Fruit Crispness Retention of ‘Honeycrisp’ Apple and Its Progeny

Fruit Crispness Retention of ‘Honeycrisp’ Apple and its Progeny

A DISSERTATION SUBMITTED TO THE FACULTY OF UNIVERSITY OF THE UNIVERSITY OF MINNESOTA BY

Hsueh-Yuan Chang

IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY

Cindy B. S. Tong

July 2020

Let me first express my gratitude to Dr. Cindy Tong. Without her, I probably would not have had the opportunity to pursue a Ph.D. degree. Cindy always believed in my ability and gave me the freedom to explore and develop my research passion. It has been a wonderful journey to work with Cindy on different aspects of postharvest research.

I am grateful to my committee members: Dr. Zata Vickers, Dr. Jim Luby, Dr. Jim

Bradeen and Dr. Matt Clark. They have been very generous with their time and advice.

Many thanks for the guidance on this project and the comments on the publications. I have learned so much about being a scientist.

There are many professors and researchers who provided various support to this project. I thank Dr. Joshua Baller, Dr. Thomas Kono, and Dr. Nelson Garcia for bioinformatic advice and help with developing the pipeline for RNA-Seq data analyses, I also like to thank Dr. Nicholas Howard and Dr. Soon Li Teh for sharing their knowledge and experience in QTL analyses. Furthermore, I thank Dr. Camila Alves for help in conducting qRT-PCR experiment, Jenna Brady for managing the sensory evaluation of apple fruit, Dr. Meng-Hsuan Wu for statistical support, and Dr. Theodore Labuza for use of texture analyzer and his lab.

Lastly, I want to thank my colleagues in the office 358, the postharvest lab, and the fruit breeding lab in the Department of Horticultural Science for their valuable discussion and selfless support over the years.

Abstract

Crispness retention is a unique postharvest trait that ‘Honeycrisp’ apple possesses.

With adequate temperature and moisture conditions, ‘Honeycrisp’ fruit retain their highly crisp texture during long-term storage. This research project aims to extend understanding of the molecular mechanisms underlying crispness retention through studying a breeding population derived from ‘Honeycrisp’ x MN1764. In Chapter1, current knowledge regarding postharvest textural changes in apple fruit was reviewed. In

Chapter 2, sensory evaluation and instrumental methods were applied to quantify fruit crispness of the breeding population. In Chapter 3, transcriptomes of the selected progeny individuals differed in their ability to retain postharvest crispness were compared to identify genes associated with crispness retention. By combining a genetically-related apple population, an improved phenotyping method for measuring fruit crispness, and transcriptomic analyses (RNA-Seq and nCounter®), we were able to identify novel candidate genes for crispness retention of ‘Honeycrisp’ fruit.

Table of Contents

Acknowledgements ...... i Abstract ...... ii List of Tables ...... v List of Figures ...... vii Chapter 1: Literature Review ...... 1 1.1 Introduction ...... 1 1.2 Crispness and firmness ...... 1 1.3 Fruit cell walls ...... 2 1.4 Cell wall-modifying enzymes ...... 3 1.5 Chromosome regions...... 7 1.6 Transcriptomic approaches ...... 8 1.7 Honeycrisp ...... 11 1.8 Conclusion ...... 13 Chapter 2: Correlations Between Sensory and Instrumental Crispness of a ‘Honeycrisp’ Apple Breeding Population ...... 16 2.1 Introduction ...... 16 2.2 Materials and Methods ...... 20 2.2.1 Plant Materials ...... 20 2.2.2 Sensory Evaluation ...... 20 2.2.3 Instrumental Tests ...... 22 2.2.4 Data Analyses ...... 23 2.3 Results ...... 26 2.3.1 Sensory Crispness ...... 26 2.3.2 Instrumental Measurements ...... 27 2.3.3 Correlations between Sensory Crispness and Instrumental Measures ...... 28 2.3.4 Application of Principal Component Analysis to Instrumental Measures.. 29 2.3.5 Multiple Linear Regression Models...... 30 2.4 Discussion ...... 30 2.4.1 Comparison of the Instrumental Methods...... 30

iii

2.4.2 Identification of Individuals with Fruit that Retain or Lose Crispness ...... 32 2.4.3 Instrumental Prediction of Crispness and Change in Crispness ...... 33 2.5 Conclusions ...... 34 Chapter 3: Transcriptome analyses identifying genes associated with crispness retention of ‘Honeycrisp’ fruit and its progeny...... 46 3.1 Introduction ...... 47 3.2 Materials and Methods ...... 47 3.2.1 Plant materials ...... 49 3.2.2 RNA sample preparation and RNA sequencing ...... 50 3.2.3 Differential expression analysis ...... 51 3.2.4 Gene validation using NanoString nCounter® and qRT-PCR ...... 52 3.3 Results ...... 53 3.3.1 Phenotype and transcriptome variations among the individuals...... 54 3.3.2 Functional analyses of differentially-expressed genes ...... 54 3.3.3 The expression patterns of auxin- and ethylene-related genes ...... 56 3.3.4 The expression patterns of cell wall-related genes ...... 59 3.3.5 RNA-Seq results validation using nCounter® technology ...... 59 3.4 Discussion ...... 61 3.4.1 Fruit ripening and crispness retention ...... 61 3.4.2 Cell wall-related genes and crispness retention ...... 61 3.5 Conclusion ...... 66 Bibliography ...... 93 Appendices ...... 106

List of Tables

Table 2-1 Sensory texture attributes and their definitions...... 35

Table 2-2 Instrumental measures generated by puncture, mechanical-acoustic, and snapping tests...... 36

Table 2-3 Sensory crispness, puncture force (PF), force linear distance (FLD), maximum force (F2), and maximum acoustic pressure (AUX1) of the 20 individuals and the two parents from the ‘Honeycrisp’ x MN1764 family at harvest and after 8-week storage. Statistical significance of differences (diff.) between fresh and stored fruit of each individual were determined by ANOVA...... 37

Table 2-4 Correlation coefficients (r) between sensory crispness and the instrumental measures, and change in sensory crispness and change in instrumental measures at harvest and after 8-week storage. Twenty individuals and the 2 parents of the ‘Honeycrisp’ x MN1764 family, for which data at harvest and after storage were collected, were used in the first sensory crispness correlation analysis, while 18 crisp individuals from the ‘Honeycrisp’ x MN1764 family and ‘Honeycrisp’ were used in the second correlation analysis. Bold values are significant coefficients (p < 0.05)...... 39

Table 2-5 Factor loadings of the instrumental measures and the sensory attributes on the first four principal components, and the proportion of the total variance explained by each component...... 40

Table 2-6 The best models for predicting sensory crispness from instrumental measures as determined by stepwise multilinear regression. Twenty-two individuals were used in the prediction models of sensory crispness, while the 18 individuals from the ‘Honeycrisp’ x MN1764 family and ‘Honeycrisp’ were used in the prediction model of change in crispness. PF = puncture force, F3 = final force, FLD = force linear distance, Y= Young’s modulus, F2 = maximum force, and AUX1 = maximum acoustic pressure.41

Table 3-1 Differentially-expressed genes (DEGs) with functions associated with cell wall synthesis or modification identified at harvest, or after 2-month cold storage. The expression levels of selected cell wall-genes were compared between (A) “Retain” vs. “Non-crisp”, (B) “Retain” vs. “Lose”, and (C) ‘Honeycrisp’ vs. MN1764. - = no DEGs...... 72

Table 3-2 The numbers of gene counts in the fruit samples at harvest and after 8-week ® cold storage measured using NanoString nCounter technology. (A) Genes that were highly expressed in ‘Honeycrisp’ (HC) and/or the “Retain” group fruit, and (B) genes that were highly expressed in MN1764 (MN) and/or “Lose” group fruit. The genes that were differentially-expressed in both parent and progeny samples were the primary candidates, while the genes that were only differentially-expressed in the parent or the progeny samples were the secondary candidates...... 75

Table A1 Alignment summary of the RNA-Seq reads to the apple reference genome. The numbers shown are the averages of three biological replicates...... 106

Table A2 nCounter® CodeSet Design...... 107

Table A3 Primers for genes tested in qRT-PCR...... 112

List of Figures

Figure 2-1 Force (black line) and acoustic (red line) curves generated by the mechanical- acoustic test for ‘Honeycrisp’ (top) and MN1764 (bottom) fruit at harvest (left) and after 8 weeks of storage (right). Representative plots are shown...... 42

Figure 2-2 Force-deformation curves generated by the snapping test for ‘Honeycrisp’ (top) and MN1764 (bottom) fruit at harvest (left) and after 8 weeks of storage (right). Representative plots are shown...... 43

Figure 2-3 Principal component analysis of all 19 instrumental measures generated by puncture, mechanical-acoustic, and snapping tests related to fruit texture. Data measured at harvest and after 8-week storage were included. A) Instrumental measure and sensory attribute plot, and B) individual individual plot. The abbreviations for the different instrumental measures can be found in Table 2. HC = ‘Honeycrisp’; MN = MN1764. ... 44

Figure 2-4 Principal component analysis of changes in the instrumental measures between harvest and 8 weeks of storage. The changes were calculated as data at harvest minus data after storage. A) Instrumental measure and sensory attribute plot, and B) individual individual plot. The abbreviations for the different instrumental measures can be found in Table 2. HC = ‘Honeycrisp’...... 45

Figure 3-1 Three instrumental texture measures, including (A) puncture force, (B) force linear distance, and (C) acoustic pressure, of the parents (Honeycrisp and MN1764) and the progeny individuals at harvest and after 2-month cold storage. The results were obtained from three-year measurements (2016-2018). For each time point, five fruit were sampled from each parent and individual. The symbols indicate statistical significances: ns = not significant, * p < 0.05, ** p < 0.01, *** p < 0.01, **** p < 0.001...... 79

Figure 3-2 Multidimensional scaling (MDS) plot based on the expression levels of the top 500 most divergent genes. The distance between each pair of samples is the root- mean-square deviation for the top genes. HCH = ‘Honeycrisp’ at harvest, HCS = ‘Honeycrisp’ after 8-week cold storage, MNH = MN1764 at harvest, MNS = MN1764 after 8-week cold storage. Each symbol represents a replicate sample...... 80

Figure 3-3 Venn diagram showing the number of differentially-expressed genes (DEGs) commonly identified between the three comparisons. (A) DEGs highly expressed in

vii

‘Honeycrisp’ (HC) and “Retain” individuals, and (B) DEGs highly expressed in MN1764 (MN), “Lose”, and “Non-crisp” individuals...... 81

Figure 3-4 Enriched Gene Ontology (GO) terms associated with the differentially- expressed genes (DEGs) distinguished between the (A) “Retain” and “Non-crisp” groups, (B) “Retain” and “Lose” groups, and (C) ‘Honeycrisp’ and MN1764 at harvest and after 2-month storage. A false discovery rate (FDR) < 0.05 was used as the threshold for identifying significantly enriched GO terms...... 84

Figure 3-5 The expression patterns of the differentially-expressed genes (DEGs) involved in the auxin-activated signaling pathway. Two genes, (A) ARF and (B) AUX/IAA, that are associated with auxin signaling, and two genes, (C) SAUR, and (D) GH3, that are associated with auxin response were studied. The relative expression is the ratio of gene expression compared to the average. H = at harvest, and S = after 8-week cold storage...... 88

Figure 3-6 The expression patterns of the differentially-expressed genes (DEGs) involved in (A) ethylene biosynthesis, (B) signaling, and (C) response. The relative expression is the ratio of gene expression compared to the average. H = at harvest, and S = after 8-week cold storage...... 91

Figure 3-7 Correlation between gene expression levels using RNA-Seq and nCounter® platforms...... 92

Figure A1 A schematic description of the workflow of this study...... 113

Figure A2 Comparisons among the expression of selected genes using RNA-Seq, nCounter®, and qRT-PCR methods. The error bars are standard error of the mean (n = 9). RH = “Retain” individuals at harvest, RS = “Retain” individuals after 8-week cold storage, LH = “Lose” individuals at harvest, LS = “Lose” individuals after 8-week cold storage...... 114

viii

Chapter 1: Literature Review

1.1 Introduction

Apple (Malus domestica Borkh.) fruit go through a series of physiological changes during ripening, such as conversion of starch to sugars, accumulation of anthocyanins, production of aromatic compounds, reduction of organic acid contents, and losses in firmness and crispness. These changes are essential for making fruit edible and appealing to seed-dispersing animals. On the other hand, fruit ripening, mainly due to loss of crispness, can be unfavorable for long-term postharvest storage. While consumers associate apple fruit crispness with texture liking and freshness (Hampson et al., 2000;

Péneau et al., 2006; Verma, 2014; Cliff et al., 2016), losing crispness could lead to important economic losses. According to a 2018 FAO report, apple is the fourth fruit crop after citrus, grape, and banana in production quantity (FAOSTAT, http://www.fao.org/faostat). More than 86 million tons of apple were produced worldwide, of which 45% were produced in China, followed by 5% in the United States.

The annual production value of apple fruit in the USA was estimated as 5 billion USD, with a 15 billion USD downstream economy (US Apple Association, https://www.usapple.org/). How to more effectively manage ripening and maintain fruit crispness during long-term storage is a major focus of postharvest research on apple fruit.

1.2 Crispness and firmness

Most studies on factors affecting postharvest texture investigate fruit softening, and measure loss of firmness. Firmness is defined as the force needed to bite through fruit

(Harker et al., 2002), and can be reliably measured instrumentally, usually using a penetrometer. For fruit that soften greatly as they ripen (e.g. tomato, peach, kiwifruit),

1 firmness can be an adequate indicator to reflect consumers’ sensation and acceptability

(Stec et al., 1989; Batu, 2004; Miceli et al., 2010). Apple fruit show moderate softening during ripening and storage, and crispness rather than firmness is the major sensory attribute accounting for consumers’ texture appreciation (Hampson et al., 2000).

Crispness is primarily an acoustical sensation, which is defined as the sound generated when fruit is bitten with the front teeth (Harker et al., 2002). A direct and effective way to measure apple fruit crispness is using sensory evaluation. However, sensory evaluation requires a well-trained panel, and can be time consuming and expensive to perform.

Researchers tend to choose instrumental methods to evaluate the texture of apple fruit.

Apple fruit crispness and firmness are distinguishable, but correlated, attributes

(McKay et al., 2011). Firmness measured with a penetrometer has been shown to be positively correlated with sensory crispness, but the accuracy of the correlation depends on the apple germplasm (King et al., 2000; Harker et al., 2002; Brookfield et al., 2011).

More reliable instrumental data can be obtained by including the acoustic properties of apple fruit in the analyses, such as the contact acoustic emission method developed by

Zdunek et al. (2010), and the mechanical-acoustic method developed by Costa et al.

(2011). Since apple fruit exhibit a wide range of phenotypic variations, the extent to which these methods can be used as proxy measures of apple sensory properties needs further investigation.

1.3 Fruit cell walls

The structural integrity of the cell wall is considered the main factor contributing to the crispness of fruit. Plant cell walls are comprised of three classes of polysaccharides: cellulose, hemicellulose and pectin. Cellulose forms the framework of

2 the cell walls, while hemicellulose attaches to cellulose microfibrils forming the cellulose-matrix network that keeps the cell wall rigid (Carpita and Gibeaut, 1993).

Pectin is the main component of middle lamella, a region considered important for maintaining cell-to-cell adhesion (Harker et al. 1997; Wakabayashi 2000).

Many studies of fruit softening are based on the hypothesis that cell wall breakdown causes loss of firmness. One way to describe cell wall breakdown is increases in depolymerization and/or solubilization of cell wall components during the softening process. The pronounced biochemical changes in the cell walls of ripening apple fruit usually occur in the pectin layers, including an increase in water-soluble pectin accompanied with a loss of galactose and arabinose, the major sugar residues in pectin

(Yoshioka et al., 1992; Massiot et al, 1996). Enzymes whose biochemical activities relate to the observed wall changes have been the foci of many scientific investigations.

1.4 Cell wall-modifying enzymes

Polygalacturonase (PG) catalyzes the hydrolysis of polygalacturonan, the major constituent of pectin. Significant increases in PG gene expression and enzyme activity were found in different apple varieties during ripening and postharvest storage, which coincided with softening of the fruit (Abeles and Biles, 1991; Goulao et al., 2007;

Gwanpua et al., 2016A). The role of MdPG in softening was further demonstrated by

Wakasa et al. (2006) who showed that the 14 apple cultivars with fruit that retained firmness after 12 days of harvest and stored at room temperature had relatively low expression of MdPG compared to other cultivars with fruit that softened. Moreover,

‘Royal Gala’ trees transformed by antisense silencing of MdPG, produced fruit showing minor changes in firmness after a 32 week-storage period at 5 °C (Atkinson et al., 2012).

Despite a lot of research on PG, its role in fruit softening is not fully understood. Low enzyme activity of PG has been found in some apple varieties during ripening, concomitant with increases in pectin solubilization and fruit softening (Yoshioka et al.,

1992; Gwanpua et al., 2014). These inconsistencies have led to the investigation of other cell wall-related enzymes.

It has been hypothesized that the cleavage of sugar residues from the side chains of pectin polymers could be the first step in pectin modification during fruit ripening, involving the enzymes α-L-arabinofuranosidase (α-AF) and β-galactosidase (β-GAL)

(Gwanpua et al., 2017). The expression of these two genes and their enzyme activities have been correlated with apple fruit softening. For example, Wei et al., (2010) found that gene expression and enzyme activity of α-AF were higher in ‘Golden Delicious’ than in ‘Fuji’ fruit, which is associated with the rapid softening of ‘Golden Delicious’ fruit.

Nobile et al. (2011) identified an apple α-AF gene, Mdα-AF3, which was highly expressed in mealy fruit of individuals of a breeding population. Ng et al., (2015) showed that the low β-GAL activity of ‘Scifresh’ apple fruit could contribute to its slow softening rate. Yang et al. (2018) showed that the expression levels of three apple β-Gals (Mdβ-

GAL1, Mdβ-GAL2, and Mdβ-GAL5) were higher in the softer ‘Fuji’ fruit than in the firmer ‘Qinguan’ fruit during ripening.

Pectin methylesterase (PME), an enzyme that removes the methyl groups from pectin, could facilitate the degradation of pectin (Brummell and Harpster, 2001).

Yoshioka et al. (1992) observed that water-soluble polyuronide had a lower degree of methylation, which supported the idea that demethylation of pectin could enhance pectin solubilization in apple fruit. Also, increased PME activity has been reported in apple

4 varieties with fruit that soften during storage, including Jonagold and Fuji (Wei et al.,

2010; Gwanpua et al., 2014). On the other hand, Ng et al. (2013) found that the highly esterified-pectin and low PME activity observed in ‘Scifresh’ fruit during the early stage of fruit development were related to its slow softening, contrasted with that of ‘Royal

Gala’ apple, which softens.

There are pectin-modifying enzymes identified in other fruit crops, but their roles in apple ripening and softening have not been established. For example, pectate lyase

(PL), a pectin-degrading enzyme, has been found to be critical in the softening processes of berry fruits such as tomato, strawberry and grape (Nunan et al., 2001; Figueroa et al.,

2008; Uluisik et al., 2016). In apple, the highest enzyme activity of pectate lyase was found at fruit set in ‘Mondial Gala’, which may not be directly related to fruit softening

(Goulao et al., 2007). Another study showed that lower pectate lyase activities were observed in firmer apple fruit stored under low oxygen and treated with calcium dips

(Ortiz et al., 2011). Galacturonosyltransferase (GAUT) is an enzyme associated with pectin biosynthesis. In a study using eight progeny individuals from an experimental cross, upregulation of GAUT gene was observed in apple fruit at harvest and during 2- month storage, which suggests that new pectic components can still be incorporated into cell walls after fruit development is completed (Dheilly et al., 2016).

Although the contents of cellulose and hemicellulose remain relatively constant in the cell wall during apple fruit ripening (Bartley, 1976), enzymes involved in reorganizing the cellulose–hemicellulose network have been proposed to affect fruit texture by loosening the cell wall (Feng et al., 2008; Muñoz-Bertomeu et al., 2013).

Xyloglucan is the most abundant hemicellulose in plant cell walls (Hayashi, 1989), and

5 xyloglucan endotransglucosylase/hydrolase (XTH) is the major enzyme affecting cell wall xyloglucan composition and amount. XTHs can act as transglucosylases or hydrolases, integrating newly secreted xyloglucan to cell walls, restructuring existing cell wall materials, or hydrolyzing xyloglucan (Rose et al., 2002). There is a high number of genes encoding XTH in apples (Zhang et al., 2017), and the members of the XTH gene family could function differently during fruit growth and development. The overall activity of XTH is highest during development of ‘Braeburn’ fruit, declining after harvest as the fruit soften (Percy et al., 1996). The expression of MdXTH2, MdXTH10, and

MdXTH11 are the most abundant among the MdXTHs in ripe ‘Granny Smith’ fruit

(Atkinson et al., 2009), and ethylene treatment enhanced the expression of MdXTH10 in

‘Golden Delicious’ fruit (Muñoz-Bertomeu et al., 2013).

Expansins act as wall-loosening proteins by disrupting the non-covalent bonds between hemicellulose and cellulose (Cosgrove, 1998). According to a recent report, expansins target the sites where cellulose microfibrils are in close contact with one another (Cosgrove, 2015). A possible role for expansins in fruit softening has been proposed (Hayama et al., 2006; Schlosser et al., 2008). Forty-one members of the expansin gene (MdEXP) family were identified in the apple genome (Zhang et al., 2014), and MdEXP1, 2, 3, 5, 8 were found to express during the ripening stage of apple fruit

(Ireland et al., 2014). Wakasa et al. (2006) examined the expression of MdExp3 in 14 apple cultivars with the trait of fruit softening, but their results did not show a strong correlation between the expression of the MdExp3 gene and fruit softening.

1.5 Chromosome regions

To investigate the genetic complexity of fruit texture, a quantitative trait loci

(QTL) approach has been applied where the association between traits and chromosome regions were determined. In apple studies, QTL were commonly identified in single full- sib populations composed of hundreds of individuals from a crossing of two parents.

Using a population derived from ‘Prima’ x ‘Fiesta’, King et al. (2000) identified three linkage groups (LGs), LG01, LG10, and LG16, associated with fruit texture. Specifically,

LG01 and LG10 were associated with firmness measured instrumentally, while LG16 was associated with sensory crispness. Based on a mechanical-acoustic method that measures crispness, Longhi et al., (2012) detected QTL on LG10 and LG16 in ‘Fuji’ x

‘Delearly’ and ‘Fuji’ x ‘Pink Lady’, respectively, for crispness after 2-month storage. In addition to crispness, studies on the discovery of texture QTL also including the identification of two QTL located on LG10 for mealiness in ‘Orin’ × ‘Akane’ (Moriya et al., 2015), and a QTL on LG14 for firmness in ‘Telamon’ x ‘Braeburn’ (Kenis et al.,

2008).

