Physiological, biochemical and transcriptional analysis of onion bulbs during storage

Abstract

Background and Aims

During the transition from endo-dormancy to eco-dormancy and subsequent growth, the onion bulb undergoes the transition from sink organ to source, to sustain cell division in the meristematic tissue. The mechanisms controlling these processes are not fully understood. Here, a detailed analysis of whole onion bulb physiological, biochemical and transcriptional changes in response to sprouting is reported, enabling a better knowledge of the mechanisms regulating post-harvest onion sprout development.

Methods

Biochemical and physiological analyses were conducted on different cultivars (‘Wellington’, ‘Sherpa’ and ‘Red Baron’) grown at different sites over 3 years, cured at different temperatures (20, 24 and 28 °C) and stored under different regimes (1, 3, 6 and 6 → 1 °C). In addition, the first onion oligonucleotide microarray was developed to determine differential gene expression in onion during curing and storage, so that transcriptional changes could support biochemical and physiological analyses.

Key Results

There were greater transcriptional differences between samples at harvest and before sprouting than between the samples taken before and after sprouting, with some significant changes occurring during the relatively short curing period. These changes are likely to represent the transition from endo-dormancy to sprout suppression, and suggest that endo-dormancy is a relatively short period ending just after curing. Principal component analysis of biochemical and physiological data identified the ratio of monosaccharides (fructose and glucose) to disaccharide (sucrose), along with the concentration of zeatin riboside, as important factors in discriminating between sprouting and pre-sprouting bulbs.

Conclusions

These detailed analyses provide novel insights into key regulatory triggers for sprout dormancy release in onion bulbs and provide the potential for the development of biochemical or transcriptional markers for sprout initiation. Evidence presented herein also suggests there is no detrimental effect on bulb storage life and quality caused by curing at 20 °C, producing a considerable saving in energy and costs.

INTRODUCTION

Onions bulbs have evolved as a storage organ to allow the plant to over-winter. During the transition from dormancy (endo-dormancy; dormancy reliant on conditions or factors within the bulb) to sprout suppression (eco-dormancy; dormancy reliant on external environmental factors) and subsequent growth, the bulb undergoes the transition from sink organ to source, to sustain cell division in the meristematic tissue. The mechanisms controlling these processes are not fully understood.

Many physiological and biochemical characteristics change during onion bulb storage, including water content and the concentration of flavour compounds, carbohydrates, minerals and plant growth regulators (PGRs). Changes in these characteristics are likely to be linked to respiration and remobilization of carbohydrates to provide energy for the growing sprout. All nutrients required for growth of the sprout must come from within the bulb; therefore, changes in the concentrations of key metabolites could be used to predict the onset of sprouting. Different onion cultivars vary in their innate storage life, but relatively little is known of the fundamental basis of this variation other than largely empirical correlations with traits such as dry matter, pungency, skin quality and degree of polymerization (DP) of non-structural carbohydrates (NSCs). Fluctuations in certain metabolites are known to coincide with sprouting, but there is currently no biochemical or molecular assay that accurately predicts sprouting.

Dry weight, NSCs, pyruvate and flavonols can be used to assess onion bulb quality. These characteristics differ both between cultivars and during storage. Onion bulbs with high dry weight are more suitable for longer term storage and tend also to contain higher concentrations of fructans (Darbyshire and Henry, 1979). Fructans represent a major carbohydrate reserve in most onion bulbs, and, in general, are enzymatically hydrolysed to fructose during the storage period, accounting for a concomitant increase in fructose concentration (Hurst et al., 1985). Soluble sugars are required to provide energy for sprout growth, and so the concentration of soluble sugars decreases when sprouting occurs (Rutherford and Whittle, 1982). Quercetin 4′-glucoside and quercetin 3,4′-diglucoside are the dominant flavonol compounds in onion flesh, and have been shown to increase during curing (Downes et al., 2010).

