5 Primer sequences, melting temperatures, and extension times for BdCESA antisense transcript survey...... 100 Table 6 BdCESA smRNA counts mined from brachypodium data ...... 103 Table 7 Comparison table of phylogenetically similar CESAs ...... 107 Table 8 BdCESA qPCR Primers ...... 108 Table 9 Primer sequences, melting temperatures, and extension times for AtCESA antisense transcript survey ...... 121 Table 10 Necessary controls for tagged SS-RT-PCR ...... 135

LIST OF FIGURES

Page

Figure 1. Radial dendrogram of the CESA superfamily ...... 21 Figure 2. AtCESAs show clustered expression with other cell wall biosynthesis genes. .. 22 Figure 3. Network visualization of hormone and transcription factor pathways...... 38 Figure 4. Consolidated miRNA processing mechanism ...... 43 Figure 5. General mechanism of siRNA Formation from Double Stranded RNA ...... 46 Figure 6. Generalized mechanism for producing dsRNA via RDR activity...... 50 Figure 7. Potential crossover point between nonsense mediated mRNA decay and siRNA synthesis pathways...... 58 Figure 8: Expression pattern of barley CESA genes in developing leaf tissue ...... 73 Figure 9. Survey of antisense transcript gels for the HvCESA gene family ...... 75 Figure 10. Developmental time course gels of antisense transcripts and smRNAs for the HvCESA1 ...... 76 Figure 11. Quantification of HvCESA1 sense and antisense transcript levels over developmental time course ...... 77 Figure 12. Map of HvCESA1 RPA probe and HvCESA1 antisense transcripts ...... 78 Figure 13. Quantification of HvCESA1 antisense smRNA levels during a developmental time course ...... 79 Figure 14. Map of VIGS target site and qPCR primer locations based on HvCESA1 representative sequence ...... 80 Figure 15. Examples of VIGS photobleaching effect ...... 81 Figure 16. Relative expression of HvCESA1 ...... 82 Figure 17. Predicted mechanism for RDR based synthesis of antisense transcripts ...... 88 Figure 18. Domain map of BdCESA proteins ...... 101 Figure 19. Survey of antisense transcripts for the BdCESA gene family...... 102 Figure 20. Developmental time courses of antisense transcripts for the BdCESA family...... 104 Figure 21. Absolute quantification of BdCESA antisense transcript levels over developmental time course...... 105 Figure 22. Phylogenetic tree of Clustal Omega multiple alignment of BdCESA and HvCESA translated amino acid sequences...... 106 Figure 23. Relative expression of primary cell wall related BdCESA genes...... 109 Figure 24. Relative expression of secondary cell wall related BdCESA genes...... 110 14

Figure 25. Relative qPCR-measured expression of all BdCESA mRNAs concentrations ...... 110 Figure 26. Relative expression of other BdCESA genes...... 111 Figure 27: Phylogenetic tree of Clustal Omega of select HvCESA, BdCESA, and AtCESA translated sequences...... 120 Figure 28. Partial survey of antisense transcript gels for the AtCESA gene family, maximum amplification conditions...... 122 Figure 29: Examination of AtCESA1 antisense transcripts in RDR mutants...... 124 Figure 30. Evidence of cDNA reverse transcription occurring without primers...... 129 Figure 31. RNase H treatment and cDNA spin purification condition testing ...... 130 Figure 32. Effects of thermal cycler selection on BdCESA1 antisense transcript signal. 131 Figure 33. Effects of cycle melt times and instrument preheating on PCR...... 132 Figure 34. Potential mechanisms for unprimed cDNA synthesis ...... 133 Figure 35. Stepwise mechanism of tagged SS-RT-PCR ...... 134 Figure 36. Example gel showing detection of an antisense transcript ...... 136 Figure 37. Mechanism of action of RNase H in degrading the RNA component of a RNA-DNA hybrid...... 137 Figure 38. Depiction of effects of cDNA cleanup via spin column...... 137

LIST OF ABBREVIAITONS (ALPHABETIZED)

Note: All proteins will be represented in uppercase, and all genes will be uppercase and italicized.

ABRC Arabidopsis Biological Resource Center

AGA Apiogalacturonan

AGO Argonaute (ribonuclease)

AGP Arabinogalactan Protein

ATAF Arabidopsis Transcription Activation Factor

CBP Cap Binding Protein

CDS Coding Sequence

CESA Cellulose Synthase

CPL CTD Phosphatase-like

CSC Cellulose Synthase Complex

CSL Cellulose Synthase-Like

Ct Cycle Threshold

CUC Cup Shaped Cotyledon

CWB Cell Wall Biosynthesis

DCB 2,6-dichlorobenzonitrile

DCL Dicer-Like (ribonuclease)

DDL Dawdle dsRNA Double Stranded RNA

GAX Glucuronoarabinoxylan 16 gDNA Genomic DNA

Glc Glucose

GSP Gene-Specific Primer

GT Glycosyltransferase hc-siRNA Heterochromatic siRNA

HEN HUA Enhancer (methyltransferase)

HESO HEN1 Suppressor (nucleotidyltransferase protein

HG Homogalacturonan

HRGP Hydroxyproline-Rich Glycoproteins

HS-GAX Highly Substituted GAX

HSP Heat Shock Protein

HYL HYPONASTIC LEAVES (dsRNA-binding protein)

JA Jasmonic Acid lncRNA Long Non-Coding RNA miRNA MicroRNA

MLG Mixed-linkage Glucan

NAC NAM-ATAF1/2-CUC2 Domain

NAM No Apical Meristem ncRNA Non-Coding RNA

NEB New England Biolabs

NOT Negative on TATA-less (TF)

NPC No Primer Control 17

NRT No Reverse Transcriptase (Control)

NST NAC Secondary Wall Thickening Promoting Factor

PCR Polymerase Chain Reaction phasiRNA Phased, Secondary siRNA

Pol II RNAP

Pri-miRNA Primary miRNA

PRP Proline Rich Protein

PTGS Post Transcriptional Gene Silencing qPCR Quantitative RT-PCR

RACE Rapid Amplification of cDNA Ends

RDR RNA-dependent RNA-polymerase

RGI Rhamnogalacturonan I

RGII Rhamnogalacturonan II

RISC RNA Induced Silencing Complex

RNAi RNA Interference

RNAP (DNA-dependent) RNA Polymerase

RPA Ribonuclease Protection Assay

RT Reverse Transcriptase

RT-PCR Reverse Transcriptase PCR

SDN Small RNA Degrading Nuclease

SE SERRATE (zinc-finger domain protein) siRNA Small Interfering RNA 18 smRNA Small RNAs

SND Secondary Wall Associated NAC Domain Protein

SQN SQUINT (cyclophilin-like protein) ssRNA Single Stranded RNA

SS-RT-PCR Strand-Specific Reverse-Transcriptase PCR tasiRNA Trans-acting siRNA

TF Transcription Factor

TGH TOUGH (TATA-box binding protein)

VND Vascular-Related NAC Domain Protein

WT Col0 Wild Type Columbia 0

XG Xyloglucan

XRN Exoribonuclease

Xyl Xylose

CHAPTER 1: INTRODUCTION

A wealth of evidence that climate change is driven by the mass combustion of non-renewable fossil fuels has spurred development of renewable energy sources that reduce dependency on fossil fuels. While solar and wind power have significant capacity for energy generation without producing any carbon dioxide, the options for storing that energy are still comparatively poor. The energy density of widely used lithium ion batteries currently has a practical energy density of 0.540 MJ/kg (Thackeray, Wolverton,

& Isaacs, 2012), compared to approximately 30.1 MJ/kg for ethanol and 45.2 MJ/kg for gasoline, depending on the blend (Davis, Williams, & Boundy, 2016). This has created an economically viable niche for the development of liquid-based biofuels that can be produced from renewable feedstocks. A major potential feedstock is plant biomass, which contains large amounts of stored energy in the form of a cell wall.

The cell wall is the defining feature of the plant, as it shapes the growth and function of the organism (Carpita, 1996; Carpita & Gibeaut, 1993). A plant cell wall is a complex, integrated network of many biopolymers which are roughly divided into cellulose (Brown & Saxena, 2000; Delmer, 1999; Doblin, Kurek, Jacob-Wilk, & Delmer,

2002), hemicelluloses (Scheller & Ulvskov, 2010), pectins (Mohnen, 2008), structural proteins (Cosgrove, 2015; Lamport, Kieliszewski, Chen, & Cannon, 2011; Showalter,

1993), and (in some cases) lignins (Boerjan, Ralph, & Baucher, 2003). All five classes of molecule have specific characteristics that work cooperatively to generate a robust and flexible structure that is capable of rapid growth, immense strength, and a large variety of specialization (Cosgrove, 2005). 20

Of the five classes, cellulose is central to the structure of the cell wall. Cellulose is a long-chain, para-crystalline polymer of 1,4-β linked glucose units (Kolpak &

Blackwell, 1976), and makes up much of the cell wall material by weight (Pauly &

Keegstra, 2008). Cellulose is synthesized by plasma membrane localized (Kimura et al.,

1999), microtubule guided (Gardiner, Taylor, & Turner, 2003), multimeric cellulose synthase complexes (Taylor, Laurie, & Turner, 2000) (CSCs) that are composed of a heterotrimeric mixture of cellulose synthase (CESA) subunits shown to have a 1:1:1 stoichiometric ratio in arabidopsis models (Hill, Hammudi, & Tien, 2014). Eudicots and monocots generally have genes for multiple CESAs, and specific members are expressed differentially depending on tissue or developmental stage (R. A. Burton, Shirley, King,

Harvey, & Fincher, 2004; McFarlane, Döring, & Persson, 2014).

CESAs are part of the larger cellulose synthase superfamily (Figure 1) (Penning et al., 2009; Richmond & Somerville, 2000), whose members generally function as glycosyltransferases (GTs) to assemble complex carbohydrate chains. Many of these chains are used in the synthesis of the cell wall.

Examination of CESA genes across many plant species shows that cell wall synthesis-related genes are tightly co-expressed with CESAs (Figure 2) (Persson, Wei,

Milne, Page, & Somerville, 2005). This indicates that specific CESA family members are highly expressed in specific sets, along with other specific cofactors likely involved in cell wall biosynthesis. The established expression patterns of CESAs are mainly split between primary and secondary cell wall development, with some outliers thought to act in specialized cases (R. A. Burton et al., 2004; Persson et al., 2005). Depending on 21 developmental stage, different sets of CESAs combine to make up the CSCs (Tanaka et al., 2003; Taylor, Howells, Huttly, Vickers, & Turner, 2003; Taylor et al., 2000).

Figure 1. Radial dendrogram of the CESA superfamily This dendrogram depicts clade groupings of CESAs and CSLs from barley (Hv), brachypodium (Bd), and arabidopsis (At). Clades are separated by color. Scale bar is number of substitutions per unit length in the alignment (excluding gaps). 22

Figure 2. AtCESAs show clustered expression with other cell wall biosynthesis genes. Adapted from Persson et al. (2005). Members of the cellulose synthase gene family have clustered co-expression with other specific cellulose synthases, as well as many other genes involved in the production of the cell wall.

During early growth of the wall, expression of “primary-wall” CESAs is elevated, and the cell walls produced are thin and flexible to accommodate rapid cell growth (Burn,

2002; Fagard et al., 2000). Once the primary wall is laid down, primary wall-linked

CESA expression is down-regulated, and “secondary wall” CESAs show a tightly coordinated increase in expression that correlates with other secondary cell wall biosynthetic genes (Taylor et al., 2003). Both primary and secondary wall CESAs produce the same cellulose biopolymer, albeit with variation in the polymer length

(McNeil, Darvill, Fry, & Albersheim, 1984) and para-crystalline nature (Kataoka &

Kondo, 1998). Currently, the only observed difference between the primary and 23 secondary wall CSCs is a difference in the rate of cellulose production (Y. Watanabe et al., 2015). This is unusual given that a single set of CESAs for a single CSC should be sufficient to produce all the necessary cellulose, and suggests that there may be a purpose for the conservation of two unique sets of functional homologs.

Given the presence of seemingly redundant CESA genes and ubiquitous usage of its cellulose product, understanding the regulatory mechanisms involved in controlling these genes and their corresponding proteins is essential toward the development of plant based biotechnologies. Here, we attempt to clarify the role(s) of post-transcriptional regulation on CESA mRNA levels, specifically involving antisense RNAs and small

RNAs. A wide array of proposed regulatory mechanisms exist in this area of study, and understanding any single component requires interlocking knowledge of the cell wall and the regulation of its development. Therefore, a thorough background of both topics has been laid out in the following chapter.

CHAPTER 2: BACKGROUND

The Plant Cell Wall

The cell wall is a structure present in plants, fungi, and some species of bacteria and Archaea. In plants, it is a complex and dynamic network of cellulose microfibrils, hemicelluloses, pectins, structural proteins, and lignins that function cooperatively to create a highly adaptive ultrastructure (Carpita & Gibeaut, 1993). This allows the cell wall to serve many purposes. It acts as a defensive barrier for the cell, protecting it against physical and biological damage. It also gives structure to the plant, and allows the plant to use turgor to create a hydrostatic skeleton, which controls both tissue position and growth in response to changing conditions. Plant cell walls can be highly specialized to let plants respond to both biotic and abiotic stresses which is critical to their survival, as plants are fixed in a single location for their entire lifetime.

Cell wall structure and function varies widely based on tissue type and developmental stage. Cell walls can be highly flexible, and are capable of great tensile strength while still allowing cellular expansion and elongation. As plant cells develop and differentiate, cell walls can thicken, become more rigid in structure, and resist changes in size and shape. This change in cell wall structure also allows for unique cellular functions to be created. For example, hollow and waterproof vascular elements (i.e. xylem) can be produced to convey water and solutes through the plants (Zhou, Lee, Zhong, & Ye,

2009), while endothecium tissues can create uneven tension distributions that break and fling pollen away from the plant to facilitate reproduction (C. Yang et al., 2007). 25

The diverse and adaptable nature of cell walls comes from their complex interwoven network structure. Each of these networks can change independently, synergistically interacting with the others. The backbone of the cell wall is generally composed of cellulose microfibrils. Each microfibril is composed of 30 to 50 parallel

(1,4)-β-D-glucan chains collectively called cellulose (Albersheim, Darvill, Roberts,

Sederoff, & Staehelin, 2011). Individual cellulose chains begin and end at random points along the microfibril, creating a long continuous strand, like fibers in a thread. The cellulose backbone is coated with hemicelluloses such as xyloglucans (XGs) or glucuronoarabinoxylans (GAXs) that interlock parallel microfibrils (Carpita & Gibeaut,

1993). Hemicelluloses can hydrogen bond to the surface of cellulose microfibrils and create cross links between fibers that increase the strength of the wall (Wilder &

Albersheim, 1973).

This framework of cellulose and hemicellulose sits inside a matrix of pectic polysaccharides. These pectic polysaccharides come in several varieties, with the main four being apiogalacturonan (AGA), rhamnogalacturonan I (RG-I), rhamnogalacturonan

II (RG-II), and homogalacturonan (HG) (Caffall & Mohnen, 2009). The functions of pectin include the regulation of wall rheology through regulation of the gel matrix via polymer modifications and structural cross linking (Carpita & Gibeaut, 1993). Pectic polysaccharides contain methyl esters that can be (reversibly) de-esterified by pectin methylesterases. De-esterification changes the net charge of the pectin and creates regions that can cross link in the presence of Ca2+ ions (Tepfer & Taylor, 1981). This flexibility in cross-linking allows the pectin matrix to regulate many aspects of the cell 26 wall, including wall porosity, pH, ionic strength, and ionic composition, responding to biotic and abiotic stresses (Caffall & Mohnen, 2009; McNeil et al., 1984). For example, modification of gel pore size can allow enzymes to contact substrates that would otherwise be inaccessible (Baron-Epel, Gharyal, & Schindler, 1988). By controlling these aspects of the cell wall, regulatory function is obtained.

After the cell wall initially develops, more permanent cross linking is necessary to solidify it. This is accomplished in part through the function of structural proteins.

Structural proteins generally have a basic pH (Mellon & Helgeson, 1982), and variable amounts of glycosylation (Showalter & Basu, 2016). The structural proteins in the cell wall consist of several subgroups, the hydroxyproline-rich glycoproteins (HRGPs), the proline-rich proteins (PRPs) (Bradley, Kjellbom, & Lamb, 1992), and the glycine-rich proteins (GRPs). The most well studied group are the HRGPs, and generally include extensins (Lamport, 1967) and arabinogalactan proteins (AGPs) (Clarke, Anderson, &

Stone, 1979). Extensins are some of the most well-characterized wall proteins, and are thought to be involved in self-assembled cross-linking that strengthens the wall as it extends. The mechanism for the assembly and cross-linking of HRGPs is not entirely understood, but certain aspects of their functions are. For example, extensins are thought to strengthen the cell wall by the cross linking of their tyrosine residues via peroxidase enzymes. AGPs are suspected to balance the effects of extensins, restoring plasticity to the cell wall (Lamport et al., 2011).

The last major component of the cell wall is lignin. Lignin is a phenolic polymer that contributes compressive strength and hydrophobicity to the cell wall, creating rigid, 27 waterproof cell walls (Boerjan et al., 2003). These features are necessary for maintaining vertical growth in plants, facilitating structural support and water transport. Lignin is also recalcitrant to biodegradation and highly stable, which is problematic when using cell walls for bioenergy purposes. Deposition of lignin in the wall can occur as part of natural development, but is also found in response to physical damage and pathogen invasion

(Vance, Kirk, & Sherwood, 1980).

Differences between Primary and Secondary Cell Wall

Cell wall composition can vary drastically based on the organism and cell type.

To better identify the structural and functional differences between cell walls, classifications have been developed to describe common wall characteristics. The main division between cell wall types is developmental. Early developing cell walls are called primary cell walls, and cell walls with a more advanced development are called secondary cell walls (Atalla, Hackney, Uhlin, & Thompson, 1993; Bailey, 1938).

Specific characteristics have been assigned to each classification.

Primary Cell Wall

The primary cell wall is a thin and flexible layer composed mostly of cellulose, hemicellulose, pectins, and structural proteins. Lignin is generally absent from this type of cell wall, rendering it more flexible, but less resistant to damage. The major structural backbone is a cellulose-XG framework. In arabidopsis models of Type I primary cell walls, this framework takes up about 50% of the wall by weight, embedded in a matrix of pectic polysaccharides which constitutes about 30% of the wall (Carpita & Gibeaut,

1993). This minimally rigid combination is capable of rapidly adapting to the expansion 28 and elongation cell growth in early developing tissues. Primary cell walls are also used to direct the early growth of a cell, controlling the balance between expansion and elongation. New cell wall material is laid down evenly across the entire cell wall during this time, creating a relatively uniform thickness, despite the directional nature of growth

(Setterfield & Bayley, 1961).

Secondary Cell Wall

The secondary cell wall is deposited inside the primary cell wall after cellular expansion and elongation have ceased. It is more rigid than the primary wall due to its increased thickness, and has more crosslinks than the primary wall (Carpita & Gibeaut,

1993). However, not all cells develop secondary walls. Of the three classes of plant ground tissue (parenchyma, collenchyma, and sclerenchyma), only sclerenchyma tissues develop a secondary wall (Zhong, Lee, Zhou, McCarthy, & Ye, 2008). Parenchyma and collenchyma stop developing wall tissue after the cell reaches full size, with no further thickening.

The wall composition of secondary cell walls generally differs from that of primary walls as well (Carpita & Gibeaut, 1993). In wood, pectin production in fiber cells and vessel elements stops once secondary wall biosynthesis occurs (Aspeborg et al.,

2005), and in cotton, pectin degrading enzymes are upregulated (Gou, Wang, Chen, Hu,

& Chen, 2007). Lignin production is generally upregulated secondary wall, where it is usually absent in primary wall (Donaldson, 2001). Deposition of the secondary cell wall often precedes programmed cell death (Fukuda, 1997). It should be noted that while the terms primary and secondary cell wall are broadly used, strict definitions of their 29 characteristics are not established. The biochemical characteristics laid out here are relatively common, but can vary depending on plant species. Only the most general and universal descriptions are used here unless otherwise noted.

Differences between Type I and Type II Primary Cell Wall

Primary cell walls have an additional dichotomy, as they are separated into Type I and Type II primary cell walls based on phylogeny (Carpita & Gibeaut, 1993). Both types of primary cell wall fulfill the same general function, e.g. providing a high tensile strength wall that is flexible and capable of expanding to allow growth. Both types of primary wall use the same general structure, consisting of a cellulose microfibril framework interlinked with hemicelluloses seated in a pectin matrix. The differences between the two types come at the chemical level, as each uses a specific set of polysaccharides to achieve functionality. Type I primary cell walls are found in all dicotyledons and some monocotyledons. Type II primary cell walls are found only in the monocotyledon family Poaceae, which contains grasses such as barley (Hordeum vulgare), wheat (Triticum aestivum), and false purple brome (Brachypodium distachyon)

(Carpita & Gibeaut, 1993). This distinction is important, as many cereal crops and candidates for biofuel feedstocks are grasses, while most research has been done on eudicots due to the popularity of Arabidopsis thaliana as a model organism, creating a deficit in knowledge about some of society’s most important domestic crops.

Type I Primary Cell Wall

Type I wall uses the common cellulose microfibril backbone, but differs in the hemicellulosic polysaccharides it produces to cross link the microfibrils. The main 30 hemicellulose in Type I walls is XG, which is a (14)-β-D-glucan chain with xylose

(Xyl) residues at the O-6 position of some of the glucosyl (Glc) units (Bauer, Talmadge,

Keegstra, & Albersheim, 1973; Darvill, McNeil, Darvill, & Albersheim, 1980). The Type

I walls contain roughly equal amounts of XG and cellulose (Carpita & Gibeaut, 1993).

XG also has characteristic substitutions of galactose (Gal) and arabinose (Ara) at the O-2 position of some Xyl units (Fry et al., 1993). One face of XG can hydrogen bond, mainly forming links with other XG and cellulose molecules, which allows cross linking between microfibrils to form. However, unlike cellulose, no stacking of XG chains into microfibrils can occur because only one face of the XG molecule can hydrogen bond

(Valent & Albersheim, 1974). Type I walls also contain lower levels of the other heteroxylan hemicelluloses. While celluloses and hemicelluloses make up most of the wall, structural proteins such as extensins make up the remaining 20% of the mass

(Carpita & Gibeaut, 1993).

Type II Primary Cell Wall

Type II primary cell walls also use cellulose as the major structural component, but differ in how the cellulose is held together. Instead of XG, GAX is the major hemicellulose in Type II walls. GAX is a β-(14)-D-xylan chain with Ara and GlcA substitutions. Ara will substitute at the O-3 position of Xyl, and GlcA will substitute at the O-2 position. Ara substitutions occur more often than GlcA. GAX has been detected in the Type I primary wall, but at minimal levels, and diagnostic digests indicate Type I

GAX has different substitution patterns (Darvill et al., 1980). 31

GAX’s large range of potential modifications can significantly impact wall character. The simplest forms of GAX are branched at only 10% of the backbone, while the most complex forms are branched and substituted at almost every residue (Carpita,

1983; Carpita & Gibeaut, 1993). The degree of branching changes the binding character of GAX. Unbranched GAX has enough free hydroxyl groups to hydrogen bond via its glucan backbone, like XG in Type I walls. Substitutions at the O-2 and O-3 positions of xylose backbone of GAX sterically hinder hydrogen bonding. Addition of sugar side chains at the O-2 and O-3 position also increases solubility (Carpita, 1983). A form of highly substituted GAX (HS-GAX) is defined by having substitutions at 6 out of every 7 xylose residues (Carpita, 1983). HS-GAX is commonly found during development in elongating and dividing cells. HS-GAX is gradually replaced with GAX as elongation stops (Carpita, 1984a; Gibeaut & Carpita, 1991).

Type II cell walls also differ from Type I walls in pectin content, as the quantity of pectin in Type II walls is much lower than that of Type I walls. Type II walls also show a preference for the accumulation of esterified or un-esterified pectins in different tissue types. There are even variations in character between Type II walls in different tissues. The walls in vascular tissue prefer the esterified pectins, those in cortical cells will accumulate the non-esterified form. Parenchyma cells in seedlings and coleoptiles also accumulate the non-esterified form (Knox, Linstead, King, Cooper, & Roberts,

1990). This variation in esterification may have implications in ion binding and wall cross linking, although the specific nature of these interactions is still unclear (Willats et al., 2001). 32

Unlike Type I walls, Type II cell walls are enriched in phenolic lignin precursors such as ferulate and p-coumarate. These hydroxycinnaminic acids can be used to cross link GAX polymers in the cell wall via ether linkages (Kato & Nevins, 1986). This occurs at much higher levels when secondary wall synthesis becomes active. Such linkages are recalcitrant to degradation by alkali, making these cells walls more difficult to degrade. This causes difficulty when attempting to break down Type II cell walls for bioenergy purposes. When subjected to 4M alkali treatment, 20% of the hemicellulosic materials (which are relatively easy to degrade into digestible sugars for ethanol production) remain bound to the cellulose microfibrils in Type II walls. In Type I cell walls, 4M alkali treatment removes nearly all hemicellulosic content (Carpita, 1983;

Kanabus, Bressan, & Carpita, 1986).

Cellulose and Hemicellulose Biosynthesis

Having reviewed the ultrastructure of the plant cell wall, examining how the wall is built can now be examined. The major component in wall biosynthesis is again cellulose, which is assembled by the cooperative action of CESA proteins. Cellulose synthesis in higher plants is performed at the plasma membrane by structures called

“rosettes”. Newly synthesized cellulose microfibrils can be seen extending from small, hexameric nodules on the surface of the plasma membrane via freeze-fracture electron microscopy (Kimura et al., 1999; Moor, 1971). Gene expression data indicates that CESA genes are coordinately expressed in groups of three or more during cell wall growth

(Doblin et al., 2002; Taylor et al., 2000). Expression data detailing this is available for 33 arabidopsis, barley, maize, and rice (Appenzeller et al., 2004; R. A. Burton et al., 2004;

Tanaka et al., 2003).

Different members of the CESA family are responsible for cellulose synthesis at different developmental stages. In arabidopsis, AtCESA1, AtCESA3, and AtCESA6 have been identified to be coordinately upregulated during primary cell wall synthesis, indicating that these subunits are tied to primary cell wall synthesis (Taylor et al., 2003).

In barley, HvCESA1, HvCESA2, and HvCESA6 have been shown to have similar coordinate expression. Such evidence has also been found for secondary wall biosynthesis with HvCESA4, HvCESA7, and HvCESA8 for both arabidopsis and barley

(R. A. Burton et al., 2004). Thus, the expression of specific sets of CESA isoforms for both primary and secondary wall biosynthesis seems to be common among all plants.

Of special interest is the variety of GT-2 enzymes responsible for the synthesis of cellulosic and hemicellulosic compounds. GT-2 enzymes are identified by common D,

DxD, D, and QXXRW amino acid motifs in the Carbohydrate Active Enzymes database

(Lombard, Golaconda Ramulu, Drula, Coutinho, & Henrissat, 2014), and include members of the cellulose synthase superfamily (Pear, Kawagoe, Schreckengost, Delmer,

& Stalker, 1996). Enzymes classified as CESAs are strictly responsible for cellulose synthase activity. Highly related GT-2 members, known as the cellulose-like synthases

(CSLS), are thought to synthesize related hemicellulosic polysaccharides (Richmond &

Somerville, 2000).

Synthesis of hemicellulosic compounds has been attributed to several members of the CSL family. Characterization of members of this family have been carried out in 34 several organisms, including arabidopsis, barley, rice (Oryza sativa), and maize (Zea mays). CSLA has been related to mannan synthase activity (Dhugga et al., 2004;

Liepman, Wilkerson, & Keegstra, 2005; Yin et al., 2011). CSLC has been related to the synthesis of XG backbone and its protein has been localized to the plasma membrane.