QTL studies can provide information useful in selecting candidate genes associated with traits of interest. Genes with functions related to fruit texture have been found within, or close to, specific QTL regions. For example, MdACO1 and MdACS1, two genes involved in ethylene biosynthesis, are located within QTL on LG10 and LG15

(Costa et al., 2005). Two cell wall-modifying genes, MdExp7 and MdPG were found to lie within QTL on LG1 and LG10 (Costa et al., 2008; Costa et al., 2010).

To discover more comprehensive genomic targets, pedigree-based analysis and genome-wide association studies (GWAS) have been applied. By including six full-

7 sibling families, Di Guardo et al. (2017) identified QTL for mechanical and acoustic properties of fruit on LG02, LG10 and LG14. The GWAS by Di Guardo et al. (2017) was based on 387 apple accessions, and identified associations between fruit acoustic properties and chromosomes 2, 14, and 15. In a study of mealiness, Moriya et al. (2017), discovered significant marker-trait associations on chromosomes 2, 9, and 10 using 82 apple accessions. Different genomic regions associated with crispness on chromosomes 5 and 13 were reported by Amyotte et al. (2017), who combined sensory evaluation and

GWAS using 170 cultivars.

1.6 Transcriptomic approaches

Several genes and chromosome regions throughout the genome have been implicated in affecting texture of apple fruit, but their roles have not been precisely determined. Transcriptomic approaches that provide comprehensive profiles of gene expression in samples can be applied to clarify the relationships among genes involved in this quantitative trait, as well as identify novel candidates. Microarray and RNA-Seq experiments are the two major transcriptomic techniques widely used to analyze and quantify complete sets of transcripts in samples.

In apple, microarrays have been used to profile the gene expression patterns in fruit of ‘Golden Delicious’ and ‘Granny Smith’, two apple varieties that differ in the amount of ethylene production and rate of cell wall-dismantling during ripening (Tadiello et al., 2016). With this approach, 15 ethylene genes were identified as differentially- expressed between the two varieties, including those involved in ethylene biosynthesis and signal transduction. After application of postharvest 1-methylcyclopropene (1-MCP), an ethylene action inhibitor, a group of auxin-related genes were found to be highly

8 expressed, suggesting involvement of auxin in apple fruit ripening. In a comparison among the transcriptomes of 3 mealy and 3 non-mealy fruit of apple individuals from an experimental hybrid, a total of 53 differentially-expressed genes (DEGs) were identified including one cell wall gene, MdPME2, and one wall-associated kinase gene (Segonne et al., 2014). Hypotheses regarding fruit texture typically focus on cell wall-modifying enzymes causing softening of fruit, but Segonne et al. (2014) found that a relatively high and stable expression of MdPME2 was observed in all the non-mealy individuals at harvest and during storage, whereas its expression decreased in mealy fruit. Expression of other cell wall-related genes, such as MdPG1, were found to be inconsistent among the different types of fruit.

The release of a reference apple genome has facilitated sequence-based approaches in apple research. The first apple reference genome was completed in 2010 by

Velasco et al. based on ‘Golden Delicious’. However, due to its relatively large genome size (~750 Mb), high level of repeat content, and heterozygosity, the first apple reference genome remained incomplete with inaccurate contig positions (Khan et al. 2012). A more complete and more accurate apple genome sequence was obtained by Daccord et al.

(2017), who used new sequencing and assembly technologies, and a double-haploid

‘Golden Delicious’ apple (GDDH13) to overcome these issues.

As a newer technology, RNA-Seq has not been extensively applied to study the ripening-related textural changes of apple fruit. Qi et al. (2017) used RNA-Seq to investigate the transcriptomic regulation of fruit ripening in ‘Royal Gala’. A total of 860

DEGs were identified between mature and ripe fruit, and an enrichment of genes involved in ethylene biosynthesis and secondary metabolism was observed. Several cell

9 wall-modifying genes, including MdPG1, XTHs, MdExp3, were differentially expressed, as well as other genes with functions relevant to cell wall structure and phosphorylation.

RNA-Seq experiments have been performed for Rosaceae fruit other than apple.

A transcriptomic study of pear (Pyrus communis L. cv Abate Fetel) by Busatto et al.

(2019) revealed similar types of ripening- and softening-related genes to those identified in apple and other climacteric fruit. For example, a set of ethylene and cell wall- modifying genes were positively related to ripening and fruit softening, while auxin- related genes were mostly down-regulated during fruit ripening. In a comparison between the transcriptome of fast- and slow-melting peach varieties (Prunus persica L.), several ethylene and cell wall-modifying genes were also distinguished (Li et al., 2015). Besides the core genes encoding essential metabolic functions, the DEGs identified by Li et al.

(2015) using RNA-Seq also included a number of transcription factors and novel genes that could potentially be involved in fruit softening.

Expression levels of genes identified in transcriptomic analyses are usually validated using quantitative reverse transcriptase polymerase chain reaction (qRT-PCR).

This validation process has been particularly important for apple. For example, the genomic polymorphisms between the studied apple varieties and the reference genome,

‘Golden Delicious’, can cause inaccurate estimation of gene expression (Hargarten et al.,

2018). Restricted by the lower number of genes to be tested at once using qRT-PCR, there are several medium-throughput methods developed to fill the gap between microarray/RNA-Seq and qRT-PCR. The BioMark® RT-PCR system (Fluidigm; http://www.fluidigm.com), FlexMAP® detection system (Luminex; http://www.luminexcorp.com), and Nanostring nCounter® system (NanoString;

10 https://www.nanostring.com) are the examples. Unlike most RNA quantification methods, nCounter® directly counts the number of targeted mRNA in a sample, using a capture probe and a reporter probe, without the reverse transcription or PCR (Kulkarni,

2011), and thus reduces the potential variations caused by the conversion of mRNA to cDNA and amplification of cDNA. nCounter® has been applied in some transcriptomic studies of horticultural crops to validating the microarray (Shalom et al., 2012; Więsyk et al., 2018) or RNA-Seq results (Gálvez et al., 2016; Neilson et al., 2017; Barra et al.,

2019).

1.7 Honeycrisp

‘Honeycrisp’ is an apple cultivar with fruit that retains crispness over long-term storage, and researchers have been interested in understanding the mechanisms underlying this trait. In studying cellular changes, Tong et al. (1999) showed that higher turgor pressure and cell wall integrity during storage were associated with crispness retention of ‘Honeycrisp’ fruit. Degradation of middle lamella and decreases in arabinose were found in fruit of the other two varieties, ‘Honeygold’ and ‘Macoun’, that lose crispness, while the arabinose content of Honeycrisp fruit remained constant after 6 months of storage.

From a physiological point of view, ‘Honeycrisp’ fruit exhibits low and stable ethylene production compared to that of ‘McIntosh’, which exhibits a typical climacteric pattern where significant increases of ethylene trigger ripening processes (Harb et al.,

2012). The low amount of ethylene produced by ‘Honeycrisp’ fruit could be one of the factors that contributes to their crispness retention. MdACS1, a molecular marker with two allelotypes, MdACS1-1 and MdACS1-2, has been associated with ethylene production

11 rate and fruit softening (Harada et al., 2000; Dougherty et al., 2016). ‘Honeycrisp’ possesses a heterozygous MdACS1-1/2 allelotype, which is associated with low ethylene production and slow softening, while ‘McIntosh’ has a homozygous MdACS1-1/1 allelotype, associated with high ethylene production and rapid softening (Sato et al.,

2004; Trujillo et al., 2012).

Several cell wall-modifying genes and their relationships with crispness retention of ‘Honeycrisp’ fruit were investigated. Mann et al. (2008) conducted a suppression- subtractive hybridization experiment to detect differentially-expressed genes between

‘Honeycrisp’ and ‘Macoun’. In this study, four cell wall-modifying genes were differentially-expressed; expressions of MdPG and MdExp2 were lower and down- regulated in ‘Honeycrisp’ fruit compared to that of ‘Macoun’. Similar results were reported by Harb et al. (2012), who found that the expression of MdPG and MdExp2 was lower in ‘Honeycrisp’ than in ‘McIntosh’ fruit after 10 days postharvest storage at room temperature. The expression of three cell wall-modifying genes, MdPL, Mdα-AF and

MdXTH2, were also lower, while that of Md β-GAL and MdXTH10 were higher, in

‘Honeycrisp’ than ‘McIntosh’ fruit.

The activities of four relevant cell wall-enzymes, including α-AF, β-GAL, expansin, and PG, in ‘Honeycrisp’ has been studied by Trujillo (2012). In addition, the enzyme activities in the fruit of its progeny were also measured at harvest and after 8- week cold storage. The results indicated that ‘Honeycrisp’ had relatively low PG activity, which could be associated with its crispness retention. However, among the other progeny individuals, there were no significant relationships between the enzyme activities and the crisp/non-crisp traits.

QTL mapping for fruit texture traits has been done using five ‘Honeycrisp’- derived families comprised of more than 600 trees (McKay, 2010). By analyzing each family separately, QTL that associated with the sensory data of crispness, firmness and juiciness were identified across all linkage groups except for LG08 and LG15. These results reveal that the nature of fruit texture is controlled by multiple genes across the genome. To identify more decisive QTL, Schmitz (2013) adopted the sensory data, refined the linkage map of the five ‘Honeycrisp’ families by McKay (2010) using the

RosBREED apple 8K SNP array (Chagné et al., 2012), and carried out pedigree-based analyses. In this case, the number of LGs that associated with the sensory data were narrowed down to three. While LG10 was associated more with firmness than crispness,

LG03 and LG16 were related to both firmness and crispness. Significant two-way

Interactions among the three LGs were also observed.

1.8 Conclusion

Apple fruit crispness, as a sensory attribute that is controlled by multiple genes, is a complex trait at both phenotypic and genotypic levels. There is considerable scope for further work toward understanding the gene regulation of postharvest changes in apple fruit texture. To extend the current knowledge, improvements in the methodology used to study fruit texture can combine the following approaches:

1. Reliable phenotyping methods

High-quality phenotyping is essential in any genetic study. Although several instrumental methods have been developed, sensory evaluation should be incorporated to ensure that these methods truly reflect sensory crispness of specific apple varieties.

2. Genetically-related apple materials

In most of the reviewed studies, candidate genes were identified based on comparisons between unrelated varieties with different softening rates. Considering that variety-specific genes unrelated to the trait of interest can also be differentially- expressed, it will be difficult to parse out the gene-phenotype relationships. The use of genetically-related apple materials can reduce this genetic background noise.

3. RNA-Seq analyses

With its ability to profile comprehensive gene expression, RNA-Seq is a powerful tool to identify genes underlying quantitative traits. While RNA-Seq has not been widely used to study the postharvest textural changes in apple fruit, a more comprehensive understanding of the molecular mechanisms underlying postharvest transformations can be obtained using this next-generation technology.

‘Honeycrisp’, as one of the major apple cultivars and breeding parents in North

America, is a particularly useful individual for studying genes associated with crispness retention. To extend current knowledge, improvements in the methodology used to study fruit texture can be done by combining (1) a genetically-related apple population, (2) an improved phenotyping method for measuring fruit crispness, and (3) RNA-Seq analyses.

In this study, phenotyping using a mechanical-acoustic method to measure crispness, coupled to sensory evaluation, and an RNA-Seq experiment were conducted using a population derived from ‘Honeycrisp’ x MN1764. Fruit crispness of the population at harvest and after 2-month storage were characterized using both instrumental methods and sensory evaluation. Candidate genes were identified by comparing the gene expression levels between individuals with fruit that were determined to retain and lose crispness based on the phenotyping methods. With these approaches, the molecular

14 mechanisms underlying the crispness retention of ‘Honeycrisp’ fruit can be better understood.

Chapter 2: Correlations Between Sensory and Instrumental Crispness of a

‘Honeycrisp’ Apple Breeding Population

Loss of crispness in apple fruit during storage reduces the fruit’s fresh sensation and consumer acceptance. Apple varieties that maintain crispness thus have higher potential for longer-term consumer appeal. To efficiently phenotype crispness, several instrumental methods have been tested, but variable results were obtained when different apple varieties were assayed. To extend these studies, we assessed the extent to which instrumental measurements correlate to and predict sensory crispness, with a focus on crispness maintenance. We used an apple breeding population derived from a cross between ‘Honeycrisp’ and ‘MN1764’, which differed in their ability to retained postharvest crispness. Three types of instrumental measurements (puncture, snapping and mechanical-acoustic tests) and sensory evaluation were performed on fruit at harvest and after 8 weeks of cold storage. Overall, 20 individuals from the family and the two parents were characterized by 19 force and acoustic measures. In general, crispness was more related to force than to acoustic measures. Force linear distance and maximum force as measured by the mechanical-acoustic test were best correlated with sensory crispness and change in crispness, respectively. The correlations varied by apple individual. The best multiple linear regression model to predict change in sensory crispness between harvest and storage of fruit of this breeding family incorporated both force and acoustic measures.

2.1 Introduction

Crispness is a determinant of apple fruit quality, as it is highly related to texture liking and the perception of freshness by consumers (Hampson et al., 2000; Péneau et al.,

2006; Verma, 2014). Thus, loss of crispness in apple fruit during ripening and postharvest storage is an economically important concern. ‘Honeycrisp’ apple, known for its high crispness as well as its ability to retain crispness over long-term storage (Luby and

Bedford, 1992; Tong et al., 1999), provides one solution for this concern. Many apple breeding programs have used ‘Honeycrisp’ as a parent to improve texture and storage characteristics of new apple varieties.

The most direct measurement of crispness is through sensory evaluation.

However, the cost and time required for sensory evaluation to assess many fruits in a breeding population can be an issue. Several instrumental methods have been used as proxy measures of apple sensory properties. Among them, puncture tests (by a penetrometer) have been the most widely used in fresh produce (Bourne, 1965; Harker et al., 1997). Brookfield et al. (2011) observed a correlation of 0.70 between puncture force and sensory crispness for the fresh fruit of nine apple cultivars. A correlation of 0.54 was observed by King et al. (2000), for 115 individuals from a cross between ‘Prima’ and

‘Fiesta’. In addition to the difference among varieties, Harker et al. (2002) showed that a range of correlations from 0.75 to 0.99 were obtained for different sensory panelists where eight cultivars were tested. These varying results suggest that a puncture test alone is not enough to explain variation in sensory crispness of apple fruit.

The snapping test is another instrumental method that has been related to crispness. In a snapping test, an apple cylinder is subjected to three-point bending, and the fracture toughness of the sample is measured. In a study including 10 apple varieties,

Mann et al. (2005) showed that the work required to fracture apple cylinder samples can be used to predict crispness of stored fruit (R2 = 0.82) and change in crispness after

17 storage (R2 = 0.76). McKay et al. (2011) found a correlation coefficient (r = 0.47) between crispness of fresh fruit and the work required to fracture apple cylinders when the test was applied to a large breeding population. Hence, the snapping test may be effective depending on the test population. A similar instrumental method, the single- edge notched bend (SENB) test, was applied by Harker et al. (2006) to investigate change in apple texture during ripening and storage. Different from the snapping test, a small notch was cut on the bottom side of each sample. In comparing the puncture test to the

SENB test, Harker et al. (2006) concluded that the SENB test performed better in detecting the crispness change of shriveled fruit due to water loss.

The use of texture analyzers that simultaneously measure both force-deformation and acoustic attributes is a newer approach towards instrumental measurements of fruit texture (Zdunek et al. 2010 and Costa et al. 2011). Vickers and Christensen (1980) noted that crispness is a complex sensation having both auditory and oral tactile components.

Demattè et al. (2014) showed that sound was critical to crispness, as panelists perceived the same apple as less crisp if the intensity of the biting sound was reduced through a headphone. Thus, it should be anticipated that more accurate prediction of crispness could be obtained by including both attributes. Zdunek et al. (2010) studied texture of three apple varieties, and developed a linear regression model explaining crispness with acoustic events, concluding that their method was useful for specific apple varieties.

Costa et al. (2011) developed a mechanical-acoustic test to study the instrumental textural properties of 86 apple cultivars. A partial least squares (PLS) model of measures derived from the force-deformation and acoustic curve was successfully used to predict sensory crispness of stored fruit. By applying the same method to 27 apple cultivars, Corollaro et

18 al. (2014) showed that the PLS model could predict “crunchiness” of stored fruit with

R2= 0.86. Crunchiness was highly correlated with crispness (r = 0.99) according to a previous study by Corollaro et al. (2013). None of the studies mentioned above specifically tested the abilities of mechanical-acoustic measurements to predict change in sensory crispness. This ability is important in studying fruit in breeding studies because not all individuals will have equal sensory crispness at harvest.

Although research has demonstrated positive correlations between sensory crispness and instrumental measurements, the magnitudes of the correlations depend on the apple varieties or populations studied, and are high when only stored fruit (Costa et al., 2011 and Corollaro et al., 2014) are analyzed. The differing results from these studies suggest that the use of only one instrumental method may not obtain reliable information of sensory crispness. To extend previously published studies, we investigated crispness of

20 individuals of an apple breeding family derived from a cross between ‘Honeycrisp’ and MN1764 and the two parents, at harvest and after storage. Unlike ‘Honeycrisp’,

MN1764 is an advanced selection with fruit that are not very crisp at harvest and that soften during storage (McKay, 2010). The ‘Honeycrisp’ x MN1764 progeny segregated for crispness and demonstrated a range in change in crispness according to our preliminary tests.

In this experiment, we applied puncture, snapping, and mechanical-acoustic tests, as well as sensory evaluation to document the textural properties of this breeding family.

The objective of our study was to assess the extent to which the instrumental measurements correlate to and predict sensory crispness and changes in crispness with

19 storage. We hypothesized that both force-deformation and acoustical measurements would be necessary to predict changes in fruit crispness.

2.2 Materials and Methods

2.2.1 Plant Materials

The apple breeding family under study was derived from a cross between

‘Honeycrisp’ and MN1764. Twenty individuals selected from the progeny and the two parents were used for further analysis. The individuals were selected to include a broad range of crispness levels based on our preliminary sensory and instrumental data. Fruit were harvested at optimum maturity when the background color turned to light green, and when the starch index was higher than 4 (based on Blanpied and Silsby, 1992). Half of the harvested fruit were used as fresh samples, while the other half were stored at the recommended storage conditions for most apples, 0 ± 1 °C at 95% relative humidity, for

8 weeks (Watkins et al. 2016). Six fruit of uniform size (ranging from 100 to 150 g) and free from external defect were selected and used for each individual, and the same fruit were used for sensory and instrumental analyses. Fruit were harvested from trees in a breeding population, in which there were usually only two trees per individual in the population, thus limiting fruit availability. Before sensory evaluation and instrumental measures, stored fruit were removed from cold storage and placed at room temperature for 24 hours.

2.2.2 Sensory Evaluation

Participants were members of the trained panel from the Sensory Center at the

University of Minnesota. Twelve members were recruited based on expressing interest in participating in sensory testing. All recruiting and testing procedures were approved by

20 the University of Minnesota Institutional Review Board. Participants were paid for both the training and the testing sessions. A one-hour training session was performed before the actual testing. References for crispness intensity were provided to panelists: 1) water- soaked celery was used for a high level of crispness, 2) raw zucchini was used for a low level of crispness, and 3) ‘Honeycrisp’ apple was used for a medium-high level of crispness. The panelists set the values for these references during training. These three references were presented as calibration samples at the start of all testing sessions. After discussion of reference standards, the crispness intensities of water-soaked celery,

‘Honeycrisp’ apple and raw zucchini were determined to be 18, 12, and 5, respectively, based on a 20-point scale with 0 = not crisp and 20 = very crisp. During the training session we demonstrated, and panelists practiced, the testing protocol: placing the apple wedge skin down on their lower incisors, biting it with their incisors, and rating the crispness and firmness, then chewing the sample twice and rating its crunchiness, hardness, juiciness and mealiness. We provided the panelists with definitions of each of the sensory attributes. (Table 2-1).

One-quarter of each apple from each treatment was cut off and used for sensory testing. Each of those quarters was further divided into four identical wedges by one horizontal and one vertical cut. Samples were not peeled. A second set of apples from each treatment made up a replicate. The remaining ¾ apple was used for instrumental texture tests. Panelists took their wedge of apple, bit through it with their incisors, and rated its crispness and firmness. Then they chewed the sample twice and rated its crunchiness, hardness, juiciness and mealiness. The peel was not expected to interfere

21 with the evaluation of crispness since the peel rested against the lower incisors and would not have affected the crispness judgments.

A total of six sensory evaluation sessions were performed. The sessions were held biweekly; the first three sessions were performed to test fresh fruit, while the last three sessions were performed to test stored fruit. During each session, 5 to 15 individuals were evaluated. The number of individuals evaluated in each of the first set of sessions depended on the number of individuals that had achieved the necessary stage of maturity.

The number of individuals evaluated in a session after storage was the same number evaluated at the fresh session.

The three references were presented as calibration samples at the start of all testing sessions. This calibration should have helped restrain any drifting in ratings from before to after the break between the fresh and the stored sessions. Each panelist evaluated two samples of each apple. Replicate samples were evaluated after evaluating the first set of all individuals. Within a session and a replicate, serving orders were balanced for position and carryover effects.

2.2.3 Instrumental Tests

Puncture force was measured using a drill press-mounted penetrometer (FT 30,

Wagner Instruments, Greenwich, CT) equipped with a 11.1 mm probe with a convex tip.

A small area of apple skin was removed from the equatorial region, and the maximum force generated during puncture was recorded. The probe was driven 8 mm into flesh. In this test, six halved fruit of each individual were used to determined average puncture force.

A snapping test was conducted using a TA.XT plus texture analyzer (Stable

Micro Systems, Scarsdale, AZ) equipped with a three-point rig (Mann et al., 2005). The probe used was a knife blade with 45° chisel end (type TA-42), and the two bottom supports were placed 2 cm apart. Two cylinders with a diameter of 8 mm and length of 3 cm were generated from cortex tissue parallel to the apple core using a cork borer. A total of 12 cylinders from each individual were measured for both fresh and stored fruit.

The test speed was 2 mm/sec. The fracture force and the distance moved by the probe to reach the fracture point were recorded. The work and the slope of the force-deformation curve were calculated using Exponent software v.4 (Stable MicroSystem).