During over-winter storage in temperate climates, a gradual change in the relative composition of PGRs occurs as the concentrations of growth inhibitors decreases and the concentrations of growth promoters increases. PGR activity in ‘Rijnsberger’ (long-storing) and ‘Lancastrian’ (short-storing) bulbs showed peaks in the concentration of gibberellins (GAs), cytokinins and auxins, with the higher concentrations of auxins persisting as sprouting continued (Thomas, 1969; Thomas and Isenberg, 1972). However, the application of exogenous GAs and auxin failed to stimulate sprouting (Thomas, 1969). Abscisic acid (ABA) has been identified as part of the inhibitory complex present in onion bulbs (Tsukomoto et al., 1969) and has been demonstrated to play a functional role in maintenance of dormancy in Allium wakegi (a cross between Japanese bunching onion and shallot) through the application of exogenous ABA and fluridone, an inhibitor of ABA biosynthesis (Yamazaki et al., 1999a). Other studies have related a reduction in ABA concentration to loss of dormancy in onion (Matsubara and Kimura, 1991; Chope et al., 2006, 2007a) and A. wakegi (Yamazaki et al., 1999b).

Sprout growth, and the suppression thereof, is a major factor in determining the storage life of onions. To date, most strategies to delay sprouting and prolong storage have focused on crop husbandry, the storage environment, breeding programmes and maleic hydrazide. In maritime climates such as the UK, onions for storage are brought inside to be artificially cured and dried after harvest because of the risk of skin staining caused by wet weather during this period (O'Connor, 1979; Gubb and MacTavish, 2002) and to reduce disease incidence during storage. Curing dries the thin outer layers of the bulb to form one or more complete outer skins, which act as a barrier against water loss and pathogen attack (O'Connor, 1979; Maude et al., 1984). Onions are then stored under refrigeration to suppress sprout development further. Current UK curing and drying is based on a method developed in the 1970s on cultivars that are very different from those used today (Shipway, 1977; O'Connor and Shipway, 1978). Improvements to current methods, such as a reduction in the temperature and/or duration of the curing and drying periods, could deliver benefits in the form of energy savings and reduced greenhouse gas emissions, whilst still producing a continuous supply of onion bulbs of a satisfactory quality.

Advances in biochemical and transcriptional analysis methods now enable detailed analyses of sprout development during storage. Onion has an estimated nuclear genome size of 15 290–15 797 Mbp per 1C, and a 2C DNA amount of 31·7–33·2 pg (Arumuganathan and Earle, 1991; Ricroch and Brown, 1997), which is approx. 107 times larger than that of Arabidopsis (King et al., 1998). Consequently, there is a paucity of publically available genetic information on onion and Allium species, yet some key genes in the sulfur assimilation pathway have been cloned (McCallum et al., 2002; Eady et al., 2008) and a collection of >10 000 onion expressed sequence tags (ESTs) is available (Kuhl et al., 2004).

Here we utilize these advances to assess the effect of curing at different temperatures in combination with different storage temperatures on whole onion bulb physiology, transcriptome and targeted metabolites. These data show novel changes in metabolites and PGRs, supported by changes in transcripts with homology to genes involved in the metabolism of these metabolites at harvest, during curing, after dormancy break and following the onset of sprouting. These advances will contribute to improved storage methods for onions.

MATERIALS AND METHODS

Plant material

Experiments were carried out over three growing seasons at multiple sites using three UK-grown onion (Allium cepa L.) cultivars with varying storage potential, i.e. ‘Red Baron’ (red, average-storing), ‘Wellington’ (brown, long-storing) and ‘Sherpa’ (brown, average-storing), grown according to normal commercial practice. In 2007, onions were grown on sandy soil (Elveden Farms, Thetford, Norfolk, UK) and on sandy clay loam (A. Findlay's, Cardington, Bedfordshire, UK). The Norfolk site was drilled on 9 March 2007 at a rate of 47–48 plants m⁻² and bulbs were hand-harvested at 100 % fall-over into storage bins on 2 September 2007. The Bedfordshire site was drilled on 28 February 2007 at a rate of 37–38 plants m⁻² and bulbs were machine-harvested at 80 % fall-over on 16 September 2007. In 2008, onions were grown on sandy soil (A. W. Mortier Farms Ltd, Woodbridge, Suffolk, UK) and sandy clay loam (A. Findlay's). The Suffolk site was drilled on 13 March 2008 at a rate of 57 seeds m⁻² and bulbs were hand-harvested at 90 % fall-over into bins on 1 September 2008. The Bedfordshire site was drilled on 5 March 2008 at a rate of 57 seeds m⁻² and bulbs were machine-harvested at 100 % fall-down on 17 September 2008. In 2009, onions were grown at A. Findlay's from seeds drilled on 16 March 2009 at a rate of 54 seeds m⁻². Onions were harvested on 7 September 2009 at 100 % fall-down. All onions were free from maleic hydrazide.