CSLC is thought to contain multiple other polysaccharide synthase activities as well, subject to what combination of CSLC subunits are combined at the time of synthesis

(Chou, Pogorelko, & Zabotina, 2012; Cocuron et al., 2007; Dwivany et al., 2009). Glucan synthase activity by CSLC proteins has been reported in arabidopsis Golgi (Chou et al.,

2012). CSLD function is unclear, though it has been suggested be involved in mannan synthase activity (Yin et al., 2011). CSLF and CSLH have been related to mixed-linkage

(13),(14) β-D-glucan (MLG) synthesis (R. a Burton et al., 2006; Taketa et al., 2012).

CSLB, CSLE, CSLG, and CSLJ have not yet been characterized (Scheller & Ulvskov,

2010).

MLG is commonly found in Type II cell walls of the grasses of the Poaceae

(Carpita & Gibeaut, 1993). It is detected in early developmental stages, and has been identified as a cell expansion related polysaccharide (Carpita, 1984b; Carpita & Gibeaut,

1993). Early cell division sees the accumulation of 5-linked arabinans, but these dissipate after cell division ends, and MLG accumulates in their place (Carpita, 1984b). MLG consists of β-(14) linked glucose, interspersed with β-(13) linkages that cause

“kinks” in the polymer. Isotope studies indicate that MLG accumulates rapidly early in cell expansion, followed by reduction in MLG as expansion finishes. Isotope studies also indicate that the turnover rate of MLG is very rapid. If the synthesis of MLG is 35 interrupted, the pool that accumulates in the peripheral cell wall is used up prematurely

(Luttenegger & Nevins, 1985). MLG synthesis is also sensitive to the growth regulator hormone auxin, indicating its involvement in development. The function of MLG is not entirely clear, but it is thought that MLG may interlock microfibrils in Type II walls, similar to the function of XG (Carpita & Gibeaut, 1993).

Regulatory Control Mechanisms for Cell Wall Biosynthesis

Transcriptional Control Network

Hormonal Controls

Regulation of plant cell wall biosynthesis (CWB) is controlled at several levels.

The broadest control of CWB seems to occur at the hormonal level, and examples of this occurring have been established for a long time. Following are several examples of such, and a summary can be found in Table 1.

Table 1

Summary table of hormone effects on cell wall biosynthesis Hormone Action

Auxin Direct induction of cell wall growth and cell elongation

Abscisic Acid Inhibition of cell wall loosening, inhibition of cell elongation and growth Gibberellins Stimulation of cell enlargement

Cytokinins Stimulation of cell enlargement

Jasmonic Acid Inconclusive. Synthesis pathway mutants inhibit the action of cellulose synthesis inhibitor DCB. Ethylene Inconclusive. Constitutively produced in AtCESA3 mutants.

The auxin class of hormones directly induce cell wall growth and cell elongation, while abscisic acid inhibits cell elongation indirectly by inhibiting cell wall loosening

(Kutschera & Schopfer, 1986). Gibberellins and cytokinins can stimulate cell enlargement, which can indirectly drive changes in the cell wall (McNeil et al., 1984).

Jasmonic acid (JA) has been implicated in more direct CWB control, as suppression of the JA pathways inhibits the function of the cellulose synthesis inhibitor 2,6- dichlorobenzonitrile (DCB). The more recently studied hormone, ethylene, may be related to regulation of CWB due to it being constitutively produced in mutants with

AtCESA3 disrupted (Ellis, 2002). This is of interest in controlling the ripening of fruit, which has agricultural implications. 37

Transcription Factor Controls

At a level below hormonal control, there are also transcriptional controls identified to more finely tune CWB regulation. Multiple networks of transcriptional control exist, each with specific transcription factors (TFs) (Figure 3) (Hussey, Mizrachi,

Creux, & Myburg, 2013). Many TFs controlling CWB contain NAC and MYB domains

(Y. Nakano, Yamaguchi, Endo, Rejab, & Ohtani, 2015; H.-Z. Wang & Dixon, 2012). The

NAC domain is short for NAM (No Apical Meristem), ATAF1/2 (Arabidopsis

Transcription Activation Factor), CUC2 (Cup-shaped Cotyledon) domain, which are independently identified, homologous transcriptional regulation domains (Lombard et al.,

2014). TFs containing the NAC domain are further subdivided by function. NAC

Secondary Wall Thickening Promoting Factor (NST) is a TF class specific to thickening secondary walls, specifically in anther endothecium. Secondary Wall Associated NAC domain protein (SND) is associated with secondary wall thickening in fiber cells (Zhong,

Richardson, & Ye, 2007). Vascular-related NAC domain (VND) protein is specific to secondary wall thickening in vascular tissues. All these NAC domain TFs have the common function of regulating secondary wall thickening. The NAC domain TFs are also commonly master regulators that activate secondary wall biosynthesis. The MYB domain TFs are usually secondary regulators that are under transcriptional control of

NAC domain proteins (Zhong, Lee, & Ye, 2010). They are named MYB after the myeloblastosis cancer tissues where they were first identified (Lombard et al., 2014). 38

Figure 3. Network visualization of hormone and transcription factor pathways. This map shows the interactions of most of the identified plant cell wall TFs, and how their layers of regulation interact. The production of these factors is tissue specific, and subject to upstream effects of hormones, miRNAs, and other TFs. Tiers 1, 2, and 3 contain mostly NAC and MYB domain TFs, which then control a broad suite of downstream cell wall biosynthesis genes. Adapted from Hussey et al. (2013). 39

Many NAC TFs have been identified in plant species as transcriptional regulators of CWB. SND1 and NST1 are redundant master regulators of secondary wall biosynthesis, and eliminating both factors stops the deposition of secondary walls in fiber and vascular tissues. An additional factor, NST2, is found in anther endothecium tissue and seems to be locally redundant to the functions of SND1 and NST1 (Mitsuda et al.,

2007; Zhong, Demura, & Ye, 2006; Zhong et al., 2007). VND6 and VND7 are both regulators of secondary wall biosynthesis in vascular tissues. Suppression of VND6 and

VND7 reduces secondary wall deposition in protoxylem and metaxylem tissues respectively (Kubo et al., 2005) . Collectively, this indicates that NAC-domain TFs have broad control over individual tissue types, but do not show much fine control of CWB.

The high number of NAC domain proteins with similar function is thought to have occurred through multiple gene duplication events. These duplications gradually evolved into genes with paralogous and orthologous function (Zhong et al., 2010). This common evolution allows the secondary wall related NACs to act as master regulators of the biosynthetic pathways for individual CWB components such as cellulose, xylans, and lignin (Zhong et al., 2008).

As noted earlier, transcriptional control of CWB is different for fibrous, vascular, and endothecium tissues. In fiber tissue, SND1 directly interacts with both MYB46 and

MYB83 (McCarthy, Zhong, & Ye, 2009). Both SND1 and NST1 directly interact with

MYB46, MYB103, SND3, and KNAT7 (Zhong et al., 2007). These factors interact with other downstream factors, including SND2, MYB69, MYB20 MYB52, MYB54,

MYB42, and MYB43 (Zhong et al., 2008). The factors MYB85, MYB58, and MYB63 40 are also downstream targets that are specific to lignin biosynthesis (Zhong et al., 2008;

Zhou et al., 2009).

In vascular tissue, the only difference is that VND6 and VND7 are the master

NAC switches for the listed lower order TFs (Kubo et al., 2005). In anther endothecium, the NAC master switches are NST1 and NST2, but they can be regulated from upstream by MYB26. The lignin specific TFs MYB58 and MYB63 have not been characterized in this tissue (McCarthy et al., 2009; Zhong et al., 2010, 2008; Zhong & Ye, 2007). TFs that directly regulate cellulose biosynthesis have also been identified in MYB46 and BES1

(Jae Heung Ko, Kim, Kim, Ahn, & Han, 2012; L. Xie, Yang, & Wang, 2011).

Post Transcriptional, Small RNA Control Networks

Regulatory control of CWB is also influenced at the transcriptional and post- transcriptional levels by small RNAs (smRNA) and RNA interference (RNAi) mechanisms. At this level of regulation, control may be achieved either via interruption of translation, or by direct, processive cleavage of the mRNA transcript. A number of smRNAs have been identified as regulatory elements with post-transcriptional and transcriptional control (Arikit, Zhai, & Meyers, 2013; Vaucheret, 2006). The exact mechanics vary, but common traits between the different mechanisms exist.

SmRNAs have been indicated as regulatory molecules in both plants and animals

(Carrington & Ambros, 2003; Yoo et al., 2004). Their functions include the regulation of tissue development, stress resistance, and phase transition, indicating smRNAs play vital roles in development (Ruiz-Ferrer & Voinnet, 2009). smRNA is a broad classification 41 including any noncoding RNA (ncRNA) 19 to 24 nucleotides (nt) in length. These smRNA are processed from longer RNA transcripts.

SmRNAs have been divided into more and more classes as different biosynthetic and functional characteristics are determined. Specific identification of smRNAs is important, as different types of smRNAs use unique pathways and have unique regulatory effects. Axtell 2013 describes the currently identified classes in detail, giving preference to biosynthetic origin and then subdividing by regulatory mechanism (Axtell,

2013). The first level of smRNA classification is by strand origin. SmRNA derived from a single strand with a hairpin fold is a hairpin RNA, and smRNA derived from two complementary RNA strands is a siRNA. The hairpin RNAs are subdivided into miRNAs and all other non-miRNA hairpin RNAs (Dunoyer et al., 2010).

SmRNA production generally starts with the transcription of long, non-protein coding RNA (lncRNA) by a DNA-dependent RNA polymerase (RNAP). There are at least five different RNAPs in plants that transcribe different RNAs depending on several factors. The major three polymerases involved in smRNA-based regulation are RNAP II,

IV, and V. RNAP II transcribes most mRNAs and miRNA precursors. RNAP IV transcribes most siRNA precursors. The plant-specific RNAP V also transcribes siRNA precursors, but for a separate function than those generally sourced from RNAP IV. Each precursor is shunted into a specific processing pathway (examined below).

Synthesis of MicroRNAs

MiRNAs are well characterized in plants and animals (Bartel, 2004; He &

Hannon, 2004), partially due to the highly specific secondary structure that pre-miRNAs 42 form prior to processing (Bartel, 2004). These sequence characteristics can be identified by computer programs, and many putative miRNA sequences have been identified by algorithmic searches of deep sequencing data. The function of miRNAs in plants and animals is similar. Both are generally 21-22 nt in length, but plant miRNAs require higher sequence complementarity to function (Rhoades et al., 2002), especially in bases

2-13. Only a single incorrect base may be tolerated in this region. No more than 5 mismatches are acceptable between the miRNA and its target. These sites are generally found at the terminal ends of mRNAs (Mallory et al., 2004; Schwab et al., 2005). This generally causes miRNAs to have far more specific targeting than siRNAs. A majority of plant miRNAs require Dicer-like (DCL) and Argonaute (AGO) family proteins for their production and functions, although there are exceptions (Axtell, 2013).

The transcriptional complex for RNAP II (Pol II in Figure 4) transcribes lncRNAs precursors which are capped, polyadenylated, and spliced similarly to mRNAs. This precursor is called a primary-miRNA (pri-miRNA), and is subjected to several processing steps required to generate a mature miRNA (Figure 4). Pri-miRNAs spontaneously fold into a hairpin RNA structure of approximately 70 bases in length, which are recognizable to the miRNA processing components that form a complex including binding proteins

(SERRATE (SE), HYPONASTIC LEAVES1 (HYL1), DAWDLE (DDL)), RNA cleavage proteins (Dicer-Like 1 (DCL1), and other cofactors (TOUGH (TGH), CAP-

BINDING PROTEINS (CBP20/CBP80), and NOT2a/b). 43

Figure 4. Consolidated miRNA processing mechanism. Adapted from Rogers & Chen 2013. 44

The hairpin is thought to be stabilized the stem-loop complex (Figure 4) (Rogers

& Chen, 2013; Bin Yu et al., 2008). Two of the complex members (HYL1 and SE) are a double-stranded RNA (dsRNA) binding protein and a zinc-finger domain protein, and together are thought to promote accurate miRNA processing (Dong, Han, & Fedoroff,

2008; Han, Goud, Song, & Fedoroff, 2004; Y Kurihara, Takashi, & Watanabe, 2006;

Lobbes, Rallapalli, Schmidt, Martin, & Clarke, 2006; Vazquez, Gasciolli, Crété, &

Vaucheret, 2004; L. Yang, Liu, Lu, Dong, & Huang, 2006).

Once the pri-miRNA hairpin is stabilized, the RNase III family endoribonuclease

DCL1 repeatedly travels over the stem-loop, cutting out the 21 base RNA duplex

(Henderson et al., 2006; Yukio Kurihara & Watanabe, 2004; Park, Li, Song, Messing, &

Chen, 2002). This duplex has a 2-base overhang on each 3’ end, making it vulnerable to degradation by the SMALL RNA-DEGRADING NUCLEASE (SDN) family. The protein HUA ENHANCER 1 (HEN1) impedes this degradation by attaching a methyl group to the 2-O of each 3’ end (J. Li, Yang, Yu, Liu, & Chen, 2005; Z. Yang, Ebright,

Yu, & Chen, 2006; B Yu et al., 2005). This prevents uridylation of the 3’ end, which is the targeting mechanism for SDN degradation (Ramachandran & Chen, 2008). The duplex is then integrated into other AGO complexes (with Heat-Shock Protein 90

(HSP90)/SQUINT (SQT)), likely in the cytoplasm (Czech & Hannon, 2011), which selectively picks one strand of the miRNA to retain for targeting and mRNA cleavage.

Synthesis of Small Interfering RNAs

While miRNAs have been well characterized, siRNAs are considerably less so, likely due to the myriad ways siRNAs can be produced and function. The siRNAs can be 45 subdivided into heterochromatic siRNA (hc-siRNA), secondary siRNAs, and natural antisense transcript siRNAs (NAT-siRNA). Further subdivisions are also noted based on unique functional character. For example, a more recently identified class of siRNA termed diRNA are thought to be involved in mediation of double strand break induced

DNA repair (Wei et al., 2012).

All known smRNA originates from processing of longer dsRNA precursors into short fragments by class III endoribonucleases such as DCL (Figure 5) (Arikit et al.,

2013). However, there are several identified sources of these dsRNA. One of the first mechanisms was miRNA directed cleavage of TAS transcripts (Howell et al., 2007). TAS are primary transcripts slated for tasiRNA production, transcribed by RNAP II, and are cleaved once by specific miRNA targeting to create free 5’ and 3’ ends. The cleaved products are used as substrate by RDR6 to make dsRNA, which is then processively cleaved into siRNA by DCL4.

Aside from TAS based siRNA generation, specific sources of siRNAs have not been identified, but it is commonly accepted that most double stranded RNA could be converted into siRNA. This could occur in several ways. Formation of a (long) RNA hairpin due to base complementarity of inverted repeats has been shown to produce small

RNA populations via DCL2, 3, and 4 (Dunoyer et al., 2010). Overlapping complementary RNAs have also been shown to produce dsRNA used to produce siRNA in response to salt stress (Borsani, Zhu, Verslues, Sunkar, & Zhu, 2005). Based on this information, it seems possible that any mechanism resulting in a stable dsRNA could produce siRNAs. Once the dsRNA has been produced, DCL activity cleaves dsRNA into 46 fragments in a sequential manner, creating short, dsRNA duplexes which are selectively separated and used as siRNA.

Figure 5. General mechanism of siRNA Formation from Double Stranded RNA. © User: Singh135 Wikimedia Commons/CC-BY-SA-4.0

Mechanisms of miRNA Silencing

Plant miRNAs are capable triggering silencing through multiple mechanisms, all of which are based on miRNA targeting. The most common mechanism is direct degradation of target RNA by the “Slicer” activity of AGO (Axtell, 2013; Baumberger &

Baulcombe, 2005). Slicer activity is defined as RNA cleavage that leaves behind an RNA 47 fragment with a 5’ phosphate and a 3’ OH group (Hammond, 2005). AGO finds specific slicing targets by carrying a miRNA fragment, which recognizes mRNA targets via

Watson-Crick base pairing. Once a complementary strand is found, the Slicer activity of

AGO takes over. The degraded mRNA product of this reaction can no longer be translated, reducing the pool of translatable transcripts.

In a similar manner, miRNA targeting can also cause translational suppression without transcript cleavage. If the miRNA is not a perfect match for the mRNA target,

AGO may not cleave, but its presence on the transcript prevents the action of the ribosome. The degree of specificity required for cleavage versus translational suppression is currently unclear. While plant miRNAs were originally thought to require near perfect complementarity for cleavage, more recent work has discovered a number of miRNAs that cleave despite breaking the canonical pairing rules (F. Wang, Polydore, & Axtell,

2015). This implies that there may be other factors involved in this decision.

Regulation further upstream is accomplished by miRNA-based heterochromatin regulation, which can cause suppression at the transcriptional level (M. A. Matzke &

Mosher, 2014). For example, in rice, a class of long miRNAs produced by an alternative

DCL3-AGO4 pathway will direct DNA methylation to occur at their production location.

These miRNA were also shown to act in trans on other similar target locations, as well as in cis (Wu et al., 2010).

Post-transcriptional miRNA silencing can also trigger cascades of other smRNA silencing types. This occurs via the production of secondary siRNAs, initiated by miRNA cleavage of its primary target. Cleaved miRNA targets can be processively degraded into 48 siRNAs, creating cascades of regulation at the post-transcriptional level (Manavella,

Koenig, & Weigel, 2012). These siRNAs are also referred to as trans acting siRNAs

(tasiRNAs) (Vazquez, 2006) or phased secondary siRNA (phasiRNA) (Fei, Xia, &

Meyers, 2013), and will be described in more detail below.

Mechanisms of siRNA Silencing

SiRNAs are like miRNA in structure and function, but their production methods are less rigid. This gives rise to a wider possible spread of targets for siRNAs. The sources of siRNA are also less well understood than those of miRNAs, in part due to their variety. While miRNAs all transcriptional sources creating cleavable hairpins, siRNAs seem only to require creation of a dsRNA for their synthesis to occur. However, like the mechanics of miRNA silencing, siRNAs can affect gene silencing at multiple control points through the actions of AGO, DCL, and RNA dependent RNA polymerases

(RDRs).

At the transcriptional level, the hc-siRNAs regulate the heterochromatin/ euchromatin balance (Chapman & Carrington, 2007; Chellappan et al., 2010; M. Matzke,

Kanno, Daxinger, Huettel, & Matzke, 2009; Simon & Meyers, 2011). The source transcripts for these siRNA are usually sequence repeats that result in slightly longer 23-

24 nt fragments, and are transcribed by RNAP IV or V. The transcripts are then processively cleaved by DCL4 and loaded into AGO4. At this point the mechanism becomes unclear (F. Wang et al., 2015). These hc-siRNAs are commonly associated with the de novo deposition of heterochromatin (Cao et al., 2000; Law & Jacobsen, 2010), leading to the suppression of transcription and gene silencing. Full characterization of 49 how this all occurs has been hindered by the embryo lethality introduced by mutation of genes involved in heterochromatin regulation.

At the post transcriptional level, siRNAs function to selectively attenuate steady state transcript levels in the cell (Axtell, 2013; Czech & Hannon, 2011; Held et al., 2008).

This occurs similarly to the mechanism for miRNA, as siRNAs duplexes are picked up by AGO4. These duplexes consist of a guide strand and a passenger strand (Czech &

Hannon, 2011). Only the guide strand of a dsRNA duplex can be picked up by the AGO protein, which is part of the RNA induced silencing complex (RISC). The AGO component of RISC uses the base pairing of the fragment to target specific single- stranded RNAs for degradation or disruption. The AGO subunit can have multiple activities like in miRNA silencing, including direct single-stranded RNA degradation

(Slicer activity) and interference in translation (Brodersen et al., 2008; Czech & Hannon,

2011). The variety of available smRNA gives AGO and RISC a broad range of sequence- specific target control (Axtell, 2013).

The Slicer activity of RISC can also give rise to further RNA-based (secondary) silencing. AGO degradation products are of ideal size for incorporation into new AGO-

RNA complexes. This systemic response can cause large amounts of silencing, both in cis and in trans, as silencing can propagate for as long as the target transcript is still present to be degraded into fresh smRNA (Manavella et al., 2012). This RNA-induced regulatory mechanism can create a cascading effect, drastically increasing the effect of a small number of inducing RNA molecules (Figure 6) (Axtell, 2013). 50

Figure 6. Generalized mechanism for producing dsRNA via RDR activity. Sliced, but incompletely degraded mRNAs can be targeted by RDRs to synthesize complementary RNA strands, producing dsRNA. This dsRNA is targeted by RISC degradation, and the resultant siRNAs are used to target additional related family members. A small number of initial targets can create a silencing cascade targeting close and distantly related family members. Adapted from Axtell 2013.

Secondary siRNAs and Silencing Propagation

Secondary siRNAs are produced by the degradation of primary RNA targets of miRNAs or siRNAs. Primary RNA (either single or double stranded) targets are cut into 51 small (20 to 24 depending on the DCL) nucleotide fragments by DCLs. These siRNA fragments are paired with AGO proteins to assemble RISCs (Manavella et al., 2012), which will cut any single stranded RNA of complementary sequence. The precise sequence complementarity required for this is unclear. It has been previously posited that while miRNAs required near-perfect base pairing, siRNAs may have less strict requirements. However, a detailed analysis of targeting requirements for siRNAs has not been completed at this time.

From this point, two possibilities arise. The first is that the RISC processively degrades the target of the secondary siRNA producing additional secondary siRNAs through sequential degradation of the single stranded RNA (Allen, Xie, Gustafson, &

Carrington, 2005; Montgomery et al., 2008). Another possibility is that the target is cleaved once, making it a target for RDR to act as a template for dsRNA synthesis and processive cleavage by DCLs. The total number of siRNAs increases in this way.

Secondary siRNAs produced from upstream/downstream sequence in an RNA strand can go on to target different regions than the initial silencing complex (Axtell, 2013). This allows secondary siRNAs to form a regulatory cascade, as the number of potential target sequences increases (Chen, Li, & Wu, 2007). Secondary siRNAs can be further used by

AGO proteins to form silencing complexes and perpetuate the silencing activity for as long as RNA sequences complementary to the siRNA are present (Chapman &

Carrington, 2007).

PhasiRNAs and TasiRNAs Further subdivisions of secondary siRNAs can be identified. Secondary siRNAs are separated into the phased siRNAs (phasiRNA) and the trans-acting siRNAs

(tasiRNA) (Axtell, 2013). PhasiRNA are initiated by a siRNA molecule binding to a complementary transcript (Zhai et al., 2011). A RDR is recruited and synthesizes the complementary RNA strand (Figure 6). This strand is then degraded by Dicer activity and the phasiRNA product perpetuates itself, similar to secondary siRNA (Axtell, 2013).

TasiRNAs are siRNAs that are specifically produced from dsRNA that forms between two complementary sequences transcribed from completely different loci.

NAT-siRNAs

A specialized class of siRNAs that are of significant interest are the NAT (natural antisense transcript)-siRNAs (Borsani et al., 2005; Held et al., 2008). NAT-siRNAs are siRNAs produced by the expression of transcript from both DNA strands in the same region. In general, this occurs when two genes are positioned on opposite DNA strands with overlapping sequence between them. Transcription of an mRNA from DNA uses the coding strand, here indicated as the “sense” strand, is complemented by concurrent transcription occurring on the “antisense” strand. The antisense strand is in turn labeled the NAT, and the overlapping region between the two transcripts can spontaneously hybridize. This forms a dsRNA hybrid which can be targeted by DCLs for siRNA synthesis.

Antisense RNAs produced at the same locus that they target are called cis NATs.

A second class of NATs termed trans-NATs have been predicted bioinformatically in mouse oocytes, but have not been characterized experimentally (T. Watanabe et al., 53

2008). No research into this class of NATs has been published for plants currently.

Hypothetical trans-NATs are transcribed from a different locus than the mRNA that they target, but still perform the same hybridization with the target, resulting in the same siRNA effects. It is important to note that these hypothetical trans NATs are assumed to be transcribed, potentially processed, then hybridized to their complementary mRNAs similarly to cis-NATs. This is different from tasiRNAs which are much shorter, and are derived from an RNA source instead of a DNA source, although their RNA degradation results may be similar.

Antisense RNA and siRNA Regulation of CESA Genes

Inquiry into antisense transcript and siRNA based regulatory control of CESAs and cellulose biosynthesis was first examined in Held et al. (2008). This paper provided the first evidence of endogenous NATs in barley, specifically NATs that targeted members of the cellulose synthase superfamily. Quantitative PCR analysis of a time course of developing barley tissue (cv. Black Hulless) showed that the expression of genes in the cellulose synthase superfamily was differentially regulated. This expression pattern was followed by the differential deposition of cellulose (HvCESA) and MLG

(HvCSLF) in the barley tissue. Early in the time course, relative quantities of cellulose peaked in the tissue. Later in the time course, MLG had a larger relative percentage of cell wall mass. However, expression of HvCESA6 and HvCSLF genes peak at the same time. This indicates that there was an unknown post-transcriptional regulatory mechanism changing the relative expression levels of HvCESA6 and HvCSLF. This directly affected cell wall composition in the developing barley tissue. 54

To determine the nature of this regulatory mechanism, several experiments were used to identify possible post-transcriptional regulatory components. Strand specific reverse-transcriptase polymerase chain reaction (SS-RT-PCR) was employed to show that HvCESA6 antisense transcripts were present in developing barley leaves. SS-RT-

PCR also indicated that the NATs were not full-length complements of the mRNA from the same locus. Evidence of smRNAs was found via ribonuclease protection assay

(RNP), showing that 21 and 24 nucleotide fragments were present during development.

The relative quantities of these smRNAs increased over the course of leaf development with increasing quantities of HvCESA6 antisense RNAs, indicating a correlation between the two. Semi-quantitative RT-PCR analysis of both the HvCESA6 derived NATs and ribonuclease protection assays of NAT-siRNA indicated a peak in NATs directly before a peak in NAT-siRNA. Both peaks also occur during the period when cellulose content in the cell wall is diminished and MLG content is peaking. Collectively, this experimental evidence indicates that cis acting NATs are transcribed from at least one HvCESA locus, and can trigger NAT-siRNA production that post-transcriptionally silences gene expression in an endogenous system.

To test the effects of manipulating an endogenous NAT-siRNA system, virus- induced gene silencing (VIGS) was employed. A set of sequences common to all members of the HvCESA gene family was cloned into a barley stripe mosaic virus system shown to initiate VIGS in barley (Holzberg, Brosio, Gross, & Pogue, 2002; Scofield,

Huang, Brandt, & Gill, 2005). Developing barley was infected with this vector, and qPCR analysis of the RNA collected from the third leaves of plants showing the 55 photobleaching effects of successful VIGS treatment was done. Relative expression levels indicated that all members of the HvCESA family were partially silenced, as well as other members of the cellulose synthase superfamily. This included HvCSLF1,

HvCSLH, HvCSLC, HvCSLA, and distantly related glycosyltransferases in the GT8 family. The silencing was not uniform, with different targets being silenced to differing degrees. This indicated that the silencing of a single gene family could have far reaching effects, similar to those observed in the phasiRNA and tasiRNA systems detailed above.

Therefore, coordinate regulation of a whole network of genes is possible through silencing of a small number of initiation points via the generation of CESA-derived small

RNAs.