Mechanical-acoustic tests were performed following the method developed by

Costa et al. (2011) using a TA.XT plus texture analyzer equipped with a 4 mm cylinder puncture probe (type TA-54). Two 1-cm thick apple slices were generated from the equatorial region of each fruit, and two punctures were made in the cortex region of each slice for a total of 24 readings from each individual. The probe speed was 5 mm/sec, and the penetrating depth was 90% in the flesh. The sound emission during the puncture was recorded using a Brüel & Kjær microphone (type 2827, Naerum, Denmark) placed 2 cm from the samples. Ten mechanical and four acoustic parameters were obtained through

Exponent software v.4 (Table 2-2).

2.2.4 Data Analyses

Data of sensory crispness were analyzed using a three-way factorial design analysis of variance (ANOVA) with the factors of individual (20 individuals + 2 parents), storage (at harvest and 8-week storage) and panelist (12 panelists). The order that each

23 individual was tested during sensory evaluation was included in the ANOVA as a block.

The model was:

퐶푟푖푠푝푛푒푠푠 = 퐼푛푑푖푣푖푑푢푎푙 × 푆푡표푟푎푔푒 × 푃푎푛푒푙푖푠푡 + 푂푟푑푒푟

ANOVA tests were also used to determine differences in sensory crispness and instrumental measures between fresh and stored fruit of each individual.

Three types of analyses including Pearson’s correlation, principal component, and multiple linear regression were performed to study the relationships between sensory crispness and instrumental measures. Averaged data on an individual basis were used.

These analyses were also applied to investigate changes in crispness and in instrumental measures between harvest and storage. For the analyses of change in crispness, only 18 individuals and Honeycrisp were chosen based on sensory crispness scores. Three individuals, CL127, EF121 and MN1764, with fresh fruit crispness scores lower than one standard deviation below the mean were excluded in this analysis, as their sensory crispness scores indicated that their fruit were not crisp at harvest, and because the goal of these analyses was to distinguish those individuals with fruit that retain crispness from those that lose crispness on storage. The inclusion of only individuals with similar fruit crispness levels at harvest allowed better identification of the capacity for crispness retention based on changes in crispness. Changes in crispness and changes in instrumental measures were calculated as the values at harvest minus the corresponding values after storage.

To understand which instrumental measures related to sensory crispness and to change in crispness, two correlation analyses were performed using RStudio version

1.0.136 software (RStudio Team, 2015). Data from both fresh and stored fruit were

24 combined to develop the correlation matrix between sensory crispness and instrumental measures, while data from the 18 crisp individuals and ‘Honeycrisp’ were used to develop the correlation matrix between change in crispness and change in instrumental measures.

Principal component analysis (PCA) was applied using XLSTAT® version 10.0

(Addinsoft, NY, NY) to summarize the data and detect relationships among measures.

Two sets of PCA plots were generated: 1) instrumental measures including all 20 individuals and the two parents at both harvest and storage, and 2) change in instrumental measures (fresh minus stored) including only the 18 crisp individuals and ‘Honeycrisp”.

Sensory attributes and changes in the sensory attributes were plotted as supplemental variables to understand their relationships with the principal components. Varimax rotations of the first four factors were applied to all the principal component analyses to spread out the variance explained by each principal component. The individuals were then categorized into three groups based on their crispness scores - 1/3 as crisp (sensory scores > 11), 1/3 as intermediate (sensory scores 9.9 – 10.9), and 1/3 as non-crisp

(sensory scores < 9.9). Also, the individuals were categorized into two groups, fresh and stored, to understand the effect of storage on the instrumental properties.

To identify instrumental predictors for sensory crispness and change in crispness, four multiple linear regression models were developed with stepwise forward selection.

The 20 original individuals and the two parents were included in the models for predicting fresh and stored crispness, while the 18 crisp individuals and ‘Honeycrisp’ were included in the models for predicting change in crispness. Model selection was based on Aikake Information Criteria (AIC), stepping to a final model with the lowest

AIC. All the instrumental measures were used as inputs. Also, instrumental measures that were insignificant in the models were removed. The multiple linear regression analyses were done using RStudio software version 1.0.136.

2.3 Results

2.3.1 Sensory Crispness

Analysis of variance revealed highly significant effects of individual, storage, and panelist factors as well as their interactions on sensory crispness. P-values were less than or equal to 0.001 for all main effects and interactions, except for the 3-way interaction.

Panelist effects were expected. Mean sensory crispness ratings by the 12 panelists ranged from 7.2 ± 0.25 SE to 11.6 ± 0.15 SE, aggregated across all individuals and storage treatments. Six panelists judged stored fruit as less crisp than fresh fruit; mean crispness ratings were lower for stored compared to fresh fruit. The other 5 panelists produced similar mean crispness ratings for fresh and stored fruit (one panelist did not score stored fruit). Panelists varied in mean crispness ratings given to the different individuals, with

50-60% of the panelists giving similar ratings for any individual. Most panelists rated

‘Honeycrisp’ (7 of 10 panelists) and EG015 (10 of 12 panelists) fruit as highly crisp, while rating the crispness of MN1764 fruit as low (11 of 12 panelists).

Due to the significant interaction between Storage and Individual, effects of storage were tested separately for each individual. The parental individuals, ‘Honeycrisp’ and MN1764, differed in sensory crispness at harvest and in their capacity to retain crispness (Table 2-3). The mean sensory crispness scores of ‘Honeycrisp’ and ‘MN1764’ fruit were 11.7 and 8.8, respectively, at harvest. After 8 weeks of storage, the crispness

26 score was retained at 12.2 in ‘Honeycrisp’ (p = 0. 4 for difference between fresh and stored fruit), but decreased to 7.2 in MN1764 (p < 0.01).

2.3.2 Instrumental Measurements

The parent individuals could be distinguished instrumentally. At harvest, higher force was required, and more discernable peaks were generated, for ‘Honeycrisp’ fruit compared to MN1764 fruit, when the probe went through cortex tissue during the mechanical-acoustic test (Figure 2-1A and Figure 2-1C). The average force (F4) and the force linear distance (FLD) generated in ‘Honeycrisp’ fruit at harvest were 0.6 Kg ± 0.02

SE and 2329.8 ± 108.4 SE. For fresh MN1764 fruit, mean F4 and FLD values were 0.5

Kg ± 0.01SE and 1537.2 ± 14.0 SE, respectively. Mean F4 and FLD values for fresh

‘Honeycrisp’ fruit differed from that of fresh MN1764 fruit, with p = 0.014 and p <

0.001, respectively. ‘Honeycrisp’ fruit also showed less change than MN1764 fruit on force-deformation curves after storage (Figure 2-1A and 1B, and Figure 2-1C and 1D, respectively). The mean FLDs were 2329 ± 108 SE and 2115 ± 177 SE for fresh and stored ‘Honeycrisp’, respectively. The mean FLD for stored ‘Honeycrisp’ fruit was 91% of the fresh fruit mean (p = 0.32). In contrast, the mean FLDs for fresh and stored

MN1764 fruit were 1537 ± 14 SE and 984 ± 67 SE (64% of the fresh fruit mean), respectively. Mean FLD values for fresh and stored MN1764 fruit were different, with p

< 0.001.

Acoustic characteristics were similar between ‘Honeycrisp’ and MN1764 fruit at harvest (red lines in Figures 1A and 1C, respectively). Decreases in the length and the peak number of the acoustic curves were observed for both ‘Honeycrisp’ and ‘MN1764’ fruit between harvest and storage, but the difference was larger for MN1764 fruit than for

‘Honeycrisp’ fruit. There was a drop of about 1000 acoustic linear distance units

(AUXLD) and 25 peak number units (AUXP) for ‘Honeycrisp’ fruit (fresh and stored

AUXLD and AUXP values differed, with p = 0.002 and p = 0.002, respectively) compared to a drop of about 1500 AUXLD and 35 AUXP for MN1764 fruit (fresh and stored AUXLD and AUXP values differed, with p < 0.001 and p <0.001, respectively).

The pattern of force-deformation curves generated by snapping tests were also similar between ‘Honeycrisp’ and ‘MN1764’ fruit at harvest (Figures 2A and 2C, respectively). After storage, a flat peak with lower intensity was observed for MN1764 fruit (Figure 2-2D) compared to a sharp peak obtained for ‘Honeycrisp’ fruit (Figure 2-

2B).

2.3.3 Correlations between Sensory Crispness and Instrumental Measures

Most of the force and acoustic measures were positively related to sensory crispness (Table 2-4). The highest correlations with sensory crispness were found between the force linear distance (FLD) by the mechanical-acoustic test and puncture force (PF) by the puncture test (r = 0.83 and 0.75, respectively). Snap test slope (SF/SPL) and maximum acoustic pressure (AUX1) were also related to crispness (r = 0.66 for

SF/SPL and crispness, and r = 0.68for AUX1 and crispness). Correlations between sensory crispness and force differences (F1-F3), force ratio (F1/F3), and peak location

(PL) were very small (each r < 0.15).

To ascertain which instrumental measures were most useful in predicting change in crispness, correlation analysis was performed based on changes between data measured at harvest and after storage (Table 2-4). Generally, instrumental measures that correlated to sensory crispness were also correlated to change in crispness. The highest

28 correlation between change in crispness and a change in an instrumental measure was found on maximum force (F2) by the mechanical-acoustic test (r = 0.79). Change in maximum acoustic pressure (AUX1) had the highest correlation (r = 0.72) among the acoustic measures with change in sensory crispness. Instrumental measures by puncture test and snapping test had lower correlation coefficients than those of the mechanical- acoustic test to sensory crispness.

2.3.4 Application of Principal Component Analysis to Instrumental Measures

Principal component 1 explained 49% of the instrumental data variation, and was mainly associated with force measures (Figure 2-3A and Table 2-5A). Hardness and firmness were the sensory attributes most highly correlated with component 1.

Component 2 explained 21% of the instrumental data variation, and was associated mainly with acoustic measures. Crispness and Juiciness were the sensory attributes most highly correlated with component 2. Mealiness was negatively related to both component

1 and component 2, as well as to the other five sensory attributes.

Crisp individuals were mainly grouped in the first (northeast) quadrant of the biplot, which was associated with higher force and acoustic properties (Figure 2-3B).

‘Honeycrisp’ was on the edge of the crisp cluster, and was characterized by high acoustic values with medium-to-low mechanical values. All individuals were more similar to one another in instrumental properties for fresh fruit than for stored fruit. Fresh individuals mainly grouped in the first quadrant of the PCA plot, while stored individuals were more widespread across the other three quadrants. There was overlap between some of the crisp and non-crisp individuals at harvest, but these classes were clearly separated for

29 stored fruit. After storage, both crisp and non-crisp individuals moved towards the south quadrants due to decreases in acoustic values.

Component 1 and component 3 were the two principal components most related to change in crispness (Figure 2-4A and Table 2-5B). Component 1 was best described by change in force measures (F1, F2, F3, and F4), while component 3 was most related to change in maximum acoustic pressure (Table 2-5B). Individuals with fruit that retained crispness were grouped separately from those that lost crispness on the individual plot

(Figure 2-4B). Overlap occurred between individuals CL143 and CM143, which trended similarly in instrumental changes, but differed in crispness retention.

2.3.5 Multiple Linear Regression Models

Puncture force (PF) as measured by the puncture test, coupled with final force (F3) and force linear distance (FLD) by the mechanical-acoustic test were required to adequately predict (R2 = 0.71) fresh fruit crispness (Table 2-6). High force linear distance (FLD) combined with low Young’s modulus (Y) indicated crisper stored fruit (R2 = 0.90). FLD and Y were included in the prediction model for data combining both fresh and stored fruit, but the accuracy of the combined model was lower than that for stored fruit alone

(R2 = 0.75). Change in maximum force (F2) combined with change in maximum acoustic pressure (AUX1) best predicted change in sensory crispness (R2 = 0.73).

2.4 Discussion

2.4.1 Comparison of the Instrumental Methods

No single instrumental method was able to completely capture sensory crispness, but each method had its advantage in measuring variation in textural properties not captured by the other methods. The puncture test was the only method that measured the

30 surface portion of fruit flesh, and a relatively high correlation (r = 0.75) was found between puncture force (PF) and sensory crispness aggregated for all individuals and storage treatments (Table 2-4). This was similar to the correlation between sensory crispness and puncture force of r = 0.70 reported by Brookfield et al. (2011), but higher than that found by King et al. (2000) where r = 0.54 based on different apple varieties. In our study, the high correlation between PF and crispness may be attributed to the use of genetically related individuals.

Although crispness has been associated with the amount of sound emission when produce is bitten with the front teeth (Vickers and Bourne, 1976; Vickers and

Christensen, 1980; Harker et al., 2002), the acoustic measures we obtained were not as highly correlated with sensory crispness (r = 0.39 to 0.68) compared to the force measures (r = 0.68 to 0.83). This could be due to high similarity in acoustic properties among the individuals, especially at harvest, and thus force properties became the main factors of determination of crispness (Figure 2-3B). Also, panelists in this study were asked to focus on higher pitched sounds for their judgments of crispness and lower pitched sounds for their judgments of crunchiness (Vickers, 1984), but the instrumental acoustic test measured total sound pressure and did not discriminate between high- and low-pitched sounds. The importance of high-pitched sounds to crispness perception of apple fruit was demonstrated by Demattè et al. (2014). They attenuated high frequencies generated during biting to about 50% of the control intensity. Panelists perceived significantly lower crispness of fruit when this was done compared to the control, in which no frequency adjustment was made when panelists bit fruit. Thus, without taking the frequency of sounds into consideration in the instrumental tests, key acoustic

31 properties associated with sensory crispness may have been lost. Harker et al. (2002) also found that sound measures generated during biting and chewing were not good indicators of sensory crispness compared to puncture force.

Instead of providing a direct relationship with crispness, the snapping test could be useful in characterizing the fracture mode associated with soft and mealy fruit.

According to Harker et al. (2006), cell fracture is the predominant fracture mode of fresh apple observed when cortex tissue is broken by a bending force. Both fresh ‘Honeycrisp’ and MN1764 fruit exhibited high adhesion among cells, as evidenced by the sharp force peaks recorded upon fruit fracture (Figure 2-1A). For stored MN1764 fruit, a wide and blunt peak was generated upon fracture (Figure 2-1D). This type of peak may indicate that the mode of tissue failure had changed from cell fracture to cell debonding due to hydrolysis of middle lamellae. Unlike the brittle nature of cell walls, middle lamellae are viscous (Harker et al., 1997). Therefore, a smoother force-deformation curve could be generated during a snapping test when the breakage is associated with cell debonding.

2.4.2 Identification of Individuals with Fruit that Retain or Lose Crispness

A combination of change in force and change in acoustic values was necessary in indicating crispness retention. According to Costa et al. (2012), a dramatic loss in force or acoustic properties was associated with cultivars with mealy fruit. They pointed out that ‘Fuji’ fruit retained (and even slightly increased) the instrumental properties which were associated with good storability. Our results suggest that both force and acoustic properties should be considered to identify crispness retention instrumentally. If only one of the principal components (PC1 or PC3) was used in data analysis, overlap between individuals with fruit that retain and lose crispness occurred (Figure 2-4B). The addition

32 of acoustic components may improve discrimination between individuals that retained or lost crispness (Table 2-5, column B, PC3). In general, individuals with fruit that retained crispness were well-separated from those that lost crispness. The exception was CL143, which was judged as retaining sensory crispness but grouped with individuals having fruit that lost crispness (Figure 2-4 and Table 2-3).

Degrees of change in instrumental properties may not have been large enough to be detected as changes in sensory crispness, as evidenced by decreases in force and/or acoustic values that occurred with some of the individuals that were judged to retain sensory crispness. Harker et al. (2002) found that a force difference greater than 6 N as measured with a puncture test was required for a sensory panel to detect a textural difference in apple fruit. In our study, individuals that significantly lost sensory crispness during storage exhibited changes in mean puncture force between harvest and storage larger than 1.3 Kg (Table 2-3).

2.4.3 Instrumental Prediction of Crispness and Change in Crispness

The crisp sensations perceived by panelists might not be due to the same factors for both fresh and stored fruit. From the instrumental perspective, Mann et al. (2005) found that, among the measures of a snapping test, snap force (SF) was the only predictor for crispness of fresh fruit, while work (W) was the only predictor for crispness of stored fruit. In agreement with Mann et al. (2005), our multiple linear regression analyses included different instrumental measures in the prediction models for fresh compared to stored fruit (Table 2-6). The principle component analysis further showed that individuals that retained sensory crispness differed in the compositions of instrumental properties before and after storage, as stored fruit were separated from corresponding fresh fruit on

33 the individual plot (Figure 2-3B). It may be difficult to model change in crispness for all apple individuals because the instrumental measurement components that best predict fresh fruit crispness are not the same as those that best predict stored fruit crispness.

Despite this difficulty, we were able to model change in crispness using both force and acoustic measurements with an R2 = 0.73 (Table 2-3), possibly because we used individuals from one breeding family.

2.5 Conclusions

Several instrumental measures obtained by puncture, snapping, and mechanical acoustic tests were related to sensory crispness of fruit from an apple family segregating for these traits. Among the measures, the highest correlations were observed between force linear distance (FLD) and crispness, and between change in maximum force (F2) and change in crispness. According to principal component analyses, force and acoustic measures complemented each other in measuring different instrumental characteristics.

Crisp fruit were associated with a combination of higher force and acoustic values.

Decreases in both force and acoustic values after storage were associated with decreases in sensory crispness. Multiple linear regression analyses showed that crispness of stored fruit could be predicted accurately with a combination of two instrumental force measures. However, instrumental measures were less effective in predicting change in crispness, highlighting the complexities in measuring changes in fruit texture. Models to predict change in crispness require the incorporation of both force and acoustic measurements, and may be individual dependent. Future models may be improved if panelist variability in scoring crispness can be further minimized and frequency discrimination in acoustic measurements is incorporated.

Table 2-1 Sensory texture attributes and their definitions.

Attribute Definition

Crispness The amount of relatively high-pitched sound produced upon biting through the apple piece with incisors (lips open). Firmness The amount of force needed to bite through the apple piece with incisors. Crunchiness The amount of relatively low-pitched sound produced upon chewing the apple piece twice with molars. Hardness The amount of force needed to chew the apple piece twice with molars. Juiciness The amount of liquid expressed from the sample during the 1st and 2nd chews. Mealiness The feeling of cells on the interior surfaces of the mouth. It is accompanied by dryness.

Table 2-2 Instrumental measures generated by puncture, mechanical-acoustic, and snapping tests.

Code Instrumental Measures Definition

Puncture Test PF Puncture Force (Kg) Force measured by a penetrometer

Mechanical-acoustic Test (Mechanical Part) F1 Yield Force (Kg) First force peak F2 Max Force (Kg) Max force recorded over the probe's travel F3 Final Force (Kg) Force measured at the end F4 Mean Force (Kg) Average force recorded over the probe's travel FLD Force Linear Distance The length of force curve FP Force Peak Number of force peaks F1-F3 Delta Force (Kg) Calculated as yield force - final Force F1/F3 Force Ratio Calculated as yield force/ final force PL Peak Location (mm) The location of yield force Y Young's Modulus (Kg/mm) Elasticity modulus. Calculated as yield force/ first peak location.

Mechanical-acoustic Test (Acoustic Part) AUX1 Max Acoustic Pressure The highest acoustic peaks detected on the (dB) sound pressure wave AUX2 Mean Acoustic Pressure Mean value of the sound pressure recorded (dB) AUXL Acoustic Linear Distance Length of acoustic curve AUXPD Acoustic Peaks Number of acoustic peaks above the threshold of 10 dB

Snapping Test SF Snapping Force (Kg) The force required to snap the apple sample SPL Peak Location (mm) The location of probe where snapping event occurred SF/SPL Slope (Kg/mm) Calculated as snapping force/ peak location W Work (J) Calculated as force × distance

Crispness PF (Kg) FLD F2 (Kg) AUX1 (dB) Individual Fresh Stored diff. Fresh Stored diff. Fresh Stored diff. Fresh Stored diff. Fresh Stored diff.

CD148 10.0 10.7 -0.7NS 5.2 3.8 1.4* 2199 1909 291* 1.0 1.1 -0.1NS 73.5 70.6 2.9NS CD150 9.7 9.6 0.1NS 5.6 3.0 2.7** 2340 1852 488* 1.0 0.9 0.1NS 73.1 70.8 2.3NS CF104 11.7 10.1 1.6** 6.6 4.9 1.7** 2623 2018 605** 1.1 0.9 0.2* 74.5 70.3 4.2** CF117 9.9 8.4 1.5* 4.6 2.2 2.4** 1974 1309 666** 0.9 0.7 0.2NS 71.3 69.3 2.0NS CF123 9.5 9.1 0.4NS 4.7 3.9 0.8* 1807 1527 280** 1.0 1.0 0.0NS 71.2 66.9 4.3* CL125 11.0 10.8 0.2NS 4 3.0 1.1** 2031 1797 235** 0.8 0.7 0.1NS 72.4 72.8 -0.5NS CL127 9.3 10.6 -1.3NS 4.4 3.5 0.9* 2091 1749 342* 1.0 0.9 0.1NS 73.2 71.8 1.5NS CL142 9.9 6.9 3.0** 4.4 3.1 1.3* 1504 1155 349** 0.8 0.7 0.1NS 70.2 66.6 3.7NS CL143 11.1 10.4 0.7NS 5.4 4.9 0.6NS 2402 1901 501* 1.2 1.1 0.1NS 73.4 69.6 3.8* CL146 11.0 10.3 0.7* 6.1 4.5 1.6* 2333 2182 151NS 1.2 1.0 0.2NS 71.1 69.8 1.2NS CM028 9.8 9.4 0.4NS 3.7 3.2 0.5NS 1844 1555 289* 0.8 0.7 0.1NS 73.4 71.2 2.1NS CM143 11.4 10.3 1.1* 5.7 4.3 1.4** 2234 1814 421* 1.0 1.0 0.0NS 71.7 67.9 3.8* EF120 11.9 11.2 0.7NS 6.0 5.2 0.7NS 2318 2208 110NS 1.1 1.0 0.1NS 72.6 72.5 0.1NS EF121 8.3 9.8 -1.5* 4.5 4.6 -0.1NS 2083 1682 401* 1.0 0.8 0.2* 73.4 68.7 4.8** EF127 11.3 7.9 3.4** 5.4 3.0 2.4** 2182 1220 962** 0.9 0.6 0.3** 73.9 65.3 8.6** EF128 9.4 7.9 1.5* 4.1 2.9 1.3* 1719 1296 422* 0.8 0.6 0.2NS 71.1 66.6 4.5NS EF129 10.3 11.5 -1.2NS 5.5 5.9 -0.4NS 2274 2154 121NS 1.1 1.1 0.0NS 73.9 73.0 0.9NS EF138 10.4 10.9 -0.5NS 5.5 5.8 -0.3NS 2028 2036 -8NS 1.1 1.0 0.1NS 70.6 73.1 -2.5NS EG014 10.8 11.0 -0.2NS 5.2 5.4 -0.3NS 2443 2124 319** 1.1 1.0 0.1NS 74.2 72.7 1.4NS EG015 12.2 12.1 0.1NS 6.3 5.9 0.4NS 2865 2471 394* 1.2 1.1 0.1NS 73.3 71.8 1.6NS Honeycrisp 11.7 12.2 -0.5NS 5.1 5.0 0.1NS 2330 2115 215NS 0.9 0.9 0.0NS 72.7 72.7 0.0NS MN1764 8.7 7.1 1.6** 3.2 2.8 0.4NS 1537 984 553** 0.7 0.5 0.2** 71.6 65.1 6.5** Average 10.4 9.9 0.5 5.1 4.1 0.9 2144 1775 368 1.0 0.9 0.1 72.6 70.0 2.6 SD 1.0 1.5 1.3 0.9 1.1 0.9 331 395 213 0.1 0.2 0.1 1.3 2.6 2.4

Sensory Change in Code Instrumental Measures Crispness Crispness

Puncture test

PF Puncture Force (Kg) 0.75 0.66

Mechanical-acoustic Test (Mechanical Part) F1 Yield Force (Kg) 0.68 0.61 F2 Max Force (Kg) 0.71 0.79 F3 Final Force (Kg) 0.68 0.74 F4 Mean Force (Kg) 0.69 0.76 FLD Force Linear Distance 0.83 0.71 FP Force Peak 0.50 0.44 F1-F3 Delta Force (Kg) 0.08 0.02 F1/F3 Force Ratio 0.01 -0.13 PL Peak Location (mm) 0.15 0.20 Y Young's Modulus (Kg/mm) 0.63 0.27

Mechanical-acoustic Test (Acoustic Part) AUX1 Max Acoustic Pressure (dB) 0.68 0.72 AUX2 Mean Acoustic Pressure (dB) 0.58 0.54 AUXLD Acoustic Linear Distance 0.39 0.35 AUXP Acoustic Peaks 0.39 0.41

Snapping Test SF Snapping Force (Kg) 0.61 0.54 SPL Peak Location (mm) 0.46 0.29 SF/SPL Slope (Kg/mm) 0.66 0.64 W Work (J) 0.54 0.40

Table 2-5 Factor loadings of the instrumental measures and the sensory attributes on the first four principal components, and the proportion of the total variance explained by each component.