Experimental design

At each growing site, the plot was divided into three sections, with onions harvested from each of the three sections being kept separate and treated as replicate blocks to be taken from the field to the store. In 2007, three different curing temperatures were used, i.e. 20, 24 and 28 °C, while only 20 and 28 °C were used in 2008 and 2009 (Table 1). All curing treatments were carried out at a relative humidity of 65–75 % at Sutton Bridge Experimental Station (Lincolnshire, UK). Experimental bulbs were placed in nets among loose bulbs in 1 t wooden boxes for 6 weeks. Following the post-harvest curing treatments, the nets were removed from the boxes and transported to Cranfield University within 3 h. Dry aerial parts and roots were removed, and diseased and damaged bulbs were discarded prior to storage in open plastic trays. Onions were stored at 0 ± 1 °C in 2007, at 1, 3 or 6 °C in 2008, and at 1, 6 or transferred from 6 to 1 °C after 13 weeks storage in 2009. Samples were taken straight after harvest (day 0), after 6 weeks post-harvest curing treatment and at regular intervals during cold storage for up to 7 months (4 months in 2007, 3 months in 2008 and 2 months in 2009).

Table 1.

Experimental design for the 3 year experiment

Year	2007	2008	2008	2008	2009
Site	Norfolk/Beds	Suffolk	Suffolk	Beds	Beds
Cultivar	Red Baron	Red Baron		Red Baron	Red Baron
	Sherpa		Sherpa	Sherpa	Sherpa
	Wellington	Wellington		Wellington	Wellington
Curing temperature (°C)	20, 24, 28	20, 28	20, 28	20, 28	20, 28
Storage temperature (°C)	1	1, 3, 6	1	1	1, 6, 6 → 1
Sample times	0 (at harvest), 1 (after curing), 2, 3, 4 (dormancy break), 5 (sprouting)	0 (at harvest), 1 (after curing), 2, 3 (dormancy break), 4 (sprouting)	0 (after harvest), 1 (after curing), 2, 3 (dormancy break), 4 (sprouting)	0 (at harvest), 1 (after curing), 2, 3 (dormancy break), 4 (sprouting)	0 (at harvest), 1 (after curing), 2 (dormancy break), 3 (sprouting)

Onions were sampled at harvest (0), after curing at 20, 24 or 28 °C (1), and at intervals during cold storage (2, 3, 4 and 5) at various temperatures (20 and 28 °C).

Four bulbs were pooled for each replicate and treatment combination. Each bulb was cut in half from top to bottom. One half of each bulb was used to provide tissue for pungency analysis (2007 only). From the remaining half, two 5 g vertical wedges were cut and immediately snap-frozen in liquid nitrogen, one of which was stored at –40 °C prior to lyophilization, and the other stored at –80 °C. Dry weight, sprout growth, sprout length, root growth, and fructose, sucrose and glucose concentrations were measured in all samples. Samples selected in triplicate for microarray analysis were ‘Sherpa’ and ‘Wellington’ from 2007 (Norfolk site), cured at 20 and 28 °C, and sampled at time 0 (harvest), 1 (after curing), 3 (before sprouting) and 4 (sprouting). Fructans were analysed in these samples, and in all samples from 2008 and 2009. Flavonols were measured in the microarray samples for 2007. The PGRs ABA, zeatin riboside (ZR) and isopentenyladenosine (IPA) were also measured in the microarray samples and in all samples from Suffolk in 2008.