RDRs and Dicer-Like Endonucleases

While the work in Held 2008 believed the antisense transcripts they detected were

NATs transcribed from a potentially cryptic DNA promoter, they did raise the point that another possible source of such antisense transcripts would be an enzyme capable of using a mRNA template to transcribe the antisense strand. Several enzymes with that activity are known, and have been established here as the RDRs. In the arabidopsis model, there are three functionally annotated RDR proteins, labeled RDR1, RDR2, and

RDR6. Each RDR has been partially characterized, although there does seem to be some overlap in activities between them that makes that difficult.

RDR1 is largely thought to be involved in plant pathogen defense and production of viral siRNAs, and does not appear to have significant involvement in endogenous smRNA regulation (D. Yu, Fan, MacFarlane, & Chen, 2003). Mutants of RDR1 in 56 tobacco are deficient in systemic RNA silencing defense against Potato Virus X, indicating a deficiency in siRNA production (Z. Xie, Fan, Chen, & Chen, 2001).

Transcripts produced by RDR1 are generally cleaved by DCL4. While RDR1 is associated with viral defense, RDR2 and RDR6 seem to be far more involved in endogenous regulation of the plant.

RDR2 generally synthesizes dsRNAs of single stranded RNAs (ssRNAs) transcribed by RNAP IV, and that dsRNA is most often cleaved by DCL3. This preferentially leaves 24 nucleotide siRNAs (Baulcombe, 2004) that are targeted to heterochromatin, known as hc-siRNAs. The hc-siRNA products of RDR2 are largely involved in translational gene silencing and transitivity of chromatin silencing via DNA methylation and histone modifications (Chan et al., 2004). RDR2 has been shown to initiate second strand synthesis in vitro by both primed and unprimed mechanisms, indicating that RDR can be guided by siRNAs, but that it is not necessary for function

(Devert et al., 2015).

RDR6 may be involved in the widest range of functions among the RDRs. It is known to be involved in viral siRNA biogenesis, as well as production of several classes of endogenous siRNAs involved in post transcriptional gene silencing (PTGS). The source transcripts for RDR6 activity are initially transcribed by RNAP II (Allen et al.,

2005), although it may act on transcripts from other sources as well. The dsRNA products of RDR6 are generally cleaved by DCL4 or DCL2 to make 21 or 22 nucleotide siRNAs.

This system is known to be involved in the production of NAT-siRNAs involved in salt stress response (Borsani et al., 2005). Like RDR2, RDR6 has been shown to initiate 57 second strand synthesis by both primed and unprimed mechanisms in vitro, although it seems to be more sensitive to the reaction environment. While it may be guided by siRNAs, it appeared less likely to do so than RDR2 (Devert et al., 2015).

A BLAST search of the barley genome (Altschul, Gish, Miller, Myers, & Lipman,

1990; Mayer et al., 2012) shows multiple RDR orthologs present in barley. Assuming these orthologs have conserved function, it may be possible that the NATs previously identified for HvCESA6 were generated by RDRs instead.

It is also of note that recent developments in the area of mRNA turnover have identified that there seems to be crossover with production of antisense and siRNA.

Canonically, the lifetime of a mRNA is dictated by the gradual degradation of its polyadenyl tail. Eventually the tail becomes short enough to trigger the activity of a decapping complex that removes the protective 7-methylguanosine cap and allows the mRNA to undergo cleavage and degradation (Liu & Chen, 2016). However, mutants of the XRN4/EIN5 ribonuclease show a strong upregulation in the production of siRNAs

(Gregory et al., 2008). Given that siRNAs generally are produced by the action of RDR6 and DCLs 2 and 4, it seems likely that the deadenylated, decapped mRNAs from the turnover pathway can instead be targeted by RDR6 for dsRNA production and cleavage into siRNAs (Figure 7). The exact extent of this crossover has yet to be elucidated

(Tsuzuki, Motomura, Kumakura, & Takeda, 2017).

Figure 7. Potential crossover point between nonsense mediated mRNA decay and siRNA synthesis pathways. Deadenylated mRNAs are targeted by the decapping complex, which acts to remove the protective 5’ cap from the mRNA. This opens the RNA to attack by exoribonucleases (XRN) that processively cleave the transcript, but this species is also a substrate for RNA-dependent RNA-polymerases (RDRs), which can attach and synthesize a complementary RNA strand, targeting the new dsRNA for degradation and siRNA production.

Experimental Aims

The working hypothesis for this project is that cellulose synthase genes are regulated in part by small RNAs derived from endogenous antisense transcripts. The literature suggests that antisense transcripts and small RNAs for CESAs like those found for HvCESA6 can modify the expression of cell wall producing genes via post- 59 transcriptional mechanisms. This work looks to expand evidence for this hypothesis, and the following specific aims are covered.

 Specific Aim 1. Identify if there are endogenous antisense transcripts and small

RNAs for barley CESAs beyond HvCESA6. While antisense transcripts and small

RNAs have been identified for HvCESA6, it is not known whether this occurs for

other members of the HvCESA gene family. Here, the developing barley third leaf

model will be used to assay for antisense transcripts in each HvCESA gene family

member. If such antisense transcripts are found, assessment corresponding small

RNAs and sense transcripts will be made. A comparison between the behaviors of

sense transcripts, antisense transcripts, and small RNAs would allow verification

of whether the antisense phenomena for HvCESA6 were an isolated event. If not,

this could indicate conservation of a cell wall biosynthesis regulatory mechanism.

 Specific Aim 2. Identify if there are antisense transcripts and small RNAs for

non-barley monocot CESAs. Examination of CESA antisense transcripts and

small RNAs in barley should be expanded to other organisms to determine if

close phylogenetic conservation of these species is occurring. A closely related

monocot grass Brachypodium distachyon will be used to survey for CESA

antisense transcripts, following previously established methods.

 Specific Aim 3. Identify if there are CESA antisense transcripts in non-monocot

plants. If CESA antisense transcripts and small RNAs are shown to be conserved

across both barley and brachypodium, then examination of this model will be

expanded to include a eudicot system. An optimal model would be Arabidopsis 60 thaliana, which large established bodies of literature, expression data, and mutant libraries. Identification of CESA antisense transcripts and small RNAs in arabidopsis would indicate broad phylogenetic conservation of these species is occurring.

CHAPTER 3: DETECTION OF CESA ANTISENSE TRANSCRIPTS AND SMALL

RNAS IN THE GRASS SPECIES HORDEUM VULGARE

Introduction

Given the tight regulation of cellulose synthases and their associated networked genes, understanding the mechanisms behind the up and down-regulation of CESA family members is an important avenue of research (McNeil et al., 1984). Efforts in that area have revealed extensive evidence for how these gene networks are triggered and upregulated, primarily by the actions of hormones and TFs. Auxin (Kutschera &

Schopfer, 1986), jasmonic acid (Ellis, 2002), abscisic acid (Kutschera & Schopfer, 1986), and other hormones (Clouse & Sasse, 1998) have been tied to changes in cell wall structure in response to germination (Taiz & Jones, 1970), disease response (Bari &

Jones, 2009; Vorwerk, Somerville, & Somerville, 2004), damage response (León, Rojo,

& Sánchez-Serrano, 2001), and senescence (Bleeckerl & Patterson, 1997). TFs with NAC and MYB motifs have been identified as triggers for primary and secondary-wall biosynthesis in response to the same biotic and abiotic stresses (Grant, Fujino, Beers, &

Brunner, 2010; J. H. Ko, Jeon, Kim, Kim, & Han, 2014; Jae Heung Ko et al., 2012; Kou et al., 2014; Mitsuda et al., 2007). However, these phenomena are only well understood in how they activate and upregulate expression of cell wall synthesizing components. The corresponding mechanisms that would selectively downregulate or turn off the same gene networks are still largely unidentified (H.-Z. Wang & Dixon, 2012).

Without evidence of a negative regulator from hormonal or TF studies, the search has moved further down the path from transcriptional regulation to post-transcriptional 62 regulation. At the RNA level, a wealth of evidence exists for mechanisms that can specifically attenuate and downregulate gene expression (Bologna & Voinnet, 2014;

Carthew & Sontheimer, 2009; Gregory et al., 2008; Pumplin & Voinnet, 2013; Voinnet,

2005). While most of these processes were identified from work on disease and viral defense, it has been shown that those models could be repurposed for regulation of endogenous gene expression (Held et al., 2008; Vazquez, Vaucheret, et al., 2004).

Central to these pathways are the mechanisms for RNA-induced gene silencing (RISC), which specifically cleaves RNA transcripts based on sequence similarity with 20 to 24 nucleotide small RNAs (smRNAs) (Baumberger & Baulcombe, 2005; Czech & Hannon,

2011).

Previous work showing the downregulation of endogenous, CESA-centric gene networks via RISC mechanisms has been done utilizing virus-induced gene silencing

(VIGS) in barley (Hordeum vulgare) (Held et al., 2008). A silencing construct specifically targeting the HvCESA family resulted in widespread downregulation of cell wall synthesis related transcripts, even for genes with minimal sequence similarity to the original VIGS target. The same VIGS treatment increased accumulation of endogenous,

CESA-specific, 21-nt smRNAs that are characteristic of the small interfering RNAs

(siRNAs) commonly found during RISC degradation of mRNA targets. Additionally, a developmental time course of the same smRNAs showed staggered correlation with antisense transcripts complementary to the HvCESA6 gene, and antisense correlation with the sense HvCESA6 transcript, suggesting specific endogenous smRNAs and antisense transcripts function together to downregulate HvCESA6. 63

Here, I expand on the initial evidence gathered for an antisense and smRNA- based regulatory mechanism in the CESA gene family. I found selective expression of antisense transcripts for additional genes in the HvCESA gene family. A developmental time course of one of these HvCESA antisense transcripts and it’s corresponding smRNAs over time also indicated a corollary relationship. This all complements a custom microarray that indicates close and distant targeting of cell-wall-related genes moderated by smRNA mechanisms. My results add to prior data indicating that specific members of the CESA gene family can be targeted and negatively regulated by endogenous RNA silencing mechanisms.

Materials and Methods

Construction and Use of BSMV Vectors

Construction of the BSMV vectors and rub inoculation of the barley plants was carried out as described previously (Held et al., 2008).

Viral Inoculation of Barley Plants

Plant inoculations were carried out as described previously. In brief, in vitro transcripts of the pBSMVα, pBSMVβ, pBSMVγ-PDSas, and the relevant gene silencing construct (pBSMVγ for empty vector control (Holzberg et al., 2002), and pBSMVγ-

HvCESA-CR2) were combined in a 1.0:1.0:0.5:0.5 ratio by volume, respectively. For each plant, 3 µl of construct mixture was combined with 22.5 µl of FES in a master mix, and applied to 6-7 day old seedlings by gentle rub-inoculation of the first leaf, repeating the rub up to 3 total times per plant to get full leaf coverage. The pBSMVγ-PDSas transcripts co-inoculated with the silencing targets served as visual indicators of silencing 64 progression. Infected plants were covered with a plastic dome for 24 hours after inoculation, then uncovered and grown as described until the white patches indicative of viral silencing were evident in the third leaves, generally on days 13-15.

Harvest of VIGS Tissue and Preparation of RNA

Third-leaf tissues from plants visibly demonstrating photobleaching were harvested 7 to 13 days after inoculation, with a typical plant showing maximal photobleaching at 8 days after inoculation. Senescent tissue was trimmed from the leaf tip if present, followed by snap-freezing in liquid nitrogen. Snap-frozen VIGS infected tissues were pulverized using a mortar and pestle under liquid nitrogen, and then combined with TRIzol® reagent (Invitrogen, Carlsbad CA). RNA was then prepared per the TRIzol® protocol.

qPCR Verification of HvCESA Silencing

QPCR was performed using total RNA collected from VIGS and control tissues.

A 1 µg aliquot of each RNA sample was used to synthesize cDNA using the Qiagen

Quantitect Reverse Transcription Kit (Qiagen, Hilden, Germany) per the manufacturer’s instructions with an extended reverse transcription time. cDNAs were diluted 1:5 by volume with sterile water and then assayed for the expression of HvCESA1 as a marker for HvCESA family silencing using the primers from Burton 2004 (R. A. Burton et al.,

2004) and reference gene HvUBI with primers from Held 2008 (Held et al., 2008).

Reactions (20 l) were performed using the Qiagen Rotor-Gene SYBR Green PCR Kit, on a Rotor Gene Q platform, per the manufacturer’s instructions. Cycle threshold (Ct) values were identified with the Rotor Gene Q software, Version 9, and used to calculate 65 relative expression for each sample using the Pfaffl method (Pfaffl, 2001) to account for amplification efficiency.

Construction of Custom Microarray

A custom, single-channel, Agilent (Santa Clara, CA) microarray based on the

8x16K architecture was designed to identify genes regulated in response to cellulose synthase silencing, enriched in sequences involved in cell wall biosynthesis, stress response, and RNA regulation. Each slide contains 8 arrays, with approximately 16K probes per array. A total of 3778 60-mer probes were selected from a list of candidate genes by the Agilent eArray service, with four technical replications of each probe per array.

Microarray Hybridization and Data Extraction

VIGS-treated barley RNA was supplied to the Ohio University Genomics

Facility, verified for quality by a Bioanalyzer 2100 instrument and hybridized to the custom 8X15K microarray per the manufacturer’s protocol (Agilent). Sixteen total samples were hybridized, one per array, with 6 BSMV-EV treated samples (negative control) and 10 BSMV-CR2 treated samples. Hybridized arrays were imaged with an

Agilent Technologies Scanner G2505B, and signals were extracted using the Agilent

Feature Extraction Tool (Version 10.7.3.1 using protocol GE1_107_Sep09).

Processing of Microarray Data

Extracted microarray data was processed using the limma package from

Bioconductor. Backgrounds were corrected using the normexp method with a +50 offset

(Ritchie et al., 2007). Arrays were normalized between each other using the quantile 66 method. All signals within 110% of the 95th percentile of the negative controls for 6 or more arrays were ignored. Signals from replicate probes for each array were then averaged and used to identify differentially expressed genes (adjusted p < 0.05).

Design of Gene-Specific Primers for Tagged-SS-RT-PCR of HvCESAs

Gene-specific primers for HvCESA gene family antisense transcript detection were designed using the OligoAnalyzer 3.1 software (orthologous to Held et al. (2008)).

Primers were BLASTed against the NCBI barley transcript library, and then aligned against every member of the HvCESA gene family to ensure specificity. To improve PCR specificity and eliminate the potential for artefacts and off-target, sense-derived transcripts, a tag was added to the 5’ end of each gene-specific cDNA synthesis primer

(Craggs, Ball, Thomson, Irving, & Grabowska, 2001).

Preparation of Barley Tissue

Seeds of Hordeum vulgare cv. Black Hulless were imbibed in aerated water for

24 hours to stimulate germination. Imbibed seeds were transferred to moist vermiculite and placed in the dark at 28°C until hypocotyls emerged, generally 3-5 days. Seedlings were then transferred to autoclaved soil (Promix BX) supplemented with Osmocote

(Scotts) 14-14-14 slow release fertilizer (1.8 g/L). Seedlings were grown in a Percival

E36HOX growth chamber under high intensity fluorescent lamps (450-700 μmol m-2 sec-

1) programmed for a 16-hour photoperiod (25 °C day, 20°C night). Third-leaf tissue from

≥ 3 plants was excised, measured for length, and pooled in liquid nitrogen at daily intervals, 10-16 days post imbibition. 67

Preparation of Barley RNA

Pooled third-leaf samples were pulverized using a mortar and pestle under liquid nitrogen, and then homogenized under TRIzol® reagent using a glass-glass homogenizer

(Steadystir analog). RNA was then prepared per the TRIzol® protocol for day 10-16 in parallel. Aliquots of each RNA sample were treated for DNA contamination using the

TURBO DNA-free KIT (Invitrogen AM1907) per the manufacturer’s instructions for rigorous treatment. 3 μl of DNase was added in two aliquots, the first at the beginning of the incubation, and the second halfway through. 500 ng of each RNA sample was separated on a 0.7% agarose gel and visualized with ethidium bromide dye to check for

RNA degradation.

Preparation of HvCESA Antisense cDNA

First strand cDNAs for antisense transcripts of HvCESAs 1, 2, 4, 5/7, 6, and 8 were synthesized from 1.7 μg of DNase-treated total RNA extracted from barley leaves,

(13 days post imbibition) using the SuperScript III First-Strand Synthesis System

(Invitrogen 18080-051), using tagged gene-specific primers. Control cDNAs were prepared as follows; Oligo-dT-primed (OdT) cDNA; No primer control (NPC) cDNA with the primer replaced with nuclease free water; No reverse transcriptase control (NRT) cDNA with the RT enzyme replaced with nuclease free water. cDNA reactions were then treated with RNase H to remove residual complementary RNA per the manufacturer’s protocol, and then diluted in a 1:9 ratio of cDNA with nuclease free water. 68

Amplification of HvCESA Antisense cDNA

First-strand cDNAs synthesized for each HvCESA antisense transcript were amplified by PCR using the corresponding sense primer and the tag2 primer. Control cDNA samples were amplified in the same manner. OdT primed cDNA was also amplified individually with each pair of HvCESA sense and antisense primers (no tag) as control for amplicon size and sense RNA presence. All PCR amplifications were assembled on ice in 25 μl reactions using 5 μl of 5X Green GoTaq buffer (Promega

M3001), 0.5 μl of each primer (10 μM), 0.5 μl of dNTPs (10 μM each), 4 µl of diluted cDNA template, and 1.25 units of GoTaq polymerase.

Cycling conditions for reactions were optimized for melting temperature and extension time. All reactions were cycled with 2 minutes of activation at 95°C, followed by 35 cycles of 95°C for 1 min, optimized annealing temperature for 1 min, and 72°C for the optimized extension time. Final elongation was 72°C for 5 minutes. ≥ 3 technical replicates of each reaction were amplified for each antisense cDNA sample.

Characterization of HvCESA Amplicons

Equal volumes of each PCR product for each sample and control reaction were separated by agarose gel electrophoresis. Gels were imaged by a Chemidoc EQ camera using Quantity One software (Version 4.5.2 Build 070). Gel images were analyzed using

ImageJ (Version 1.49E).

Antisense amplification products were excised and purified with the Zymoclean

Gel DNA Recovery Kit (Zymo), and cloned into the pGEM T-Easy vector kit (Promega).

Clones were fully sequenced and confirmed as the targeted sequence. Inclusion of the 69 tag1 sequence confirmed that cDNA samples synthesized from the antisense-tag1 or primer were unique from cDNA samples synthesized from the OdT primer.

RNA Loading Control for HvCESA1 Antisense Time Course

500 ng of DNase-Treated total RNA (based on spectrophotometrically measured concentrations) for all time points were separated on a 1% agarose gel to check for degradation and equal RNA loading. Gels were imaged by a Chemidoc EQ camera using

Quantity One software (Version 4.5.2 Build 070). Gel images were analyzed using

ImageJ (Version 1.49E). All time course measurements were normalized to the RNA loading.

Preparation of HvCESA1 Antisense Time Course cDNA

First strand cDNA for each time point was per the methods for HvCESA1 used in the initial detection survey (described above in Amplification of HvCESA Antisense cDNA).

Amplification of HvCESA1 Antisense Time Course cDNA

First-strand cDNAs synthesized using the HvA1-antisense-tag1 were used as templates for PCR following the same assembly as the initial detection survey. Cycling conditions for reactions using A1-sense primer and tag1 primer included 2 min of activation at 95°C, followed by 34 cycles of 95°C for 1 min, 60°C for 1 min, and 72°C for 45 sec. Final elongation was 72°C for 5 minutes. Antisense transcript cycling conditions were optimized to terminate amplifications during the mid/late-log phase so that semi-quantitative densitometry could be performed. Cycling conditions for reactions using HvA1-sense primer and HvA1-antisense primer included 2 min of activation at 70

95°C, followed by 24 cycles of 95°C for 1 min, 55°C for 1 min, and 72°C for 45 sec.

Final elongation was 72°C for 5 minutes. At least 3 technical replicates of each reaction were amplified for each cDNA sample.

Characterization of HvCESA1 Time Course Amplicons

Three replicates of equal volumes of antisense PCR products for each time point were separated by agarose gel electrophoresis. Gels were imaged by a Chemidoc EQ camera using Quantity One software (Version 4.5.2 Build 070). Gel images were analyzed using ImageJ (Version 1.49E). Background subtraction was performed with a rolling ball radius of 50.0 pixels. Densitometry was performed, and then normalized to the densitometry results from the RNA loading gel.

Design of HvCESA1 RPA Probes

Primers for a 400-base pair region inside the sequence of the antisense transcript identified via SS-RT-PCR were designed using the PrimerQuest software (IDT DNA).

The fragment was amplified by RT-PCR from an oligo dT primed cDNA and then cloned into the pGEM T-Easy vector kit (Promega). α-32P-UTP (Perkin Elmer Health Sciences) radiolabeled probes were prepared from linearized plasmid templates (SpeI or NcoI) having 5’ overhangs from either T7 or SP6 RNA polymerase using the MAXIscript Kit

(Ambion) to produce the HvCESA1 antisense-targeting (466 bp) and HvCESA1 sense- targeting (506 bp) riboprobes respectively. A 61-base portion of the HvCESA1 sense- targeting riboprobe and an 82-base portion of the HvCESA1 antisense-targeting riboprobe were derived from the pGEM T-Easy vector, so empty vector probes were similarly prepared for both as negative controls. 71

HvCESA1 Time Course RPA

Ribonuclease protection assays were performed by using the Ribonuclease

Protection Assay (RPA) III kit (Ambion). Labeled riboprobes were gel-purified by 5%

PAGE containing 8 M urea in 1XTBE buffer per kit instructions, and hybridized with

10–20 μg total RNA from either barley, yeast, or mouse for 16–18 h at 42 °C. Reaction mixtures were digested with RNase A/T1 (1:100) for 30 min at 37 °C, then stopped with inactivation buffer (Ambion) and protected fragments were precipitated by using 10 μg yeast RNA as a carrier. The protected fragments were separated by 12.5% PAGE containing 8 M Urea in 1X TBE buffer. γ-32ATP (Perkin Elmer Health Sciences) end- labeled Decade Marker (Ambion), prepared per manufacturer’s protocol, served as the size standard.

Characterization of HvCESA1 Time Course RPA Gels

Autoradiograms of RPA gels were uniformly scanned at 600 dpi grayscale in a lossless format. The intensity of bands in the 22 and 24 nt regions were analyzed using

ImageJ (Version 1.49E).

Results

Antisense Transcripts Corresponding to Multiple Cellulose Synthase Genes Are

Detectible in Developing Barley Leaf

Expanding on work from Held et al. (2008), strand-specific RT-PCR (Craggs et al., 2001; S. Li, Liberman, Mukherjee, Benfey, & Ohler, 2013) (SS-RT-PCR) was adapted to survey for the presence of other antisense cellulose synthase (CESA) transcripts in barley (Hordeum vulgare). RNA was isolated from pools of rapidly 72 growing, barley third-leaf, which has been shown to be elevated in expression of CESAs

(R. A. Burton et al., 2004). First-strand cDNAs were synthesized using gene-specific primers targeting the antisense RNA strand for HvCESA1 (MLOC_55153.1), HvCESA2

(MLOC_62778), HvCESA4 (MLOC_66568.3), HvCESA5/7 (MLOC_43749), HvCESA6

(MLOC_64555.1), and HvCESA8 (MLOC_68431.4) listed in Table 2. HvCESA3

(MLOC_61930.2) was omitted from this study because its expression does not cluster with primary or secondary-wall expression (Figure 8) (R. A. Burton et al., 2004). Primers for cDNA synthesis targeted the region 800 to 1700 base pairs upstream of the known 3’ end of the sense transcript, near regions homologous to the primers used in the original publication. Care was made in selecting for unique sequences to differentiate between the similar HvCESA sequences.

Table 2

Primer sequences, melting temperatures, and extension times for HvCESA antisense transcript survey and ribonuclease protection assay Primer Name Primer Sequence Ta (°C) Extension Time (s) HvA1-antisense-Tag1 CTTATTCGCCACCATGACCGGTGTTGAAGGTGCTGGGTTT HvA1-antisense GTGTTGAAGGTGCTGGGTTT 60 45 HvA1-sense CTGTTGATGGCGTAGGAGGT HvA2-antisense-Tag1 CTTATTCGCCACCATGACCGAAGAAGCCACCGTCAAGGAC HvA2-antisense AAGAAGCCACCGTCAAGGAC 55 75 HvA2-sense AACCGCATTCTTGCCTTACA HvA4-antisense-Tag1 CTTATTCGCCACCATGACCGGGGCTCCTTGGGTTCTACA HvA4-antisense GGGCTCCTTGGGTTCTACA 60 45 HvA4-sense GATCAGCAGGGTTGTCCACT HvA5/7-antisense-Tag1 CTTATTCGCCACCATGACCGACGGGAAATCGACAACTACG HvA5/7-antisense ACGGGAAATCGACAACTACG 55 45 HvA5/7-sense ACCCAGAGGAGGGAGAAGAC HvA6-antisense-Tag1 CTTATTCGCCACCATGACCGAAAACCCGCATGATGAAGAG HvA6-antisense AAAACCCGCATGATGAAGAG 48 45 HvA6-sense GACTGGTCCACTTGAACACG HvA8-antisense-Tag1 CTTATTCGCCACCATGACCGGGAGCAGATGATGTCCCAAA HvA8-antisense GGAGCAGATGATGTCCCAAA 60 45 HvA8-sense CGGACCAGATGATGACGATG Tag1 CTTATTCGCCACCATGACCG HvCESA1 RPA Probe Forward Primer (Antisense) TAAGCGCCCAGCTTTCAA HvCESA1 RPA Probe Reverse Primer (Sense) GATACCTCCAATGACCCAGAAC

Figure 8: Expression pattern of barley CESA genes in developing leaf tissue. The pattern of HvCESA3 gene does not match that of the other HvCESAs. Leaf descriptions along the X-axis (A-E) indicate region of barley first leaf measured via qPCR. Leaf D indicates the region 5-6 cm from the base of the leaf. Leaf B indicates the region 2-3 cm from the tip of the leaf. Adapted from Burton et al. (2004). 74

PCR amplification of antisense cDNAs yielded antisense amplicons for

HvCESA1, HvCESA4, and HvCESA6, with lengths of 913, 966, and 898 base pairs respectively (Figure 9). DNA sequencing of each antisense control transcript confirmed that all transcripts were complementary to the corresponding exonic sequence with no introns or indels. Further, all three amplicons included the tag primer from the cDNA synthesis step on the correct end of the transcript. This confirms that the PCR product was the direct product of an antisense-transcript primed by my tagged primer. Control sense amplicons of the same sizes (minus the length of the tag) were detected for each

(Figure 9), and showed much brighter bands, despite being cycled under the same conditions, indicating that the relative quantity of antisense transcripts present is very low compared to the mRNA-derived sense transcripts. No antisense transcripts for the remaining HvCESAs were detected despite the presence of the control sense amplicons.

In conjunction with the exhaustive negative controls, the HvCESA antisense transcript amplicons indicate that several specific members of the HvCESA gene family have corresponding antisense transcripts complementary to the 3’ end of the sense coding region.

Figure 9. Survey of antisense transcript gels for the HvCESA gene family. Antisense transcripts were detected for HvCESA1, HvCESA4, and HvCESA6. Digital images of the ethidium bromide stained gels were adjusted for brightness and contrast for visibility.

Expression of HvCESA1 Antisense Transcripts Are Dynamic during Leaf Growth

An additional primary-wall synthase (HvCESA1) antisense transcript was found in my survey, and was further examined to see if the same expression pattern occurred as previously seen for HvCESA6. A developmental time-course of barley third-leaf was tracked and collected from the first development of third-leaf tissue (10 days post imbibition) until after the growth of the third-leaf began to plateau 6 days later (16 days).