A) PCA of the measures B) PCA of changes in the measures

PC1 PC2 PC3 PC4 PC1 PC2 PC3 PC4 (49%) (21%) (14%) (11%) (28%) (27%) (17%) (16%)

Instrumental Measures Puncture Force 0.83 0.3 -0.05 -0.02 0.33 0 0.33 0 Yield Force 0.96 0.06 0.24 0.08 0.63 0.25 0 0.04 Max Force 0.96 0.13 -0.05 0.09 0.86 0 0.05 0 Final Force 0.98 0.08 -0.12 0.07 0.85 0.04 0.02 0.05 Mean Force 0.99 0.09 0.02 0.01 0.87 0.01 0.02 0.02 Force Linear Distance 0.82 0.51 0.02 -0.08 0.49 0.06 0.16 0.23 Force Peak 0.16 0.78 -0.13 -0.39 0 0.17 0.57 0.04 Delta Force 0.04 -0.04 0.99 0.02 0.01 0.87 0.03 0 Force Ratio -0.06 -0.06 0.99 0.04 0 0.82 0.07 0 Peak Location 0.3 -0.41 0.53 0.61 0.13 0.78 0 0.02 Young's Modulus 0.81 0.39 -0.15 -0.35 0.11 0.63 0 0.14 Max Acoustic Pressure 0.48 0.79 -0.03 0.12 0.29 0.08 0.26 0.1 Mean Acoustic Pressure 0.2 0.53 -0.03 0.53 0.07 0 0.69 0.07 Acoustic Linear 0.2 0.92 -0.06 0.05 0.02 0.05 0.14 0.67 Distance Acoustic Peaks 0.18 0.93 -0.07 0.04 0.06 0.05 0.15 0.66 Snapping Force 0.93 0.2 0.04 0.24 0.55 0.06 0 0.26 Peak Location 0.7 0.17 -0.03 0.54 0.29 0.03 0.03 0.42 Slope 0.94 0.22 0.06 0.08 0.49 0.03 0.12 0.06 Work 0.87 0.16 0.03 0.35 0.46 0.07 0.02 0.32

Sensory Attributes Crispness 0.64 0.42 0.08 0.04 0.48 0 0.21 0 Firmness 0.74 0.39 0.1 0.03 0.7 0.02 0.17 0.02 Hardness 0.8 0.35 0.03 0.02 0.72 0.01 0.16 0 Crunchiness 0.71 0.38 0.07 0.06 0.59 0.01 0.21 0.14 Juiciness 0.29 0.47 0.17 -0.23 0.34 0.02 0.46 0.01 Mealiness -0.47 -0.39 0.05 -0.12 0.39 0.28 0.02 0.19

Data Type Regression Equation R2 Adjusted R2 F-value P-value RMSE

At Harvest Y = 5.81 + 0.97 PF – 5.4 F3 + 0.0019 FLD 0.71 0.66 14.37 < 0.001 0.62 After Storage Y = 4.58 + 0.0049 FLD – 6.3 Y 0.90 0.89 83.36 < 0.001 0.51 Combined* Y = 5.70 + 0.0042 FLD – 6.4 Y 0.75 0.74 61.05 < 0.001 0.68 Change** Y = -0.50 + 8.4 F2 + 0.18 AUX1 0.73 0.69 21.19 < 0.001 0.66

*Combined = sensory and instrumental data at harvest and after 8-week storage were combined in the model development. **Change = data at harvest minus data after 8-week storage.

Figure 2-1 Force (black line) and acoustic (red line) curves generated by the mechanical-acoustic test for ‘Honeycrisp’ (top) and MN1764 (bottom) fruit at harvest (left) and after 8 weeks of storage (right). Representative plots are shown.

Chapter 3: Transcriptome analyses identifying genes associated with crispness retention of ‘Honeycrisp’ fruit and its progeny

Crispness retention is a postharvest trait that fruit of the ‘Honeycrisp’ apple and some of its progeny possess. To investigate the molecular mechanisms of crispness retention, progeny individuals derived from a ‘Honeycrisp’ × MN1764 population with fruit that either retain crispness (named “Retain”), lose crispness (named “Lose”), or that are not crisp at harvest (named “Non-crisp”) were selected for transcriptomic comparisons. Differentially-expressed genes (DEGs) were identified using RNA-Seq, and the expression levels of the DEGs were validated using nCounter®. Functional analyses of the DEGs revealed distinct ripening behaviors between fruit of the “Retain” and “Non- crisp” individuals, characterized by opposing expression patterns of auxin- and ethylene- related genes. However, both types of genes were highly expressed in the fruit of “Lose” individuals and ‘Honeycrisp’, which led to the potential involvements of genes encoding auxin-conjugating enzyme (GH3), ubiquitin ligase (ETO), and jasmonate O- methyltransferase (JMT) in regulating fruit ripening. Cell wall-related genes also differentiated the phenotypic groups; greater numbers of cell wall synthesis genes were highly expressed in fruit of the “Retain” individuals and “Honeycrisp” when compared with “Non-crisp” individuals and MN1764. On the other hand, the phenotypic differences between fruit of the “Retain” and “Lose” individuals could be attributed to the functioning of fewer cell wall-modifying genes. A cell wall-modifying gene, MdXTH, was consistently identified as differentially-expressed in those fruit over two years in this study, so is a major candidate for crispness retention.

3.1 Introduction

‘Honeycrisp’ (Malus domestica Borkh.), known for its exceptional fruit crispness, is an apple cultivar developed by the University of Minnesota (Luby and Bedford, 1992).

Since its release in 1991, ‘Honeycrisp’ has become the fourth most produced apple cultivar in Washington state, the major USA apple producer, with the highest increase in growing acreage over years (NASS, 2017). A crisp apple fruit texture is desired by consumers (Hampson et al., 2000; Péneau et al., 2006). Unlike most commercial apple cultivars with fruit that lose crispness during ripening and postharvest storage,

‘Honeycrisp’ fruit retain their crisp texture for at least six-months of cold storage (Tong et al., 1999).

Ethylene, a plant hormone essential for the ripening of climacteric fruit (Barry and Giovannoni, 2007), could be related to ‘Honeycrisp’ fruit crispness retention.

Typically, a burst of ethylene production can be observed in apple fruit at the onset of fruit ripening, which triggers a series of physiological changes, including losses in two related traits, firmness (compression force) and crispness (fracturability). (Johnston et al.,

2002). However, ethylene concentrations in ‘Honeycrisp’ fruit are relatively low and stable during ripening, especially compared with ‘McIntosh’, a variety that exhibits the usual climacteric ethylene production and rapid fruit softening (Harb et al., 2012). Low ethylene production is a feature of apple varieties other than ‘Honeycrisp’ with fruit that retain higher firmness (less softening) during postharvest storage (Costa et al., 2005;

Costa et al., 2010). MdACS1 and MdACO1, the major genes responsible for climacteric ethylene production in apple fruit, are also correlated with postharvest changes in fruit firmness (Costa et al., 2005). In particular, MdACS1, with two allelotypes, MdACS1-1

47 and MdACS1-2 associated with different ethylene production rates and fruit softening patterns, has been used as a molecular marker for long shelf life apples (Harada et al.,

2000; Dougherty et al., 2016).

The cell wall, responsible for the crisp texture of apple fruit (Harker et al., 1997), is another focus of crispness retention research. Tong et al. (1999) showed that

‘Honeycrisp’ fruit were able to maintain integrity of the cell wall after long-term storage without obvious degradation of the middle lamella, a region important for cell-to-cell adhesion and fruit crispness (Harker et al. 1997; Wakabayashi 2000). Pectin is the main component of middle lamella, and thus lack of middle lamella degradation of

‘Honeycrisp’ fruit can be related to genes involved in pectin degradation. As the most structurally complex plant cell wall polysaccharide (Caffall and Mohnen, 2009), the degradation of pectin is accomplished by the synergistic functions of various genes. In apple, the expression patterns of several pectolytic genes, such as Mdα-AF1, Mdα-AF2,

Mdα-AF3, Mdβ-GAL2, MdPG1, and MdPME1, have been correlated with fruit softening in specific varieties (Wakasa et al. 2006; Goulao et al., 2007; Nobile et al., 2011;

Gwanpua et al., 2016A; Gwanpua et al., 2016B). Among these genes, MdPG1 may have major effects on crispness retention, since its gene expression was consistently low in

‘Honeycrisp’ fruit during ripening and storage (Mann et al., 2008; Harb et al., 2012).

Also, the correlation between MdPG1 and crispness retention was supported by its low enzyme activity (Trujillo, 2012).

In addition to pectin, cellulose and hemicellulose are other major structural polysaccharides in the plant cell wall. Non-pectolytic genes, such as MdEXPs and

MdXTHs, involved in reorganizing the cellulose–hemicellulose network, could also affect

48 the texture of apple fruit (Atkinson et al., 2009; Muñoz-Bertomeu et al., 2013; Ireland et al., 2014), but their roles in ‘Honeycrisp’ crispness retention have not been clearly established. In a genetic study, Harb et al. (2012) showed that MdEXP2 and MdXTH2 had lower gene expression in ‘Honeycrisp’ fruit during fruit ripening compared with

‘McIntosh’ fruit, while MdXTH10 exhibited an opposite expression pattern. Trujillo et al.

(2012), on the other hand, found that MdEXP2 was not strongly related to crispness retention by examining the gene expression and effects of allelotype in fruit of several apple varieties and ‘Honeycrisp’ progeny individuals. So far, there has been no additional study to validate the functions of MdXTHs in ‘Honeycrisp’ fruit, and thus the MdXTH gene family is a potential target for further analysis.

Toward a more comprehensive understanding of the molecular mechanisms behind apple fruit crispness retention, the transcriptomes of fruit from a population derived from ‘Honeycrisp’ × MN1764 in which the fruit of progeny individuals differed in their ability to retain postharvest crispness (Chang et al., 2018) were studied. RNA-

Seq, as a high-throughput sequencing-based method (Wang et al., 2009), was first applied to identify differentially-expressed genes (DEGs) associated with the crispness retention of the ‘Honeycrisp’ × MN1764 population. The expression patterns of the DEGs were then validated using nCounter®, a mid-throughput hybridization-based method (Kulkarni,

2011), by including second-year samples. With the uses of a genetically-related apple population and transcriptomic approaches, candidate genes for crispness retention can be more reliably identified.

3.2 Materials and Methods

3.2.1 Plant materials

The cultivar ‘Honeycrisp’ and an unnamed breeding selection, MN1764, were the parents of the breeding population studied. ‘Honeycrisp’ fruit retain crispness during storage, while MN1764 fruit lose crispness. This breeding population is comprised of 170 progeny individuals, and each individual has 2 to 4 replicated trees. The trees were grown at the University of Minnesota Horticultural Research Center in Victoria, Minnesota.

Fruit were harvested from late August to early October 2018 and 2019 when the starch index reached the score of four based on an eight-point starch-iodine index chart

(Blanpied and Silsby, 1992). Fruit were peeled, and cortex samples were collected at two time points, at harvest and after 8-week cold storage (0 ± 1 °C and 95% relative humidity). Diced fruit pieces from individual fruit were packaged in aluminum foil and frozen with liquid nitrogen, then kept at -80 °C for later RNA extraction.

Eight progeny individuals derived from the ‘Honeycrisp’ × MN1764 population were identified from each of three groups: 1) individuals with fruit that either retained crispness through 8-week of cold storage (named “Retain” individuals); 2) individuals producing fruit that lost crispness after two months of cold storage (named “Lose” individuals); or 3) individuals that had non-crisp fruit at harvest (named “Non-crisp” individuals) were selected for transcriptomic comparisons. The crispness of fruit was determined using a penetrometer (FT 30, Wagner Instruments, Greenwich, CT) mounted on a drill press and equipped with a FT716 size plunger, and a TA.XT plus texture analyzer (Stable Micro Systems, Hamilton, MA) that measures force and acoustic properties of the fruit (Costa et al., 2011; Chang et al., 2018). The selection of the individuals was based on 3-years of instrumental data (2016-2018).

3.2.2 RNA sample preparation and RNA sequencing

RNA was extracted from 10 g of cortex tissue from each fruit. An RNA extraction method developed by Lopez-Gomez and Gomez-Lim (1992) for fruit with high polysaccharides was used with modification (Mann et al., 2008; Trujillo et al., 2012). The extracted RNA was further purified with the RNeasy Midi kit (Qiagen, Valencia, CA) following manufacturer protocols. DNase I (New England Biolabs, Ipswich, MA) was applied to eliminate genomic DNA contamination in the RNA extracts.

RNA-Seq of samples harvested in 2018 was performed at the University of

Minnesota Genomic Center. The RNA integrity and concentration were measured with the RNA ScreenTape System (Agilent Technologies, Santa Clara, CA) and RiboGreen

RNA quantification kit (Molecular Probes, Eugene, OR). The concentrations of the RNA samples were from 40 to 100 ng/μl, with RNA integrity number (RIN) larger than eight. cDNA libraries were constructed using Illumina TruSeq RNA sample preparation protocol. The parent samples were sequenced in 2014 on an Illumina HiSeq 2000 platform with eight libraries pooled into one lane of an Illumina flow cell. The progeny samples were sequenced in 2018 on an Illumina HiSeq 2500 platform with 12 libraries pooled in one lane. A total number of 60 paired-end RNA-Seq datasets were generated including the two parents and the eight individuals sampled at two time-points (at harvest and after storage) with three biological replicates of each individual and time-point

(Table A1 and Figure A1).

Quality control of the sequence reads was performed using FastQC (Andrews,

2010; version 0.11.9). Adapter contamination and low-quality reads (with Phred quality score < 30) were filtered using Trimmomatic (Bolger et al., 2014; version 0.39). The filtered reads were mapped to the apple reference genome GDDH13 v1.1 (Daccord et al.,

2017) using HISAT (Kim et al., 2015; version 2.1.0). SAMTools (Li et al., 2009; version

1.9) was used to sort the aligned reads based on their locations in the genome, while featureCounts (Liao et al., 2014; version 1.5.2) was used to count the number of the sorted reads as the expression level of each gene. The results of RNA sequencing and read alignments are shown in the Supplementary material (Table A1).

3.2.3 Differential expression analysis

Statistical analyses of gene expression were performed using edgeR software

(Robinson et al., 2010; version 3.11). Genes with low counts across all datasets were removed using default settings in the “filterByExpr” function. The trimmed mean of M- value (TMM) method was applied for data normalization using the “calNormFactor” function to reduce technical variations. A multidimensional scaling (MDS) plot was generated from the top 500 genes with the largest standard deviation between samples using the limma R package (Smyth, 2005; version 3.11) to examine the relationships among samples.

Gene expression was normalized to counts per million mapped reads (CPM). The significance level was set at FDR < 0.05, and genes with log2-fold difference > 1 and

CPM values > 1 were considered as differentially-expressed. Three comparisons were made: (1) ‘Honeycrisp’ vs. MN1764, (2) “Retain” vs. “Lose”, and (3) “Retain” vs. “Non- crisp” (Figure A1). To identify differentially-expressed genes (DEGs) that commonly occurred in all the comparisons, a Venn diagram was generated using the “vennDiagram” function in the limma R package. Heatmaps visualizing the expression patterns of the

DEGs was generated using the pheatmap R package (Kolde and Kolde, 2015; version

1.0.12).

Gene functional annotation was obtained from the Genome Database of Rosaceae

Species (GDR, https://www.rosaceae.org/species/malus/all). Gene Ontology terms for each gene were assigned using the PANNZER2 webservice with default settings

(Koskinen et al., 2015). GO enrichment analyses were conducted using goseq software

(Young et al., 2010; version 1.40.0), which was specifically designed to minimize length- derived bias that may affect RNA-Seq data. GO terms with FDR < 0.05 were considered as significantly enriched.

3.2.4 Gene validation using NanoString nCounter® and qRT-PCR

A subset of the DEGs that showed > 1.5 log2-fold difference were further validated using the nCounter® analysis system (NanoString Technology, Seattle, WA) with a customized CodeSet designed and created to target the 47 genes of interest (Table

A2). Three housekeeping genes including CKL (casein kinase 1 isoform delta like,

MD09G119011), TMp1 (type 1 membrane protein like, MD04G1005300) and DLD

(dihydrolipoamide dehydrogenase, MD16G1145800) that were consistently expressed in apple fruit over storage were selected for normalizing expression of the genes of interest

(Perini et al., 2014). To evaluate the consistency of the genes, RNA samples from two different harvest years were included in this experiment. A total of 96 samples consisting of the two parents and six progeny individuals (three each from the “Retain” and “Lose” individuals), each collected at two time-points (at harvest and after two months of cold storage) and from two years (2018 and 2019) with three biological replicates each, were analyzed. The nCounter® data was adjusted using the manufacturer-provided spiked positive and negative controls.

Eight genes, including two cell wall-related genes and six genes that showed inconsistent expression between RNA-Seq and nCounter® results, were further examined using qRT-PCR. DLD (MD16G1145800) was selected as the housekeeping gene used to compare to genes of interest that were tested. The primers for the genes were designed

(Table A3) using the Integrated DNA Technologies (IDT) online tool

(https://www.idtdna.com). Reverse transcription reactions were performed using

GoScriptTM Reverse Transcriptase (Promega, Madison, WI) following manufacturer protocols using Oligo(dT)15 and random primers. Real-time PCR was performed on a

CFX96TM thermal cycler (Bio-Rad, Hercules, CA), with SYBR® Green Supermix (Bio-

Rad, Hercules, CA) as the fluorescence reagent. Reactions for the target and housekeeping genes were performed in duplicate with a total volume of 20 μl. PCR was conducted with the following conditions: initial incubation at 95 °C for 5 min, followed by 40 cycles of denaturation at 95 °C for 20 s, annealing at 60 °C for 30 s, extension at 72

°C for 30 s, and finishing with 72 °C for 5 min. Gene expression levels were calculated using the 2-ΔΔCt method (Livak and Schmittgen, 2001).

3.3 Results

3.3.1 Phenotype and transcriptome variations among the individuals

Fruit of the selected progeny individuals and parents were distinct in puncture force (PF), force linear distance (FLD), and acoustic pressure (AUX) (Figure 3-1). Crisp fruit were characterized with higher force (FLD and PF) and acoustic (AUX) values compared with the non-crisp fruit. As reported in our previous study (Chang et al., 2018), the average PF, FLD and AUX of this breeding population at harvest were 5.1 kg, 2144, and 72.6 dB. Using these numbers as the thresholds, fruit with all three instrumental

54 measures higher than these numbers were considered as crisp. Among the six individuals with crisp fruit at harvest, three of them retained their force and acoustic values after 8- week cold storage while the others did not. Based on the instrumental measures, three phenotypes were distinguishable: 1) individuals with fruit that retained crispness (CL136,

EF129, and EF138), 2) individuals with fruit that lost crispness (CF117, CL156, and

EF117), and 3) individuals with fruits that were not crisp at harvest (CL121 and CL124).

The terms “Retain”, “Lose”, and “Non-crisp” were used to designate the three phenotypes. Based on the instrumental measures, ‘Honeycrisp’ fruit retained, while

MN1764 fruit lost, crispness.

Differences among the “Retain”, “Lose”, and “Non-crisp” individuals were observed at the transcriptomic level. A multidimensional scaling (MDS) plot (Figure 3-2) showed that samples with the same phenotypes had similar gene expression patterns. A separation was detected between freshly-harvested and stored samples. Freshly-harvested

“Non-crisp” samples clustered apart from the “Retain” and “Lose” individuals, but after storage, the individuals were not as widely separated as at harvest. ‘Honeycrisp’ samples clustered with other crisp individuals, while MN1764 replicates were spread between crisp and non-crisp samples, both at harvest and after storage.

There were 4,129; 2,030; and 6,870 DEGs identified in comparisons between

‘Honeycrisp’ and MN1764 (e.g., in Figure 3-3, sum of

1634+101+67+488+1499+84+40+216), the “Retain” and “Lose” individuals, and the

“Retain” and “Non-crisp” individuals, respectively (Figure 3-3). A total number of 107 genes were found in all three comparisons. Of these, 67 genes were commonly expressed in fruit of ‘Honeycrisp’ and the “Retain” individuals, and therefore potentially related to

55 the trait of retaining crispness (Figure 3-3A). Additionally, 40 genes were commonly expressed in fruit of MN1764 and the “Non-crisp” and “Lose” individuals and are therefore possibly related to losing crispness (Figure 3-3B).