Sprout length, rooting, disease incidence and dry weight

The presence or absence of roots was recorded, and then bulbs were assessed for the presence of a sprout after being cut in half. If a sprout was present, the length of the sprout, and the length and height of the bulb were recorded (Chope et al., 2006). Dry weight was recorded in lyophilized samples.

Pungency

Pungency was measured in all samples from 2007. Pyruvate concentration was measured according to Abayomi and Terry (2009), with slight modifications. Pooled samples (30 g) were homogenized in 45 mL of distilled water using a hand-held blender (Braun, Type 4192, Spain). A second identical 30 g sample was left to stand at 4 °C for 1 h in 5 % trichloroacetic acid (TCA) and homogenized to estimate the endogenous pyruvate concentration. The background pyruvate concentration as calculated from the TCA samples was subtracted from the water samples and expressed in μmol pyruvate g⁻¹ f. wt.

Non-structural carbohydrates

Non-structural carbohydrates were extracted according to Downes and Terry (2010). Briefly, fructans were initially extracted from freeze-dried onion powder (150 mg) with high-performance liquid chromatography (HPLC)-grade water, then fructose, glucose and sucrose were extracted from the same sample using aqueous methanol. All NSCs were quantified using HPLC coupled with evaporative light scattering detection. In 2007 and 2008, fructans were analysed separately from fructose, sucrose and glucose using a gradient of ethanol in water (Downes and Terry, 2010), or water (Davis et al., 2007), as a mobile phase, respectively. In 2009, all NSCs were analysed together in a single run using a gradient of acetonitrile in water (Downes and Terry, 2010).

Flavonols

Flavonols were extracted and quantified according to Downes et al. (2010). Briefly, flavonols were extracted from freeze-dried onion powder (150 mg) with acidified methanol and incubated at 37 °C for 1·5 h. Flavonol extracts were loaded onto an Agilent 1200 series HPLC system (Agilent, Berks, UK). The extracts were injected into an Agilent ZORBAX Eclipse XDB-C18 column. The mobile phase consisted of HPLC water with 0·5 g L⁻¹ trifluoroacetic acid (TFA) (A) and acetonitrile with 0·5 g L⁻¹ TFA (B). The program involved a linear increase/decrease of solvent B: 5–10 %, 5 min; 10–25 %, 5 min; 25–85 %, 6 min; 85–5 %, 4 min; 5 %, 5 min at a flow rate of 0·8 mL min⁻¹. The compounds were detected using a diode array detector set to a wavelength of 370 nm. The sample peak areas were compared against the peak area of authentic standards of known concentration.

Abscisic acid, zeatin riboside and isopentenyladenosine

Lyophilized bulb samples (200 mg) were extracted in HPLC-grade water (5 mL) for 12–18 h on a shaker at 4 °C. Before extraction, 50 ng of deuterated internal standards for ABA, ZR and IPA (d₄-ABA, d₃-DHZR and d₆-IPA, respectively, National Research Council of Canada, SK, Canada) were added. Samples were centrifuged and the pellet re-extracted with 1 mL water, and the supernatants were then pooled. Samples were passed through a pre-conditioned SepPak C18 cartridge (Waters, MA, USA), and eluted with 80 % (v/v) methanol (Hou et al., 2008), then evaporated to dryness and resuspended in 200 µL of 100 % methanol.

Samples were analysed using a Waters Alliance 2795 HPLC coupled to a Micromass Quattro quadrupole tandem mass spectrometer (Waters) with an electrospray ion source. Both the HPLC and the mass spectrometer were operated by MassLynx v4·0 SP3 software (Waters). Samples (10 µL) were separated on a Zorbax Eclipse XDB-C18 analytical column (3·5 µm, 2·1 × 100 mm, Agilent Technologies Inc., Santa Clara, CA, USA) with a 1 mm C18 guard column (Optiguard, Optimize Technologies, OR, USA) maintained at 25 °C. The mobile phase consisted of HPLC-grade methanol (A), water (B) and 5 % acetic acid (C). The gradient involved an increase/decrease in solvent A; 10–60 %, 15 min; 60–99·2 %, 15 min; 99·2–10 %, 2 min; 10 %, 8 min, at a constant proportion of solvent C (0·8 %) at a flow rate of 0·2 mL min⁻¹. Mass spectrometry was carried out using multiple reaction monitoring (MRM), with a capillary potential of 2·75 kV, a source temperature of 120 °C, a desolvation temperature of 350 °C, cone gas and desolvation gas flow rates of 50 and 950 L h⁻¹, respectively, and a collision gas (Ar) pressure of 0·5 Pa. The MRM transition, cone voltage, collision energy and retention window for each compound and internal standard are shown in Supplementary Data Table S6. Calibration curves were prepared using a range of standard solutions containing an increasing amount of ABA, ZR and IPA with a constant amount of deuterated internal standards, and the amount of endogenous compounds was quantified in relation to the internal standard using the calibration curves generated.