The third-leaf model is directly comparable to prior work with VIGS treated tissues (Held et al., 2008).

HvCESA1 antisense transcripts were monitored during this time using the tagged

SS-RT-PCR method introduced above, optimized for endpoint signal detection. The quantity of HvCESA1 antisense transcript was lowest on day 10, then increased over the 76 course of development to a maximum on days 15 and 16 (Figure 10A) increasing approximately by a factor of 2.5 (Figure 11). HvCESA1 sense signal over the same period was highest on days 10 to 13, then fell by approximately half on days 14 to 16 (Figure 10 and Figure 11).

Figure 10. Developmental time course gels of antisense transcripts and smRNAs for the HvCESA1.A) Barley third leaf time course tracking antisense and sense expression of HvCESA1. B) Parallel barley third leaf time course tracking levels of antisense HvCESA1 smRNA. HvCESA1 antisense and smRNA signals increase as the sense signal decrease. Digital photograph of the ethidium bromide-stained gel is adjusted for brightness and contrast for visibility. Quantification of signal strengths reported in Figure 11 and Figure 13. 77

Figure 11. Quantification of HvCESA1 sense and antisense transcript levels over developmental time course. Signal was measured from densitometry of representative agarose gels, normalized for RNA loading by dividing signal for each antisense or sense band by that of the RNA signal for the corresponding day. All normalized signals for each sample set (sense or antisense) were divided by the lowest signal to give relative normalized signal for each day. Source gel in Figure 10A. Absolute comparisons between sense and antisense transcript quantities are not applicable. Overlaid are the average leaf blade lengths (mm).

HvCESA1 antisense PCR conditions for the time course were optimized at 34 cycles to show the largest dynamic range, but the antisense transcript was visible after a minimum of 32 cycles (unpublished data). However, the HvCESA1 sense transcript was brightly visible after 24 cycles of amplification (Figure 10A). This was consistent across all 7 days of the trial, indicating that the expression of the HvCESA1 antisense transcript is consistently at least 8-fold lower than the sense transcript. 78

Dynamic Expression of 21 and 24 Nucleotide Antisense smRNAs

A parallel third leaf time course examining antisense smRNAs was performed.

Antisense HvCESA1 smRNAs (21 and 24 nucleotide sizes) were identified via a ribonuclease protection assay (RPA) using a sense probe. The sense probe was designed to be internal to the known antisense region of HvCESA1 (Figure 12 and Table 2), so only antisense smRNAs with the same sequence as the selected region within the

HvCESA1 antisense transcript would be detected. Comparison between the 21 and 24 nt signals indicates that there is consistently a stronger signal from the 21 nt size, indicating that the population of 21 nt RNAs is consistently larger than that of the 24 nucleotide

RNAs.

Figure 12. Map of HvCESA1 RPA probe and HvCESA1 antisense transcripts. PCR amplicons and RPA probes were designed internal to the coding region of HvCESA1. The untranslated regions (UTR) at the 5’ and 3’ ends are indicated in green, with the coding sequence (CDS) indicated in tan. The region amplified to detect HvCESA1 antisense transcript is in purple, and the sequence region used to probe for antisense HvCESA1 smRNAs is indicated in red.

A RPA time course of RNA from 11 to 16 days using this probe identified 21 and

24 nucleotide-sized smRNAs present on all 6 days of the trial (Figure 10). The signal intensity of the smRNAs varied over time, showing an overall increase in intensity from day 11 to day 16 (Figure 13). The largest single changes occurred on between days 11 and 12 and again between days 15 and 16. These changes roughly correlate with the 79 acceleration and deceleration in the growth of the third-leaf. On days 10 and 11, the growth of the leaf begins to increase, staying consistent from day 11 to day 15, and then dropping off again from day 15 to 16. The overall dynamic increase of the signal was by a factor of ~2.5 for both the 21 nt and 24 nt smRNAs.

Figure 13. Quantification of HvCESA1 antisense smRNA levels during a developmental time course. Signal measured from densitometry of autoradiography films, normalized for RNA loading by dividing signal for each RPA band by that of the RNA signal for the corresponding day, relative to day 11 (lowest signal). Antisense smRNA levels rise as leaf growth slows. Source gel in Figure 10B. Overlaid are the average leaf blade lengths.

Microarray Analysis of Specific HvCESA Silenced Tissues Shows Changes in Other Cell

Wall Related Genes

To examine the effect of targeted CESA gene silencing on cell wall gene networks, a microarray study of VIGS treated barley tissues was performed. Probes for a 80 custom microarray were constructed from nucleotide sequences for barley and the related grasses, rice and brachypodium (Kellogg, 2015). Known cell wall biosynthetic genes and

TFs thought to be involved in that process were emphasized. The resulting Agilent

8X16K microarray (Wolber, Collins, Lucas, De Witte, & Shannon, 2006) contained 3778 unique 60-mer probes, enriched for my targets of interest. Control VIGS (virus induced gene silencing) samples and HvCESA-targeting samples were labeled empty vector (EV) and CR2 respectively. The HvCESA-CR2 region was positioned at a conserved site in the

CESA family, indicated in Figure 14. Samples were screened for photobleaching (Figure

15) and/or silencing of HvCESA1 transcript levels via qPCR (Figure 16) prior to microarray analysis to confirm a silenced state.

Figure 14. Map of VIGS target site and qPCR primer locations based on HvCESA1 representative sequence 81

Figure 15. Examples of VIGS photobleaching effect. The leftmost leaf was from a plant mock inoculated with FES media, while the middle and right leaves were inoculated with BSMV-γ-PDS and BSMV-γ-PDS /BSMV-γ-HvCESA-CR2 respectively. Coinfection with the PDS (phytoene desaturase) silencing construct prevents production of chlorophyll, leading to pale or white leaves. Note the strong green color of the mock leaf, while both BSMV infected leaves show a much lighter average leaf color. This indicates both inoculated samples have systemic BSMV infections. No visual distinction of CESA gene silencing can be made.

Figure 16. Relative expression of HvCESA1 in selected unsilenced and silenced samples used in microarray analysis HvCESA1 is representative indicator of HvCESA gene family silencing. Samples shown are from barley third leaves infected with either the EV negative control BSMV construct or with the HvCESA-CR2 silencing construct (see Figure 14). Samples sets were grown, infected, and processed in batches, with sample sets from three batches shown here. While there is inter-sample and inter-batch variation, silenced (CR2, red) samples show significantly reduced HvCESA1 transcript quantities compared to negative control (EV, green) samples. HvCESA1 expression was normalized against the HvUBI10 reference gene. Error bars are normalized, relative standard deviation of n=3 technical replicates.

The results from the microarray indicate that 91 probes showed significant values

(adj. p ≤ 0.05), with a distribution of annotated functions (Table 3 and Table 4). 70 probes showed downregulated expression and 21 showed upregulated expression.

Approximately 43 of those probes are specific to genes annotated for cell wall modification activity, cell wall structural proteins, glycosyltransferase activity, and glycosylhydrolase activity. 83

Table 3

Groupings of significantly up and downregulated genes following VIGS of the HvCESA gene family Protein Function Number Cell Wall Modifying Proteins 16 Transcription Factor 16 Cell Wall Structural Proteins 12 Glycosyl Transferase 8 Glycosyl Hydrolases 7 Stress Response 6 Cytoskeleton 4 Lignin Biosynthesis 4 Metabolism 4 Promoter Binding 3 Transport 3 Epigenetic Modulator 2 Photosynthesis 2 Ribosomal 2 Unknown 1

Table 4

Statistically significant microarray results. This list contains differentially expressed microarray probes, along with their log2 fold-change (log2FC) and adjusted p-values (adj P. Val). Each probe accession was annotated with the most likely function tracked by that probe, along with the grouping it was assigned to (Table 3). Probe Source Accession log2FC adj.P.Val Curated Annotations Curated Grouping AK252852 -2.21 0.002 Extensin family protein Cell Wall Structural Proteins AK253095 -2.05 0.005 Extensin family protein Cell Wall Structural Proteins AK367663 -2.04 0.005 Xyloglucan endotransglucosylase/hydrolase family protein Glycosyl Hydrolases AK360797 -1.81 0.010 Classical AGP 9-like Cell Wall Structural Proteins AK357812 -1.74 0.002 peroxidase superfamily protein Cell Wall Modifying Proteins AK363764 -1.73 0.013 FASCICLIN-like arabinogalactan-protein 11 Cell Wall Structural Proteins AK362474 -1.72 0.002 LTPL65 - Protease inhibitor/seed storage/LTP family protein Transport AK357303 -1.65 0.002 Polygalacturonase Glycosyl Hydrolases Barley1_00444 -1.63 0.013 Extensin family protein Cell Wall Structural Proteins AK374224 -1.63 0.006 FASCICLIN-like arabinogalactan-protein 7 Cell Wall Structural Proteins Barley1_04319 -1.58 0.002 Leucine-rich repeat (LRR) family protein Cell Wall Structural Proteins AK251106 -1.57 0.003 Expansin Cell Wall Modifying Proteins AK251384 -1.49 0.007 Extensin family protein Cell Wall Structural Proteins AK372172 -1.47 0.006 zinc finger C-x8-C-x5-C-x3-H type family protein Transcription Factor AK376221 -1.46 0.006 Expansin Cell Wall Modifying Proteins AK356748 -1.45 0.038 Glycerophosphoryl diester phosphodiesterase Cell Wall Modifying Proteins Barley1_34235 -1.42 0.006 Transcription Factor bHLH48-like Transcription Factor AK249636 -1.40 0.007 Expansin Cell Wall Modifying Proteins AK251033 -1.39 0.002 Beta-tubulin Cytoskeleton AK251033 -1.38 0.002 Beta-tubulin (6) Cytoskeleton AK357092 -1.37 0.032 Xyloglucan endotransglucosylase Glycosyl Hydrolases AK373472 -1.36 0.012 Hydroxyproline-rich glycoprotein family protein Cell Wall Structural Proteins AK357691 -1.34 0.002 peroxidase superfamily protein Cell Wall Modifying Proteins AK374669 -1.32 0.028 Ribosomal Protein S17 Ribosomal AK370931 -1.29 0.020 Histone deacetylase HD2 isoform 1 Epigenetic Modulator AK366434 -1.28 0.009 Squamosa promoter binding protein 3 Promoter Binding AK251810 -1.27 0.033 Alpha-tubulin (4) Cytoskeleton AK250129 -1.27 0.003 Alpha-1,4-glucan-protein synthase Glycosyl Transferase AK356471 -1.27 0.013 Glucan endo-1,3-beta-glucosidase Glycosyl Hydrolases AK365601 -1.26 0.033 Pectinesterase Cell Wall Modifying Proteins AK361442 -1.25 0.018 MYB family transcription factor Transcription Factor AK368621 -1.25 0.002 ATP binding cassette subfamily B1 Transcription Factor AK252349 -1.25 0.006 SAM-dependent methyltransferase Cell Wall Modifying Proteins AK374737 -1.24 0.030 Beta-galactosidase Glycosyl Hydrolases AK355270 -1.23 0.002 Leucine-rich repeat (LRR) family protein Cell Wall Structural Proteins AK355499 -1.23 0.002 peroxidase superfamily protein Cell Wall Modifying Proteins AK363620 -1.22 0.037 LIM domain protein Transcription Factor AK362138 -1.20 0.002 RNA Binding Protein-Defense Related Stress Response AK361417 -1.19 0.003 Arabinogalactanprotein 16 Cell Wall Structural Proteins AK375167 -1.15 0.013 Pectate lyase Cell Wall Modifying Proteins Barley1_11939 -1.14 0.015 Rapid Alkalinization Factor Family Protein 23 Stress Response AK364850 -1.13 0.020 Glycosyltransferase Glycosyl Transferase AK252202 -1.12 0.014 AAA-type ATPase family protein Cell Wall Modifying Proteins AK356323 -1.09 0.035 MYB-family transcription factor Transcription Factor AK248424 -1.07 0.026 Choice-of-anchor C domain protein (potential GPI anchor) Cell Wall Modifying Proteins

Table 4

Statistically significant microarray results, continued. Probe Source Accession log2FC adj.P.Val Curated Annotations Curated Grouping AK361610 -1.05 0.003 Squamosa promoter binding protein Promoter Binding AK371287 -1.04 0.009 Growth regulator related protein (kinase?) Photosynthesis AK355696 -1.03 0.017 NAC-family transcription factor (103) Transcription Factor AK355954 -1.02 0.013 Glucan endo-1,3-beta-glucosidase Glycosyl Hydrolases AK251810 -1.00 0.004 Alpha tubulin (3) Cytoskeleton AK375789 -0.99 0.022 SAM-dependent methyltransferase Cell Wall Modifying Proteins AK353584 -0.99 0.024 E2F Transcription Factor-Like Transcription Factor AK361520 -0.95 0.009 Endomembrane protein 70 protein family Transport AK371158 -0.94 0.049 Histone deacetylase HD2 isoform 1 Epigenetic Modulator AK366434 -0.94 0.020 Squamosa promoter binding protein 3 Promoter Binding AK357056 -0.93 0.010 HvCESA6 Glycosyl Transferase AK358361 -0.92 0.025 Auxin response factor 8 Transcription Factor AK358127 -0.92 0.027 HMG CoA Reductase Lignin Biosynthesis AK374669 -0.91 0.017 Ribosomal Protein S40 Ribosomal AK354932 -0.89 0.026 Galacturonosyltransferase Glycosyl Transferase AK364583 -0.89 0.043 Zinc finger DNA binding domain containing protein Transcription Factor AK374683 -0.89 0.043 Galactosyl transferase GMA12/MNN10 family protein Glycosyl Transferase AK360719 -0.89 0.022 MATE efflux family protein Stress Response AK353678 -0.87 0.047 O-methyltransferase Lignin Biosynthesis AK357503 -0.86 0.024 S-formylglutathione hydrolase Metabolism AK249902 -0.83 0.027 40S ribosomal protein S3A Ribosomal AK366245 -0.82 0.028 Homeobox associated leucine zipper Transcription Factor AK356786 -0.81 0.037 Calcium ion-binding protein Cell Wall Structural Proteins AK361971 -0.78 0.018 Homeobox-leucine zipper protein PROTODERMAL FACTOR 2 Transcription Factor Barley1_14102 -0.72 0.043 PAM68-like protein Photosynthesis AK373066 0.64 0.043 Xylosyltransferase Glycosyl Transferase AK369083 0.64 0.040 basic helix-loop-helix (bHLH) DNA-binding superfamily protein Transcription Factor AK366125 0.64 0.046 O-fucosyltransferase family protein Glycosyl Transferase Barley1_15001 0.69 0.032 NAD(P)-binding Rossmann-fold superfamily protein Metabolism Barley1_05497 0.77 0.020 Apoptosis-inducing factor 2 Stress Response AK363783 0.78 0.022 WRKY transcription factor 19 Transcription Factor AK252924 0.83 0.033 Cycling-DOF factor 2 Transcription Factor Barley1_26368 0.89 0.028 Glycosyl hydrolase family protein Glycosyl Hydrolases AK364649 0.92 0.026 NAC-family transcription factor (6) Transcription Factor Barley1_30495 0.95 0.033 No Annotation ? Barley1_12794 0.96 0.038 Laccase Lignin Biosynthesis Barley1_45347 1.10 0.019 Cytochrome C-type biogenesis protein Metabolism Barley1_16179 1.11 0.044 Concanavalin A-like lectin protein kinase family protein Cell Wall Modifying Proteins Barley1_15070 1.15 0.039 SPFH/Band 7/PHB domain-containing membrane-associated protein family Transport AK354068 1.19 0.017 Fructose-bisphosphate aldolase Metabolism Barley1_50245 1.26 0.023 cysteine-rich receptor-like protein kinase 35 Cell Wall Modifying Proteins AK376662 1.35 0.024 DNA K Family Protein Stress Response AK364970 1.50 0.002 Xyloglucan xyloglucosyl transferase Glycosyl Transferase Barley1_04056 1.63 0.007 Thaumatin Stress Response AK359449 1.87 0.002 peroxidase superfamily protein Cell Wall Modifying Proteins AK365008 1.93 0.022 O-methyltransferase Lignin Biosynthesis

Discussion

Novel candidates for molecules acting in the downregulation of active cell wall components during the primary-secondary wall transition generally fall into the category of regulatory RNAs. Prior work (Held et al., 2008) has shown antisense transcripts and 86 small RNAs (smRNAs) for the primary-wall synthase (R. A. Burton et al., 2004)

HvCESA6 have dynamic and potentially linked expression over a leaf developmental time-course. This period exhibits rapid growth of the length of the leaf (Figure 11

Also, coexpression of gene networks centered around CESAs and corresponding co-suppression seen in virus induced gene silencing models lends weight to this idea

(Held et al., 2008). This hypothesis was reinforced by the results of our custom cell-wall targeting microarray. In barley, VIGS been shown to increase the amounts of 21 and 24 nucleotide smRNAs complementary to the target sequence (Held et al., 2008). This has been previously shown to down regulate specific members of the HvCESA family, along with other glycosyltransferases potentially involved in cell wall biosynthesis. By artificially creating an increase in CESA smRNAs via VIGS, I triggered a wide range of changes in cell wall gene expression in response to specific targeting of the HvCESA gene family (Table 3). This suggests that specific targeting of the HvCESA gene family can trigger changes in both directly and indirectly related cell wall proteins (Table 4).

The difficulties lie in determining the specific mechanisms involved in the regulation. Different classes of RNAs such as miRNAs (Lu et al., 2008; Rajagopalan,

Vaucheret, Trejo, & Bartel, 2006; Zhu et al., 2008), siRNAs (Vazquez, Vaucheret, et al.,

2004), lncRNAs (Jabnoune et al., 2013; Swiezewski, Liu, Magusin, & Dean, 2009; Yuan et al., 2014), and natural antisense transcript pairs (Borsani et al., 2005) have all been implicated in the process. A previously proposed mechanism (Held et al., 2008) suggested that the antisense HvCESA transcripts detected were a product of a (potentially cryptic) promoter on the opposite strand producing a product that was at least partially 87 complementary (Held et al., 2008). However, in the time since that mechanism was proposed, a full barley genome and several transcriptomes have been published (Mayer et al., 2012).

The most recent assembly of the barley genome (Build Hv_IBSC_PGSB_v2,

Ensembl Release 36, 2017-06-07) does not predict the presence of nearby overlapping genes on the opposite strand near HvCESA1 and HvCESA6. This does not preclude the possibility of transcriptional overlap, but the genomic map does not match the predicted antisense overlap model. HvCESA1 does have another gene predicted < 1000 bp downstream of it (HORVU6Hr1G050740.2), but it is positioned on the same strand.

HvCESA6 is positioned very close to the end of its contig assembly and the region downstream of the gene is not continuously sequenced, so the region downstream of that gene is uncertain. Transcription of an overlapping gene may still be a possibility for

HvCESA4, as it does show a predicted gene on the complementary strand

(HORVU1Hr1G039240.1) that could converge with the 3’ end of the HvCESA4 transcript. However, there is a ~300 nucleotide gap predicted between the two transcript ends that would likely preclude transcriptional overlap if the prediction is correct. 88

Figure 17. Predicted mechanism for RDR based synthesis of antisense transcripts. The updated model for antisense transcript and smRNA based CESA regulation is based on RNA-dependent RNA-polymerase (RDR) activity. A CESA mRNA targeted by a smRNA (tasiRNA example used here), recruiting a RDR to synthesize a complementary natural antisense transcript (NAT) strand starting at the 3’ end of the mRNA. The product double stranded RNA (dsRNA) could recruit Dicer-like (DCL) cleavage enzymes, which produce short RNA duplexes. These duplexes could feed into downstream RNA silencing machinery for use in targeted silencing.

A possible alternative can be found in a class of enzymes called RDRs. RDRs are capable of synthesizing complementary RNA transcripts from RNA templates, and 89 generally initiate synthesis from the 3’ end of the template strand (Figure 17) (Devert et al., 2015; Jauvion, Rivard, Bouteiller, Elmayan, & Vaucheret, 2012; Peragine,

Yoshikawa, Wu, Albrecht, & Poethig, 2004). This RNA biosynthesis mechanisms could fit the several pieces of evidence laid out here.

First, it could explain why all amplified antisense transcripts show perfect complementarity with the sense strand, even over exon junctions. The exonic nature of these antisense transcripts also indicates that they are not likely derived from genomic

DNA, but rather a mature (spliced) mRNA source. By directly synthesizing an antisense transcript from the mRNA, it bypasses any sources of alternative splicing that would occur if the antisense transcript came from a genomic source.

Second, it could explain why antisense transcripts have only been detected complementary to the 3’ end of the mRNA. Antisense transcripts complementary to the

5’ end of a CESA transcript have not yet been detected, either here or in Held et al.

(2008). Assuming the antisense transcripts are RDR-derived, the RDR may have lower processivity, leading to transcripts never extending far from the canonical 3’ initiation site. However, processivity of RDRs has not been established in the literature. This also may explain why multiple iterations of 5’ RACE were unsuccessful in determining a fixed 5’ end of these transcripts (data not shown), as there may be no single consensus end to find.

Third, it could explain the repeated failure of 3’ RACE in attempting to find the end of the transcript (data not shown). As this technique is dependent on the presence of a poly-A tail, I suspect that these may be missing or obscured in some way. If these 90 antisense transcripts are indeed synthesized by RDRs in the cytoplasm, then de novo polyadenylation may be unlikely, as no mechanism has been established in the literature for cytoplasmic polyadenylation outside of vertebrate translational control mechanisms

(Lange, Sement, Canaday, & Gagliardi, 2009).

Fourth, the mechanisms of VIGS and Dicer/RISC could work synergistically with

RDR activity. RDRs are known to function in the absence of an initiating primer, but they can also be targeted with a smRNA primer (Devert et al., 2015). A miRNA or siRNA hypothetically could serve this purpose. As the smRNA initiated RDR generates dsRNA, this could activate Dicer-like activities which would processively cleave the dsRNAs to create the smRNAs seen here. This would mesh well with evidence that multiple cell wall related transcripts are downregulated when only a specific gene motif is targeted, seen in Held et al. (2008) and here. If a single gene is targeted by a miRNA or siRNA for post-transcriptional degradation, the resultant degradation products could serve to initiate RDR-based elongation and trans silencing of other complementary targets (Axtell, 2013).

Finally, this mechanism could explain the extremely low levels of antisense transcript compared to the sense signal. Production of a dsRNA by RDR could immediately initiate RISC-based RNA degradation processes, so any accumulation of an antisense signal would be low. It seems probable that the antisense transcripts described here could very well be transient molecules present only in low numbers.

This still leaves the question of why only some CESAs show evidence of antisense amplicons. For barley, two primary-wall synthases (HvCESA1 and HvCESA6) 91 and one secondary-wall synthase (HvCESA4) show antisense transcripts, and for brachypodium (see Chapter 4) a similar pattern is seen. This phenomena may be tied to the mechanistic crossover between smRNA-based degradation pathways, and mRNA turnover pathways (Gregory et al., 2008). It may also be possible that antisense transcript production is a more global phenomenon, but I cannot detect them due to the detection threshold limits of PCR.

During the time courses tested here, it is assumed that primary-wall development is more active than secondary-wall development due to rapid leaf growth. In tissue where leaf growth is most active, mRNA levels for primary wall HvCESAs are elevated (R. A.

Burton et al., 2004) which also suggests that the quantity of mRNA in turnover is similarly increased. It is possible that the antisense transcripts seen here may (in part or whole) be by-products of that process. This would make it more likely that antisense transcripts for more highly expressed synthases be more detectable than those with lower mRNA expression levels. This roughly matches the detected offset between mRNA and antisense RNA expression levels (Held et al., 2008). The exact nature of this mechanism will be a subject of further study.

Conclusion

While the precise purpose and mechanism of action of these HvCESA- complementary antisense transcripts and smRNAs remains uncertain, several important points have been established here. Antisense transcripts occur for multiple CESA genes in barley, suggesting they may be part of a more highly conserved mechanism.

Additionally, transcripts are perfectly complementary to 3’ regions of their mRNA 92 complements, indicating they are sourced directly from a spliced RNA template. They also appear to have a relationship with smRNAs corresponding to the same transcript regions, although the precise nature of that relationship is unclear. However, it was shown that regulation of endogenous genes using a mechanism harnessing both antisense transcripts and smRNAs is possible. It remains to be seen how all the individual parts observed here precisely come together, but it opens an important avenue into potential endogenous gene regulation mechanisms for cell wall biosynthesis.

CHAPTER 4: DETECTION OF CESA ANTISENSE TRANSCRIPTS IN THE GRASS

SPECIES BRACHYPODIUM DISTACHYON

Introduction

In order to expand on the evidence for antisense transcript production identified in barley, the closely related grass species Brachypodium distachyon (Bd) was examined

(Kellogg, 2015). Brachypodium and barley are both members of the Poales order in the

Pooideae subfamily, which also contains several crop species that humans use as food sources, such as wheat, rye, and oats. This is in line with the original selection of barley as a model organism for its widespread usage as a crop species. Brachypodium is therefore also monocotyledonous and can be expected to contain a similarly developed

Type II primary cell wall. Brachypodium leaf growth also shows a similar developmental pattern compared to barley, where new leaves develop at intervals internal to the previous leaf showing rapid growth followed by a slowdown and thickening of the leaf. These similar characteristics allow me to look for signs of evolutionary conservation of antisense transcripts in the CESA gene family.

Here, I identify four antisense transcripts in the BdCESA gene family. A developmental time course tracking the expression of these BdCESA antisense transcripts and their corresponding mRNAs indicate a tight correlation between them, but shows a differing pattern from barley. These results add to the pool of evidence that antisense

CESA transcripts are involved in cell wall biosynthesis regulation. 94

Materials and Methods

Preparation of Brachypodium Tissue

Brachypodium distachyon seeds were imbibed in aerated water for 48 hours to stimulate germination, then transferred to damp vermiculite and incubated at 22°C in the dark for 7 days to stimulate cotyledon growth. On day 9, seedlings were transferred to autoclaved soil (Promix BX) supplemented with Osmocote (Scotts) 14-14-14 slow release fertilizer (1.8 g/L). Seedlings were grown in a Percival E36HOX growth chamber under high intensity fluorescent lamps (180-200 μmol m-2 sec-1) programmed for a 20- hour photoperiod (22 °C constant). Third-leaf tissue from ≥ 5 plants was excised, measured for length, and pooled in liquid nitrogen at 17, 19, 21, 24, and 27 days after imbibition.

Preparation of Brachypodium RNA

Pooled third-leaf sample was pulverized using a mortar and pestle under liquid nitrogen, and then homogenized under TRIzol® reagent. RNA was prepared per the

TRIzol® protocol. Aliquots of each RNA sample were treated for DNA contamination using the TURBO DNA-free KIT (Invitrogen) per the manufacturer’s instructions. 3 μl of

DNase was added in two aliquots, the first at the beginning of the incubation, and the second halfway through. Extra care was taken during DNase inactivation to avoid transfer of the DNase Inactivation Reagent into the final RNA sample. 500 ng of each

RNA sample was separated on a 1% agarose gel and visualized with ethidium bromide dye to check for RNA degradation. 95

Design of Gene-Specific Primers for Tagged SS-RT-PCR of BdCESAs

Gene-specific primers for the BdCESA gene family were designed using the

OligoAnalyzer 3.1 software (orthologous to Held et al. (2008)). Primers were BLASTed against the NCBI Brachypodium distachyon transcript library and then aligned against every member of the BdCESA gene family to ensure specificity. As before, a tag was added to the 5’ end of each gene-specific cDNA synthesis primer to eliminate artifact signals during PCR amplification. The LbB1.3 primer (tag2) was used (Alonso, 2003).