3.3.2 Functional analyses of differentially-expressed genes

The larger transcriptomic variations between fruit of the “Non-crisp” and

“Retain” individuals also were reflected in the higher number of enriched Gene Ontology

(GO) terms (Figure 3-4A). Most differentially-expressed genes (DEGs) highly expressed in fruit of the “Retain” but not in the “Non-crisp” individuals were associated with the oxidation-reduction process (GO:0055114) after storage, and at harvest, signal transduction (GO:0007165) and peptidyl-tyrosine modification (GO:0018212). For the enriched GO terms with more specific biological functions, the response to auxin

(GO:0009733), response to light stimulus (GO:0009416), and carbohydrate transport

(GO:0008643) represented about 2% of the after storage differentially-expressed genes

(DEGs). On the other hand, DEGs highly expressed in fruit of the “Lose” but not in the

“Retain” individuals were mainly enriched in the defense response (GO:0006952), rRNA processing (GO:0006365), response to karrikin (GO:0080167), and fruit ripening

(GO:0009835) (Figure 3-4A).

A different group of GO terms were identified between fruit of the “Retain” and

“Lose” individuals at harvest compared with the “Retain” versus “Non-crisp” individuals

(Figure 3-4B). The most enriched GO terms were the RNA modification (GO:0009451), and microtubule severing (GO:0051013) for the “Retain” fruits, and the DNA rewinding

(GO:0036292) and oligosaccharide catabolic process (GO:0009313) for the “Lose” fruits.

Functional similarity of the DEGs were observed after storage. Two of the significantly

56 enriched GO terms, the oxidation-reduction process (GO:0055114) and response to karrikin (GO:0080167), for fruit of the “Retain” and “Lose” individuals respectively, were also identified between fruit of the “Retain” and “Non-crisp” individuals.

There were only three GO terms identified between ‘Honeycrisp’ and MN1764 fruit (Figure 3-4C). The DEGs highly expressed in ‘Honeycrisp’ and MN1764 fruit were consistently enriched in signal transduction (GO:0007165) and carbohydrate transport

(GO:0008643) at harvest and storage, and the number of genes within each term were similar. These two GO terms were also identified in comparison of fruit of the “Retain” and “Non-crisp” individuals (Figure 3-4A).

3.3.3 The expression patterns of auxin- and ethylene-related genes

Genes involved in the auxin signaling pathway, due to their potential roles in fruit ripening, were analyzed to extend the results of the functional analyses where genes associated with response to auxin (GO:0009733) were enriched (Figure 3-5A). In the comparison between fruit of the “Retain” and “Non-crisp” individuals, most auxin- related genes, including ARF, AUX/IAA, SAUR, and GH3, were highly expressed in the

“Retain” fruits, especially at harvest (Figure 3-5). Higher expression of auxin-related genes was also found in ‘Honeycrisp’ fruit compared with MN1764 fruit, but some of them were differentially-expressed after storage (Figure 3-5). In contrast, fewer auxin- related genes were differentially expressed between fruit of the “Retain” and “Lose” individuals, and most of them were highly expressed in the “Lose” fruits (Figure 3-5A-C) except for the GH3 groups, which were highly expressed in the “Retain” fruits (Figure 3-

5D). IAA/AUX (MD02G1057200) and GH3 (MD05G1092300) were the only auxin- related genes identified in all three comparisons. Additionally, three SAUR genes,

MD02G1133900, MD04G1082600 and MD14G1152100, were differentially-expressed in two of three comparisons.

For ethylene-related genes, the major difference was the higher number of 1- aminocyclopropane 1-carboxylate synthases (ACSs) genes expressed in fruit of the “Non- crisp” and “Lose” individuals compared to that of the “Retain” individuals (Figure 3-6A).

Among them was MdACS1 (MD15G1302200), the ACS responsible for climacteric ethylene production in apple fruit (Figure 3-6A). However, the number of ACSs differentially expressed between the parents were similar and MdACS1 was highly expressed in ‘Honeycrisp’ fruit (Figure 3-6A). Two types of genes displayed specific expression patterns in the “Non-crisp” fruit: low expression of two ethylene overproducer genes (ETOs) (Figure 3-6A) and increased expression between harvest and after storage of one ethylene response sensor (ERS) gene (Figure 3-6B). The other two types of ethylene- related genes, 1-aminocyclopropane 1-carboxylate oxidases (ACOs) and ethylene response factors (ERFs), were differentially-expressed across comparisons, but there was no consistent pattern in terms of the number of DEGs identified in specific individuals (Figure 3-6A and C). For example, about half of the differentially-expressed

ACOs were highly expressed in “Non-crisp” fruit, while another half were highly expressed in “Retain” fruits. MdACO1 (MD10G1328100), another gene involved in climacteric ethylene production, was not differentially-expressed among the individuals in this study.

Two ACS genes, MdACS6 (MD06G1090600) and MdACS3a (MD15G1203500), were identified in all three comparisons (Figure 3-6A). MdACS6 was highly-expressed in fruit of MN1764 and the “Lose”, and “Non-crisp” individuals, while MdACS3a was

58 highly-expressed in ‘Honeycrisp’ fruit and those of the “Retain” individuals. Three ERF genes, ERF1B-like (MD13G1213100), MdERF3 (MD14G1226300), and SHN1

(MD17G1209000) were differentially expressed in two of the three comparisons (Figure

3-6C).

3.3.4 The expression patterns of cell wall-related genes

Major cell wall-related genes associated with the texture of apple fruit were studied (Table 3-1). Based on the total number of DEGs, cell wall-related genes were more highly-expressed in fruit of the “Retain” individuals compared with the “Non-crisp” individuals (Table 3-1A). At harvest, cellulose synthase, galacturonosyltransferase

(GAUT), pectin methylesterase (PME), and xyloglucan endotransglucosylase/hydrolase

(XTH) were more highly-expressed in fruit of the “Retain” individuals compared to

“Non-crisp”, while α-arabinofuranosidases (α-AFs) were highly-expressed in fruit of the

“Non-crisp” individuals (Table 3-1A). After storage, most cellulose synthases and

GAUTs maintained higher expression in fruit of the ‘Retain’ than ‘Non-crisp’ individuals

(Table 3-1A). Similar patterns of gene expression were detected between ‘Honeycrisp’ and MN1764. There was higher expression of cellulose synthases and GAUTs in

‘Honeycrisp’ fruit and α-AF in MN1764 fruit, but these patterns were observed only after storage and not at harvest (Table 3-1C).

In the comparison between fruit of the “Retain” and “Lose” individuals, expression patterns of the cell wall-genes were not as clear as in the comparison between those of the “Retain” and “Non-crisp” individuals (Table 3-1B). At harvest, most XTHs were identified in fruit of the “Retain” individuals but not in those of the “Lose” individuals, while β-galactosidase (β-GAL), and cellulose synthase genes were mostly

59 identified in the “Lose”, but not “Retain”, individual’s fruit. After storage, one GAUT was identified in fruit of the “Retain”, but not “Lose”, individuals, while expansin genes and PGs were mainly identified in the fruit of the “Lose”, but not “Retain”, individuals.

3.3.5 RNA-Seq results validation using nCounter® technology and qRT-PCR

To validate the RNA-Seq results, nCounter® data were generated from the same

RNA samples used for RNA-Seq analyses, as well as samples harvested in 2019. Results depicted in Figure 3-7 showed that gene expression measured by RNA-Seq (in FPKM) and nCounter® (in gene count) were correlated (r = 0.8). Higher correlations were obtained for genes with high expression levels and inconsistencies occurred mostly for genes with low expression levels.

qRT-PCR was performed to further probe gene expression discrepancies between

RNA-Seq and nCounter® results. Figure 2 shows that RNA-Seq and nCounter® both failed to detect gene expression in some cases, in contrast with qRT-PCR results. For

MD02G1057200 (auxin-responsive protein), MD10G1315100 (XTH), and

MD11G1230200 (unknown gene), qRT-PCR results agreed with the RNA-Seq data, but for MD05G1098700 (AMP-dependent synthetase) and MD12G1164900

(pentatricopeptide repeat-containing protein), qRT-PCR results agreed with nCounter® data. One of the tested genes encoding a CASP-like protein 1F2 (MD14G1150200) was not correctly detected using RNA-Seq and nCounter® technology. Expressions of two cell wall-modifying genes, XTH (MD16G1091200) and PG (MD10G1179100), were well correlated among the three methods.

Among the 47 tested genes using nCounter® (Table A2), 26 genes were differentially-expressed in the parent and progeny samples (primary candidate gene) and

14 genes were differentially-expressed only in the parent or progeny samples (secondary candidate gene) (Table 3-2). The other 7 genes identified as differentially-expressed by

RNA-Seq were not differentially-expressed over both years in which fruit were harvested. The validated candidate genes included those involved in cell wall modification, as well as plant hormone signaling and biosynthesis. The cell wall- modifying genes were Mdα-AF3 (MD16G1158300), GAUT (MD17G1141200), MdPG1

(MD10G1179100), and XTHs (MD10G1315100, MD16G1091200), and the hormone- related genes were AUX/IAA (MD02G1057200), GH3.1 (MD05G1092300), ACSs

(MD06G1090600, MD15G1203500), and jasmonate methyltransferase (JMT,

MD15G1023600).

3.4 Discussion

3.4.1 Fruit ripening and crispness retention

Different ripening characteristics could be one factor causing crisp and non-crisp phenotypes at harvest. Based on the transcriptomic data, fruit of the “Non-crisp” individuals exhibited the typical climacteric ripening processes of apple fruit, in which ethylene plays a crucial role. Genes highly expressed in the “Non-crisp” fruit were enriched in expression of fruit ripening (Figure 3-4A) genes, and ACS genes encoding the rate-limiting enzyme of ethylene biosynthesis (Kende, 1993) were the major DEGs in this

GO term (Figure 3-6A). One member of the ACS family, MdACS1 (MD15G1302200), which is essential for increased ethylene production during apple fruit ripening (Harada et al., 2000), was highly expressed in the “Non-crisp” fruit. Another important ACS in apple fruit, MdACS3a (MD15G1203500), associated with relatively stable ethylene production before climacteric ripening (Tan et al., 2013), was highly expressed in the “Retain” fruit.

MdACS6 (MD06G1090600) also displayed greater expression in “Non-crisp”, “Lose”, and MN1764 fruit compared to “Retain” and ‘Honeycrisp’ fruit. This gene was reported by An et al. (2018) and Zhao et al. (2020) to be up-regulated during fruit ripening or following ethylene production due to salt stress.

The differential-expression patterns of several auxin-related genes further indicated that the fruit of the “Retain” and “Non-crisp” individuals are under different physiological control mechanisms (Figure 3-5). Auxins have been reported to both promote and inhibit fruit ripening. For example, exogenous auxin application induces ethylene production in ‘Golden Delicious’ apple fruit before harvest (Yue et al., 2020). In contrast, the delayed-ripening phenotype of the transgenic MADS8/9-suppressed apple was attributed to the maintenance of high auxin concentrations in the fruit (Schaffer et al.,

2013). Although the two reports are seemingly contradictory, they both hypothesize that auxin is critical at the onset of fruit ripening. A fruit ripening model for apple established by Busatto et al. (2017) demonstrates that after fruit enter advanced ripening stages, auxin-related genes are downregulated along with increased expression of ethylene- related genes. Thus, the higher expression of auxin-related genes observed in this study suggests that fruit of the “Retain” individuals were harvested at relatively early ripening stages.

Gretchen Hagen 3 (GH3) genes, which encode auxin-conjugating enzymes, could be alternative ripening indicators for fruit of the “Retain” individuals wherein ethylene- related genes did not show a differential response. It has been proposed that GH3 initiates the ripening processes of apple fruit due to its capacity to convert biologically active auxin to inactive amino acid (Shin et al., 2016). Among the GH3 genes identified in this

62 study (Figure 3-5D and Table 3-2A), GH3.1 (MD05G1092300) was also observed to increase in expression in ‘Royal Gala’ apples during fruit maturation and ripening

(Devoghalaere et al., 2012), and homologous genes have been correlated with the ripening of other fruits, such as grape (Vitis vinifera L.) and pungent pepper (Capsicum chinense L.) (Liu et al., 2005; Böttcher et al., 2010).

The softening phenotype of fruit of “Lose” individuals, characterized as being crisp at harvest and non-crisp after storage, could result from interactions between auxin and ethylene. Ethylene appeared to be involved in ripening of the “Lose” fruits, since several ACS genes were highly expressed in fruit of the “Lose” individuals compared with those of the “Retain” individuals (Figure 3-6A). MdACS1 (MD15G1302200) was not differentially-expressed at harvest but during storage of the “Lose” fruits, which correlated with the period during which they started to lose crispness. The delayed- ripening characteristic of the “Lose” fruits might be attributed to auxin. In the comparison between the “Retain” and “Lose” fruits, auxin related-genes, except for the GH3 group, were more highly-expressed in fruit of the “Lose” individuals. It is possible that the lower expression of GH3 genes caused higher concentrations of active auxin in fruit of the

“Lose” individuals, which in turn inhibited climacteric ethylene biosynthesis and fruit ripening.

Defense response genes also differentiated the postharvest ripening statuses of the individuals. Fruit ripening and defense responses are closely connected. For example, Shi et al. (2014) found that proteins differentially-expressed in stored ‘Golden Delicious’ fruit mostly belonged to stress and defense response categories. Zheng et al. (2013) also reported increases in defense response proteins in ‘Golden Delicious’ fruit during

63 ripening and after ethylene treatment. In this study, most of the DEGs between “Retain” and “Non-crisp” fruits were enriched in the GO term “defense response”. Three of the

DEGs, including chitinase (MD01G1213100), glucan endo-1,3-β-glucosidase

(MD11G1189000), and jasmonate methyltransferase (JMT, MD15G1023600), showed consistently higher expression in fruit of the “Lose” compared with the “Retain” individuals (Table 3-2B). Chitinases and glucan endo-1,3-β-glucosidases are fungi and bacteria cell wall-degrading enzymes (Leubner-Metzger and Meins, 1999), while JMTs are involved in the biosynthesis of methyl jasmonate (MeJA), a plant volatile that regulates defense responses (Bai et al., 2013). In addition to roles in defense responses,

MeJA is also related to fruit ripening. Exogenous application of MeJA enhanced climacteric ethylene production and softening of ‘Golden Delicious’ fruit (Lv et al.,

2018), as well as expression of the transcription factors, MdMYC2 (MD16G1274200) and

MdERF3 (MD14G1226300), that regulate ethylene biosynthesis (Li et al., 2017). In this study, MdMYC2 was not differentially-expressed, but MdERF3 displayed matching expression patterns (Figure 3-6C) to the JMT gene (Table 3-2B). Because of its biological function, and the timing of expression, the JMT gene is one of the candidate genes associated with fruit ripening of MN1764 fruit and that of the “Lose” individuals.

‘Honeycrisp’ fruit has a mechanism of ripening distinct from that of the phenotypically-similar “Retain” individuals based on high expression of MdACS1

(MD15G1302200) (Figure 3-6A). In apple, there are two MdACS1 allelotypes, MdACS1-

1 and MdACS1-2. MdACS1-1 is associated with high ethylene production and softer fruit, while MdACS1-2 is associated with low ethylene production and firmer fruit (Dougherty et al., 2016). The high expression of MdACS1 in ‘Honeycrisp’ fruit corresponds to its

64 heterozygous allelotype, including both MdACS1-1 and MdACS1-2, in contrast to that of the homozygous (MdACS1-2/2) MN1764 (Trujillo et al., 2012), but is inconsistent with its ripening behavior, including crispness retention. Despite exhibiting high MdACS1 expression, ‘Honeycrisp’ fruit produce low amounts of ethylene during ripening compared with ‘McIntosh’ fruit, that show climacteric ripening characteristics (Harb et al., 2012). Thus, MdACS1 expression is unrelated with the crispness retention of

‘Honeycrisp’ fruit, but low ethylene production could still be an important factor in its ripening characteristic.

An explanation for low ethylene production in ‘Honeycrisp’ fruit is post- transcriptional regulation of MdACS1. Among genes involved in ethylene biosynthesis, ethylene overproducers (ETOs) were DEGs identified in fruit of ‘Honeycrisp’ and the

“Retain” individuals as having higher expression levels than fruit of MN1764 or “Non- crisp” individuals (Figure 3-6A). ETOs are ubiquitin ligases, which target type 2 ACSs to the proteasome for degradation (Chae and Kieber 2005; Yoshida et al., 2005). The classification of ACSs is based on the C-terminal amino acid sequences, and MdACS1 belongs to the type 2 ACSs (Shi et al., 2013). In Arabidopsis, overexpression of AtETO1 inhibits AtACS5 enzyme activity and promotes its degradation (Wang et al., 2004), but the roles of ETOs in fruit ripening have not been reported. Without differential expression of the other ethylene synthesis genes, for example MdACO1 (MD10G1328100), the

ETOs could be candidates for further elucidating the relationship between ethylene and ripening of ‘Honeycrisp’ fruit.

3.4.2 Cell wall-related genes and crispness retention

The continuing expression of cell wall synthesis genes during storage could be related to the crispness retention of ‘Honeycrisp’ fruit. Cellulose synthase and

Galacturonosyltransferase (GAUT) are the two types of cell wall synthesis genes investigated in this study, and there were greater transcript numbers of the two genes highly expressed in fruit of ‘Honeycrisp’ and the “Retain” individuals after storage when compared with those of MN1764 and the “Non-crisp” individuals (Table 3-1A and C).

The cellulose synthase gene family in plants includes cellulose synthase and cellulose synthase-like genes, which mediate the synthesis of cellulose and hemicellulose, respectively (Delmer and Amor, 1995; Pauly et al., 2013). In climacteric apple fruit, the amount of cellulose and hemicellulose are relatively constant during ripening (Bartley,

1976), and cellulose synthase and cellulose synthase-like genes have not been the main foci to postharvest textural changes. In a recent study by Win et al., 2019, degradation of cellulose and hemicellulose, correlating with fruit softening, was observed in fruit of

‘Summer King’ and ‘Green Ball’ apples during long-term storage. Whether the continuous expression of cellulose synthase and cellulose synthase-like genes in fruit of

‘Honeycrisp’ during storage contributes to crispness retention can be further evaluated in future research by measuring the cellulose and hemicellulose content in ‘Honeycrisp’ fruit. From another point of view, the higher expression of cellulose synthases during

‘Honeycrisp’ fruit storage is an additional indicator showing the slow ripening behaviors of ‘Honeycrisp’ fruit, because cell wall synthesis genes were usually active during fruit growth and are down-regulated as fruit mature (Dheilly et al., 2016). Unlike typical climacteric ripening, a lack of clear transition between development and ripening could

66 result in minimal changes in the expression patterns of cell wall genes—a possible mechanism that results in ‘Honeycrisp’ fruit crispness retention.

GAUTs, key enzymes in pectin biosynthesis, are potentially involved in crispness retention by acting against the pectin-modifying enzymes that cause fruit softening. One of the GAUT genes (MD17G1141200) was identified as a candidate favorable for crispness retention (Table 3-2A), showing consistently higher levels of gene expression in ‘Honeycrisp’ than MN1764 fruit. Dheilly et al. (2016) showed that this GAUT was more active before than after harvest, similar to the expression patterns of most cell wall synthesis genes. The high expression levels of several GAUTs in ‘Honeycrisp’ fruit at harvest and after storage further emphasized the close relationships between the ripening behaviors of ‘Honeycrisp’ fruit and the expression of cell wall-related genes. It should be pointed out that the GAUT gene (MD17G1141200) was not differentially-expressed between fruit of “Retain” and “Lose” individuals (Table 3-2A), implying the involvement of additional genes in crispness retention.

The expression patterns of cell wall-modifying genes were more consistent than those of cell wall synthesis genes in differentiating fruit of the “Retain” and “Lose” individuals, since the cell wall synthesis genes, cellulose synthases and GAUTs, were not specifically expressed in fruit of the “Retain” individuals (Table 3-1B). Several cell wall- modifying genes that have been previously reported to affect postharvest fruit texture were observed in the current study to differentiate crispness phenotypes of fruit of the

‘Honeycrisp’ × MN1764 population. Among them, MdPG1 (MD10G1179100) was the most prominent gene associated with losing crispness, as significantly high expression of

MdPG1 was consistently observed in fruit of MN1764, the “Lose”, and the “Non-crisp”

67 individuals (Table 3-1 and Table 3-2B). Polygalacturonases (PGs) catalyze the hydrolysis of pectin, and the expression of MdPG1 during fruit ripening has been correlated with increased pectin solubilization, which leads to softening of apple fruit (Brummell, 2006;

Gwanpua et al., 2016; Win et al., 2019). Many studies have illustrated the roles of

MdPG1 through different approaches, including the examination of MdPG1 in fruit of varieties with different firmnesses (Wakasa et al., 2006) and the study of transgenic apple with suppressed MdPG1 (Atkinson et al., 2012). In this study, while high expression of

MdPG1 correlated with softening of the non-crisp fruit of the studied population, low or no expression of this gene alone cannot completely explain crispness retention. Despite differential expression, there was also an abundance of MdPG1 mRNA accumulation in fruit of the “Retain” individuals after storage (Table 3-2B), which did not result in a significant decrease in fruit crispness, suggesting that PG protein synthesis or activity may differ between fruit with different crispness retention phenotypes.

Other pectin-modifying enzymes, such as pectin methylesterase (PME), α- arabinofuranosidase (α-AF), and β-galactosidase (β-GAL), could have partial effects on fruit crispness retention. The function of PMEs is to remove the methyl groups from pectin (Brummell and Harpster, 2001), while α-AFs and β-GALs are responsible for the cleavage of sugar residues from the side chains of pectin polymers (Yoshioka et al.,

1995). Together, these enzymes have been hypothesized to facilitate pectin degradation during fruit ripening (Gwanpua et al., 2016A). MdPME1 (MD08G1195600), Mdα-AF3

(MD16G1158300), and Mdβ-GAL2 (MD02G1079200) were the genes corresponding to the pectin-modifying proteins identified in this study (Table 3-1). Each of these genes has been previously reported to relate to fruit texture. For example, MdPME1 in ‘Jonagold’

68 apple is related to its postharvest softening (Gwanpua et al., 2016B), Mdα-AF3 is highly expressed in individuals with mealy fruit (Nobile et al., 2011), and Mdβ-Gal2 is associated with the softer texture of ‘Fuji’ compared with ‘Qinguan’ fruit (Yang et al.,

2018). However, in our study, the relationships between the three pectin-modifying genes and fruit crispness were not strong. After including samples from a second year in our analyses, the only significant difference observed in our study was the expression levels of Mdα-AF3 between fruit of the “Retain” and “Lose” individuals (Table 3-2). Because of the inconsistencies between the parents and progeny individuals and year-to-year variations, MdPME1 (MD08G1195600), Mdα-AF3 (MD16G1158300), and Mdβ-GAL2

(MD02G1079200) were deemed as low- or non-priority candidate genes.