Extraction of RNA

Total RNA was extracted according to Cools et al. (2011) from frozen, ground onion tissue (100 mg) homogenized in 1 mL of extraction buffer using a pestle and mortar. The mixture was transferred to a 2 mL microtube and incubated at 65 °C for 10 min, then allowed to return to room temperature. Chloroform (1 mL) was added and mixed before being centrifuged at 16·2 g for 5 min at room temperature. The aqueous phase was removed to a clean tube and an equal volume of precipitation buffer mixed and incubated on ice for 30 min. Samples were centrifuged at 16·2 g for 20 min at 4 °C and the supernatant removed. The pellet was resuspended in SSTE [1·0 m NaCl, 0·5 % SDS, 10 mm Tris–HCl (pH 8·0), 1 mm Na₂EDTA (pH 8·0)] and briefly incubated at 37 °C, before being allowed to return to room temperature. A chloroform liquid–liquid extraction was carried out according to Cools et al. (2011). The pellet containing total nucleic acid was washed with 1 mL of 70 % (v/v) ethanol, then left to air dry and finally resuspended in 50 µL of diethylpyrocarbonate (DEPC)-treated water. Then, 30 µL of 8 m lithium chloride solution was added and the samples were incubated on ice overnight to precipitate RNA selectively. Samples were centrifuged at 16·2 g for 30 min at 4 °C, the supernatant was removed, and the pellet was washed with 0·5 mL of 70 % ethanol and resuspended in 15 µL of RNase-free water. Sample purity and integrity were verified using the RNA 6000 Nano Assay on the Agilent 2100 BioAnalyzer (Agilent Technologies Inc.) and then treated with Baseline Zero DNase (Epicentre, Madison, WI, USA) according to the supplier's instructions. Each sample was extracted three times to replicate the microarray data.

Microarray analysis

A total of 13 310 onion nucleotide sequences were available for the construction of a 60-mer oligonucleotide custom A. cepa microarray. The majority were obtained from public databases; 13 154 from the Onion Gene Index (http://compbio.dfci.harvard.edu/cgi-bin/tgi/gimain.pl?gudb=onion) and 102 from GenBank (http://www.ncbi.nlm.nih.gov/genbank/), with the remaining 54 sequenced directly from onion bulb tissue. Microarrays were designed using the Agilent Technologies e-array microarray design platform (https://earray.chem.agilent.com/earray/). Initially, a prototype chip was designed in a 4 × 44 K format where 60-mer oligonucleotide probes for ESTs and singletons were designed to both sense and antisense. Test hybridizations of RNA from a range of onion tissues (root, shoot, bulb and leaf) were used to orientate these probes, thus reducing the number of probes, so the final format was 8 × 15 K, consisting of eight independent arrays of 15 K probes, on a single glass slide. Each array consisted of 15 736 60-mer oligonucleotide probes in total, representing 536 internal control probes and 15 200 probes representing 13 310 unique onion sequences. In order to further our analyses of onion gene expression, the annotation for individual probes was populated from the closely related, fully sequenced, genome of rice (Oryza sativa). Translated blastx alignments were made between onion sequences downloaded from the Onion Gene Index (Release 2·0; http://compbio.dfci.harvard.edu/tgi/plant.html) and rice cDNA sequences from the Rice Genome Annotation project (Version 6·1; http://rice.plantbiology.msu.edu/index.shtml). The tblastx alignments were performed with an E-value cut-off of 0·01 (Altschul et al., 1997). Annotations, including descriptions and Gene Onotology (GO) assignments were then cross-referenced from rice sequences with significant homology to onion sequences, allowing GO analysis and more informative descriptions on the putative role of onion genes.