Preparation of BdCESA cDNA

First strand cDNAs for antisense transcripts of BdCESAs were synthesized from

1.7 μg of DNase treated total RNA extracted from Brachypodium distachyon third leaves, using the SuperScript III First-Strand Synthesis System (Invitrogen), using oligo dT

(OdT) and tagged gene-specific antisense primers. Control cDNAs were prepared as follows; OdT-primed (OdT) cDNA; No primer control (NPC) cDNA with the primer replaced with nuclease free water; No reverse transcriptase control (NRT) cDNA with the

RT enzyme replaced with nuclease free water. cDNA reactions were then treated with

RNase H to remove residual complementary RNA per the manufacturer’s protocol, and then diluted in a 1:9 ratio of cDNA with nuclease free water.

Amplification of Brachypodium cDNA

First-strand cDNAs synthesized for each BdCESA antisense transcript were amplified by PCR using the corresponding sense primer and the tag2 primer. Negative control cDNA samples were amplified in the same manner. OdT primed cDNA was also amplified individually with each pair of BdCESA sense and antisense primers (no tag) as 96 control for amplicon size and sense RNA presence. All PCR amplifications were assembled on ice in 25 μl reactions using 5 μl of 5X Green GoTaq buffer (Promega

M3001), 1.0 μl of each primer (10 μM), 0.5 μl of dNTPs (10 μM each), 4 µl of diluted cDNA template, and 1.25 units of GoTaq polymerase.

Cycling conditions for reactions were optimized for melting temperature and extension time. All reactions were cycled with 2 minutes of activation at 95°C, followed by 37 cycles of 95°C for 30 s, optimized annealing temperature for 30 s, and 72°C for the optimized extension time. Final elongation was 72°C for 5 minutes. ≥ 3 technical replicates of each reaction were amplified for each antisense cDNA sample.

Characterization of BdCESA Survey Amplicons

Equal volumes of PCR product for each sample and control reaction were separated by agarose gel electrophoresis. Gels were imaged by a Chemidoc EQ camera using Quantity One software (Version 4.5.2 Build 070). Gel images were analyzed using

ImageJ (Version 1.49E).

Antisense amplification products were excised and purified with the Zymoclean

Gel DNA Recovery Kit (Zymo), and cloned into the pGEM T-Easy vector kit (Promega).

Clones were fully sequenced and confirmed as the targeted sequence. Inclusion of the tag2 sequence confirmed that cDNA samples synthesized from the antisense-tag2 primer were unique from cDNA samples synthesized from the OdT primer.

RNA Loading Control for BdCESA Antisense Time Courses

500 ng masses of DNase-Treated total RNA (based on spectrophotometrically measured concentrations) for all time points were separated on a 1% agarose gel to check 97 for degradation and equal RNA loading. Gels were imaged by a Chemidoc EQ camera using Quantity One software (Version 4.5.2 Build 070). Gel images were analyzed using

ImageJ (Version 1.49E).

Preparation of BdCESA Antisense Time Course cDNA

First strand cDNAs synthesized using the corresponding BdA#-antisense-tag2 were used as templates for PCR following the same assembly as the initial detection survey in barley.

Amplification of BdCESA Antisense Time Course cDNA

Cycling conditions for reactions were also the same as for the initial detection survey. Three technical replicates of each reaction were amplified for each antisense cDNA sample.

Characterization of BdCESA Time Course Amplicons

Three replicates of equal volumes of antisense PCR products for each time point were separated by agarose gel electrophoresis. Gels were imaged by a GelDoc MP

System using the Image Lab software (Version 5.2.1 Build 11). Gel images were analyzed using ImageJ (Version 1.50i). Densitometry was performed, and then normalized to the densitometry results from the RNA loading gel.

qPCR Analysis of BdCESA mRNA Expression

A 1 µg aliquot of each RNA sample was used to synthesize cDNA using the

Qiagen Quantitect Reverse Transcription Kit per the manufacturer’s instructions with an extended reverse transcription time. cDNAs were diluted at a 1:5 ratio with sterile water and then assayed for the expression of BdCESA family members using custom primers 98 against the pre-optimized BdUBC18 (ubiquitin conjugating 18) reference gene (Hong,

Seo, Yang, Xiang, & Park, 2008). Reaction were performed in a 20 µl volume using the

Qiagen Rotor-Gene SYBR Green PCR Kit, on a Rotor Gene Q platform, per the manufacturer’s instructions. Ct values were identified with the Rotor Gene Q software,

Version 9, and used to calculate relative expression for each sample using the Pfaffl method (Pfaffl, 2001) to account for amplification efficiency.

CESA Phylogenetic Tree

Nucleic acid sequences for HvCESAs and BdCESAs were translated to amino acid sequences and multiple aligned using the Clustal Omega tool (Sievers et al., 2014) with default settings. Alignment file was loaded into Dendroscope 3 (Huson & Scornavacca,

2012) to generate an unrooted radial dendrogram.

Data Mining for BdCESA smRNAs

The brachypodium small RNASeq dataset OBD02 (GSM1266844) (Jeong et al.,

2013) hosted at mpss.danforthcenter.org were queried (M. Nakano et al., 2006) using

BdCESA genes. All smRNAs matching BdCESAs were BLASTed against the brachypodium genome to ensure specificity to only BdCESA genes (E-value cutoff of 1E-

10), and any sequences with alternate targets were omitted.

Results

Antisense Transcripts and smRNAs of CESAs Are Present in the Model Grass

Brachypodium distachyon

RNA pools from rapidly growing brachypodium third-leaf were assayed for antisense transcripts. First-strand cDNAs were synthesized using tagged, gene-specific 99 primers (Table 5) targeting the antisense RNA strand for BdCESA1 (Bradi2g34240),

BdCESA2 (Bradi1g04597), BdCESA4 (Bradi2g49912), BdCESA5 (Bradi1g02510),

BdCESA6 (Bradi1g53207), BdCESA7 (Bradi3g28350), BdCESA8 (Bradi1g54250), and

BdCESA9 (Bradi4g30540) (Table 5). Primers for cDNA synthesis targeted the region

1000 to 1700 base pairs upstream of the 3’ end of the sense transcript, following the same methodology as previous. BdCESA3 (Bradi1g29060) and BdCESA11 (Bradi1g36740) were not examined, as they each are missing specific motifs characteristic of cellulose synthases (Figure 18) (Handakumbura et al., 2013). BdCESA3 (Bradi1g29060.1) is missing the RING-type zinc finger domain, and BdCESA11 (Bradi1g36740.1) is missing the RING-type zinc finger domain, the aspartate residues, and the QxxRW domain. 100

Table 5

Primer sequences, melting temperatures, and extension times for BdCESA antisense transcript survey. Primer Name (paper) Primer Sequence Tm (Extension Time (s) BdA1-antisense-tag2 ATTTTGCCGATTTCGGAACAAAAACCGTATGATGAAGAG BdA1-antisense AAAAACCGTATGATGAAGAG 50 60 BdA1-sense GAGATGGAGGATCACCCAGA BdA2-antisense-tag2 ATTTTGCCGATTTCGGAACGTGTTTTTGTGGCCTCCACT BdA2-antisense GTGTTTTTGTGGCCTCCACT 55 60 BdA2-sense TCTTGTGGTGAACGGATCAA BdA4-antisense-tag2 ATTTTGCCGATTTCGGAACAGCAGGACAGACCAGAGTAT BdA4-antisense TCTACGGGAAATTGACAACTATGA 55 60 BdA4-sense AGCAGGACAGACCAGAGTAT BdA5-antisense-tag2 ATTTTGCCGATTTCGGAACTCCGAGTCTCTGCTGTACTT BdA5-antisense TCCGAGTCTCTGCTGTACTT 60 60 BdA5-sense GCTAAGCTCTGGAGTGATGAA BdA6-antisense-tag2 ATTTTGCCGATTTCGGAACACAAAAGCCAAGCCAGAGAA BdA6-antisense ACAAAAGCCAAGCCAGAGAA 55 60 BdA6-sense CCGACCAAACCTTTGAGAAA BdA7-antisense-tag2 ATTTTGCCGATTTCGGAACAAGAAGGGAGGGTCCTACAG BdA7-antisense AAGAAGGGAGGGTCCTACAG 51 60 BdA7-sense ATGACCCGTACCCATTGTTG BdA8-antisense-tag2 ATTTTGCCGATTTCGGAACCTCAGTCCTCAACTCCAGAATC BdA8-antisense CTCAGTCCTCAACTCCAGAATC 54 60 BdA8-sense CACTGACACGGGTGGTAAA BdA9-antisense-tag2 ATTTTGCCGATTTCGGAACTGGATCTATGGGTCGATCACT BdA9-antisense TGGATCTATGGGTCGATCACT 53 60 BdA9-sense CGAAATTGGTCTCCTCCCTATG Tag2 ATTTTGCCGATTTCGGAAC 101

Figure 18. Domain map of BdCESA proteins. BdCESA3 and BdCESA11 are indicated in red, and are missing specific common domains of the CESA superfamily. Adapted from Handakumbura et al. (2013). Naming conventions used here are on the left, those used in the original paper are below each diagram, and do not follow existing conventions.

PCR amplification of the antisense cDNAs yielded antisense amplicons for

BdCESA1, BdCESA4, BdCESA6, and BdCESA8, with lengths of 1059, 1078, 1107, and

1009 base-pairs respectively (Figure 19). Like in barley, DNA sequencing of each antisense control transcript confirmed that all transcripts were complementary exonic sequence with no introns or indels, and that all four amplicons included the tag2 primer from the cDNA synthesis step on the correct end of the transcript. Control sense amplicons of the same sizes (minus the length of the tag) were detected for each (Figure

19), and showed much brighter bands despite being cycled under the same conditions, indicating similarly to barley that the relative quantity of antisense transcripts present is 102 very low compared to the sense mRNAs. No antisense transcripts for the remaining

BdCESAs were detected despite the presence of the control sense amplicons.

Figure 19. Survey of antisense transcripts for the BdCESA gene family. Antisense transcripts were detected for BdCESA1, BdCESA4, BdCESA6, and BdCESA8. Digital images of the ethidium bromide stained gels were adjusted for brightness and contrast for visibility.

To survey for the presence of BdCESA smRNAs, existing small RNASeq databases were queried. Third leaf tissue data sets were not available, but similar tissue from 6-week old leaf and stem was considered comparable. RNASeq data showed all

BdCESAs had smRNA populations. Antisense transcript associated BdCESA1, 4, and 8 has relatively elevated smRNA counts, while BdCESA6 had a relatively lower count

(Table 6). Of the BdCESAs not associated with antisense transcripts, BdCESA2, 7, and 9, 103 had relatively lower counts, and BdCESA5 had counts similar to BdCESA1 and 4. In general, antisense transcript-associated CESAs had elevated smRNA counts compared to those where antisense transcripts were not detected.

Table 6

BdCESA smRNA counts mined from brachypodium data

Expression of BdCESA Antisense Transcripts Decrease Over Time

In following with the work done in barley, a time course of developing brachypodium third-leaf was prepared to assay the expression of BdCESA antisense transcripts over the course of leaf development. As with barley, this period exhibits rapid leaf growth, indicative of rapid cell wall development and high CESA activity.

For the primary wall group, the maximum amount of antisense signal occurs on the first day of the time course (day 17) and drops severely throughout, sometimes below the limit of detection (Figure 20). The relative drop in signal differs between transcripts 104 without a discernable pattern and is quantified in Figure 21. For the secondary wall group, HvCESA4 antisense transcript shows elevated levels on day 17 and 21 with a complete loss of signal between them (Figure 20). The signal for the last two days of the time course are almost below the limit of detection and below the limit of detection respectively.

Figure 20. Developmental time courses of antisense transcripts for the BdCESA family. Antisense amplicons for four BdCESAs were tracked over 11 days. RNA loading gels shown above. Antisense transcript levels for BdCESA1, BdCESA6, and BdCESA8 immediately decrease as the time course progresses. BdCESA4 antisense transcript levels peak on days 17 and 21, and then decrease. Digital photographs of the ethidium bromide- stained gels were adjusted for brightness and contrast for visibility.

105

Figure 21. Absolute quantification of BdCESA antisense transcript levels over developmental time course. Signal measured from densitometry of representative agarose gels, normalized for RNA loading, relative to lowest signal day for each transcript. Antisense transcript levels for BdCESA1, BdCESA6, and BdCESA8 immediately decrease as the leaf grows. BdCESA4 antisense transcript levels peak on days 17 and 21, and then decrease. Source gels in Figure 20. Signal differences between separate transcripts should not be compared. Overlaid are the average leaf blade lengths (mm) ± SD (n ≥ 5).

106

Expression of BdCESA mRNAs Show Correlation with Antisense Transcripts

The entire BdCESA family was assayed to get a complete picture of the behaviors of these genes in brachypodium leaf, which has not been previously characterized, but were predicted based on orthology with HvCESAs and AtCESAs (Figure 22 and Table 7).

Figure 22. Phylogenetic tree of Clustal Omega multiple alignment of BdCESA and HvCESA translated amino acid sequences. Sequences circled in red have been shown to have antisense transcripts. Scale bar is number of substitutions per unit length in the alignment (excluding gaps).

107

Table 7

Comparison table of phylogenetically similar CESAs Barley Brachypodium Arabidopsis Wall (Hordeum (Brachypodium (Arabidopsis Association vulgare) distachyon) thaliana) HvCESA1 BdCESA2/8 AtCESA3 Primary HvCESA2 BdCESA5/6 AtCESA2/5/6/9 Primary HvCESA4 BdCESA7 AtCESA4 Secondary HvCESA5/7 BdCESA4 AtCESA8 Secondary HvCESA6 BdCESA1 AtCESA1/10 Primary

HvCESA8 BdCESA9 AtCESA7 Secondary

Expression levels of BdCESA mRNA levels were assayed using custom gene- specific primers (Table 8). Signals from the BdCESA mRNAs could be grouped into three categories, clustering into primary cell wall associated, secondary cell wall associated, and unassigned. This almost perfectly matches the predictions of behavior based on the phylogenetic tree of BdCESAs and HvCESAs, as well as the expected behavior of

CESAs during plant development. 108

Table 8

BdCESA qPCR Primers Primer Name Primer Sequence BdCESA1 -F CGTTGATGACCTGGACAATG BdCESA1 -R GTTCATGGCGAGAGGATGAT BdCESA2 -F GAAACTCGTCAGCCGCTATC BdCESA2 -R CATTACGCACAGGATTGGTG BdCESA4 -F TTTGCCTGCTCTTCCAAAGT BdCESA4 -R TTCTCGTCGGGAGTCTCTGT BdCESA5 -F AGTTCAACTGGAGGGACAGG BdCESA5 -R TTGGTGAGCAAGGGAACATT BdCESA6 -F TAGACGGCAAGCGTTGTATG BdCESA6 -R TTCTCTGGCTTGGCTTTTGT BdCESA7 -F CTCGGGAAGAAGCTCTGCTA BdCESA7 -R AGCCCCTTCATGTTGATGTC BdCESA8 -F TCCCATGACTAATGGCACAA BdCESA8 -R TAGACAGAGGCTGCCGAGTT BdCESA9 -F ATGGTGACCTGTGACTGCTG BdCESA9 -R ATTCCGTCACCGACACTCTC

Four of the five primary cell wall associated transcripts had their highest expression early in the time course (Figure 23), dropping off significantly as the leaf growth slowed, with an overall relative decrease from a factor of 2.5 (BdCESA5) to a factor of 7 (BdCESA6).

109

Figure 23. Relative expression of primary cell wall related BdCESA genes. qPCR quantification of BdCESA genes orthologous to HvCESA1, HvCESA2, and HvCESA6, normalized against BdUBC18. Brachypodium third-leaf growth curve overlay is the same curve from Figure 21.

For the secondary cell wall associated BdCESA transcripts, their general pattern is to show elevated signals on days 17 and 21 with a dip between them, followed by a halving of overall signal from the maxima over days 24 through 27 (Figure 24). This matches the pattern for HvCESA8 laid out in Held et al. (2008), which maintained an elevated relative expression of secondary wall associated CESAs for an extended period after the primary cell wall associated CESA signal dropped. Again, this matches the pattern for the antisense transcript, as BdCESA4 shows peak levels on days 17 and 21, and very low signals on days 19, 24, and 27 (Figure 21). It should also be noted that relative concentration of the secondary wall associated BdCESA mRNAs are much lower on day 17 (and throughout) than those of the primary wall (Figure 25). 110

Figure 24. Relative expression of secondary cell wall related BdCESA genes. qPCR quantification of BdCESA genes orthologous to HvCESA4, HvCESA7, and HvCESA8, normalized against BdUBC18. Brachypodium third-leaf growth curve overlay is the same curve from Figure 21.

Figure 25. Relative qPCR-measured expression of all BdCESA mRNAs concentrations on day 17 of time course.

111

For the primary wall association-predicted BdCESA2, expression deviated from the expectation. It shows low expression early during the time course, spiking on day 21, and then gradually falling back to baseline on day 27 (Figure 26). The dynamic range of this change is at maximum a factor of two, and it does not seem to correlate with any distinct morphological change in the plant. The overall expression of BdCESA2 is also extremely low compared to all other examined BdCESA mRNAs (Figure 25).

Figure 26. Relative expression of other BdCESA genes. qPCR quantification of BdCESA genes not orthologous to any HvCESA, normalized against BdUBC18. Brachypodium third-leaf growth curve overlay is the same curve from Figure 21.

Discussion

Identification of additional CESA antisense transcripts (Figure 19) and smRNAs

(Table 6) in brachypodium increases evidence for the possibility that antisense transcripts are involved in the regulation of cell wall biosynthesis. BdCESA1, BdCESA6, and

BdCESA8 are orthologous to barley primary-wall synthases and BdCESA4 is orthologous 112 to barley secondary-wall synthases (Figure 22 and Table 7). This roughly matches the patterning for barley, with detection of antisense transcripts favoring primary wall associated CESAs.

The exon-only nature of these also seems to hold true, and has been confirmed against published genomic and transcriptional sequences from the brachypodium genome. Usage of tagged SS-RT-PCR with an alternate tag sequence (tag2) also shows the robustness of the assay under varying conditions. This indicates that this class of antisense transcripts is consistently derived from a mature mRNA source. A limited preliminary assay examining the length of the antisense transcript for BdCESA1

(unpublished data) also seem to confirm that the starting position of this type of antisense transcript lies somewhere between the 5’ end and the midpoint of the coding region of the mRNA sequence. However, this aspect of CESA antisense transcripts has only been examined for HvCESA6 (Held et al., 2008) and BdCESA1.

Additionally, while follow up RPA studies were not performed in brachypodium, mining of previously published brachypodium stem and leaf smRNA datasets (Table 6) shows that smRNA reads are present for all members of the BdCESA gene family. The tissue samples used in their studies were approximately 6 weeks old, and as such cannot be assumed to match my data directly, but still provide evidence that CESA smRNAs are detectable in brachypodium.

Time courses for these antisense transcripts (Figure 20) diverge from the patterns established for barley, calling into question the fundamental nature of the time course in brachypodium. For brachypodium, the growth pattern of the leaf is slightly different from 113 that of barley (Figure 21). Brachypodium grass is physically smaller than barley, and starts from a much smaller, lighter seed. This causes early brachypodium seedlings to be very small comparatively. By the time the third-leaf is large enough to see and manipulate, the leaf is already growing at its maximum rate, so the growth curve does not have an initial slow period of growth. Brachypodium is also a much slower growing grass, requiring approximately ten days for the third-leaf to reach maximum length

(Figure 21), opposed to barley completing growth in approximately 7 days.

Together, this raises the likelihood that brachypodium may not be a suitable model to compare to barley. The lack of an early measured slow leaf growth period like that seen in barley suggests that if the same antisense developmental pattern is occurring, it may start before I can even access the leaf. My earliest time points may be too late in the growth phase to see the expected behavior. This is supported by the fact that the brachypodium growth curve looks like the latter half of the barley growth curve. This may require selection of a different model organism such as rice or wheat to pursue in further studies.

Alternatively, it is possible that the patterning of CESA expression in brachypodium leaf during development is fundamentally different from that of barley.

This hypothesis is augmented by the increased cycle number required to detect antisense transcripts in brachypodium. If antisense transcripts are indeed part of a regulatory pathway, then differences in the genes they regulate could cause the differences seen.

Alternatively, the additional cycles may suggest absolute quantity of these transcripts may differ between the two species. 114

The mRNA expression patterns for the BdCESAs mostly seem to match expectations laid out for Arabidopsis and Barley. Expression of BdCESAs 1, 5, 6, and 8 seem to peak when primary wall synthases are predicted to be active (rapid growth), and fall off rapidly as growth slows. Expression of BdCESAs 4, 7, and 9 remains elevated through the midpoint of the time course, only falling as the leaf finishes maturing, hallmarks of secondary wall growth. However, downregulation of both groups seems to occur relatively early compared to the HvCESA6 (primary) and HvCESA8 (secondary) time course expression levels in Held et al. (2008), again suggesting that this time course may be too late in development. Expression the of BdCESA2 is not as predicted, and it may act in some specialized capacity that does not track with a developmental pattern.

The close matching of the sense and antisense expression patterns is also of interest (Figure 21, Figure 23, and Figure 24), because it provides a mechanistic insight.

As mentioned in Chapter 3 a potential source of these antisense transcripts may be RDR transcription of CESA mRNA templates guided by smRNAs. The close tracking of relative sense and antisense transcript levels would indicate that an RDR source mechanism is possible. An increased pool of mRNA as seen for most of the CESAs early in the time course would provide a larger number of templates for RDR to bind to and complement. However, this calls into question how the antisense transcripts for

HvCESA1 and HvCESA6 show alternative patterns from previously established mRNA expression data (Held et al., 2008). It may suggest that there are other factors involved in controlling how much antisense transcript is produced, or that BdCESA and HvCESA 115 antisense transcripts come from different sources. Pursuit of these questions via silencing or mutant studies should be a focus for future research on this topic.

Conclusion

Discovery of CESA antisense transcripts and smRNAs outside the barley model shows that these components may share a conserved mechanism. However, deviation of the behavior of the BdCESA antisense transcripts from those predicted by the HvCESAs somewhat complicate mechanistic predictions. It suggests that the production methods of antisense transcripts may vary between species.

116

CHAPTER 5: DETECTION OF CESA ANTISENSE TRANCRIPTS IN ARABIDOPSIS

AND ATTEMPTS AT IDENTIFYING ANTISENSE SOURCE

Introduction

To this point, all model work was done in monocot grasses, which while highly relevant to environmental, agricultural, and biofuel purposes, suffer from a relative lack of literature data and mature bioinformatic resources. The genomes for these organisms were only published relatively recently (Mayer et al., 2012; Vogel et al., 2010), and are currently not well annotated. Therefore, it makes sense to pursue my antisense transcript model in Arabidopsis thaliana (At), which is the most researched plant model to date.

Arabidopsis has a large quantity of publicly available data sets (Winter et al., 2007), and a well annotated genome (The Arabidopsis Genome Initiative, 2000), which can be useful for ongoing examination of antisense transcripts. Arabidopsis also has the benefit of a large, well-curated library of mutants, which includes mutants in the small RNA silencing pathways, allowing for examination of pathways that could produce antisense transcripts.

Here, a targeted survey of CESA antisense transcripts showing sequence homology with antisense transcripts in both HvCESAs and BdCESAs identified two AtCESA antisense transcripts.

Materials and Methods

CESA Phylogenetic Tree

A phylogenetic tree was prepared as described previously in Chapter 4 using all

AtCESA sequences and CESAs with associated antisense transcripts.

117

Arabidopsis thaliana Liquid Culture

Liquid cultures of bulk arabidopsis seedlings were prepared as follows. Sterile flasks of MSG + 1% sucrose media with 100 μl of sterilized, cold-treated (3 to 5 days at

4°C) Arabidopsis seed from wild type Columbia 0 were prepared. Flasks were incubated under the condition of 16-hour days, 21°C constant temperature, photon flux at 100 μmol photons per meter squared per second, shaking at 100 rpm. Once the seeds germinated and formed a moderately sized bolus of tissue (10 to 14 days), tissue was drained and washed to remove media solution. Tissue bolus was then flash frozen in liquid nitrogen for RNA isolation.

Preparation of Arabidopsis RNA

Arabidopsis liquid cultures were homogenized into powder under liquid nitrogen with a pre-chilled mortar and pestle. Powdered homogenate was added to TRIzol® reagent per the manufacturer’s instructions and further homogenized in a glass-glass grinder. Homogenized samples were immediately processed into RNA per the manufacturer’s instructions. Total RNA aliquots of each RNA sample were treated for

DNA contamination using the TURBO DNA-free KIT (Invitrogen AM1907) per the manufacturer’s instructions. 3 μl of DNase was added in two aliquots, the first at the beginning of the incubation, and the second halfway through. 1 μg of each RNA sample was separated on a 1% agarose gel and visualized with ethidium bromide dye to check for RNA degradation. 118

Preparation of AtCESA Antisense cDNA

First strand cDNAs synthesized using the corresponding AtA#-antisense-tag1 were used as templates for PCR following the same assembly as the initial detection survey in barley.

Amplification of AtCESA Antisense cDNA

First-strand cDNAs synthesized for each AtCESA antisense transcript were amplified by PCR using the corresponding sense primer and the tag1 primer. Negative control cDNA samples were amplified in the same manner as the antisense cDNAs. OdT primed cDNA was also amplified individually with a pair of AtCESA sense and tagged antisense primers as control for amplicon size and sense RNA presence. All PCR amplifications were assembled on ice in 50 μl reactions using 10 μl of 5X Green GoTaq buffer (Promega M3001), 1 μl of each primer (10 μM), 1 μl of dNTPs (10 μM each), 32

µl of diluted cDNA template, and 2.5 units of GoTaq polymerase. Some reactions were performed with a 25 µl volume using the same ratios. All reactions were cycled with 2 minutes of activation at 95°C, followed by 40 cycles of 95°C for 30 seconds, 56°C for 30 seconds, and 72°C for the optimized extension time. Final elongation was 72°C for 5 minutes.

Characterization of AtCESA Antisense Amplicons

Antisense amplification products were separated in a 1.5% agarose gel for 40 minutes at 100 V, and correctly sized fragments were excised and purified with the

Zymoclean Gel DNA Recovery Kit (Zymo), and cloned into the pGEM T-Easy vector kit 119

(Promega). Clones were sequenced and confirmed for inclusion of the tag1 sequence on the correct end.

Growing Arabidopsis thaliana Inflorescences

Arabidopsis seeds were sterilized and cold treated for 3 days at 4°C, then plated onto ½ MSG + 1% sucrose media plates to synchronize germination. Plates were incubated under 16 or 20-hour day conditions at 21-22 °C, with approximately 100-175

µmol photons per meter squared per second until primary inflorescences developed to approximately 15 cm in height. Primary inflorescences were severed at the rosette to induce secondary inflorescences to develop. Secondary inflorescences were allowed to develop until they were 10 to 20 cm in height (7 to 10 days). Secondary inflorescences were then severed at the base and collected in liquid nitrogen.

Results

Targeted Detection of Antisense AtCESA Transcripts

To survey for the presence of antisense transcripts of AtCESAs a phylogenetic approach was taken in the selection of AtCESAs to examine. Nucleotide sequences for

CESAs in barley and brachypodium that have antisense transcripts were translated and then multiple aligned using the Clustal Omega tool (Sievers et al., 2014). This identified three AtCESAs closely related to at least two other CESAs with antisense transcripts.

These were AtCESA1 (AT4G32410), AtCESA3 (AT5G05170), and AtCESA10

(AT2G25540) (Figure 27). From that pool, AtCESA1 and AtCESA3 are both canonical primary wall associated CESAs (Taylor, 2008), and AtCESA10 is seed and silique 120 specific CESA that is not detected in the rest of the plant (Schmid et al., 2005; Winter et al., 2007).