MdXTH (MD16G1091200), with consistently higher expression in fruit of

‘Honeycrisp’ and the “Retain” individuals over two years (Table 3-1 and 3-2A), has the characteristics of a good candidate gene underlying crispness retention. Xyloglucan is the most abundant hemicellulose in plant cell walls (Hayashi, 1989), and xyloglucan endotransglucosylase/hydrolase (XTH) is a key enzyme controlling wall strength and extensibility through its modification of the interactions between cellulose and hemicellulose xyloglucan (Rose et al., 2002). There are eleven identified members of

MdXTH, named MdXTH1 to MdXTH11 (Miedes and Lorences, 2004), but the specific

MdXTH (MD16G1091200) identified in this study, with sequence similarity to the

Arabidopsis AtXTH33, has not been formally named. Although not a focus in previous studies, there are clues suggesting that this MdXTH may have a role in regulating fruit crispness. In a study comparing ripening-associated gene expression between ‘Golden

Delicious’ and ‘Fuji’ fruit, a consistently higher expression of this MdXTH was observed

69 in ‘Fuji’ fruit, which are crisper at harvest, with better crispness retention than ‘Golden

Delicious’ fruit (Busatto et al., 2016). Also, the physical position of MdXTH (6.3 Mb) on chromosome 16 is within a QTL region (3.2 to 6.9 Mb) previously identified using five

‘Honeycrisp’ populations including ‘Honeycrisp’ × MN1764 and associated with apple fruit crispness (Schmitz, 2013). This particular MdXTH was favorable for retaining crispness texture properties, in contrast to those associated with the prevailing hypothesis that MdXTHs cause fruit softening by loosening the cell wall (Feng et al., 2008; Muñoz-

Bertomeu et al., 2013). In the current study, the highest expression of MdXTH was at harvest (Table 3-2A), which indicated that this MdXTH was activated before the difference in crispness occurred between fruit of the “Retain” and “Lose” individuals.

Based on its expression patterns, previously reported QTL results, and its biological functions, MdXTH could contribute significantly to both at harvest crispness and crispness retention of ‘Honeycrisp’ fruit. It is not clear how MdXTH is regulated, but regulation may be independent of gene sequence differences, since fruit of the “Retain” and “Lose” individuals shared the same QTL allelotype.

3.5 Conclusion

Genes involved in the auxin-activated signaling pathway and ethylene biosynthesis (i.e. MdACSs) differentiated fruit of the “Retain” and “Non-crisp” individuals, while GH3.1 and JMT might be more important in determining the different ripening outcomes of fruit of the “Retain “and “Lose” individuals. ‘Honeycrisp’ fruit differed from those of the “Retain” individual by higher expression of several MdACS1 genes, which suggests a possible post-transcriptional regulation of MdACS1 by ETOs.

Several cell wall-related genes were differentially-expressed in the phenotypic group comparisons. A great number of cell wall synthesis genes, such as cellulose synthases and GAUTs, were highly expressed in fruit of the “Retain” individuals and

‘Honeycrisp’ compared to those of the ‘Non-crisp’ individuals and MN1764. Among the cell wall-modifying genes, the expression patterns of MdPG1 and the MdXTH were the most closely correlated with crispness retention.

As newer technologies, RNA-Seq and nCounter® have not been extensively applied in studies of the ripening-related textural changes of apple fruit. By using these transcriptomic approaches, we were able to extend current understanding of the crispness retention of ‘Honeycrisp’ fruit and discover novel candidate genes.

At harvest After Storage Cell wall-gene Retain Non-crisp Retain Non-crisp

α-Arabinofuranosidase - MD08G1221800, MD16G1158300 - - β-Galactosidase MD08G1023600, MD15G1251100 MD00G1018700, MD02G1079200 MD15G1251100 - Cellulose synthase/ MD01G1236500, MD01G1236600, MD03G1028900, MD03G1178600, MD01G1236500, MD01G1236600, MD03G1028900, MD08G1147200 Cellulose synthase- like MD02G1311600, MD03G1133700, MD13G1209200, MD15G1150300 MD08G1076900, MD08G1126200, MD07G1309200, MD08G1147200, MD09G1037900, MD15G1064400, MD10G1029800, MD15G1123200, MD15G1415100, MD15G1415200, MD15G1340200, MD17G1099800 MD17G1144800 Expansin MD00G1125400, MD01G1166700, MD03G1090700, MD05G1130300, MD00G1125400, MD01G1166700, - MD04G1129800, MD07G1233100, MD06G1041000, MD06G1195100 MD04G1052600 MD09G1279500 Galacturonosyltransferase MD04G1181600, MD05G1363900, - MD09G1041100, MD10G1140000, - MD09G1041100, MD09G1061900, MD13G1084900, MD16G1084000, MD11G1318000, MD13G1084900, MD17G1141200 MD17G1141200, MD16G1084000 Pectate lyase MD14G1167100 - - MD01G1100600 Pectin methylesterase MD00G1105300, MD04G1198000, MD08G1195600 MD02G1104600, MD13G1149800, MD06G1064700, MD07G1255000, MD07G1289200, MD12G1198000, MD15G1222000 MD08G1195600, MD09G1054900, MD15G1222000 MD16G1150200 Polygalacturonase MD01G1068900, MD01G1069000, MD00G1140300, MD03G1162500, MD00G1140300, MD06G1105300 - MD06G1105300, MD09G1290500, MD07G1279000, MD09G1030100, MD12G1064100, MD16G1161800 MD09G1030200, MD10G1179100 Xyloglucan endotransglucosylase/ MD02G1192600, MD09G1102600, MD16G1145200, MD17G1140000 MD04G1020100, MD15G1303500, MD13G1237300, MD16G1278900 hydrolase MD10G1315100, MD15G1303500, MD16G1091200 MD16G1014000, MD16G1091200

(B)

At harvest After storage Cell wall-gene Retain Lose Retain Lose

α-Arabinofuranosidase - - - MD16G1158300 β-Galactosidase - MD00G1018700, MD02G1079200 - - Cellulose synthase/ MD05G1296600, MD13G1209200 MD03G1028900, MD15G1415100, MD05G1296600, MD13G1209200 MD03G1028900, MD17G1099500, Cellulose synthase-like MD15G1415200, MD16G1145200, MD17G1099600 MD17G1038900 Expansin MD05G1130300 MD11G1054500, MD16G1070600 MD05G1130300, MD17G1271500 MD01G1166700, MD06G1195100, MD07G1233100, MD13G1070200 Galacturonosyltransferase - - MD17G1141200 - Pectate lyase - - - MD06G1161400 Pectin methylesterase MD06G1191000 MD13G1149800 - MD08G1195600 Polygalacturonase MD06G1105300 MD10G1179100 MD13G1092000 MD00G1087900, MD09G1030100, MD09G1030200 Xyloglucan endotransglucosylase/ MD10G1315100, MD13G1237300, MD16G1145200 MD15G1303500, MD16G1091200 MD13G1268900, MD16G1145200 hydrolase MD15G1303500, MD16G1091200

(C)

At harvest After storage Cell wall-gene Honeycrisp MN1764 Honeycrisp MN1764

α-Arabinofuranosidase - - - MD08G1221800, MD16G1158300

β-Galactosidase - - MD08G1139000, MD09G1192500 MD11G1133400

Cellulose synthase/ MD03G1029100, MD03G1178600, MD03G1029000, MD03G1133700, MD03G1029100, MD03G1178600, MD03G1133700 Cellulose synthase-like MD04G1173700, MD15G1340200 MD15G1415100, MD15G1415200 MD04G1173700, MD11G1156200, MD13G1209200, MD15G1340200, MD17G1099600 Expansin - MD16G1070600 - MD04G1052600, MD10G1133200

Galacturonosyltransferase MD10G1140000, MD17G1141200 MD00G1136600, MD04G1181600 MD09G1093100, MD10G1140000, - MD11G1318000

Pectate lyase - MD05G1179500 - - Pectin methylesterase MD01G1220700, MD06G1191000, MD06G1191000, MD08G1195600, MD01G1220700, MD09G1172600 MD11G1307500, MD16G1150200 MD09G1172600 MD11G1307500 Polygalacturonase MD03G1292400, MD15G1441700 - - MD07G1279000, MD10G1179100 Xyloglucan endotransglucosylase/ MD10G1315100, MD16G1091200 - MD16G1091200 MD09G1152600, MD09G1152700, hydrolase MD13G1237300, MD17G1140000

Table 3-2 The numbers of gene counts in the fruit samples at harvest and after 8-week cold storage measured using NanoString ® nCounter technology. (A) Genes that were highly expressed in ‘Honeycrisp’ (HC) and/or the “Retain” group fruit, and (B) genes that were highly expressed in MN1764 (MN) and/or “Lose” group fruit. The genes that were differentially-expressed in both parent and progeny samples were the primary candidates, while the genes that were only differentially-expressed in the parent or the progeny samples were the secondary candidates.

(A)

At harvest After storage At harvest After storage Gene ID Gene function HC MN Diff.1 HC MN Diff. Retain Lose Diff. Retain Lose Diff.

Primary candidate gene MD01G1062800 6052 5851 0.0NS 6731 2135 -1.7** 9720 5414 -0.8** 4488 5713 0.3NS PIP1 | aquaporin MD03G1019900 14 3 -2.1* 9 3 -1.5* 27 13 -1.1** 11 6 -0.8** RLK1 | receptor-like protein kinase MD05G1092300 128 24 -2.4* 22 6 -1.9 NS 101 9 -3.4** 12 8 -0.6 NS GH3 | auxin-responsive protein MD07G1237500 659 44 -3.9** 909 12 -6.3** 600 253 -1.2** 269 149 -0.9** RPL | ribosomal protein MD07G1247100 561 179 -1.7** 545 154 -1.8** 448 274 -0.7** 377 293 -0.4* PMSR | peptide met S-oxide reductase MD07G1259200 132 3 -5.3** 138 15 -3.2** 141 62 -1.2** 145 84 -0.8** RPM | disease resistance protein MD07G1270800 190 27 -2.8* 40 23 -0.8NS 209 129 -0.7** 8 11 0.4NS TUB | tubulin MD07G1274100 12 2 -3.0** 9 2 -2.1** 8 5 -0.8** 17 9 -0.9** SK | SKP1-like protein MD08G1106600 109 6 -4.2** 24 15 -0.7NS 58 34 -0.8** 16 20 0.3NS scpl | serine carboxypeptidase MD14G1056600 107 38 -1.5** 173 139 -0.3NS 132 39 -1.8** 253 107 -1.2** function unknown MD14G1110100 82 2 -5.4** 57 4 -3.7** 66 26 -1.3** 82 19 -2.1** function unknown MD15G1297000 91 10 -3.3** 80 6 -3.7** 123 25 -2.3** 80 24 -1.7** APK | adenylyl-sulfate kinase MD16G1091200 274 12 -4.5** 83 7 -3.5** 291 36 -3.0** 90 22 -2.0** XTH | xyloglucan endotransglucosylase

Secondary candidate gene MD01G1042500 4 11 1.4NS 7 8 0.2 NS 2 6 1.4** 4 18 2.1** ELI | elicitor-activated gene MD02G1057200 12 16 0.4 NS 17 7 -1.3* 8 7 -0.2 NS 9 8 -0.2 NS AUX/IAA | auxin-responsive protein MD05G1098700 367 111 -1.7* 490 357 -0.5 NS 140 138 0.0 NS 288 317 0.1 NS LACS | AMP-dependent synthetase MD10G1315100 155 76 -1.0 NS 23 9 -1.3* 101 120 0.2 NS 16 19 0.2 NS XTH | xyloglucan endotransglucosylase MD11G1230200 38 13 -1.5** 31 18 -0.7* 38 32 -0.3 NS 71 56 -0.3 NS function unknown MD15G1203500 2823 77 -5.2* 179 33 -2.4 NS 305 531 0.8 NS 5 15 1.6** ACS | ACC synthase MD17G1141200 92 32 -1.5** 110 43 -1.4* 88 75 -0.2 NS 80 55 -0.6 NS GAUT | galacturonosyltransferase

NS not significant, * p < 0.05, ** p < 0.01 1 Diff. = Log2 fold-difference

(B)

At harvest After storage At harvest After storage

Gene ID HC MN Diff.1 HC MN Diff. Retain Lose Diff. Retain Lose Diff. Gene function

Primary candidate gene MD00G1036800 52 123 1.2NS 24 96 2.0* 44 172 2.0** 45 92 1.0** ABCG | ABC transporter G MD01G1213100 789 1587 1.02NS 2193 7274 1.7** 859 1041 0.3 NS 2259 11492 2.3** CHIA | chitinase MD03G1108400 62 194 1.6** 124 819 2.7** 12 37 1.6** 134 444 1.7** GLTP | glycolipid transfer protein MD05G1310400 17 2357 7.1** 23 1585 6.1** 391 2023 2.4** 33 769 4.5** protein E6-like MD05G1313300 10 5 -1.0* 15 80 2.4* 4 8 0.9** 11 60 2.5** function unknown MD06G1233800 51 146 1.52NS 78 153 1.0* 90 155 0.8** 89 228 1.4** monoacylglycerol lipase-like MD08G1127900 8 34 2.02NS 15 50 1.7** 5 23 2.3** 10 65 2.8** AFR | F-box protein AFR-like MD10G1179100 3074 8577 1.52NS 1602 64580 5.3** 308 7092 4.5** 42086 76687 0.9** PG | polygalacturonase MD11G1189000 460 734 0.72NS 1626 3551 1.1* 272 922 1.8** 662 10298 4.0** BG | glucan endo-1,3-β-glucosidase MD12G1164900 41 208 2.32NS 22 161 2.9** 58 185 1.7** 30 84 1.5** PPR | pentatricopeptide repeat protein MD12G1183000 46 66 0.52NS 80 250 1.6** 35 45 0.3 NS 77 138 0.8** LURP-one-related 15-like MD13G1112700 16 8 -1.0NS 9 36 2.0** 12 26 1.1** 17 27 0.7* CYP | cytochrome P450 MD15G1023600 896 817 -0.1 NS 1433 13417 3.2* 89 500 2.5** 2485 7947 1.7** JMT | jasmonate O-methyltransferase

Secondary candidate gene MD03G1060100 72 58 -0.3 NS 44 122 1.5 NS 26 41 0.6* 27 83 1.6** LBD | LOB domain-containing protein MD05G1297900 11 40 1.9 NS 11 172 4.0** 10 39 1.9 NS 147 140 -0.1 NS EFR | EF-TU receptor MD05G1349800 1837 1358 -0.4 NS 2045 1726 -0.2 NS 1434 2284 0.7** 1382 2055 0.6** WRKY | WRKY transcription factor MD06G1090600 14 45 1.7 NS 46 112 1.3 NS 5 10 1.1 NS 76 507 2.7** ACS | ACC synthase MD16G1158300 2231 5769 1.4 NS 13467 23748 0.8 NS 5066 7463 0.6 NS 19752 27723 0.5* α -AF | α -arabinofuranosidase MD16G1277800 3 13 2.1 NS 5 10 0.9 NS 3 10 1.9** 10 17 0.8** NRT | nitrate transporter MD17G1256100 19 56 1.5 NS 34 67 1.0 NS 27 42 0.6 NS 81 286 1.8** SFBB | F-box family protein

NS not significant, * p < 0.05, ** p < 0.01 1 Diff. = Log2 fold-difference

(A)

(B)

(C)

(A) (B)

Figure 3-3 Venn diagram showing the number of differentially-expressed genes (DEGs) commonly identified among the three comparisons. (A) DEGs highly expressed in ‘Honeycrisp’ (HC) and “Retain” individuals, and (B) DEGs highly expressed in MN1764 (MN), “Lose”, and “Non-crisp” individuals.

(A)

(B)

(C)

(A) Auxin response factor (ARF)

(B) AUX/IAA

(D) Gretchen Hagen 3 (GH3)

(A) Ethylene biosynthesis 1-Aminocyclopropane 1-carboylate synthase (ACS)

1-Aminocyclopropane 1-carboylate oxidase (ACO)

Ethylene overproducer (ETO)

(B) Ethylene signaling Ethylene response sensor (ERS)

Ethylene insensitive (EIN)

Figure 3-7 Correlation between gene expression levels using RNA-Seq and nCounter® platforms.

Bibliography

Abeles, F. B., & Biles, C. L. (1991). Cellulase activity in developing apple fruits. Scientia Horticulturae, 47(1-2), 77-87. An, J. P., Yao, J. F., Xu, R. R., You, C. X., Wang, X. F., & Hao, Y. J. (2018). An apple NAC transcription factor enhances salt stress tolerance by modulating the ethylene response. Physiologia plantarum, 164(3), 279-289. Amyotte, B., Bowen, A. J., Banks, T., Rajcan, I., & Somers, D. J. (2017). Mapping the sensory perception of apple using descriptive sensory evaluation in a genome wide association study. PloS one, 12(2). Andrews, S. (2010). FastQC: a quality control tool for high throughput sequence data. Atkinson, R. G., Johnston, S. L., Yauk, Y. K., Sharma, N. N., & Schröder, R. (2009). Analysis of xyloglucan endotransglucosylase/hydrolase (XTH) gene families in kiwifruit and apple. Postharvest Biology and Technology, 51(2), 149-157. Atkinson, R. G., Sutherland, P. W., Johnston, S. L., Gunaseelan, K., Hallett, I. C., Mitra, D., ... & Schaffer, R. J. (2012). Down-regulation of POLYGALACTURONASE1 alters firmness, tensile strength and water loss in apple (Malus x domestica) fruit. BMC plant biology, 12(1), 129. Bai, S., Dong, C., Li, B., & Dai, H. (2013). A PR-4 gene identified from Malus domestica is involved in the defense responses against Botryosphaeria dothidea. Plant physiology and biochemistry, 62, 23-32. Barra, M., Meneses, C., Riquelme, S., Pinto, M., Lagüe, M., Davidson, C., & Tai, H. H. (2019). Transcriptome Profiles of Contrasting Potato (Solanum tuberosum L.) Individuals Under Water Stress. Agronomy, 9(12), 848. Barry, C. S., & Giovannoni, J. J. (2007). Ethylene and fruit ripening. Journal of Plant Growth Regulation, 26(2), 143. Bartley, I. M. (1976). Changes in the glucans of ripening apples. Phytochemistry, 15(5), 625-626. Batu, A. (2004). Determination of acceptable firmness and colour values of tomatoes. Journal of food engineering, 61(3), 471-475. Blanpied, G. D., & Silsby, K. J. (1992). Predicting harvest date windows for apples. Cornell Cooperative Extension. Bolger, A.M., Lohse, M., & Usadel, B. (2014). Trimmomatic: a flexible trimmer for Illumina sequence data. Bioinformatics 30(15), 2114-2120.

Böttcher, C., Keyzers, R. A., Boss, P. K., & Davies, C. (2010). Sequestration of auxin by the indole-3-acetic acid-amido synthetase GH3-1 in grape berry (Vitis vinifera L.) and the proposed role of auxin conjugation during ripening. Journal of Experimental Botany, 61(13), 3615-3625. Bourne, M. C. (1965). Studies on punch testing of apples. Food Technology, 19(3), 413. Brock, G., Pihur, V., Datta, S., & Datta, S. (2011). clValid, an R package for cluster validation. Journal of Statistical Software. Brookfield, P. L., Nicoll, S., Gunson, F. A., Harker, F. R., & Wohlers, M. (2011). Sensory evaluation by small postharvest teams and the relationship with instrumental measurements of apple texture. Postharvest biology and technology, 59, 179-186. Brummell, D. A., & Harpster, M. H. (2001). Cell wall metabolism in fruit softening and quality and its manipulation in transgenic plants. In plant cell walls (pp. 311-340). Springer, Dordrecht. Brummell, D. A. (2006). Cell wall disassembly in ripening fruit. Functional Plant Biology, 33(2), 103-119. Busatto, N., Farneti, B., Tadiello, A., Velasco, R., Costa, G., & Costa, F. (2016). Candidate gene expression profiling reveals a time specific activation among different harvesting dates in ‘Golden Delicious’ and ‘Fuji’ apple cultivars. Euphytica, 208(2), 401- 413. Busatto, N., Tadiello, A., Trainotti, L., & Costa, F. (2017). Climacteric ripening of apple fruit is regulated by transcriptional circuits stimulated by cross-talks between ethylene and auxin. Plant signaling & behavior, 12(1), e1268312. Busatto, N., Farneti, B., Tadiello, A., Oberkofler, V., Cellini, A., Biasioli, F., ... & Costa, F. (2019). Wide transcriptional investigation unravel novel insights of the on-tree maturation and postharvest ripening of ‘Abate Fetel’ pear fruit. Horticulture research, 6(1), 1-15. Caffall, K. H., & Mohnen, D. (2009). The structure, function, and biosynthesis of plant cell wall pectic polysaccharides. Carbohydrate research, 344(14), 1879-1900. 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. Chae, H. S., & Kieber, J. J. (2005). Eto Brute? Role of ACS turnover in regulating ethylene biosynthesis. Trends in plant science, 10(6), 291-296.