An Agilent One Color Quick Amp Labelling Kit (Agilent Technologies Inc.) was used to amplify and label target RNA with Cyanine 3-CTP to generate complementary RNA (cRNA) according to the manufacturer's instructions. Purification of the labelled cRNA was performed using RNeasy mini spin columns (Qiagen, Hilden, Germany) and quantified using a NanoDrop ND-1000 UV-VIS spectrometer. An Agilent One Color RNA Spike-In Kit (Agilent Technologies Inc.) was used as a positive control for monitoring sample amplification, labelling and microarray processing. The cRNA was fragmented and hybridized to the array using the Agilent Gene Expression Hybridisation Kit, and then washed with Gene Expression Wash Buffers 1 and 2, according to the manufacturer's instructions (Agilent Technologies Inc.). Microarrays were scanned on an Agilent G2565BA Microarray scanner equipped with Agilent Scan Control version A8·4·1 at a resolution of 5 µm, using the extended dynamic range option. Signal values for individual probes were extracted using Agilent Feature Extraction version 10·5·1·1 software (Agilent Technologies Inc.). All microarray data have been submitted to the online database Gene Expression Omnibus for public access and long-term storage (accession no. GSE28159).

Quantitative real-time PCR validation of microarray results

The transcript levels of three cDNAs (TC3845, CUST_7_P1403527117, F TTGGGGTTTAGTCAGGAGTTG, R ATTACAGGGATAGTTCAGCAGTTAG; TC4371, CUST_533_P1403527117, F ATGAACCACGACACCCAGAG, R CATTCCCACGCTTTCTCAAC; and BI095694, CUST_8043_P1404013528, F ATGCCCTGCTTTCTCTCAAG, R GATGGAGGAGGAGGATACGG) that were significantly and differentially expressed at different times in the microarray experiments were confirmed by quantitative real-time PCR. The ThermoScript™ RT-PCR System for First-Strand cDNA Synthesis kit (Invitrogen, catalogue no. 11146-024) was used to produce cDNA from total RNA samples (1 µg) using a combination of random hexamers and oligo(dT) primers (20:80 mix, respectively). Gene-specific primers were designed using Primer 3 and PrimerSelect (Lasergene) software. Transcript abundance was detected by an ABI Prism 7900HT sequence detection system (Applied Biosystems) controlled by SBS 2·1 software (Applied Biosystems) using a SensiMix SYBR Green qPCR MasterMix (Bioline, London, UK). The quantitative PCR was performed in 384-well plates using the ‘Standard Curve’ method (Wong and Medrano, 2005) for mRNA quantification with normalization to the endogenous control gene, tumour protein (TC4554 CUST_716_P1403527117; F TCCGACTACAGGAACAACCAG, R AAACTCCTCTGCCTTCTCAGC). The control gene was selected from six genes evaluated for stability within our samples using the geNorm software package (Vandesompele et al., 2002). Quantitative PCR conditions, efficiency calculations and data normalizations were as described previously (Hammond et al., 2006).