Figure 27: Phylogenetic tree of Clustal Omega of select HvCESA, BdCESA, and AtCESA translated sequences. Nucleotide sequences from CESAs known to produce antisense transcripts in barley and brachypodium (circled in blue) were translated and aligned against the translated sequences for the whole AtCESA family. Items circled in red show where conservation of antisense transcripts occurs across barley and brachypodium, indicating AtCESA1, AtCESA3, and AtCESA10 as candidates (green underline) for further antisense transcript assays. Scale bar is number of substitutions per unit length in the alignment (excluding gaps).

As the growth pattern of arabidopsis is completely different from that of grasses, liquid cultures of whole seedlings were prepared as a model. Whole seedlings can be grown in bulk, with representation of both rapidly growing aerial and root tissues, to give the greatest likelihood of detecting CESA antisense transcripts if they were present. First strand cDNAs were synthesized using gene specific primers targeting the antisense RNA strand for AtCESA1 and AtCESA3 (listed in Table 9). AtCESA10 was omitted both 121 because its expression is not expected to be found in the tissue type examined, and because it does not express in a pattern with established primary and secondary wall

AtCESAs. Primers for cDNA synthesis targeted the region 1300 to 1400 bases upstream of the 3’ end of the mRNA sequence. During this step, the 5’ tag1 sequence was added to each cDNA to eliminate artefact signals as described previously in Chapter 3. Control cDNAs were also prepared as previously described.

Table 9

Primer sequences, melting temperatures, and extension times for AtCESA antisense transcript survey Primer Name Primer Sequence Ta (°C) Extension Time (s) AtA1-antisense-Tag1 CTTATTCGCCACCATGACCGGACAAGACTGAATGGGGCAAAG 56 45 AtA1-sense AGAAGAGCTGTCCATTTGAAGATG AtA3-antisense-Tag1 CTTATTCGCCACCATGACCGGGAGTTGAAGGTGCTGGTTTT 56 45 AtA3-sense ATCCACTGTTGATAGCATAAGAGAC

PCR amplification of the antisense cDNAs identified amplicons for both

AtCESA1 and AtCESA3 with lengths of 653 and 919 base pairs respectively (Figure 28A).

DNA sequencing of these cloned transcripts again confirmed that all transcripts are complementary exonic sequence without introns or indels. Both transcripts also contain the tag1 primer on the correct end, confirming the PCR amplicon was the direct product of an antisense transcript. A consideration here is that these antisense transcripts were not detected at 37 cycles as was used for brachypodium antisense CESAs, requiring an increase to 40 amplification cycles for detection. Control sense bands were omitted from the gels in this case (Figure 28A), as the relative difference in signal strength between the 122 antisense and sense signal caused the sense signal to bleach out the antisense signal). The remaining CESAs have not yet been assayed for antisense transcripts.

Figure 28. Partial survey of antisense transcript gels for the AtCESA gene family, maximum amplification conditions. Two predicted CESA antisense transcripts were detected in arabidopsis. A) AtCESA1 and AtCESA3 antisense transcripts were detected in bulk liquid cultured arabidopsis seedlings (14 days) at 40 cycles using 16 µl of template with clean negative controls. Sense controls were omitted for a clearer visualization. B) AtCESA1 and AtCESA3 antisense transcripts separated in parallel to visualize difference in signal strength. Digital images of the ethidium bromide stained gels were adjusted for brightness and contrast for visibility.

Efforts Towards Identifying Antisense Transcript Source

A potential candidate to produce antisense transcripts for the CESA family members are RDRs. In arabidopsis, three RDRs have been identified. RDR1

(AT1G14790) is almost exclusive to viral response, but both RDR2 and RDR6 have been 123 shown to act on endogenous mRNA transcripts. A mutant line of each RDR in a

Columbia 0 ecotype background was acquired to determine if these proteins are in a pathway that affects production of antisense transcripts.

The mutant for RDR2 (AT4G11130) is rdr2-2 (SALK_059661), a T-DNA insertion mutant and a seed stock was acquired from the Arabidopsis Biological Resource

Center (ABRC) (Alonso, 2003). The original behavior of rdr2-2 in the production of tasiRNAs was characterized in Vazquez et al. (2004). Homozygous rdr2-2 genotype was confirmed by PCR. The mutant for RDR6 (AT3G49500) is rdr6-11 (CS24285), a single nucleotide nonsense mutation (C805T), also acquired by ABRC. The rdr6-11 mutation was confirmed by its curled-leaf phenotype, and by sequencing. The original behavior of rdr6-11 in the production of tasiRNAs was characterized previously (Peragine et al.,

2004).

As the AtCESA3 antisense transcript had very weak signal compared to that of the

AtCESA1 antisense transcript (Figure 28B), thus AtCESA1 antisense transcript was used as a marker to examine the effects of RDR mutation on CESA antisense transcript levels.

Young, whole inflorescences from a wild type Columbia 0 line (WT Col0), and the two

RDR mutant lines and assessed the AtCESA1 signal strength relative to the RNA loading quantity. Ideally, liquid cultured seedlings would have been used, but a combination of low seed supply and poor mutant germination rates required the change to inflorescence tissues. It was found that both rdr2-2 and rdr6-11 show a decrease in antisense signal relative to the WT Col0 sample (Figure 29). 124

Figure 29: Examination of AtCESA1 antisense transcripts in RDR mutants. AtCESA1 antisense transcript levels in whole Arabidopsis inflorescences appear to be downregulated in rdr6-11 mutants, and possibly in rdr2-2 mutants.

Discussion

By transitioning into an arabidopsis model, we expanded evidence of antisense

CESA transcripts to a third species with the addition of antisense transcripts for AtCESA1 and AtCESA3 to the list (Figure 28). While tissue targeting for these antisense transcripts is considerably less specific that for barley and brachypodium, it provides a starting point for better characterization of the transcripts. It was also noteworthy that predictions of these antisense transcripts were correct based on sequence homology with more distantly related CESAs (Figure 27). However, it should be noted that the mRNA orthologs for barley (HvCESA1 & 6) (R. A. Burton et al., 2004) and the brachypodium (BdCESA1 and

8) (Figure 25) are all relatively highly expressed relative to other CESAs in the tissues examined. If the antisense transcripts are produced from an mRNA source, the most highly expressed CESAs could be more likely to have antisense transcript production just based on population. 125

Preliminary examination of RDR mutants has provided interesting information as well, but without any clear conclusions at this point. Initially, both RDRs seem to be involved in antisense transcript production but follow up studies will be required to confirm these results.

Conclusion

Detection of CESA antisense transcripts in arabidopsis marks the first detection of this type of antisense transcripts in a eudicotyledonous plant. This suggests the mechanism of their synthesis is highly preserved, as the divergence of monocots and dicots occurred approximately 150 million years ago (Chaw, Chang, Chen, & Li, 2004).

We have also gained the first mechanistic insights into how RDRs may be involved in the synthesis of antisense transcripts, but further replication is required before any firm conclusions can be made on that topic.

126

CHAPTER 6: LESSONS LEARNED; REFINEMENT OF A STRAND-SPECIFIC

REVERSE-TRANSCRIPTION ASSAY TO DETECT ANTISENSE TRANSCRIPTS

Introduction

A major hurdle to the accurate identification of antisense transcripts via polymerase chain reaction (PCR) is verifying that only the targeted strand has been amplified during the process. Due to the double stranded and complementary nature of

DNA, traditional reverse transcriptase PCR (RT PCR) is poor at specifically amplifying only a single strand of RNA/DNA if both that strand and its complement are present in solution simultaneously. Every PCR reaction requires at least two primers, one to target the sense strand and one to target the complementary antisense strand. This means that the PCR products of amplification from both the sense and antisense templates can look identical, leading to a false positive for the detection of antisense transcripts.

The mechanisms of cDNA synthesis and PCR rely on the behaviors of enzymes, whose major traits are well studied, but have less well understood behaviors that become important to consider in edge cases such as this. During the process of reverse transcription, the enzyme reverse transcriptase (RT) is utilized to turn RNAs into complementary cDNAs, which is essential for PCR, as RNA is rapidly destroyed by the process of PCR.

Here, I show that stringent negative controls as strict precautions against cross contamination are essential for performing strand-specific RT-PCR (SS-RT-PCR), to get detectable, strand-specific, reproducible, results. 127

Materials and Methods

Design of SS-RT-PCR Tag1 Primer

The 20 base-pair tag (5’-CTTATTCGCCACCATGACCG-3’) was designed by hand starting with a chunk of sequence randomly generated by the Primer3 software using HvCESA6 as a template. The tag was then manually changed one base at a time until it resulted in no strong hits (E<1) when BLASTed against the NCBI database of barley transcripts (E threshold = 1000, word size = 7, match/mismatch scores 1, -3, deselect low complexity regions filter, deselect mask for lookup table only), and had a predicted melting temperature in the range of 60-65°C.

Total RNA Isolation and Purification

RNA isolation was performed per the manufacturer’s protocol for TRIzol® reagent. Fine tipped pipets were used to remove residual supernatant during the isopropanol precipitation and ethanol removal steps to reduce near-230 nm contamination. Reconstituted RNA was treated with TURBO DNase (Ambion) per the manufacturer’s instructions under rigorous conditions to eliminate genomic DNA.

cDNA Synthesis and Dilution

Strand specific cDNA was synthesized using the Superscript III First Strand

Synthesis system with 1700 ng of DNase-treated total RNA and a gene-specific or oligo dT primer per the manufacturer’s protocols. cDNA was diluted at a 1:9 ratio with nuclease free water after synthesis. 128

Spin Column cDNA Cleanup Method

Diluted cDNA was purified using the DNA Clean & Concentrator 5 Kit (Zymo) per the manufacturer’s instructions and cDNA was eluted off the column with the same volume of water.

Generic Method for cDNA Amplification

First-strand cDNAs were amplified by PCR using the corresponding sense primer and antisense primer, depending on the original cDNA synthesis primer. Control cDNA samples were amplified in the same manner. All PCR amplifications were assembled on ice in 25 μl reactions using 5 μl of 5X Green GoTaq buffer (Promega M3001), 0.5 μl of each primer (10 μM), 0.5 μl of dNTPs (10 μM each), 4 µl of diluted cDNA template, and

1.25 units of GoTaq polymerase.

Reactions were cycled with 2 minutes of activation at 95°C, followed by a variable number of cycles of 95°C for 1 min, primer-specific annealing temperature for 1 min, and 72°C for the amplicon-specific extension time. Final elongation was 72°C for 5 minutes. Cycle number for SS-RT-PCR in barley without tags was 38, cycle number for

SS-RT-PCR with tags in barley was 35, and cycle number for SS-RT-PCR with tags in brachypodium was 37.

Results

Identification of Strand Specificity Problems with SS-RT-PCR

Reports on errant detection of the wrong RNA strand by SS-RT-PCR in RNA viruses (Craggs et al., 2001; Kawakami et al., 2011; Purcell, Hart, Kurath, & Winton,

2006) prompted us to test for evidence that this was occurring in the antisense CESA 129 detection system (Figure 30). No-primer cDNA synthesis reactions (cDNA synthesis reaction run without a primer, so no cDNA should form) were amplified via PCR using sense and antisense primers for HvCESA6. Strong signal appeared in the short antisense, medium antisense, and medium sense no primer control (NPC) lanes (Figure 30B), where no signal should appear. No bright bands appeared in the long antisense NPC lane, similarly to the long antisense detection lane in Figure 30A. No bright bands appeared in the no reverse transcriptase (NRT) control lanes, although a dim band of very small size did appear in the short antisense NRT lane, indicating the template amplified in Figure

30B was dependent on the presence of reverse transcriptase.

Figure 30. Evidence of cDNA reverse transcription occurring without primers. A) cDNAs primed using antisense targeting primers for short, medium, and long positions were amplified, identifying putative medium antisense transcripts for the antisense template. B) However, in the NPC negative control, bands appear in the short and medium antisense lanes when no band should appear, with identical sizing to the bands in A. C) Near background level amplification of the short NRT control indicates that the DNase treatment may be insufficient in removing all genomic DNA, and indicates that the source of the contamination in B is likely created by the RT enzyme. Single gel image separated and re-spliced for clarity and page fit.

Improving Antisense Assay Signal and Reproducibility

Low reproducibility in detection of antisense signal was another hurdle during antisense assay (tagged SS-RT-PCR) development. Several technical modifications were 130 examined to mitigate this problem. RNase H treatment and silica spin column purification of the cDNAs were tested in Figure 31 with mixed results. RNase H treatment dramatically increased the signal strength for my antisense target (Figure 31 left), but purification of both RNase H-treated and untreated cDNAs on the spin columns showed no benefit (Figure 31 far-left and center-left).

Figure 31. RNase H treatment and cDNA spin purification condition testing amplifying HvCESA1 antisense transcripts in technical triplicate.

Reproducibility differences were also noted for different thermal cyclers with near-identical programming. In Figure 32, identical templates and cycling programs were used in amplifications on two different instruments a BioRad Dyad Peltier thermal cycler and a forced air heated, high-resolution Qiagen Rotor Gene Q thermal cycler. Both thermal cyclers showed the same pattern of expression with a high initial peak diminishing over time. However, samples cycled on the Qiagen machine had much stronger signal than those cycled on the BioRad machine (direct comparison in Figure 32 lower). 131

Figure 32. Effects of thermal cycler selection on BdCESA1 antisense transcript signal. Condition A and condition B gels are separated by day, with three technical replicates. A significant BdCESA1 antisense signal strength change occurred between thermal cyclers using the same programmed conditions. Asterisks indicate which representative sample from Condition A was separated with the Condition B samples for comparison.

This prompted inquiry into the source of this signal difference. Fundamentally, the two machines should function in the same way, but two possible differences were considered. The first was that the Dyad and Rotor Gene Q have different rates of temperature change, with the Dyad taking more than twice as long to change temperature between some steps. To examine if this was affecting signal strength, the same antisense reaction was cycled with shorter and longer cycle melt times on the Dyad, to see if longer exposures to high temperatures were somehow negatively impacting the reaction. It was found that a shorter melting time reduced signal (Figure 33A), eliminating that possibility.

The second difference was that the Dyad machine was preheated before loading of samples, so there was no long warm up time before cycling occurred. This differed from the Rotor Gene Q, in that it could not be preheated prior to loading, but it had a much more rapid temperature adjustment rate, such that a preheat was not necessary. To check this possibility, the preheat step was not performed for the Dyad, and both instruments were also tested for multiple cycle melt times to examine if there was a 132 combined effect. Figure 33B shows that without the preheat step, the signals from amplifications in the Dyad machine increased to the levels of the Rotor Gene Q, with minimal dependence on the cycle melt time.

Figure 33. Effects of cycle melt times and instrument preheating on PCR. A) Variation in cycle melting time duration in BioRad Dyad (Dy) thermal cycler cannot recover antisense BdCESA1 signal to Qiagen Rotor Gene Q (RGQ) levels. B) Eliminating Dyad thermal cycler preheat step before sample loading recovers signal levels to approximately RGQ levels. Variations in melting time during each cycle seems to have a variable effect.

Discussion

The amplification and detection of signal in multiple unprimed NPC control

(Figure 30B) reverse transcription templates indicated a fundamental flaw in the original

SS-RT-PCR mechanism used to detect and quantify antisense CESA transcripts. This errant signal in the NPC negative control indicated that somehow the RNA in the cDNA synthesis reaction was priming itself in the absence of a DNA oligomer. The lack of signal in the no reverse transcriptase (NRT) control (Figure 30C) indicated that the 133 template for the signal was not genomic DNA (gDNA) and was somehow added by the reverse transcriptase (RT) enzyme.

There are three potential mechanisms that could allow this self-primed cDNA synthesis to occur. The first possibility is the formation of a stable hairpin with a free 3’ end, as seen in Figure 34 (left), which is a common occurrence in RNA. The second possibility is overlap between transcripts that are antisense to each other (Figure 34, middle). In a scenario where sense and antisense complements are already suspected to exist, this seems like a likely source. The third possibility is priming via the annealing of a smRNA, which functionally could serve the same purpose as a DNA primer (Figure 34, right).

Figure 34. Potential mechanisms for unprimed cDNA synthesis From left to right, hairpin mechanism, antisense overlap mechanism, and smRNA-primed mechanism

This led me to the development of the tagged SS-RT-PCR approach depicted in

Figure 35. A 20-nucleotide sequence unique from any sequences in the barley genome was designed and appended to the 5’ end of the antisense targeting gene-specific primer

(GSP) used during cDNA synthesis (see Chapters 3-5). During cDNA synthesis, the unique tag would be incorporated at the 5’ end of the cDNA, so that during PCR amplification, only a sense GSP and a primer with the tag sequence would be used. As 134 the tag primer could only anneal to sequences derived from the cDNA, the problem with unintended amplification of sense transcripts could be avoided entirely. This mechanism had the added benefit of being confirmable by sequencing.

Figure 35. Stepwise mechanism of tagged SS-RT-PCR The mechanism follows the same basic melt, anneal, extend steps of PCR, but the inclusion of the tag sequence during initial cDNA preparation allows for an added level of specificity. The combination of the GSP and tag primer selectively omits amplification from self-primed sense cDNA molecules.

To check for all the possible errors that could occur during tagged SS-RT-PCR, three negative controls are necessary for each antisense assay (Table 10). The first negative control is the TAG Control, which verifies that the tag is not causing off-target amplification. The second negative control is the No Primer Control (NPC), which verifies that the tagged primer system is not amplifying self-primed cDNAs. The last negative control is the No Reverse Transcriptase control (NRT), which verifies that the amplicon is coming from an RNA source and wasn’t templated from genomic DNA. 135

There is also an optional positive control that checks for the approximate size of the antisense transcript, and verifies that the corresponding (sense) mRNA is present. Figure

36 shows an example of a full antisense assay gel with triplicate antisense detection lanes and clean controls.

Table 10

Necessary controls for tagged SS-RT-PCR

136

Figure 36. Example gel showing detection of an antisense transcript with triplicate detection and clean controls, per Table 10.

Further improvements to antisense detection were made based on the optimization study shown in Figure 31. RNase H treatment selectively degrade away residual RNA template from cDNA synthesis, which should make the template more available for priming and amplification (Figure 37). The silica column cleanup should allow for the removal of salts, oligos, and cofactors from the cDNA synthesis reaction, some of which are inhibitory to PCR (Figure 38). 137

Figure 37. Mechanism of action of RNase H in degrading the RNA component of a RNA-DNA hybrid. RNase H preferentially degrades only RNA:DNA hybrids, processively cleaving the RNA strand. This improves PCR reactions by freeing up priming sites otherwise occupied by RNA in the early amplification cycles.

Figure 38. Depiction of effects of cDNA cleanup via spin column. The silica spin column is preferentially bound by long nucleic acids, with shorter nucleic acids, enzymes, and salts being washed away.

Based on the results from Figure 31, it was determined that while RNase H treatment dramatically improved signal strength, silica column cleanup inhibited signal, likely because it was retaining antisense template on the column along with other filtered materials. Additionally, the dilution of template and subsequent larger loading volumes produced much better reproducibility than previous loading of smaller volumes of 138 undilute template (data not shown). The combination of these two changes made the tagged SS-RT-PCR CESA antisense assay consistently replicable.

The last complication with the antisense CESA assay came from unexpected changes in signal strength when changing between different thermal cyclers using the same programs. When initially running the large-scale presence/absence antisense transcript assays, a BioRad Dyad (Peltier) thermal cycler was used. Despite the low antisense signal, the machine produced usable results (see Figure 9 or Figure 19 for examples).

However, later work on detection of antisense transcripts over a time course showed signal variability between technical replicates (data not shown) prompting concerns that the thermal cycling conditions may not have been uniform across the block.

This prompted repeat assays in a Qiagen Rotor Gene Q thermal cycler designed for quantitative RT-PCR (qPCR) and high-resolution melt curves (Figure 32 and Figure 33).

This instrument utilized a centrifugal rotor design such that all samples would constantly be spun through the same, closely-tracked temperature environment. Amplifications done in this machine did show improved reproducibility (Figure 32 upper) but also had signal strength much higher than that from the Dyad (Figure 32 lower).

Follow up studies examining effects of the length of the melting step at each cycle

(Figure 33A) and the effect of preheating the thermal cycler (Figure 33B) showed that the major factor was the presence/absence of an initial preheat step. This assay was designed with a non-Hot-Start enzyme, the original purpose of the preheat was to cut down on off- target or short amplicons produced by premature cycling. However, these results indicate 139 that the preheat step on the Dyad was somehow degrading or inhibiting the PCR amplification in addition to the original purposes. The specific mechanistic causes involved in that inhibition or degradation are still uncertain.

Conclusion

While PCR and RT-PCR are generally considered to be fully mature biochemical techniques, the work here has shown that the mechanisms of PCR are not so robust for edge cases where low-population transcripts must be detected or complementary sequences must be selectively detected. Most of the modifications and controls used here are unlikely to be necessary for everyday PCR, but the fact that they were necessary at all shows that we may not understand the behaviors of RNA, DNA, reverse transcriptase, and DNA polymerase as well as we think. As further work goes into assessing the RNA level of gene regulation, reliance on PCR based techniques are going to require scrutiny to avoid the generation of questionable results via improper experimental design.

The overall effect of these findings on the research presented in Chapters 3-5 is mixed. Chronologically, the development of the tag system and the usage of RNase H occurred while the HvCESA antisense detection was being optimized, so all antisense

CESA results for barley, brachypodium, and arabidopsis used this methodology.

However, discovery of the effects of thermal cycler preheating were not discovered until time courses in brachypodium were almost complete, and indicate that we may have been observing artificially low antisense transcript signals. While this does not invalidate the results seen here, follow up studies may be able to use lower cycle numbers and less stringent methods to obtain the same type of results, possibly with better quantification. 140

CHAPTER 7: FUTURE DIRECTIONS

The CESA antisense transcripts and corresponding siRNAs were originally detected for CESA6 in barley. I have shown that the presence of antisense transcript is not unique to HvCESA6, and that antisense transcripts appear in two additional species, for 7 additional CESA genes. As identification of CESA antisense transcripts moves phylogenetically further from barley, determination of their source becomes more important. The exonic nature of my antisense transcripts points to them being templated from mature mRNA transcripts, though this is still uncertain. If the production of transient antisense transcripts from an RNA source is a conserved mechanism, it could occur for a much wider range of genes than just CESAs.

Future inquiry into the source of antisense transcripts and their broader network effects will require a hard look at recent developments in the fields of siRNA production and mRNA turnover. The mechanisms and behaviors of these pathways resemble some of the characteristics seen for my antisense transcripts. For example, research into the

RDRs that help trigger siRNA production pathways now indicates that the RDRs can act broadly, as they may produce long dsRNA from any target that can form a hairpin or be annealed to a complementary RNA (Devert et al., 2015; Gregory, Malley, et al., 2008;

Tsuzuki et al., 2017). This is in close accord with the proposed mechanism seen in Figure

17. The HvCESA6 genes have already shown that they can self-prime reverse transcription reactions (Figure 30), so the presence of an RNA hairpin or smRNA primer that could prime RDR is already likely present. The hypothetical resulting RDR-produced antisense transcript could have all the same characteristics that we have already seen for 141

CESA antisense transcripts. Additionally, work by Gregory et al. (2008) showed that in an exoribonuclease (XRN) mutant, loss of XRN4 that normally degrades deadenylated mRNA triggered a resulting increase in the amount and type of smRNAs produced, which suggests that mRNAs that escape aspects of the RNA turnover pathway can be shunted into RDR-based siRNA production pathways. This exact model is recently suggested in Tsuzuki et al. (2017).

How do these antisense transcripts and smRNAs fit in the greater cell wall regulatory framework to effect change on the cell wall? I have identified evidence for the existence of these potential post-transcriptional regulators in Chapter 3 through Chapter

5, and I have shown these molecules have dynamic expression patterns that suggest interaction in Chapter 3. Along with work in Held et al. (2008), this shows that there is a broad link between the production of CESA antisense transcripts, the downregulation of corresponding CESA sense transcripts, and that it likely is driven and amplified via smRNAs. The problem that remains is we need a better characterization of fine antisense silencing behavior during the transition from primary cell wall synthesis to secondary cell wall synthesis. A model with more a uniform primary to secondary wall transitions will be necessary. Once identified, simultaneous analysis of wall character, transcript levels of mRNA, antisense RNA, smRNAs, for members of the CESA superfamily, and potentially involved smRNA pathway and RNA turnover mechanisms will be necessary to create a more definite map of how all these parts fit together.

Moving forward, the hypothesis is that CESA antisense transcripts and small

RNAs can influence behavior of cell wall biosynthesis genes. My results suggest that 142 these species can modify the production of cell wall materials via cis or trans-acting post transcriptional mechanisms. If this hypothesis is confirmed, then selective modification of the cell wall via smRNAs and antisense transcripts can be developed for the creation of bioengineered agricultural and bioenergy crops. The following specific aims are proposed for the future work.

 Specific Aim 1. Identify a more specific and granular set of tissues for examining

primary to secondary wall transition. The working hypothesis for this aim is that

an organ starting with a more asymmetric ratio of primary and secondary wall

CESAs will show a more accurate picture of CESA antisense transcript and small

RNA expression. Here, the developmental patterns of barley tissue will be studied

to identify developmental hotspots where primary to secondary cell wall

changeovers occur. Working from quantifications of CESA mRNA expression

data in Burton et al. (2004), the data predicts the most ideal transitional state to

examine is likely developing root tips or coleoptiles. By finding organs enriched

for primary or secondary wall growth, a more precise transition from primary wall

CESAs to secondary wall CESAs can be examined at the transcriptional and post-

transcriptional levels.

 Specific Aim 2. Determine a better way to get a global picture of antisense

transcripts and their relationship with antisense transcripts and small RNAs. Low

throughput methods of SS-RT-PCR, qPCR, and RPA are not suitable for

characterizing mRNAs, antisense transcripts, and siRNAs in a broad manner. By

applying RNASeq to tissue samples well characterized for primary or secondary 143

wall biosynthesis enrichment, a global comparison of the three transcript types

can be made. It is likely that these RNAs exist in equilibrium, and by shifting that

equilibrium at specific points, have regulatory effects. This also allows

examination of whether the presence of an antisense transcript can be predicted

based on the presence of a corresponding smRNA or vice-versa.

 Specific Aim 3. Examine the known RDR pathway mutants in arabidopsis for

mRNA, antisense transcript, and siRNA behavior. The current working hypothesis

for Aim 3 is that the antisense transcripts detected for the HvCESAs are produced

by smRNA production pathway RDR enzymes using CESA mRNAs as templates.

Examining mutants in the smRNA pathway such as RDRs, DCLs, AGOs, XRNs,

and DCPs will give a conclusive answer as to the specific source(s) of the CESA

antisense transcripts.

Specific Aim 1. Identify More Specific and Granular Tissues for Examining Primary to

Secondary Wall Transition

In Specific Aim 1, the focus will be on selecting candidate organs or tissues that show a more definitive transition from primary cell wall production to secondary cell wall production. Whole third leaf or whole plant tissue was previously used to maximize the chances of detecting antisense transcripts, and was successful at that. These studies also showed that HvCESA1 and HvCESA6 antisense transcript levels seem to come up as primary wall sense transcript levels go down, giving me a working hypothesis that CESA antisense transcripts are involved in this mRNA downregulation across the primary to secondary wall transition. However, detection of this transition is not optimal in whole 144 leaf. Barley whole leaf samples contain a mix of many tissues at many stages of development. Each tissue may have a different set of sense/antisense expression patterns, and simultaneous production of different antisense transcripts in adjacent tissues could average out and hide unique signals when assayed in bulk. To closely examine the antisense transcript/smRNA CESA phenomenon, a more specific model is required.