Chagné, D., Crowhurst, R. N., Troggio, M., Davey, M. W., Gilmore, B., Lawley, C., ... & Velasco, R. (2012). Genome-wide SNP detection, validation, and development of an 8K SNP array for apple. PLoS one, 7(2). Chang, H. Y., Vickers, Z. M., & Tong, C. B. (2018). The use of a combination of instrumental methods to assess change in sensory crispness during storage of a ‘Honeycrisp’ apple breeding family. Journal of texture studies, 49(2), 228-239. Cliff, M. A., Stanich, K., Lu, R., & Hampson, C. R. (2016). Use of descriptive analysis and preference mapping for early‐stage assessment of new and established apples. Journal of the Science of Food and Agriculture, 96(6), 2170-2183. Corollaro, M. L., Endrizzi, I., Bertolini, A., Aprea, E., Demattè, M. L., Costa, F., ... & Gasperi, F. (2013). Sensory profiling of apple: Methodological aspects, cultivar characterisation and postharvest changes. Postharvest Biology and Technology, 77, 111- 120. Corollaro, M. L., Aprea, E., Endrizzi, I., Betta, E., Demattè, M. L., Charles, M., ... & Gasperi, F. (2014). A combined sensory-instrumental tool for apple quality evaluation. Postharvest Biology and Technology, 96, 135-144. Cosgrove, D. J. (1998). Cell wall loosening by expansins. Plant physiology, 118(2), 333- 339. Cosgrove, D. J. (2015). Plant expansins: diversity and interactions with plant cell walls. Current opinion in plant biology, 25, 162-172. Costa, F., Stella, S., Van de Weg, W. E., Guerra, W., Cecchinel, M., Dallavia, J., ... & Sansavini, S. (2005). Role of the genes Md-ACO1 and Md-ACS1 in ethylene production and shelf life of apple (Malus domestica Borkh). Euphytica, 141(1-2), 181-190. Costa, F., Van de Weg, W. E., Stella, S., Dondini, L., Pratesi, D., Musacchi, S., & Sansavini, S. (2008). Map position and functional allelic diversity of Md-Exp7, a new putative expansin gene associated with fruit softening in apple (Malus× domestica Borkh.) and pear (Pyrus communis). Tree Genetics & Genomes, 4(3), 575-586. Costa, F., Peace, C. P., Stella, S., Serra, S., Musacchi, S., Bazzani, M., ... & Van de Weg, W. E. (2010). QTL dynamics for fruit firmness and softening around an ethylene- dependent polygalacturonase gene in apple (Malus× domestica Borkh.). Journal of experimental botany, 61(11), 3029-3039. Costa, F., Cappellin, L., Longhi, S., Guerra, W., Magnago, P., Porro, D., ... & Gasperi, F. (2011). Assessment of apple (Malus× domestica Borkh.) fruit texture by a combined acoustic-mechanical profiling strategy. Postharvest Biology and Technology, 61, 21-28.

Costa, F., Cappellin, L., Fontanari, M., Longhi, S., Guerra, W., Magnago, P., ... & Biasioli, F. (2012). Texture dynamics during postharvest cold storage ripening in apple (Malus× domestica Borkh.). Postharvest Biology and Technology, 69, 54-63. Daccord, N., Celton, J. M., Linsmith, G., Becker, C., Choisne, N., Schijlen, E., ... & Di Pierro, E. A. (2017). High-quality de novo assembly of the apple genome and methylome dynamics of early fruit development. Nature genetics, 49(7), 1099. Delmer, D. P., & Amor, Y. (1995). Cellulose biosynthesis. The plant cell, 7(7), 987. Demattè, M. L., Pojer, N., Endrizzi, I., Corollaro, M. L., Betta, E., Aprea, E., ... Gasperi, F. (2014). Effects of the sound of the bite on apple perceived crispness and hardness. Food Quality and Preference, 38, 58-64. Devoghalaere, F., Doucen, T., Guitton, B., Keeling, J., Payne, W., Ling, T. J., ... & Diak, R. (2012). A genomics approach to understanding the role of auxin in apple (Malus x domestica) fruit size control. BMC Plant Biology, 12(1), 7. Dheilly, E., Le Gall, S., Guillou, M. C., Renou, J. P., Bonnin, E., Orsel, M., & Lahaye, M. (2016). Cell wall dynamics during apple development and storage involves hemicellulose modifications and related expressed genes. BMC plant biology, 16(1), 201. Di Guardo, M., Bink, M. C., Guerra, W., Letschka, T., Lozano, L., Busatto, N., ... & van de Weg, E. (2017). Deciphering the genetic control of fruit texture in apple by multiple family-based analysis and genome-wide association. Journal of experimental botany, 68(7), 1451-1466. Dougherty, L., Zhu, Y., & Xu, K. (2016). Assessing the allelotypic effect of two aminocyclopropane carboxylic acid synthase-encoding genes MdACS1 and MdACS3a on fruit ethylene production and softening in Malus. Horticulture research, 3(1), 1-10. Feng, H. L., Zhong, Y. X., Xie, H., Chen, J. Y., Li, J. G., & Lu, W. J. (2008). Differential expression and regulation of longan XET genes in relation to fruit growth. Plant science, 174(1), 32-37. Figueroa, C. R., Pimentel, P., Gaete-Eastman, C., Moya, M., Herrera, R., Caligari, P. D., & Moya-León, M. A. (2008). Softening rate of the Chilean strawberry (Fragaria chiloensis) fruit reflects the expression of polygalacturonase and pectate lyase genes. Postharvest Biology and Technology, 49(2), 210-220. Gálvez, J. H., Tai, H. H., Lagüe, M., Zebarth, B. J., & Strömvik, M. V. (2016). The nitrogen responsive transcriptome in potato (Solanum tuberosum L.) reveals significant gene regulatory motifs. Scientific Reports, 6, 26090.

Goulao, L. F., Santos, J., de Sousa, I., & Oliveira, C. M. (2007). Patterns of enzymatic activity of cell wall-modifying enzymes during growth and ripening of apples. Postharvest biology and technology, 43(3), 307-318. Gwanpua, S. G., Van Buggenhout, S., Verlinden, B. E., Christiaens, S., Shpigelman, A., Vicent, V., ... & Geeraerd, A. (2014). Pectin modifications and the role of pectin- degrading enzymes during postharvest softening of Jonagold apples. Food Chemistry, 158, 283-291. Gwanpua, S. G., Mellidou, I., Boeckx, J., Kyomugasho, C., Bessemans, N., Verlinden, B. E., ... & Geeraerd, A. H. (2016A). Expression analysis of candidate cell wall-related genes associated with changes in pectin biochemistry during postharvest apple softening. Postharvest Biology and Technology, 112, 176-185. Gwanpua, S. G., Verlinden, B. E., Hertog, M. L. A. T. M., Nicolai, B. M., Hendrickx, M., & Geeraerd, A. H. (2016B). Understanding the regulation of texture degradation during apple softening-a kinetic modelling approach. In VIII International Postharvest Symposium: Enhancing Supply Chain and Consumer Benefits-Ethical and Technological Issues 1194 (pp. 1399-1406). Gwanpua, S. G., Verlinden, B. E., Hertog, M. L., Nicolai, B. M., Hendrickx, M., & Geeraerd, A. (2017). A transcriptomics-based kinetic model for enzyme-induced pectin degradation in apple (Malus× domestica) fruit. Postharvest Biology and Technology, 130, 64-74. Hampson, C. R., Quamme, H. A., Hall, J. W., MacDonald, R. A., King, M. C., & Cliff, M. A. (2000). Sensory evaluation as a selection tool in apple breeding. Euphytica, 111(2), 79-90. Harada, T., Sunako, T., Wakasa, Y., Soejima, J., Satoh, T., & Niizeki, M. (2000). An allele of the 1-aminocyclopropane-1-carboxylate synthase gene (Md-ACS1) accounts for the low level of ethylene production in climacteric fruits of some apple cultivars. Theoretical and Applied Genetics, 101(5-6), 742-746. Harb, J., Gapper, N. E., Giovannoni, J. J., & Watkins, C. B. (2012). Molecular analysis of softening and ethylene synthesis and signaling pathways in a non-softening apple cultivar, ‘Honeycrisp’ and a rapidly softening cultivar, ‘McIntosh’. Postharvest biology and technology, 64(1), 94-103. Hargarten, H., Waliullah, S., Kalcsits, L., & Honaas, L. A. (2018). Leveraging transcriptome data for enhanced gene expression analysis in apple. Journal of the American Society for Horticultural Science, 143(5), 333-346. Harker, F. R., Stec, M. G., Hallett, I. C., & Bennett, C. L. (1997). Texture of parenchymatous plant tissue: a comparison between tensile and other instrumental and

97 sensory measurements of tissue strength and juiciness. Postharvest biology and technology, 11(2), 63-72. Harker, F. R., Maindonald, J., Murray, S. H., Gunson, F. A., Hallett, I. C., & Walker, S. B. (2002). Sensory interpretation of instrumental measurements 1: texture of apple fruit. Postharvest biology and technology, 24(3), 225-239. Harker, F. R., White, A., Gunson, F. A., Hallett, I. C., & De Silva, H. N. (2006). Instrumental measurement of apple texture: A comparison of the single-edge notched bend test and the penetrometer. Postharvest Biology and Technology, 39, 185-192. Hayama, H., Shimada, T., Fujii, H., Ito, A., & Kashimura, Y. (2006). Ethylene-regulation of fruit softening and softening-related genes in peach. Journal of Experimental Botany, 57(15), 4071-4077. Hayashi, T. (1989). Xyloglucans in the primary cell wall. Annual review of plant biology, 40(1), 139-168. Ireland, H. S., Gunaseelan, K., Muddumage, R., Tacken, E. J., Putterill, J., Johnston, J. W., & Schaffer, R. J. (2014). Ethylene regulates apple (Malus× domestica) fruit softening through a dose× time-dependent mechanism and through differential sensitivities and dependencies of cell wall-modifying genes. Plant and cell physiology, 55(5), 1005-1016. Johnston, J. W., Hewett, E. W., & Hertog, M. L. (2002). Postharvest softening of apple (Malus domestica) fruit: a review. New Zealand Journal of Crop and Horticultural Science, 30(3), 145-160. Kende, H. (1993). Ethylene biosynthesis. Annual review of plant biology, 44(1), 283- 307. Kenis, K., Keulemans, J., & Davey, M. W. (2008). Identification and stability of QTLs for fruit quality traits in apple. Tree Genetics & Genomes, 4(4), 647-661. Khan, M. A., Han, Y., Zhao, Y. F., Troggio, M., & Korban, S. S. (2012). A multi- population consensus genetic map reveals inconsistent marker order among maps likely attributed to structural variations in the apple genome. PLoS One, 7(11). Kim, D., Langmead, B., & Salzberg, S.L. (2015). HISAT: a fast spliced aligner with low memory requirements. Nature methods 12(4), 357-360. King, G. J., Maliepaard, C., Lynn, J. R., Alston, F. H., Durel, C. E., Evans, K. M., ... & Tartarini, S. (2000). Quantitative genetic analysis and comparison of physical and sensory descriptors relating to fruit flesh firmness in apple (Malus pumila Mill.). Theoretical and applied genetics, 100(7), 1074-1084. Kolde, R., & Kolde, M.R. (2015). Package ‘pheatmap’. R Package 1(7), 790.

Koskinen, P., Törönen, P., Nokso-Koivisto, J., & Holm, L. (2015). PANNZER: high- throughput functional annotation of uncharacterized proteins in an error-prone environment. Bioinformatics, 31(10), 1544-1552. Kulkarni, M. M. (2011). Digital multiplexed gene expression analysis using the NanoString nCounter system. Current protocols in molecular biology, 94(1), 25B-10. Leubner-Metzger, G., & Meins Jr, F. (1999). Functions and regulation of plant β-(PR-2). Pathogenesis-Related Proteins in Plants; CRC Press: Boca Raton, FL, USA. Li, H., Handsaker, B., Wysoker, A., Fennell, T., Ruan, J., Homer, N., et al. (2009). The sequence alignment/map format and SAMtools. Bioinformatics 25(16), 2078-2079. Li, X. W., Jiang, J., Zhang, L. P., Yu, Y., Ye, Z. W., Wang, X. M., ... & Jia, H. J. (2015). Identification of volatile and softening-related genes using digital gene expression profiles in melting peach. Tree genetics & genomes, 11(4), 71. Li, T., Xu, Y., Zhang, L., Ji, Y., Tan, D., Yuan, H., & Wang, A. (2017). The jasmonate- activated transcription factor MdMYC2 regulates ETHYLENE RESPONSE FACTOR and ethylene biosynthetic genes to promote ethylene biosynthesis during apple fruit ripening. The Plant Cell, 29(6), 1316-1334. Liao, Y., Smyth, G.K., & Shi, W. (2014). featureCounts: an efficient general purpose program for assigning sequence reads to genomic features. Bioinformatics 30(7), 923- 930. Liu, K., Kang, B. C., Jiang, H., Moore, S. L., Li, H., Watkins, C. B., ... & Jahn, M. M. (2005). A GH3-like gene, CcGH3, isolated from Capsicum chinense L. fruit is regulated by auxin and ethylene. Plant Molecular Biology, 58(4), 447-464. Livak, K. J., & Schmittgen, T. D. (2001). Analysis of relative gene expression data using real-time quantitative PCR and the 2-ΔΔCt method. methods, 25(4), 402-408. Longhi, S., Moretto, M., Viola, R., Velasco, R., & Costa, F. (2012). Comprehensive QTL mapping survey dissects the complex fruit texture physiology in apple (Malus x domestica Borkh.). Journal of Experimental Botany, 63(3), 1107-1121. Lopez-Gomez, R., & Gomez-Lim, M. A. (1992). A method for extracting intact RNA from fruits rich in polysaccharides using ripe mango mesocarp. HortScience, 27(5), 440- 442. Luby, J. J. & Bedford, D. S. (1992). Honeycrisp Apple. Minnesota Agricultural Experiment Station. University of Minnesota Digital Conservancy. Retrieved from http://hdl.handle.net/11299/141481

Lv, J., Zhang, M., Zhang, J., Ge, Y., Li, C., Meng, K., & Li, J. (2018). Effects of methyl jasmonate on expression of genes involved in ethylene biosynthesis and signaling pathway during postharvest ripening of apple fruit. Scientia Horticulturae, 229, 157-166. Mann, H. S., Alton, J. J., Kim, S., & Tong, C. B. (2008). Differential expression of cell- wall–modifying genes and novel cDNAs in apple fruit during storage. Journal of the American Society for Horticultural Science, 133(1), 152-157. Mann, H., Bedford, D., Luby, J., Vickers, Z., & Tong, C. (2005). Relationship of instrumental and sensory texture measurements of fresh and stored apples to cell number and size. HortScience, 40, 1815-1820. Massiot, P., Baron, A., & Drilleau, J. F. (1996). Effect of storage of apple on the enzymatic hydrolysis of cell wall polysaccharides. Carbohydrate polymers, 29(4), 301- 307. McKay, S. J. (2010). The genetic dissection of fruit texture traits in the apple cultivar ‘Honeycrisp’ (Doctoral dissertation, University of Minnesota). McKay, S. J., Bradeen, J. M., & Luby, J. J. (2011). Prediction of genotypic values for apple fruit texture traits in a breeding population derived from ‘Honeycrisp’. Journal of the American Society for Horticultural Science, 136(6), 408-414. Miceli, C. D., Infante, R., & Inglese, P. (2010). Instrumental and sensory evaluation of eating quality of peaches and nectarines. European Journal of Horticultural Science, 75(3), 97-102. Miedes, E., & Lorences, E. P. (2004). Apple (Malus domestica) and tomato (Lycopersicum esculentum) fruits cell-wall hemicelluloses and xyloglucan degradation during Penicillium expansum infection. Journal of Agricultural and Food Chemistry, 52(26), 7957-7963. Moriya, S., Iwanami, H., Haji, T., Okada, K., Yamada, M., Yamamoto, T., & Abe, K. (2015). Identification and genetic characterization of a quantitative trait locus for adventitious rooting from apple hardwood cuttings. Tree Genetics & Genomes, 11(3), 59. Moriya, S., Kunihisa, M., Okada, K., Iwanami, H., Iwata, H., Minamikawa, M., ... & Matsumoto, T. (2017). Identification of QTLs for flesh mealiness in apple (Malus× domestica Borkh.). The Horticulture Journal, MI-156. Muñoz-Bertomeu, J., Miedes, E., & Lorences, E. P. (2013). Expression of xyloglucan endotransglucosylase/hydrolase (XTH) genes and XET activity in ethylene treated apple and tomato fruits. Journal of plant physiology, 170(13), 1194-1201.

100

NASS (2017). Washington Tree Fruit Acreage Report 2011. US Department of Agriculture, National Agricultural Statistics Service, Washington Field Office, Washington, DC. Neilson, J., Lagüe, M., Thomson, S., Aurousseau, F., Murphy, A. M., Bizimungu, B., ... & Tai, H. H. (2017). Gene expression profiles predictive of cold-induced sweetening in potato. Functional & integrative genomics, 17(4), 459-476. Ng, J. K., Schröder, R., Sutherland, P. W., Hallett, I. C., Hall, M. I., Prakash, R., ... & Johnston, J. W. (2013). Cell wall structures leading to cultivar differences in softening rates develop early during apple (Malus x domestica) fruit growth. BMC plant biology, 13(1), 183. Ng, J. K., Schröder, R., Brummell, D. A., Sutherland, P. W., Hallett, I. C., Smith, B. G., ... & Johnston, J. W. (2015). Lower cell wall pectin solubilisation and galactose loss during early fruit development in apple (Malus x domestica) cultivar ‘Scifresh’ are associated with slower softening rate. Journal of plant physiology, 176, 129-137. Nobile, P. M., Wattebled, F., Quecini, V., Girardi, C. L., Lormeau, M., & Laurens, F. (2011). Identification of a novel α-L-arabinofuranosidase gene associated with mealiness in apple. Journal of experimental botany, 62(12), 4309-4321. Nunan, K. J., Davies, C., Robinson, S. P., & Fincher, G. B. (2001). Expression patterns of cell wall-modifying enzymes during grape berry development. Planta, 214(2), 257-264. Ortiz, A., Graell, J., & Lara, I. (2011). Cell wall-modifying enzymes and firmness loss in ripening ‘Golden Reinders’ apples: A comparison between calcium dips and ULO storage. Food Chemistry, 128(4), 1072-1079. Pauly, M., Gille, S., Liu, L., Mansoori, N., de Souza, A., Schultink, A., & Xiong, G. (2013). Hemicellulose biosynthesis. Planta, 238(4), 627-642. Péneau, S., Hoehn, E., Roth, H. R., Escher, F., & Nuessli, J. (2006). Importance and consumer perception of freshness of apples. Food quality and preference, 17(1-2), 9-19. Percy, A. E., O'Brien, I. E., Jameson, P. E., Melton, L. D., MacRae, E. A., & Redgwell, R. J. (1996). Xyloglucan endotransglycosylase activity during fruit development and ripening of apple and kiwifruit. Physiologia Plantarum, 96(1), 43-50. Perini, P., Pasquali, G., Margis-Pinheiro, M., de Oliviera, P. R. D., & Revers, L. F. (2014). Reference genes for transcriptional analysis of flowering and fruit ripening stages in apple (Malus× domestica Borkh.). Molecular Breeding, 34(3), 829-842. Qi, Y., Lei, Q., Zhang, Y., Liu, X., Zhou, B., Liu, C., & Ren, X. (2017). Comparative transcripome data for commercial maturity and physiological maturity of ‘Royal Gala’

101 apple fruit under room temperature storage condition. Scientia horticulturae, 225, 386- 393. Robinson, M. D., McCarthy, D. J., & Smyth, G. K. (2010). edgeR: a Bioconductor package for differential expression analysis of digital gene expression data. Bioinformatics, 26(1), 139-140. Rose, J. K., Braam, J., Fry, S. C., & Nishitani, K. (2002). The XTH family of enzymes involved in xyloglucan endotransglucosylation and endohydrolysis: current perspectives and a new unifying nomenclature. Plant and Cell Physiology, 43(12), 1421-1435. Sato, T., Kudo, T., Akada, T., Wakasa, Y., Niizeki, M., & Harada, T. (2004). Allelotype of a ripening-specific 1-aminocyclopropane-1-carboxylate synthase gene defines the rate of fruit drop in apple. Journal of the American Society for Horticultural Science, 129(1), 32-36. Schaffer, R. J., Ireland, H. S., Ross, J. J., Ling, T. J., & David, K. M. (2013). SEPALLATA1/2-suppressed mature apples have low ethylene, high auxin and reduced transcription of ripening-related genes. AoB plants, 5. Schlosser, J., Olsson, N., Weis, M., Reid, K., Peng, F., Lund, S., & Bowen, P. (2008). Cellular expansion and gene expression in the developing grape (Vitis vinifera L.). Protoplasma, 232(3-4), 255. Schmitz, C. A. (2013). Enabling marker-assisted breeding for fruit texture traits in progeny of the apple cultivar Honeycrisp (Master’s thesis, University of Minnesota) Segonne, S. M., Bruneau, M., Celton, J. M., Le Gall, S., Francin-Allami, M., Juchaux, M., ... & Renou, J. P. (2014). Multiscale investigation of mealiness in apple: an atypical role for a pectin methylesterase during fruit maturation. BMC plant biology, 14(1), 375. Shalom, L., Samuels, S., Zur, N., Shlizerman, L., Zemach, H., Weissberg, M., ... & Sadka, A. (2012). Alternate bearing in citrus: changes in the expression of flowering control genes and in global gene expression in on-versus off-crop trees. PLoS One, 7(10). Shi, H., Wang, Y., Qi, A., Zhang, Y., Xu, J., Wang, A., & Zhang, Y. (2013). PpACS1b, a pear gene encoding ACC synthase, is regulated during fruit late development and involved in response to salicylic acid. Scientia horticulturae, 164, 602-609. Shi, Y., Jiang, L., Zhang, L., Kang, R., & Yu, Z. (2014). Dynamic changes in proteins during apple (Malus x domestica) fruit ripening and storage. Horticulture research, 1(1), 1-21. Shin, S., Lee, J., Rudell, D., Evans, K., & Zhu, Y. (2016). Transcriptional regulation of auxin metabolism and ethylene biosynthesis activation during apple (Malus× domestica) fruit maturation. Journal of plant growth regulation, 35(3), 655-666.

102

Smyth, G. K., Ritchie, M., Thorne, N., & Wettenhall, J. (2005). LIMMA: linear models for microarray data. In Bioinformatics and Computational Biology Solutions Using R and Bioconductor. Statistics for Biology and Health. Stec, M. G., Hodgson, J. A., Macrae, E. A., & Triggs, C. M. (1989). Role of fruit firmness in the sensory evaluation of kiwifruit (Actinidia deliciosa cv Hayward). Journal of the Science of Food and Agriculture, 47(4), 417-433. Tadiello, A., Longhi, S., Moretto, M., Ferrarini, A., Tononi, P., Farneti, B., ... & Biasioli, F. (2016). Interference with ethylene perception at receptor level sheds light on auxin and transcriptional circuits associated with the climacteric ripening of apple fruit (Malus x domestica Borkh.). The Plant Journal, 88(6), 963-975. Tan, D., Li, T., & Wang, A. (2013). Apple 1-aminocyclopropane-1-carboxylic acid synthase genes, MdACS1 and MdACS3a, are expressed in different systems of ethylene biosynthesis. Plant molecular biology reporter, 31(1), 204-209. Tong, C., Krueger, D., Vickers, Z., Bedford, D., Luby, J., El-Shiekh, A., ... & Ahmadi, H. (1999). Comparison of Softening-related Changes during Storage of ‘Honeycrisp' Apple, Its Parents, and ‘Delicious'. Journal of the American Society for Horticultural Science, 124(4), 407-415. Trujillo, D. (2012). Molecular and enzymatic examination of cell wall-modifying proteins in relation to apple crispness maintenance (Master’s thesis, University of Minnesota) Trujillo, D. I., Mann, H. S., & Tong, C. B. (2012). Examination of expansin genes as related to apple fruit crispness. Tree genetics & genomes, 8(1), 27-38. Uluisik, S., Chapman, N. H., Smith, R., Poole, M., Adams, G., Gillis, R. B., ... & Samsulrizal, N. (2016). Genetic improvement of tomato by targeted control of fruit softening. Nature biotechnology, 34(9), 950-952. Velasco, R., Zharkikh, A., Affourtit, J., Dhingra, A., Cestaro, A., Kalyanaraman, A., ... & Salvi, S. (2010). The genome of the domesticated apple (Malus× domestica Borkh.). Nature genetics, 42(10), 833-839. Verma, S. (2014). A predictive genetic knowledge for apple “fresh sensation” provides information to increase breeding efficiency in Washington (Doctoral dissertation, Washington State University). Vickers, Z. A. T. A., & Bourne, M. C. (1976). A psychoacoustical theory of crispness. Journal of Food Science, 41(5), 1158-1164. Vickers, Z. M. (1984). Crispness and crunchiness-a difference in pitch? Journal of Texture Studies, 15(2), 157-163.