Statistical analysis

All statistical analyses were carried out using Genstat for Windows 12th Edition, Version 12·1·0·3338 (VSN International Ltd, Herts, UK) unless otherwise stated. Analysis of variance (ANOVA) was used to identify factors significantly affecting variance in the biochemical and physiological data. ANOVA was performed on the data specifying a nested treatment structure of a common baseline (observation before curing treatments was the baseline). Least significant difference values (LSD; P = 0·05) were calculated from each analysis, for comparison of appropriate treatment means, using critical values of t for two-tailed tests. Results are significant to P < 0·05 unless otherwise stated. Principal component analysis (PCA) was carried out on autoscaled data on the samples for which the maximum variables were recorded; these were the samples used for microarray analysis in 2007, all samples grown at Suffolk in 2008 and all samples in 2009. Microarray data analysis was performed using Genespring GX11 (Agilent Technologies). A fully balanced experimental design was used for the microarray experiment, with three biological replicates for each time (harvest, cured, before sprouting and sprouting), curing temperature (20 and 28 °C) and cultivar (‘Wellington’ and ‘Sherpa’) combination (n = 42). Raw expression data were subject to quantile normalization, and then baseline normalization was applied to individual probes, by dividing probe signal values by the median probe signals across all samples. Significantly differentially expressed transcripts were selected using a one-way ANOVA (GeneSpring GX) with a Benjamini–Hochberg (BH)-corrected P-value <0·05 and a fold change cut-off >2. Significantly differentially expressed transcripts were then grouped using the K-means clustering algorithm in GeneSpring GX. The cluster analysis was set up to look for a number of significant groups or clusters with similar expression patterns. This helps with the data interpretation as it reduces the number of expression patterns, and the top genes in each group can be selected.

RESULTS

Storage temperature, but not curing temperature, affects sprout development

The effect of curing temperature on sprout initiation and growth was tested over 3 years, with samples cured at either 28 °C (current commercial practice) or 20 °C for 6 weeks. There was no significant effect of curing temperature on either sprout growth or sprout initiation during storage (data not shown). The effect of storage temperature on sprout development was also recorded during storage (Table 1). Onset of sprouting occurred between approx. 13 and 29 weeks of storage at 1 °C (current commercial practice). There was no significant difference in sprout incidence between the cultivars grown, but sprout growth proceeded most rapidly in ‘Sherpa’ and most slowly in ‘Wellington’. Onset of sprouting occurred significantly earlier in onions stored at 3 °C (2008) or 6 °C (2008 and 2009), approx. 13 weeks after curing, and sprouts grew most quickly in onions stored at 6 °C. In 2009, a sub-sample of onion bulbs was transferred from 6 to 1 °C after 13 weeks storage until the end of the experiment. When analysed at the end of storage (25 weeks), there was a significant effect of storage temperature, with the greatest sprout growth at 6 °C, followed by transfer from 6 to 1 °C and the least sprout growth at 1 °C (sprouts reaching 59·6, 48·7 and 34 % of bulb height, respectively).

No root growth occurred throughout the 2007 experiment. In the last two experimental years, the least root development occurred in ‘Wellington’, although this was only significant in 2009 (41·7, 31·9 and 18·1 % of ‘Red Baron’, ‘Sherpa’ and ‘Wellington’ bulbs, respectively, showed root growth after 25 weeks storage). Onions stored at 1 or 6 °C had the lowest incidence of root growth compared with 3 °C or transfer from 6 to 1 °C. Curing temperature did have an effect on root growth in 2009, with root development significantly greater in onions cured at 28 °C (22·2 %) than at 20 °C (38·9 %).

Onion metabolome was significantly affected by post-harvest storage

Targeted metabolomic profiling was used to investigate the effects of onion bulb curing and storage on carbohydrates, flavour compounds and PGRs, and enabled comparisons with changes in sprout development during storage. Distinct sugar profiles were observed for the individual cultivars, which were consistent between the 3 years (Supplementary Data Tables S1, S2, S3). ‘Red Baron’ bulb composition was dominated by sucrose (fructose 126·8, glucose 137·5, sucrose 193·5 mg g⁻¹ d. wt), while ‘Sherpa’ contained more monosaccharides (fructose 138·0, glucose 160·90 mg g⁻¹ d. wt) than disaccharide (sucrose 160·1 mg g⁻¹ d. wt), and ‘Wellington’ was between the two (fructose 140·3, glucose 164·3, sucrose 192·5 mg g⁻¹ d. wt). Overall, fructose, sucrose and glucose concentrations were 1·4-fold higher in 2007 and 2008 than in 2009 (Supplementary Data Tables S1, S3, S4). In all years, fructose increased during storage until the onset of sprouting, after which it decreased. In 2009, fructose was higher in ‘Wellington’ and ‘Sherpa’ (108·8 and 103·3 mg g⁻¹ d. wt, respectively) than in ‘Red Baron’ (95·3 mg g⁻¹ d. wt). In 2007 and 2008, glucose changed little during storage, but in 2009 both glucose and sucrose concentrations decreased dramatically (3- to 4-fold) until the onset of sprouting and then increased thereafter (Supplementary Data Tables S1, S3, S4).