Experimental Design

Identify Tissues Where Primary or Secondary Cell Wall Character Is Dominant

Publicly available literature and expression datasets should be mined for CESA expression data and microscopy images of barley tissues to identify organs or tissues that fit two specific conditions. First, the model should show initial growth that is highly biased for primary wall development, with pursuant growth transitioning to include secondary wall development. The second and more important requirement is that model show high primary wall CESA expression initially that diminishes over time, and low secondary wall CESA expression initially that rises over time. Burton et al. (2004) gives a good starting point with CESA expression data over a range of tissues and developmental stages. The expression data indicate several candidates that can be tested, including roots transitioning from growing tip to mature root, maturing young coleoptiles, or specific slices of first leaf near the base of the leaf.

Characterize CESAs in Selected Organ Model Candidates

Ideally, a whole organ tissue can be used in order to maximize tissue quantity per sample and minimize the difficulties that come with microscale applications. Candidate organs should be assayed via qPCR to examine the CESA transcript levels over time or 145 development. Putative model organs showing the desired downregulation of primary wall

CESAs and upregulation of secondary wall CESAs should be further assessed for wall character.

Examine Cell Wall Character in New Model Where Preexisting Characterization Is

Incomplete

Where such data does not already exist, the physical appearance of tissue cross sections for each candidate model will be studied using dyes known to visualize traits of the cell wall diagnostic to their primary or secondary character. Barley model samples will be fixed, mounted, and sectioned for histochemical staining with dyes known to visualize the cell wall. Primary wall is known to be thinner and un-lignified, with minimal specialization, while secondary wall is generally thicker, enriched in lignin and reduced in pectin. General wall morphology can be examined with Toluidine Blue O, and specific staining for cellulose, lignin, and pectins can be done with calcofluor white, phloroglucinol, and ruthenium red respectively. There are also several other dyes that could serve these purposes.

Models that show a mostly uniform tissue type transitioning from primary to secondary wall should be advanced for analysis in Specific Aim 2. If no organ candidates are identified due to heterogenous tissues or lack of CESA set transition, then individual tissues must be examined.

Isolate Specific Barley Tissues via Laser Capture Microdissection

If organ level studies fail to identify a suitable candidate for studying primary wall CESA to secondary wall CESA, then a tissue specific model will be required. As it is 146 known that cells involve in fiber or xylem development produce thick cell walls, mature xylem cells and their precursors could be a suitable model. To collect samples at that scale, laser capture microdissection would need to be utilized. RNAs isolated in this manner should be analyzed directly in Specific Aim 2, as low microgram yields are common. Note that if RNA yields are too low, standard RNASeq techniques may need to be supplanted for single cell sequencing techniques and linear RNA amplification may be required.

Specific Aim 2. Determine a Better Way to Get a Global Picture of Antisense Transcripts

and Their Relationship with Antisense Transcripts and Small RNAs

In Specific Aim 2, I work to improve the comprehensive detection and examination of sense transcripts, antisense transcripts, and smRNAs for CESAs as well as other genes that may show similar behaviors. The low throughput methodologies of tagged SS-RT-PCR and RPA are costly and difficult to scale up. Therefore, I intend to switch to modern high-throughput sequencing to get a global view of RNA behavior. To complement this, sequenced tissues will also be assessed for cell wall character, hopefully connecting changes at the RNA level to changes in cell wall structure.

Experimental Design

RNASeq Analysis Can Quantify Transcripts in a Strand-specific Manner in Un-pooled

Tissue

A high throughput solution for differentiating between sense and antisense transcripts is to utilize deep RNA sequencing technologies. Small starting masses of sample can be used to generate enormous quantities of high precision sequence, with 147 sufficient technical replication in each run to detect sparse transcripts. The adaptability for use with small RNA masses allows individual biological replicates to be examined instead of tissue pools.

RNASeq Identification of Sense and Antisense Transcripts

Paired-end RNASeq will be performed to detect and quantify both sense and antisense transcripts in barley/brachypodium. Paired end library synthesis preserves the original orientation of the RNA strand, circumventing the standard PCR problem of ambiguous strand origin. To allow for inclusion of antisense transcripts in the library population, synthesis should be done by ribosomal, mitochondrial, and chloroplast RNA depletion instead of poly-A tail selection. This prevents the biased sequencing of only mature mRNA strands. While most sequence reads are expected to be from the sense strand just based on relative quantity, sequencing deeply with 100 million or more reads per sample should be statistically capable of detecting quantifiable amounts of antisense transcript. This technique has the added benefit of detecting all antisense and sense transcripts simultaneously.

RNASeq Identification of smRNAs

Paired end RNASeq will also be utilized to sequence size-fractionated total RNA, selecting for low molecular weight (15 to 30 nucleotide) sRNAs. The low RNA mass required for RNASeq allows the same RNA sample to be used for sequencing both the sense/antisense RNAs and the smRNAs. This improves on the low-throughput methods of SS-RT-PCR and RPAs, which required individual time courses to be run to collect 148 sufficient RNA. As every time course and tissue is slightly different, this was a constant source of error that can now be avoided.

Differential and Relative Gene Expression Analysis of Sense, Antisense, and smRNA

Transcripts

RNASeq data sets will be used to calculate the ratios of sense, antisense, and smRNA transcript for each tissue sample, and across developmental time courses. By examining the ratios of antisense:sense:smRNA, we can see if a steady state balance of these species exists, and predict what changing that ratio might do to the cell wall. This will provide the basis for a post-transcriptional RNA regulation map that can be tested empirically.

Use smRNA Sequencing Data to Predict miRNA and tasiRNA Targets

SmRNA sequencing data will be fed into prediction tools such as psRNATarget and miRDeep2 (Dai & Zhao, 2011; Friedländer, MacKowiak, Li, Chen, & Rajewsky,

2012). miRDeep2 will predict which of the sequenced smRNAs is a miRNA, which provides a candidate list of miRNAs that could trigger trans-acting silencing mechanisms.

PsRNATarget will predict the smRNA targeting sites that siRNA candidates are likely to bind to. By cross referencing the output of these processes, strong candidates for initiator miRNAs and their potential initial and secondary RNA targets will be identified.

Identify Cell Wall Composition of Tissues Sequenced via RNASeq

Tissue mass permitting, barley model regions sequenced selected for RNASeq analysis will have their cell walls characterized. Quantification of cellulose, pectin, and lignin composition of these tissues will allow cross comparison of transcriptional and 149 post-transcriptional patters against cell wall character. This will also provide a baseline for future examinations of these tissues under mutated or drug affected conditions.

Specific Aim 3. Examine the Known RDR Pathway Mutants for mRNA, Antisense

Transcript, and siRNA Behavior

In Specific Aim 3, focus transitions from antisense transcript and smRNA detection/quantification into examination of enzymes that may be involved in their production. Here, arabidopsis is leveraged for its vast mutant library to examine enzymes in the smRNA and mRNA turnover pathways. Mutants that show altered CESA antisense transcript and siRNA levels will become candidates for future mutant studies in barley/brachypodium.

Experimental Design

Survey All Remaining AtCESAs for Antisense Transcripts

The partial survey of antisense transcripts for AtCESA1 and AtCESA3 will be expanded to assess all members of the AtCESA gene family via tagged SS-RT-PCR.

Liquid cultured arabidopsis should be split into root and shoot and assayed independently. Antisense surveys in mature plants should be performed as well in a tissue specific manner. Tissues highly expressing CESAs per existing microarray data should be targeted.

Determine If Any of the Known smRNA or RNA Turnover Pathway Enzymes Are Involved in Antisense CESA and CESA siRNA Production

Characterized mutants in smRNA pathway-affecting enzymes will be surveyed for CESA antisense transcripts and smRNAs using low throughput methods. Tissues with 150 highly expressed CESA antisense transcripts identified previously will be the top priority for this survey, as a stronger signal in WT will make it easier to identify if a mutant causes the signal to change.

Identify All Available Mutants for the Arabidopsis siRNA Synthesis Pathways

The literature will be mined for all characterized smRNA synthesis and mRNA turnover pathway mutants. This list should include but not be limited to mutants of

RDRs, AGOs, DCLs, XRNs, and DCPs. Mutants that are not embryo lethal, and that have an ortholog in barley/brachypodium will be acquired for testing.

Survey for Antisense Transcript and smRNA Levels in Mutant Lines

Selected mutants will be surveyed for AtCESA antisense transcripts and smRNAs. Mutants showing a strong increase or decrease in antisense transcript/smRNA levels will be marked as candidates for future translational mutant studies in barley.

151

REFERENCES

Albersheim, P., Darvill, A. G., Roberts, K., Sederoff, R., & Staehelin, A. (2011). Plant

cell walls. (S. Masson, Ed.) (1st ed.). New York, NY: Garland Science.

Allen, E., Xie, Z., Gustafson, A. M., & Carrington, J. C. (2005). microRNA-directed

phasing during trans-acting siRNA biogenesis in plants. Cell, 121(2), 207–21.

Alonso, J. M. (2003). Genome-wide insertional mutagenesis of Arabidopsis thaliana.

Science, 301(5633), 653–657.

Altschul, S. F., Gish, W., Miller, W., Myers, E. W., & Lipman, D. J. (1990). Basic local

alignment search tool. Journal of Molecular Biology, 215(3), 403–410.

Appenzeller, L., Doblin, M., Barreiro, R., Wang, H., Niu, X., Kollipara, K., Carrigan, L.,

et al. (2004). Cellulose synthesis in maize: isolation and expression analysis of the

cellulose synthase (CesA) gene family. Chemistry and Materials Science, 11(3/4),

287–299.

Arikit, S., Zhai, J., & Meyers, B. C. (2013). Biogenesis and function of rice small RNAs

from non-coding RNA precursors. Current Opinion in Plant Biology, 16, 1–10.

Elsevier Ltd.

Aspeborg, H., Schrader, J., Coutinho, P. M., Stam, M., Kallas, A., Djerbi, S., Nilsson, P.,

et al. (2005). Carbohydrate-active enzymes involved in the secondary cell wall

biogenesis in hybrid aspen. Plant Physiology, 137(3), 983–97.

Atalla, R. H., Hackney, J. M., Uhlin, I., & Thompson, N. S. (1993). Hemicelluloses as

structure regulators in the aggregation of native cellulose. International Journal of

Biological Macromolecules, 15(2), 109–112. 152

Axtell, M. J. (2013). Classification and comparison of small RNAs from plants. Annual

Review of Plant Biology, 64, 137–59.

Bailey, I. W. (1938). Cell wall structure of higher plants. Industrial and Engineering

Chemistry, 30(1), 40–47.

Bari, R., & Jones, J. D. G. (2009). Role of plant hormones in plant defense responses.

Plant Molecular Biology, 69(4), 473–488. Springer Netherlands.

Baron-Epel, O., Gharyal, P. K., & Schindler, M. (1988). Pectins as mediators of wall

porosity in soybean cells. Planta, 175(3), 389–395.

Bartel, D. P. (2004). MicroRNAs: genomics, biogenesis, mechanism, and function. Cell,

116(2), 281–297.

Bauer, W. D., Talmadge, K. W., Keegstra, K., & Albersheim, P. (1973). The structure of

plant cell walls: II. The hemicellulose of the walls of suspension-cultured sycamore

cells. Plant Physiology, 51(1), 174–187.

Baulcombe, D. (2004). RNA silencing in plants. Nature, 431(7006), 356–363.

Baumberger, N., & Baulcombe, D. C. (2005). Arabidopsis ARGONAUTE1 is an RNA

Slicer that selectively recruits microRNAs and short interfering RNAs. Proceedings

of the National Academy of Sciences, 102(33), 11928–33.

Bleeckerl, A. B., & Patterson, S. E. (1997). Last exit : senescence , abscission , and

meristem arrest in arabidopsis. The Plant Cell, 9(July), 1169–1179. American

Society of Plant Biologists.

Boerjan, W., Ralph, J., & Baucher, M. (2003). Lignin biosynthesis. Annual Review of

Plant Biology, 54(1), 519–546. 153

Bologna, N. G., & Voinnet, O. (2014). The diversity, biogenesis, and activities of

endogenous silencing small RNAs in Arabidopsis. Annual Review of Plant Biology,

65, 473–503. Annual Reviews.

Borsani, O., Zhu, J., Verslues, P. E., Sunkar, R., & Zhu, J. K. (2005). Endogenous

siRNAs derived from a pair of natural cis-antisense transcripts regulate salt

tolerance in Arabidopsis. Cell, 123(7), 1279–1291.

Bradley, D. J., Kjellbom, P., & Lamb, C. J. (1992). Elicitor- and wound-induced

oxidative cross-linking of a proline-rich plant cell wall protein: A novel, rapid

defense response. Cell, 70(1), 21–30.

Brodersen, P., Sakvarelidze-Achard, L., Bruun-Rasmussen, M., Dunoyer, P., Yamamoto,

Y. Y., Sieburth, L., & Voinnet, O. (2008). Widespread translational inhibition by

plant miRNAs and siRNAs. Science, 320 VN-(5880), 1185–1190.

Brown, R. M., & Saxena, I. M. (2000). Cellulose biosynthesis: A model for

understanding the assembly of biopolymers. Plant Physiology and Biochemistry,

38(1–2), 57–67.

Burn, J. E. (2002). Functional analysis of the cellulose synthase genes CesA1, CesA2,

and CesA3 in arabidopsis. Plant Physiology, 129(2), 797–807. American Society of

Plant Biologists.

Burton, R. A., Shirley, N. J., King, B. J., Harvey, A. J., & Fincher, G. B. (2004). The

CesA gene family of barley. Quantitative analysis of transcripts reveals two groups

of co-expressed genes. Plant Physiology, 134(January), 224–236.

Burton, R. a, Wilson, S. M., Hrmova, M., Harvey, A. J., Shirley, N. J., Medhurst, A., 154

Stone, B. A., et al. (2006). Cellulose synthase-like CslF genes mediate the synthesis

of cell wall (1,3;1,4)-β-D-glucans. Science, 311(March), 1940–1942.

Caffall, K. H., & Mohnen, D. (2009). The structure, function, and biosynthesis of plant

cell wall pectic polysaccharides. Carbohydrate Research, 344(14), 1879–1900.

Elsevier Ltd.

Cao, X., Springer, N. M., Muszynski, M. G., Phillips, R. L., Kaeppler, S., & Jacobsen, S.

E. (2000). Conserved plant genes with similarity to mammalian de novo DNA

methyltransferases. Proceedings of the National Academy of Sciences, 97(9), 4979–

84.

Carpita, N. C. (1983). Hemicellulosic polymers of cell walls of Zea coleoptiles. Plant

Physiology, 72(2), 515–521.

Carpita, N. C. (1984a). Fractionation of hemicelluloses from maize cell walls with

increasing concentrations of alkali. Phytochemistry, 23(5), 1089–1093.

Carpita, N. C. (1984b). Cell wall development in maize coleoptiles. Plant Physiology,

76(1), 205–12.

Carpita, N. C. (1996). Structure and biogenesis of the cell walls of grasses. Annual

Review of Plant Physiology and Plant Molecular Biology, 47, 445–476. Annual

Reviews 4139 El Camino Way, P.O. Box 10139, Palo Alto, CA 94303-0139, USA.

Carpita, N. C., & Gibeaut, D. M. (1993). Structural models of primary cell walls in

flowering plants: consistency of molecular structure with the physical properties of

the walls during growth. The Plant Journal, 3(1), 1–30.

Carrington, J. C., & Ambros, V. (2003). Role of microRNAs in plant and animal 155

development. Science, 301(July), 336–338.

Carthew, R. W., & Sontheimer, E. J. (2009). Origins and mechanisms of miRNAs and

siRNAs. Cell, 136(4), 642–55. Elsevier Inc.

Chan, S. W.-L., Zilberman, D., Xie, Z., Johansen, L. K., Carrington, J. C., & Jacobsen, S.

E. (2004). RNA silencing genes control de novo DNA methylation. Science,

303(February), 2004.

Chapman, E. J., & Carrington, J. C. (2007). Specialization and evolution of endogenous

small RNA pathways. Nature Reviews Genetics, 8(11), 884–896.

Chaw, S. M., Chang, C. C., Chen, H. L., & Li, W. H. (2004). Dating the monocot-dicot

divergence and the origin of core eudicots using whole chloroplast genomes.

Journal of Molecular Evolution, 58(4), 424–441. Springer-Verlag.

Chellappan, P., Xia, J., Zhou, X., Gao, S., Zhang, X., Coutino, G., Vazquez, F., et al.

(2010). siRNAs from miRNA sites mediate DNA methylation of target genes.

Nucleic Acids Research, 38(20), 6883–6894.

Chen, H.-M., Li, Y.-H., & Wu, S.-H. (2007). Bioinformatic prediction and experimental

validation of a microRNA-directed tandem trans-acting siRNA cascade in

Arabidopsis. Proceedings of the National Academy of Sciences, 104(9), 3318–23.

Chou, Y.-H., Pogorelko, G., & Zabotina, O. A. (2012). Xyloglucan xylosyltransferases

XXT1, XXT2, and XXT5 and the glucan synthase CSLC4 form Golgi-localized

multiprotein complexes. Plant Physiology, 159(4), 1355–1366.

Clarke, A. E., Anderson, R. L., & Stone, B. A. (1979). Form and function of

arabinogalactans and arabinogalactan-proteins. Phytochemistry, 18(4), 521–540. 156

Clouse, S. D., & Sasse, J. M. (1998). Brassinosteroids: Essential regulators of plant

growth and development. Annual Review of Plant Physiology and Plant Molecular

Biology, 49(1), 427–451.

Cocuron, J.-C., Lerouxel, O., Drakakaki, G., Alonso, A. P., Liepman, A. H., Keegstra, K.,

Raikhel, N., et al. (2007). A gene from the cellulose synthase-like C family encodes

a beta-1,4 glucan synthase. Proceedings of the National Academy of Sciences,

104(20), 8550–8555.

Cosgrove, D. J. (2005). Growth of the plant cell wall. Nature Reviews Molecular Cell

Biology, 6(November), 850–861.

Cosgrove, D. J. (2015). Plant expansins: Diversity and interactions with plant cell walls.

Current Opinion in Plant Biology, 25, 162–172. Elsevier Ltd.

Craggs, J. K., Ball, J. K., Thomson, B. J., Irving, W. L., & Grabowska, A. M. (2001).

Development of a strand-specific RT-PCR based assay to detect the replicative form

of hepatitis C virus RNA. Journal of Virological Methods, 94(1–2), 111–120.

Czech, B., & Hannon, G. J. (2011). Small RNA sorting: matchmaking for Argonautes.

Nature Reviews Genetics, 12(1), 19–31. Nature Publishing Group.

Dai, X., & Zhao, P. X. (2011). psRNATarget: a plant small RNA target analysis server.

Nucleic Acids Research, 39(Web Server issue), W155-9.

Darvill, J. E., McNeil, M., Darvill, A. G., & Albersheim, P. (1980). Structure of Plant

Cell Walls: XI. Glucuronoarabinoxylan, a second hemicellulose in the primary cell

walls of suspension-cultured sycamore cells. Plant Physiology, 66(6), 1135–9.

Davis, S. C., Williams, S. E., & Boundy, R. G. (2016). Transportation energy data book. 157

Transportation Energy Data Book (35th ed.). Oak Ridge.

Delmer, D. P. (1999). Cellulose biosynthesis: Exciting times for a difficult field of study.

Annual Review of Plant Physiology and Plant Molecular Biology, 50(1), 245–76.

Annual Reviews 4139 El Camino Way, P.O. Box 10139, Palo Alto, CA 94303-

0139, USA.

Devert, A., Fabre, N., Floris, M., Canard, B., Robaglia, C., & Crété, P. (2015). Primer-

dependent and primer-independent initiation of double stranded RNA synthesis by

purified arabidopsis RNA-dependent RNA polymerases RDR2 and RDR6. PLoS

ONE, 10(3), e0120100.

Dhugga, K. S., Barreiro, R., Whitten, B., Stecca, K., Hazebroek, J., Randhawa, G. S.,

Dolan, M., et al. (2004). Guar seed B -mannan synthase is a member of the cellulose

synthase super gene family. Science, 303(January), 363–367.

Doblin, M. S., Kurek, I., Jacob-Wilk, D., & Delmer, D. P. (2002). Cellulose biosynthesis

in plants: From genes to rosettes. Plant and Cell Physiology, 43(12), 1407–1420.

Donaldson, L. A. (2001). Lignification and lignin topochemistry - An ultrastructural

view. Phytochemistry, 57(6), 859–873.

Dong, Z., Han, M.-H., & Fedoroff, N. (2008). The RNA-binding proteins HYL1 and SE

promote accurate in vitro processing of pri-miRNA by DCL1. Proceedings of the

National Academy of Sciences, 105(29), 9970–5.

Dunoyer, P., Brosnan, C. a, Schott, G., Wang, Y., Jay, F., Alioua, A., Himber, C., et al.

(2010). An endogenous, systemic RNAi pathway in plants. The EMBO Journal,

29(10), 1699–712. Nature Publishing Group. 158

Dwivany, F. M., Yulia, D., Burton, R. A., Shirley, N. J., Wilson, S. M., Fincher, G. B.,

Bacic, A., et al. (2009). The cellulose-synthase like C (CSLC) family of barley

includes members that are integral membrane proteins targeted to the plasma

membrane. Molecular Plant, 2(5), 1025–1039.

Ellis, C. (2002). The Arabidopsis mutant cev1 links cell wall signaling to jasmonate and

ethylene responses. The Plant Cell, 14(7), 1557–1566.

Fagard, M., Desnos, T., Desprez, T., Goubet, F., Refregier, G., Mouille, G., McCann, M.,

et al. (2000). PROCUSTE1 encodes a cellulose synthase required for normal cell

elongation specifically in roots and dark-grown hypocotyls of Arabidopsis. The

Plant Cell, 12(12), 2409–2424.

Fei, Q., Xia, R., & Meyers, B. C. (2013). Phased, secondary, small interfering RNAs in

posttranscriptional regulatory networks. The Plant Cell, 25(7), 2400–2415.

Friedländer, M. R., MacKowiak, S. D., Li, N., Chen, W., & Rajewsky, N. (2012).

MiRDeep2 accurately identifies known and hundreds of novel microRNA genes in

seven animal clades. Nucleic Acids Research, 40(1), 37–52.

Fry, S. C., York, W. S., Albersheim, P., Darvill, A., Hayashi, T., Joseleau, J.-P., Kato, Y.,

et al. (1993). An unambiguous nomenclature for xyloglucan‐derived

oligosaccharides. Physiologia Plantarum, 89(1), 1–3.

Fukuda, H. (1997). Tracheary element differentiation. The Plant Cell, 9(July), 1147–

1156.

Gardiner, J. C., Taylor, N. G., & Turner, S. R. (2003). Control of cellulose synthase

complex localization in developing xylem. The Plant Cell, 15(8), 1740–1748. 159

Gibeaut, D. M., & Carpita, N. C. (1991). Tracing cell wall biogenesis in intact cells and

plants: selective turnover and alteration of soluble and cell wall polysaccharides in

grasses. Plant Physiology, 97(2), 551–561.

Gou, J.-Y., Wang, L.-J., Chen, S.-P., Hu, W.-L., & Chen, X.-Y. (2007). Gene expression

and metabolite profiles of cotton fiber during cell elongation and secondary cell wall

synthesis. Cell Research, 17(5), 422–34. Nature Publishing Group.

Grant, E. H., Fujino, T., Beers, E. P., & Brunner, A. M. (2010). Characterization of NAC

domain transcription factors implicated in control of vascular cell differentiation in

Arabidopsis and Populus. Planta, 232(2), 337–352.

Gregory, B. D., O’Malley, R. C., Lister, R., Urich, M. A., Tonti-Filippini, J., Chen, H.,

Millar, A. H., et al. (2008). A Link between RNA metabolism and silencing

affecting arabidopsis development. Developmental Cell, 14(6), 854–866.

Hammond, S. M. (2005). Dicing and slicing: The core machinery of the RNA

interference pathway. FEBS Letters, 579(26), 5822–5829.

Han, M.-H., Goud, S., Song, L., & Fedoroff, N. (2004). The Arabidopsis double-stranded

RNA-binding protein HYL1 plays a role in microRNA-mediated gene regulation.

Proceedings of the National Academy of Sciences, 101(4), 1093–8.

Handakumbura, P. P., Matos, D. A., Osmont, K. S., Harrington, M. J., Heo, K., Kafle, K.,

Kim, S. H., et al. (2013). Perturbation of Brachypodium distachyon cellulose

synthase A4 or 7 results in abnormal cell walls. BMC Plant Biology, 13(1), 131.

BioMed Central.

He, L., & Hannon, G. J. (2004). MicroRNAs: small RNAs with a big role in gene 160

regulation. Nature Reviews Genetics, 5(7), 522–531.

Held, M. A., Penning, B. W., Brandt, A. S., Kessans, S. A., Yong, W. D., Scofield, S. R.,

& Carpita, N. C. (2008). Small-interfering RNAs from natural antisense transcripts

derived from a cellulose synthase gene modulate cell wall biosynthesis in barley.

Proceedings of the National Academy of Sciences, 105(51), 20534–20539.

Henderson, I. R., Zhang, X., Lu, C., Johnson, L., Meyers, B. C., Green, P. J., & Jacobsen,

S. E. (2006). Dissecting Arabidopsis thaliana Dicer function in small RNA

processing, gene silencing and DNA methylation patterning. Nature Genetics, 38(6),

721–725.

Hill, J. L., Hammudi, M. B., & Tien, M. (2014). The Arabidopsis cellulose synthase

complex: a proposed hexamer of CESA trimers in an equimolar stoichiometry. The

Plant Cell, 26(12), 4834–42.

Holzberg, S., Brosio, P., Gross, C., & Pogue, G. P. (2002). Barley stripe mosaic virus-

induced gene silencing in a monocot plant. The Plant Journal, 30(3), 315–327.

Hong, S.-Y., Seo, P. J., Yang, M.-S., Xiang, F., & Park, C.-M. (2008). Exploring valid

reference genes for gene expression studies in Brachypodium distachyon by real-

time PCR. BMC Plant Biology, 8, 112.

Howell, M. D., Fahlgren, N., Chapman, E. J., Cumbie, J. S., Sullivan, C. M., Givan, S.

A., Kasschau, K. D., et al. (2007). Genome-Wide Analysis of the RNA-dependent

RNA polymerase 6/Dicer-like 4 pathway in arabidopsis reveals dependency on

miRNA- and tasiRNA-directed targeting. The Plant Cell Online, 19(3), 926–942.

American Society of Plant Biologists. 161

Huson, D. H., & Scornavacca, C. (2012). Dendroscope 3: An interactive tool for rooted

phylogenetic trees and networks. Systematic Biology, 61(6), 1061–1067. Springer,

Berlin/Heidelberg.

Hussey, S. G., Mizrachi, E., Creux, N. M., & Myburg, A. A. (2013). Navigating the

transcriptional roadmap regulating plant secondary cell wall deposition. Frontiers in

Plant Science, 4(August), 325. Frontiers.

Jabnoune, M., Secco, D., Lecampion, C., Robaglia, C., Shu, Q., & Poirier, Y. (2013). A

rice cis-natural antisense RNA acts as a translational enhancer for its cognate mRNA

and contributes to phosphate homeostasis and plant fitness. The Plant Cell, 25(10),

4166–82.

Jauvion, V., Rivard, M., Bouteiller, N., Elmayan, T., & Vaucheret, H. (2012). RDR2

partially antagonizes the production of RDR6-dependent siRNA in sense transgene-

mediated PTGS. PLoS ONE, 7(1), e29785. Public Library of Science.

Jeong, D.-H., Schmidt, S. A., Rymarquis, L. A., Park, S., Ganssmann, M., German, M.