103

Vickers, Z. M., & Christensen, C. M. (1980). Relationships between sensory crispness and other sensory and instrumental parameters. Journal of Texture Studies, 11(3), 291- 308. Wakabayashi, K. (2000). Changes in cell wall polysaccharides during fruit ripening. Journal of Plant Research, 113(3), 231. Wakasa, Y., Kudo, H., Ishikawa, R., Akada, S., Senda, M., Niizeki, M., & Harada, T. (2006). Low expression of an endopolygalacturonase gene in apple fruit with long-term storage potential. Postharvest Biology and Technology, 39(2), 193-198. Wang, K. L. C., Yoshida, H., Lurin, C., & Ecker, J. R. (2004). Regulation of ethylene gas biosynthesis by the Arabidopsis ETO1 protein. Nature, 428(6986), 945-950. Wang, Z., Gerstein, M., & Snyder, M. (2009). RNA-Seq: a revolutionary tool for transcriptomics. Nature reviews genetics, 10(1), 57-63. Watkins,C.B., Kupferman, E. & Rosenberger, D.A. (2016) Apple. In Gross, K., Wang, C., & Saltveit, M. (Eds.), The Commercial Storage of Fruits, Vegetables, and Floral and Nursery Crops (pp. 176-194). Retrieved from https://www.ars.usda.gov/northeast- area/beltsville-md/beltsville-agricultural-research-center/food-quality- laboratory/docs/ah66/ Wei, J., Ma, F., Shi, S., Qi, X., Zhu, X., & Yuan, J. (2010). Changes and postharvest regulation of activity and gene expression of enzymes related to cell wall degradation in ripening apple fruit. Postharvest Biology and Technology, 56(2), 147-154. Więsyk, A., Iwanicka-Nowicka, R., Fogtman, A., Zagórski-Ostoja, W., & Góra- Sochacka, A. (2018). Time-course microarray analysis reveals differences between transcriptional changes in tomato leaves triggered by mild and severe variants of potato spindle tuber viroid. Viruses, 10(5), 257. Win, N. M., Yoo, J., Kwon, S. I., Watkins, C., & Kang, I. K. (2019). Characterization of fruit quality attributes and cell wall metabolism in 1-methylcyclopropene (1-MCP) treated 'Summer King' and 'Green Ball' apples during cold storage. Frontiers in Plant Science, 10, 1513. Yang, H., Liu, J., Dang, M., Zhang, B., Li, H., Meng, R., ... & Zhao, Z. (2018). Analysis of β-galactosidase during fruit development and ripening in two different texture types of apple cultivars. Frontiers in plant science, 9, 539. Yoshida, H., Nagata, M., Saito, K., Wang, K. L., & Ecker, J. R. (2005). Arabidopsis ETO1 specifically interacts with and negatively regulates type 2 1-aminocyclopropane-1- carboxylate synthases. BMC Plant Biology, 5(1), 14.

104

Yoshioka, H., Aoba, K., & Kashimura, Y. (1992). Molecular weight and degree of methoxylation in cell wall polyuronide during softening in pear and apple fruit. Journal of the American Society for Horticultural Science, 117(4), 600-606. Yoshioka, H., Kashimura, Y., & Kaneko, K. (1995). β-D-Galactosidase and α-L- arabinofuranosidase activities during the softening of apples. Journal of the Japanese Society for horticultural Science, 63(4), 871-878. Young, M. D., Wakefield, M. J., Smyth, G. K., & Oshlack, A. (2010). Gene ontology analysis for RNA-seq: accounting for selection bias. Genome biology, 11(2), R14. Yue, P., Lu, Q., Liu, Z., Lv, T., Li, X., Bu, H., ... & Wang, A. (2020). Auxin‐activated MdARF5 induces the expression of ethylene biosynthetic genes to initiate apple fruit ripening. The New Phytologist, 226(6), 1781. Zdunek, A., Konopacka, D., & Jesionkowska, K. (2010). Crispness and crunchiness judgment of apples based on contact acoustic emission. Journal of Texture Studies, 41(1), 75-9. Zhang, S., Xu, R., Gao, Z., Chen, C., Jiang, Z., & Shu, H. (2014). A genome-wide analysis of the expansin genes in Malus× Domestica. Molecular genetics and genomics, 289(2), 225-236. Zhang, Z., Wang, N., Jiang, S., Xu, H., Wang, Y., Wang, C., ... & Wu, S. (2017). Analysis of the xyloglucan endotransglucosylase/hydrolase gene family during apple fruit ripening and softening. Journal of agricultural and food chemistry, 65(2), 429-434. Zhao, J., Quan, P., Liu, H., Li, L., Qi, S., Zhang, M., ... & Han, M. (2020). Transcriptomic and Metabolic Analyses Provide New Insights into the Apple Fruit Quality Decline during Long-Term Cold Storage. Journal of Agricultural and Food Chemistry, 68(16), 4699-4716. Zheng, Q., Song, J., Campbell-Palmer, L., Thompson, K., Li, L., Walker, B., ... & Li, X. (2013). A proteomic investigation of apple fruit during ripening and in response to ethylene treatment. Journal of proteomics, 93, 276-294.

105

Appendices

Table A1 Alignment summary of the RNA-Seq reads to the apple reference genome. The numbers shown are the averages of three biological replicates.

No. of No. of No. of raw alignment Group Samples Storage clean mapping reads rate (%) reads genes

Fresh 34,241,743 30,149,846 31,094 90.6 Honeycrisp Stored 30,984,289 29,555,786 31,211 91.4 Parent Fresh 36,064,961 34,651,990 31,955 90.9 MN1764 Stored 35,028,804 33,645,922 31,676 90.3 Fresh 21,446,684 21,109,459 32,244 90.4 CL136 Stored 20,110,557 19,088,718 31,763 91.7 Fresh 22,267,074 21,858,330 31,929 90.7 Retain EF129 Stored 21,686,882 21,336,808 31,650 91.5 Fresh 19,115,769 18,117,752 31,795 90.6 EF138 Stored 18,925,195 18,454,352 31,459 91.3 Fresh 20,353,488 19,969,759 32,645 91.6 CF117 Stored 18,313,785 17,647,405 32,472 92.1 Fresh 19,504,048 18,957,150 32,117 91.1 Lose CL156 Stored 22,039,618 21,753,675 30,980 92.7 Fresh 19,139,162 18,235,612 31,081 92.3 EF117 Stored 23,862,341 23,588,590 31,236 92.8 Fresh 19,740,478 19,508,389 31,815 91.9 CL121 Stored 19,448,799 18,886,681 31,402 92.3 Non-crisp Fresh 18,426,520 17,863,150 32,022 91.6 CL124 Stored 18,877,565 18,622,936 31,990 91.7

106

Table A2 nCounter® CodeSet Design.

Gene ID Position Probe Target Sequence

Capture AGCAGACCGGGGAGATCTTAATCAATGGCCGTAAAGAAACACTTGCTTTC MD00G1036800 631-730 Reporter GGAACATCGGCGTACGTGACTCAAGACGACACTCTAATGACAACATTAAC Capture GAAGCAAAATGGAGCTCGTGTTGGTGACTATTGTAGTGATCTCACTTAGA MD01G1042500 1354-1453 Reporter TTCTTCAAGAACTTGATCAACCAAGTCCTAGACGTACTCCTAGCAAACAA Capture GTCACATAAATCCAGCTGTGACATTTGGTCTCTTCCTAGCAAGGAAGCTC MD01G1062800 580-679 Reporter TCCCTGACCAGATCCGTGTTCTACATCATCATGCAAAGCCTTGGGGCAAT Capture ACGGCATGACCCGCGACCTTCTCTGGCTCTGAGTCCTCTCTTCGTTTCCA MD01G1087300 2211-2310 Reporter TTCTCGAGTCTCTCGACCCACGACTGCCTTTGCAAGATGAGCTTGAGCTT Capture GACGGTAATGCCGACGCTCTTTTGAGCAGTTGGAACCAATGGGCCTCGGT MD01G1213100 772-871 Reporter TCCGGCCACCCAGGTGTTCATGGGGTTACCGGCAGCTCCTGAGGCCGCTC Capture TGTCTCTACATAAAGCAGTGGAATTAACTGCGGTATTGGAATGCTAGTTT MD02G1057200 1535-1634 Reporter GGTTTTGCTTTGTAGCCTGTATTGCAAGACTTGCCAAAAAAAGGGGTAAG Capture TCTAATTCCAGGAAGAAACTTGAGACAAAGACATGCATGGAGATTAGTCA MD03G1019900 15-114 Reporter AGCCTAAACTATGGTACACATGCATTTCCATCAAGAAATTTCAAGGAAAC Capture TGCCATTGTTCATAAGGTCTTTGGTGCTAGCAATGTCAGCAAAATGTTGC MD03G1060100 207-306 Reporter AGGAGCTTCCCATTCATCAAAGAGGAGATGCTGTGAGCAGTTTGGTTTAT Capture CCCATGTTCGAGAAATGCCTTGCTTCTTAGTATAGTCTTACTGCTTCAAC MD03G1108400 689-788 Reporter AGAGGAAATTTCGGAAAACTCCCAGCAACTTATCATTGGTTGTTCTCTAT Capture TCGGCGAACTCAGAGAGTGAGTCGTTCGACGCGGTGATGGACCAGTGCTG MD05G1092300 1543-1642 Reporter CCTAACGATGGAGGAATCACTTAACTCGGTGTATCGGCAGGGTCGAGTGG MD05G1098700 80-179 Capture GTGGCGTCGGAGTTCCAAAAAACGGAAAGTTCAACTTGACCCATGCACGG

107

Reporter TTGCAGGAGCTCGTCAACCACGCTGCCTCCCTGTTAATATCCTTCAGCAT Capture CTCCCTTGCTTTTAACCAGCGAGACCAACCAGACGGCCACCTTGAATTTC MD05G1160100 711-810 Reporter AAAATGCAAACTGATCGTGCCTACTTAGGTGAAGAGCTCATCGAGGAAAT Capture CTGCACTTTGTGGTGCACCGCGACTCCATGTTCAACCATGCACAAACAGT MD05G1297900 1136-1235 Reporter ACAGAGAAACCAAATTCAAAGAAGGCTAGAAAATCTATCGTGAAGTATAT Capture ACTTTGAGGCCAATCCTAATGGCCTGAGCAACACGAGGTTCGCTCAGAGC MD05G1310400 648-747 Reporter AGCTACACTACCACACCCAGCAACTACCAGTACAACAACAACTTTGAGGC Capture TATCAAGGGAATCATGTCGGCGATTCATCAGATCCGAGAGAAGGCTCACA MD05G1313300 184-283 Reporter AGGACGGTCTGAAGAAGAACGAAGAGACGATTTCGAGTGTGGCAGCTGAG Capture TCATATTTTGGCATCTCTTCGGACGGAGTACTGAGTCTTAAGCTTTCGGA MD05G1331000 754-853 Reporter CCATGGCAGCAACTGGGGCCTTAACTGGGAGGCACCGGTGAATCCATGTG Capture GACTGATGACCAAGTTTCCAATTCTTCATCGGAAGCAAGAACCGGATCAG MD05G1349800 898-997 Reporter CTTCACCACCCCAAAACATCAATGAAGTTGCGAAAAACGATCAGATTGGT Capture CCTATGAGAAAGCTCAGAAGGCAAACATCAGAGTAAAGGGCTTGCTCATT MD06G1090600 903-1002 Reporter ACTAACCCCTCAAACCCCTTAGGTACTGTCCTTGACCGAGACACCCTCAT Capture GGACTGGATATGAGATTCTCCGAATCACATCCTACTTGCAGCAGAACCTA MD06G1233800 1312-1411 Reporter ACCAAATTGATGGTACCCTTCTTCGTTCTCCACGGAACAGCAGACACAGT Capture GGGTTCGTCTAGTTCGAGTAAACGGAATATGGTGCGTGGTATATGTGATG MD07G1237500 937-1036 Reporter TGTGAAGGCTTGTTAGACTTATCATAACTTGGGATCCAAGACTCTTCTCC Capture ACCACCACACCGCAACGGCACCTAAATTCTCTCTCTCGAACCCCTTCCTT MD07G1247100 92-191 Reporter TCTCTCTCTAAACTCCGACACAAACCCGCCTCCATTTTCCCGCAAACCCA Capture TAGGGTGGAGGATCTTCTGAGGACGGTAGTCAAAATGTTCTACAATTCAA MD07G1259200 864-963 Reporter GAAAGGAGAACTTCCCAGAGGACATTGACACAACGGACGAGGAGTCCCTG MD07G1270800 1379-1478 Capture CAGGATGCTACAGCCGATGAGGAGGGGTACGACTATGAAGAAGAGGAGGA

108

Reporter AGCTCAGGAGGAGGCTTGATCTTGGATCTCGATTCTCATCTGTTCTCTGT Capture AATACTCGACTACTGTCGGTTTCATCAAGTACCCGCTCGCTCTAATAAGA MD07G1274100 607-706 Reporter AAACTTGGAGAAAATATTGCTACCCTTTATGTGGCCGAAAAAGGGTGGTT Capture CAGAAAGCCGGGAGAGCATTGTTCTACTACTTTGTAGAGTCGCCCCAGGA MD08G1106600 413-512 Reporter TTCATCAACTAAACCTTTGGTGCTATGGTTAAATGGAGGACCAGGATGCT Capture ACCTTTGTGTATCGAACATCTACAAATCAATGGTTGACAGCCGCTCCAAT MD08G1127900 466-565 Reporter GCGGACCCCACGGTCATTTTTCGACGCTGGAAATATCAACGGAAAGATCT Capture CAAATTCGGGAAGGGACGTTTGAACTTGAAAATGTCAAGCCATGCTCGTG MD08G1195600 649-748 Reporter CAATTTATGACTCCGCTATCAATCATCGGGGAAGAAGGCTACTTCAGGTG Capture ACAGCTGTGACCTTCAACAGGTGCAATAACTTGGTGGTGAAGAATCTGAA MD10G1179100 902-1001 Reporter TATCCAAGACGCACAACAAATCCATGTCATATTCCAAAACTGCATCAACG Capture AAAGGAACAAACACACTGTATCGACCAACTGGCCTAATCCAACCATAATC MD10G1315100 72-171 Reporter CAATGATTAAAAACACCAGACAAACAAATGTGACTGGGTTCCCAGCCTTC Capture ATCTGCAAGGCCAAATCAAAGTCTCTACAGCAATTGACACAACTCTTGTG MD11G1189000 479-578 Reporter ACCAATGCCTACCCTCCCTCCGATGGAGTATACACTGACCCTGCAAACCA Capture CACACTCTCACACACACTCTCTCTCTCTAGGCATTTTCCATCCAAACCGT MD11G1230200 21-120 Reporter CACCATGTCCCTGTTTAACACCCAGTTCTCCTTCTTCTTCCTCACAGTCT Capture ATTGCAGAATAGGGAATATGGAGCATGCACGCAGTGTATTTGTGGAGATG MD12G1164900 962-1061 Reporter CTTAAAAGGGGTTTGAAACCAAACGATCATGTCTTTGATGATATGGTGGC Capture ATGTCAACGGCAACCTCATGTTCAACATCAAAGGCTCACTTTTCAGCCTT MD12G1183000 406-505 Reporter CACGACCGCCGCGTGTTGGTCGACAACGCCGGTAATCCAATCGTCTCTTT Capture CCATATGCCTACAAGTTACGTTCTTTGCATGCGGAGGCATGAGCATCGGT MD13G1108900 398-497 Reporter ATGGGCATGTCGCACAAGGTCGGAGACGCCCTGTCATATTTCACGTTCCT MD13G1112700 5-104 Capture TAAGAATAACGTGTTCTCTCTCACGCTTGGACTGGTTGCCTCCCTCTCAT

109

Reporter AGTTTCAACTAGAAAGAGAGAGCAGGCTTAAATTCAAGTTCTGTGTGTCA Capture TTTGACAGTGCTGAAGCCGCAGCTCTTGCCTATGATCAAGCTGCCTTTTC MD13G1213100 358-457 Reporter CATGCGAGGCTCTGCGGCGGTTCTCAATTTTCCAGTTGAAAGAGTCCGCA Capture GTATGCCTGATTTTCCGACAAGGCGATTCCGGCCGATGACCAGAGAACTA MD14G1056600 231-330 Reporter GACCTCGCTACTATTGGTAAACGATGCCTCTTACTCCCATGACAAATATT Capture ACTCTCTACTCTACTTCTCCTCCTTGTCTTCGTCAACCATCTTCCTCCCT MD14G1110100 153-252 Reporter CTCATATACTTCCTCTTCCTCTTCACTTTCACGCAACAGCGCCGCCGTCC Capture GCCTTTGGTCACCTTGCCTTTGGAACGACTCTGGTCGAACTTTAGCCCTC MD14G1150200 1028-1127 Reporter GTGTTATCGAACCCACTCCACTAACATGGGTTTACCAAAAGCACTTGGTT Capture ACACCGTTTTTATGTCACTCCCGGCATTCTACAACCAACTAAAGGAAGAA MD15G1023600 376-475 Reporter CATGGTGCTGGACTCGGGTCTTTCTTTGTCTCCGCCACGCCGGGCTCGTT Capture TATCTGAGTCGCACCCACAGCTTTCTTGAAGGCCAAAGTGCTTTTACAAG MD15G1203500 587-686 Reporter AAACCTAGTTATGGCTATAGATATTGAGCAGCGGCAGCAGCCTTCTCCTG Capture AGTGGAAGGGAAGTTAGGGGAGGAATTTCATGTTTTCAGTGTCATTGTGT MD15G1297000 1106-1205 Reporter AATGAAGTTGGACGTCTAGATAGTTGTCGTGGATGCTCGTGCTTGATTGG Capture AGTACAGCATCATCTGGAACAACCATCATACAGTGTTTCTAGTGGACAAC MD16G1091200 617-716 Reporter GTCCCAGTGAGGGAGTTTCAGCACAGCAGTTCATTTTTCCCATCAAAACC Capture CTTGGATTGTGGGCCATTCCTTGGCATCCACACAGAGGCGGCAGTGAGGT MD16G1158300 1026-1125 Reporter TTGGCCAGGTAAACGAGATCGACATTAACTATGCGTTGGCTAACACAATC Capture ACAAAACAAATCGCTTCTAGGATGGATTGATTGCCAACATCGGATGAGAA MD16G1277800 2087-2186 Reporter GAAAAAGAAATGAACAATCAGTCACACTCGATGTCAATGTCGATGCAAAT Capture GTTCCACCCGCGTTGATTGCATTTGAGGGTCATGTGCATCCTATTGACCC MD17G1141200 1606-1705 Reporter ATCATGTCATGTGGCTGGGCTAGGTTCTCGATCTCCAGAGGTTCCCGAAG MD17G1256100 388-487 Capture GTAATAGCAGGGAAAACTGTTATTATTTTATGCAATCCTGGAACCGGGGA

110

Reporter ATTCAGGCAACTTCCCGATTCATGCCTTCTTGTACCCCTTCCCAAGGAAA

Housekeeping gene Capture GGGATACGTTACTCTACTCTAACGTCAAGCTGGACTGAGCATGGCAGGCA MD04G1005300 1236-1335 Reporter GCTACTTGGTCAATGTTTCAGGCACAGGTTTTTCGAGCCCATCAGACGCT Capture GAAGGTAGACCAACTGGTTGGTCATTGTCAGATCCTTCTCGTAGGAGAAA MD09G1190100 1277-1376 Reporter CTCTGGACCGATTCTAAATTCTGGAAACATTTCTAGGCAAAAAAGTCCTG Capture GGTGGCAGATAAAAGCGCTTGAAAACAAGGTTGGTGTTTTGTTCACTCTT MD16G1145800 1746-1845 Reporter TGGCGATGTTAAGGGATTAACTCTCTATGCTTTCTCATCTCACTGTGAAT

111

Table A3 Primers for genes tested in qRT-PCR.

Gene ID Type Sequence Tm

Forward CAGCAATCAACCAACAGGTTTAG 62.2 MD02G1057200 Reverse CAGACCCTTGGGATTTCTTCTC 62.3 Forward AGTTCGACCTTAACCGGATTG 62.1 MD05G1098700 Reverse GGGCGATGCTGAAGGATATT 62.1 Forward AGATTGCCACACCTCCTACT 62.7 MD09G1190100 Reverse CCTCACCACCAGATTGTCTTTC 62.7 Forward TGGTAGTAGTTTCCAGGAGGT 62.1 MD10G1179100 Reverse CCTTCTATGGTGTCCGTGTATG 62.0 Forward CAGGCGTAGTCACAGCTTATT 62.1 MD10G1315100 Reverse TAAGGTTGCCCGCTTCTATTC 62.2 Forward CCATGGACCTTTCTCCTCTTTC 62.3 MD11G1230200 Reverse GTGTAGGACTTGACCTCCTTTG 62.1 Forward GTGATGATGCGTGGTTGTTTG 62.4 MD12G1164900 Reverse GCGTCAAGTTCTACCTCTCTTG 62.3 Forward CCCGCCAACTACTTCATCTT 61.9 MD14G1150200 Reverse CCGAGTGGCTATTCCCATATT 61.6 Forward GAACTCTTGGGCCACGATAA 61.9 MD16G1091200 Reverse TGTTGGGTCAAACCAGAGATAG 61.9

112

Figure A1 A schematic description of the workflow of this study.

113

114