A., Accerbi, M., et al. (2013). Parallel analysis of RNA ends enhances global

investigation of microRNAs and target RNAs of Brachypodium distachyon. Genome

biology, 14(12), R145.

Kanabus, J., Bressan, R. A., & Carpita, N. C. (1986). Carbon assimilation in carrot cells

in liquid culture. Plant Physiology, 82(2), 363–368.

Kataoka, Y., & Kondo, T. (1998). FT-IR microscopic analysis of changing cellulose

crystalline structure during wood cell wall formation. Macromolecules, 31(3), 760–

764. American Chemical Society. 162

Kato, Y., & Nevins, D. J. (1986). Fine structure of (1→3),(1→4)-β-D-glucan from Zea

shoot cell-walls. Carbohydrate Research, 147(1), 69–85.

Kawakami, E., Watanabe, T., Fujii, K., Goto, H., Watanabe, S., Noda, T., & Kawaoka, Y.

(2011). Strand-specific real-time RT-PCR for distinguishing influenza vRNA,

cRNA, and mRNA. Journal of Virological Methods, 173(1), 1–6. Elsevier B.V.

Kellogg, E. A. (2015). Brachypodium distachyon as a genetic model system. Annual

Review of Genetics, 49(1), 1–20.

Kimura, S., Laosinchai, W., Itoh, T., Cui, X., Linder, C., & Brown, R. (1999).

Immunogold labeling of rosette terminal cellulose-synthesizing complexes in the

vascular plant Vigna angularis. The Plant Cell, 11(11), 2075–2086.

Knox, J. P. P., Linstead, P. J., King, J., Cooper, C., & Roberts, K. (1990). Pectin

esterification is spatially regulated both within cell walls and between developing

tissues of root apices. Planta, 181(4), 512–521.

Ko, J. H., Jeon, H. W., Kim, W. C., Kim, J. Y., & Han, K. H. (2014). The

MYB46/MYB83-mediated transcriptional regulatory programme is a gatekeeper of

secondary wall biosynthesis. Annals of Botany, 114(6), 1099–1107.

Ko, J. H., Kim, W. C., Kim, J. Y., Ahn, S. J., & Han, K. H. (2012). MYB46-mediated

transcriptional regulation of secondary wall biosynthesis. Molecular Plant, 5(5),

961–963.

Kolpak, F. J., & Blackwell, J. (1976). Determination of the structure of cellulose II.

Macromolecules, 9(2), 273–278. American Chemical Society.

Kou, X., Wang, S., Wu, M., Guo, R., Xue, Z., Meng, N., Tao, X., et al. (2014). Molecular 163

characterization and expression analysis of NAC family transcription factors in

tomato. Plant Molecular Biology Reporter, 32(2), 501–516.

Kubo, M., Udagawa, M., Nishikubo, N., Horiguchi, G., Yamaguchi, M., Ito, J., Mimura,

T., et al. (2005). Transcription switches for protoxylem and metaxylem vessel

formation. Genes and Development, 19(16), 1855–1860.

Kurihara, Y., Takashi, Y., & Watanabe, Y. (2006). The interaction between DCL1 and

HYL1 is important for efficient and precise processing of pri-miRNA in plant

microRNA biogenesis. RNA, 12(2), 206–212.

Kurihara, Y., & Watanabe, Y. (2004). Arabidopsis micro-RNA biogenesis through Dicer-

like 1 protein functions. Proceedings of the National Academy of Sciences, 101(34),

12753–12758.

Kutschera, U., & Schopfer, P. (1986). Effect of auxin and abscisic acid on cell wall

extensibility in maize coleoptiles. Planta, 167(4), 527–535.

Lamport, D. T. a. (1967). Hydroxyproline-O-glycosidic linkage of the plant cell wall

glycoprotein extensins. Nature, 216(5122), 1322–1324. Nature Publishing Group.

Lamport, D. T. a, Kieliszewski, M. J., Chen, Y., & Cannon, M. C. (2011). Role of the

extensin superfamily in primary cell wall architecture. Plant Physiology, 156(1), 11–

Lange, H., Sement, F. M., Canaday, J., & Gagliardi, D. (2009). Polyadenylation-assisted

RNA degradation processes in plants. Trends in Plant Science, 14(9), 497–504.

Law, J. A., & Jacobsen, S. E. (2010). Establishing, maintaining and modifying DNA

methylation patterns in plants and animals. Nature Reviews Genetics, 11(March), 164

204–220. Nature Publishing Group.

León, J., Rojo, E., & Sánchez-Serrano, J. J. (2001). Wound signalling in plants. Journal

of Experimental Botany, 52(354), 1–9. Oxford University Press.

Li, J., Yang, Z., Yu, B., Liu, J., & Chen, X. (2005). Methylation protects miRNAs and

siRNAs from a 3’-end uridylation activity in Arabidopsis. Current Biology, 15(16),

1501–7.

Li, S., Liberman, L. M., Mukherjee, N., Benfey, P. N., & Ohler, U. (2013). Integrated

detection of natural antisense transcripts using strand-specific RNA sequencing data.

Genome Research, 23(10), 1730–1739. Cold Spring Harbor Laboratory Press.

Liepman, A. H., Wilkerson, C. G., & Keegstra, K. (2005). Expression of cellulose

synthase-like (Csl) genes in insect cells reveals that CslA family members encode

mannan synthases. Proceedings of the National Academy of Sciences, 102(6), 2221–

Liu, L., & Chen, X. (2016). RNA quality control as a key to suppressing RNA silencing

of endogenous genes in plants. Molecular Plant, 9(6), 826–836. Elsevier Ltd.

Lobbes, D., Rallapalli, G., Schmidt, D. D., Martin, C., & Clarke, J. (2006). Serrate: a new

player on the plant microRNA scene. EMBO Reports, 7(10), 1052–1058.

Lombard, V., Golaconda Ramulu, H., Drula, E., Coutinho, P. M., & Henrissat, B. (2014).

The carbohydrate-active enzymes database (CAZy) in 2013. Nucleic Acids

Research, 42(D1), D490–D495. Oxford University Press.

Lu, C., Jeong, D.-H., Kulkarni, K., Pillay, M., Nobuta, K., German, R., Thatcher, S. R., et

al. (2008). Genome-wide analysis for discovery of rice microRNAs reveals natural 165

antisense microRNAs (nat-miRNAs). Proceedings of the National Academy of

Sciences, 105(12), 4951–6.

Luttenegger, D. G., & Nevins, D. J. (1985). Transient nature of a (1 → 3), (1 → 4)-β-d-

glucan in Zea mays coleoptile cell walls. Plant Physiology, 77, 175–178.

Mallory, A. C., Reinhart, B. J., Jones-Rhoades, M. W., Tang, G., Zamore, P. D., Barton,

M. K., & Bartel, D. P. (2004). MicroRNA control of Phabulosa in leaf development:

importance of pairing to the microRNA 5’ region. The EMBO Journal, 23(16),

3356–3364.

Manavella, P. a, Koenig, D., & Weigel, D. (2012). Plant secondary siRNA production

determined by microRNA-duplex structure. Proceedings of the National Academy of

Sciences, 109(7), 2461–6.

Matzke, M. A., & Mosher, R. A. (2014). RNA-directed DNA methylation: an epigenetic

pathway of increasing complexity. Nature Reviews Genetics, 15(6), 394–408.

Nature Publishing Group.

Matzke, M., Kanno, T., Daxinger, L., Huettel, B., & Matzke, A. J. (2009). RNA-

mediated chromatin-based silencing in plants. Current Opinion in Cell Biology,

21(3), 367–376.

Mayer, K. F. X., Waugh, R., Brown, J. W. S., Schulman, A., Langridge, P., Platzer, M.,

Fincher, G. B., et al. (2012). A physical, genetic and functional sequence assembly

of the barley genome. Nature, 491(7426), 711–716. Nature Publishing Group.

McCarthy, R. L., Zhong, R., & Ye, Z. H. (2009). MYB83 is a direct target of SND1 and

acts redundantly with MYB46 in the regulation of secondary cell wall biosynthesis 166

in arabidopsis. Plant and Cell Physiology, 50(11), 1950–1964.

McFarlane, H. E., Döring, A., & Persson, S. (2014). The cell biology of cellulose

synthesis. Annual Review of Plant Biology, 65, 69–94.

McNeil, M., Darvill, A. G., Fry, S. C., & Albersheim, P. (1984). Structure and function

of the primary cell walls of plants. Annual Review of Biochemistry, 53(53), 625–663.

Mellon, J. E., & Helgeson, J. P. (1982). Interaction of a hydroxyproline-rich glycoprotein

from tobacco callus with potential pathogens. Plant Physiology, 70(2), 401–405.

American Society of Plant Biologists.

Mitsuda, N., Iwase, A., Yamamoto, H., Yoshida, M., Seki, M., Shinozaki, K., & Ohme-

Takagi, M. (2007). NAC transcription factors, NST1 and NST3, are key regulators

of the formation of secondary walls in woody tissues of Arabidopsis. The Plant Cell,

19(1), 270–80.

Mohnen, D. (2008). Pectin structure and biosynthesis. Current Opinion in Plant Biology,

11(3), 266–277.

Montgomery, T. A., Yoo, S. J., Fahlgren, N., Gilbert, S. D., Howell, M. D., Sullivan, C.

M., Alexander, A., et al. (2008). AGO1-miR173 complex initiates phased siRNA

formation in plants. Proceedings of the National Academy of Sciences, 105(51),

20055–20062. National Academy of Sciences.

Moor, H. (1971). Recent Progress in the Freeze-Etching Technique. Philosophical

Transactions of the Royal Society of London. Series B, Biological Sciences,

261(837), 121–131.

Nakano, M., Nobuta, K., Vemaraju, K., Tej, S. S., Skogen, J. W., & Meyers, B. C. 167

(2006). Plant MPSS databases: signature-based transcriptional resources for

analyses of mRNA and small RNA. Nucleic Acids Research, 34(Database issue),

D731-5.

Nakano, Y., Yamaguchi, M., Endo, H., Rejab, N. A., & Ohtani, M. (2015). NAC-MYB-

based transcriptional regulation of secondary cell wall biosynthesis in land plants.

Frontiers in Plant Science, 6, 288.

Park, W., Li, J., Song, R., Messing, J., & Chen, X. (2002). Carpel Factory, a Dicer

homolog, and HEN1, a novel protein, act in microRNA metabolism in Arabidopsis

thaliana. Current Biology, 12(17), 1484–95.

Pauly, M., & Keegstra, K. (2008). Cell-wall carbohydrates and their modification as a

resource for biofuels. Plant Journal, 54(4), 559–568. Blackwell Publishing Ltd.

Pear, J. R., Kawagoe, Y., Schreckengost, W. E., Delmer, D. P., & Stalker, D. M. (1996).

Higher plants contain homologs of the bacterial celA genes encoding the catalytic

subunit of cellulose synthase. Proceedings of the National Academy of Sciences,

93(22), 12637–12642.

Penning, B. W., Hunter, C. T., Tayengwa, R., Eveland, A. L., Dugard, C. K., Olek, A. T.,

Vermerris, W., et al. (2009). Genetic resources for maize cell wall biology. Plant

Physiology, 151(4), 1703–1728.

Peragine, A., Yoshikawa, M., Wu, G., Albrecht, H. L., & Poethig, R. S. (2004). SGS3

and SGS2/SDE1/RDR6 are required for juvenile development and the production of

trans-acting siRNAs in Arabidopsis. Genes and Development, 18(19), 2368–2379.

Persson, S., Wei, H., Milne, J., Page, G. P., & Somerville, C. R. (2005). Identification of 168

genes required for cellulose synthesis by regression analysis of public microarray

data sets. Proceedings of the National Academy of Sciences, 102(24), 8633–8.

Pfaffl, M. W. (2001). A new mathematical model for relative quantification in real-time

RT-PCR. Nucleic Acids Research, 29(9), 2002–2007.

Pumplin, N., & Voinnet, O. (2013). RNA silencing suppression by plant pathogens:

defence, counter-defence and counter-counter-defence. Nature Reviews

Microbiology, 11(11), 745–60. Nature Publishing Group.

Purcell, M. K., Hart, S. A., Kurath, G., & Winton, J. R. (2006). Strand-specific, real-time

RT-PCR assays for quantification of genomic and positive-sense RNAs of the fish

rhabdovirus, Infectious hematopoietic necrosis virus. Journal of Virological

Methods, 132(1–2), 18–24.

Rajagopalan, R., Vaucheret, H., Trejo, J., & Bartel, D. P. (2006). A diverse and

evolutionarily fluid set of microRNAs in Arabidopsis thaliana. Genes and

Development, 20(24), 3407–3425.

Ramachandran, V., & Chen, X. (2008). Degradation of microRNAs by a family of

exoribonucleases in Arabidopsis. Science, 321(5895), 1490–2.

Rhoades, M. W., Reinhart, B. J., Lim, L. P., Burge, C. B., Bartel, B., & Bartel, D. P.

(2002). Prediction of plant microRNA targets. Cell, 110(4), 513–520.

Richmond, T. a, & Somerville, C. R. (2000). The cellulose synthase superfamily. Plant

Physiology, 124(2), 495–498.

Ritchie, M. E., Silver, J., Oshlack, A., Holmes, M., Diyagama, D., Holloway, A., &

Smyth, G. K. (2007). A comparison of background correction methods for two- 169

colour microarrays. Bioinformatics, 23(20), 2700–2707. Oxford University Press.

Rogers, K., & Chen, X. (2013). Biogenesis, turnover, and mode of action of plant

microRNAs. The Plant Cell, 25(7), 2383–99.

Ruiz-Ferrer, V., & Voinnet, O. (2009). Roles of plant small RNAs in biotic stress

responses. Annual Review of Plant Biology, 60, 485–510.

Scheller, H. V., & Ulvskov, P. (2010). Hemicelluloses. Annual Review of Plant Biology,

61(1), 263–289. Annual Reviews.

Schmid, M., Davison, T. S., Henz, S. R., Pape, U. J., Demar, M., Vingron, M.,

Schölkopf, B., et al. (2005). A gene expression map of Arabidopsis thaliana

development. Nature Genetics, 37(5), 501–506. Nature Publishing Group.

Schwab, R., Palatnik, J. F., Riester, M., Schommer, C., Schmid, M., & Weigel, D.

(2005). Specific effects of microRNAs on the plant transcriptome. Developmental

Cell, 8(4), 517–527.

Scofield, S. R., Huang, L., Brandt, A. S., & Gill, B. S. (2005). Development of a virus-

induced gene-silencing system for hexaploid wheat and its use in functional analysis

of the Lr21-mediated leaf rust resistance pathway. Plant Physiology, 138(August),

2165–2173.

Setterfield, G., & Bayley, S. T. (1961). Structure and physiology of cell walls. Annual

Review of Plant Physiology, 12(1), 35–62.

Showalter, A. M. (1993). Structure and function of plant cell wall proteins. The Plant

Cell, 5(1), 9–23.

Showalter, A. M., & Basu, D. (2016). Extensin and arabinogalactan-protein biosynthesis: 170

glycosyltransferases, research challenges, and biosensors. Frontiers in Plant

Science, 7, 814. Frontiers Media SA.

Sievers, F., Wilm, A., Dineen, D., Gibson, T. J., Karplus, K., Li, W., Lopez, R., et al.

(2014). Fast, scalable generation of high-quality protein multiple sequence

alignments using Clustal Omega. Molecular Systems Biology, 7(1), 539–539.

Simon, S. a, & Meyers, B. C. (2011). Small RNA-mediated epigenetic modifications in

plants. Current Opinion in Plant Biology, 14(2), 148–55. Elsevier Ltd.

Swiezewski, S., Liu, F., Magusin, A., & Dean, C. (2009). Cold-induced silencing by long

antisense transcripts of an Arabidopsis Polycomb target. Nature, 462(7274), 799–

802. Nature Publishing Group.

Taiz, L., & Jones, R. L. (1970). Gibberellic acid, β-glucanase, and the cell walls of barley

aleurone layers. Planta, 84(1), 73–84. Springer-Verlag.

Taketa, S., Yuo, T., Tonooka, T., Tsumuraya, Y., Inagaki, Y., Haruyama, N., Larroque,

O., et al. (2012). Functional characterization of barley betaglucanless mutants

demonstrates a unique role for CslF6 in (1,3;1,4)-β-D-glucan biosynthesis. Journal

of Experimental Botany, 63(1), 381–392.

Tanaka, K., Murata, K., Yamazaki, M., Onosato, K., Miyao, A., & Hirochika, H. (2003).

Three distinct rice cellulose synthase catalytic subunit genes required for cellulose

synthesis in the secondary wall. Plant Physiology, 133(1), 73–83.

Taylor, N. G. (2008). Cellulose biosynthesis and deposition in higher plants. New

Phytologist, 178(2), 239–252.

Taylor, N. G., Howells, R. M., Huttly, A. K., Vickers, K., & Turner, S. R. (2003). 171

Interactions among three distinct CesA proteins essential for cellulose synthesis.

Proceedings of the National Academy of Sciences, 100(3), 1450–1455. National

Academy of Sciences.

Taylor, N. G., Laurie, S., & Turner, S. R. (2000). Multiple cellulose synthase catalytic

subunits are required for cellulose synthesis in Arabidopsis. The Plant Cell, 12(12),

2529–2540.

Tepfer, M., & Taylor, I. E. P. (1981). The interaction of divalent cations with pectic

substances and their influence on acid-induced cell wall loosening. Canadian

Journal of Botany, 59(8), 1522–1525. NRC Research Press Ottawa, Canada.

Thackeray, M. M., Wolverton, C., & Isaacs, E. D. (2012). Electrical energy storage for

transportation—approaching the limits of, and going beyond, lithium-ion batteries.

Energy & Environmental Science, 5(7), 7854. The Royal Society of Chemistry.

The Arabidopsis Genome Initiative. (2000). Analysis of the genome sequence of the

flowering plant Arabidopsis thaliana. Nature, 408(6814), 796–815.

Tsuzuki, M., Motomura, K., Kumakura, N., & Takeda, A. (2017). Interconnections

between mRNA degradation and RDR-dependent siRNA production in mRNA

turnover in plants. Journal of Plant Research, 130(2), 211–226. Springer Japan.

Valent, B. S., & Albersheim, P. (1974). The structure of plant cell walls: V. On the

binding of xyloglucan to cellulose fibers. Plant Physiology, 54, 105–108.

Vance, C. P., Kirk, T. K., & Sherwood, R. T. (1980). Lignification as a mechanism of

disease resistance. Annual Review of Phytopathology, 18(1), 259–288. Annual

Reviews 4139 El Camino Way, P.O. Box 10139, Palo Alto, CA 94303-0139, USA. 172

Vaucheret, H. (2006). Post-transcriptional small RNA pathways in plants: mechanisms

and regulations. Genes and Development, 20(7), 759–771.

Vazquez, F. (2006). Arabidopsis endogenous small RNAs: highways and byways. Trends

in Plant Science, 11(9), 460–468.

Vazquez, F., Gasciolli, V., Crété, P., & Vaucheret, H. (2004). The nuclear dsRNA

binding protein HYL1 is required for microRNA accumulation and plant

development, but not posttranscriptional transgene silencing. Current Biology,

14(4), 346–51.

Vazquez, F., Vaucheret, H., Rajagopalan, R., Lepers, C., Gasciolli, V., Mallory, A. C.,

Hilbert, J. L., et al. (2004). Endogenous trans-acting siRNAs regulate the

accumulation of arabidopsis mRNAs. Molecular Cell, 16(1), 69–79.

Vogel, J. P., Garvin, D. F., Mockler, T. C., Schmutz, J., Rokhsar, D., Bevan, M. W.,

Barry, K., et al. (2010). Genome sequencing and analysis of the model grass

Brachypodium distachyon. Nature, 463(7282), 763–768. Nature Publishing Group.

Voinnet, O. (2005). Induction and suppression of RNA silencing: insights from viral

infections. Nature Reviews Genetics, 6(3), 206–220.

Vorwerk, S., Somerville, S., & Somerville, C. (2004). The role of plant cell wall

polysaccharide composition in disease resistance. Trends in Plant Science, 9(4),

203–209.

Wang, F., Polydore, S., & Axtell, M. J. (2015). More than meets the eye? Factors that

affect target selection by plant miRNAs and heterochromatic siRNAs. Current

Opinion in Plant Biology, 27, 118–124. 173

Wang, H.-Z., & Dixon, R. A. (2012). On–off switches for secondary cell wall

biosynthesis. Molecular Plant, 5(2), 297–303.

Watanabe, T., Totoki, Y., Toyoda, A., Kaneda, M., Kuramochi-Miyagawa, S., Obata, Y.,

Chiba, H., et al. (2008). Endogenous siRNAs from naturally formed dsRNAs

regulate transcripts in mouse oocytes. Nature, 453(7194), 539–543.

Watanabe, Y., Meents, M. J., McDonnell, L. M., Barkwill, S., Sampathkumar, A.,

Cartwright, H. N., Demura, T., et al. (2015). Visualization of cellulose synthases in

Arabidopsis secondary cell walls. Science, 350(6257), 198–203.

Wei, W., Ba, Z., Gao, M., Wu, Y., Ma, Y., Amiard, S., White, C. I., et al. (2012). A role

for small RNAs in DNA double-strand break repair. Cell, 149(1), 101–12. Elsevier

Inc.

Wilder, B. M., & Albersheim, P. (1973). The structure of plant cell walls: IV. A

structural comparison of the wall hemicellulose of cell suspension cultures of

sycamore (Acer pseudoplatanus) and of red kidney bean (Phaseolus vulgaris). Plant

Physiology, 51(5), 889–893.

Willats, W. G. T., Orfila, C., Limberg, G., Buchholt, H. C., Van Alebeek, G. J. W. M.,

Voragen, A. G. J., Marcus, S. E., et al. (2001). Modulation of the degree and pattern

of methyl-esterification of pectic homogalacturonan in plant cell walls: Implications

for pectin methyl esterase action, matrix properties, and cell adhesion. Journal of

Biological Chemistry, 276(22), 19404–19413. American Society for Biochemistry

and Molecular Biology.

Winter, D., Vinegar, B., Nahal, H., Ammar, R., Wilson, G. V., & Provart, N. J. (2007). 174

An “electronic fluorescent pictograph” browser for exploring and analyzing large-

scale biological data sets. (I. Baxter, Ed.)PLoS ONE, 2(8), e718. Public Library of

Science.

Wolber, P. K., Collins, P. J., Lucas, A. B., De Witte, A., & Shannon, K. W. (2006). The

Agilent in situ-synthesized microarray platform. Methods in Enzymology, 410, 28–

57.

Wu, L., Zhou, H., Zhang, Q., Zhang, J., Ni, F., Liu, C., & Qi, Y. (2010). DNA

methylation mediated by a microRNA pathway. Molecular cell, 38(3), 465–75.

Elsevier Ltd.

Xie, L., Yang, C., & Wang, X. (2011). Brassinosteroids can regulate cellulose

biosynthesis by controlling the expression of CESA genes in Arabidopsis. Journal of

Experimental Botany, 62(13), 4495–4506.

Xie, Z., Fan, B., Chen, C., & Chen, Z. (2001). An important role of an inducible RNA-

dependent RNA polymerase in plant antiviral defense. Proceedings of the National

Academy of Sciences, 98(11), 6516–6521. National Academy of Sciences.

Yang, C., Xu, Z., Song, J., Conner, K., Vizcay Barrena, G., & Wilson, Z. a. (2007).

Arabidopsis MYB26/Male Sterile 35 regulates secondary thickening in the

endothecium and is essential for anther dehiscence. The Plant Cell, 19(2), 534–548.

Yang, L., Liu, Z., Lu, F., Dong, A., & Huang, H. (2006). Serrate is a novel nuclear

regulator in primary microRNA processing in Arabidopsis. Plant Journal, 47(6),

841–850.

Yang, Z., Ebright, Y. W., Yu, B., & Chen, X. (2006). HEN1 recognizes 21-24 nt small 175

RNA duplexes and deposits a methyl group onto the 2’ OH of the 3’ terminal

nucleotide. Nucleic Acids Research, 34(2), 667–675.

Yin, L., Verhertbruggen, Y., Oikawa, A., Manisseri, C., Knierim, B., Prak, L., Jensen, J.

K., et al. (2011). The cooperative activities of CSLD2, CSLD3, and CSLD5 are

required for normal arabidopsis development. Molecular Plant, 4(6), 1024–1037.

Yoo, B.-C., Kragler, F., Varkonyi-Gasic, E., Haywood, V., Archer-Evans, S., Lee, Y. M.,

Lough, T. J., et al. (2004). A systemic small RNA signaling system in plants. Plant

Cell, 16(8), 1979–2000.

Yu, B., Bi, L., Zheng, B., Ji, L., Chevalier, D., Agarwal, M., Ramachandran, V., et al.

(2008). The FHA domain proteins Dawdle in Arabidopsis and SNIP1 in humans act

in small RNA biogenesis. Proceedings of the National Academy of Sciences,

105(29), 10073–8.

Yu, B., Yang, Z. Y., Li, J. J., Minakhina, S., Yang, M. C., Padgett, R. W., Steward, R., et

al. (2005). Methylation as a crucial step in plant microRNA biogenesis. Science,

307(5711), 932–935.

Yu, D., Fan, B., MacFarlane, S. A., & Chen, Z. (2003). Analysis of the involvement of an

inducible arabidopsis RNA-dependent RNA polymerase in antiviral defense.

Molecular Plant-Microbe Interactions, 16(3), 206–216. The American

Phytopathological Society.

Yuan, C., Wang, J., Harrison, A. P., Meng, X., Chen, D., & Chen, M. (2014). Genome-

wide view of natural antisense transcripts in Arabidopsis thaliana. DNA Research,

22(3), 233–243. Oxford University Press. 176

Zhai, J., Jeong, D.-H., De Paoli, E., Park, S., Rosen, B. D., Li, Y., González, A. J., et al.

(2011). MicroRNAs as master regulators of the plant NB-LRR defense gene family

via the production of phased, trans-acting siRNAs. Genes and Development, 25(23),

2540–53.

Zhong, R., Demura, T., & Ye, Z.-H. (2006). SND1, a NAC domain transcription factor, is

a key regulator of secondary wall synthesis in fibers of Arabidopsis. The Plant Cell,

18(11), 3158–70.

Zhong, R., Lee, C., & Ye, Z.-H. (2010). Evolutionary conservation of the transcriptional

network regulating secondary cell wall biosynthesis. Trends in Plant Science,

15(11), 625–632. Elsevier Ltd.

Zhong, R., Lee, C., Zhou, J., McCarthy, R. L., & Ye, Z.-H. (2008). A battery of

transcription factors involved in the regulation of secondary cell wall biosynthesis in

Arabidopsis. The Plant Cell, 20(10), 2763–2782.

Zhong, R., Richardson, E. A., & Ye, Z.-H. (2007). Two NAC domain transcription

factors, SND1 and NST1, function redundantly in regulation of secondary wall

synthesis in fibers of Arabidopsis. Planta, 225(6), 1603–1611.

Zhong, R., & Ye, Z. H. (2007). Regulation of cell wall biosynthesis. Current Opinion in

Plant Biology, 10(6), 564–572.

Zhou, J., Lee, C., Zhong, R., & Ye, Z.-H. (2009). MYB58 and MYB63 are transcriptional

activators of the lignin biosynthetic pathway during secondary cell wall formation in

Arabidopsis. The Plant Cell, 21(1), 248–66.

Zhu, Q. H., Spriggs, A., Matthew, L., Fan, L., Kennedy, G., Gubler, F., & Helliwell, C. 177

(2008). A diverse set of microRNAs and microRNA-like small RNAs in developing rice grains. Genome Research, 18(9), 1456–1465.

! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! !

Thesis and Dissertation Services ! !