Onions contained fructans with DPs ranging from 3 to 8, with the concentration decreasing in proportion to the increase in size, and the following ranges recorded across all 3 years; kestose 20–160; nystose 5–130; DP5, 5–90; DP6, 2–55; DP7, 2–45; DP8, 2–30 mg g⁻¹ d. wt (Supplementary Data Tables S1, S3, S4). In 2007, fructans did not vary with cultivar, but in 2008 and 2009 ‘Red Baron’ contained the highest concentrations of fructans. In all years, fructans decreased during curing and cold storage, reaching a minimum level prior to the onset of sprouting (except kestose which increased during curing). In 2008, fructans were generally lower in onions stored at 1 °C, and the increase in the higher DPs after the onset of sprouting was greater in magnitude and occurred earlier in onions stored at 3 or 6 °C than at 1 °C, but there was no main effect of storage temperature in 2009.

Flavonols were measured in onion bulb tissue from the 2007 growing season. The major flavonols in onion flesh were quercetin 3,4′-diglucoside and quercetin 4′-glucoside (Supplementary Data Table S2). Isorhamnetin 3,4′-diglucoside, isorhamnetin 4′-glucoside, quercetin 3-glucoside and quercetin were present in very low concentrations. Quercetin 3,4′-diglucoside and quercetin 4′-glucoside were higher in ‘Sherpa’ (3·49 and 2·37 mg g⁻¹ d. wt, respectively) than in ‘Wellington’ (2·66 and 1·66 mg g⁻¹ d. wt). Quercetin 3,4′-diglucoside increased during curing at both temperatures, while quercetin 4′-glucoside increased during curing at 20 °C, but decreased at 28 °C.

Field site, but not storage, affects onion dry weight

Onion dry weight was significantly higher in bulbs from the Bedfordshire site (115·63 mg g⁻¹ f. wt) than the Norfolk site (110·27 mg g⁻¹ f. wt) in 2007, but in 2008 there was no difference between sites (Supplementary Data Tables S1, S3). In all years, dry weight was greater in the red onion ‘Red Baron’ (129·52 mg g⁻¹ d. wt) compared with the brown onions ‘Sherpa’ and ‘Wellington’ (110·47 and 110·14 mg g⁻¹ d. wt, respectively). Dry weight generally decreased during curing, although this was not significant, and subsequently remained stable during cold storage. There was no effect of curing temperature or storage temperature on dry weight. Disease incidence in all years was <1 %.

Pyruvate concentration was used as a marker for pungency as described previously by Abayomi et al. (2006). Pungency was measured in all samples in 2007; there was a significant effect of site, with onions from the Bedfordshire site (6·28 µmol pyruvate g⁻¹ f. wt) more pungent than those from Norfolk (2·64 µmol pyruvate g⁻¹ f. wt) which was unusually low; however, there was no effect of curing temperature, and therefore pungency was not measured in subsequent years.

PGRs signal sprout initiation in stored onion bulbs

In 2007, overall mean ABA concentration was greater in ‘Red Baron’ and ‘Sherpa’ (19·2 and 19·9 ng ABA g⁻¹ d. wt, respectively) than in ‘Wellington’ (15·5 ng g⁻¹ d. wt); however, there was no difference between cultivars in 2008 (Supplementary Data Tables S1, S5). Overall, ABA decreased during curing and reached a minimum prior to the onset of sprouting, after which it increased. In 2007, ABA was significantly greater in onions cured at 20 °C than at 24 or 28 °C; however, there was no difference between curing temperatures in 2008. In 2007, ‘Sherpa’ contained the most ZR (130·8 ng g⁻¹ d. wt) and ‘Wellington’ (97·8 ng g⁻¹ d. wt) the least. The same trend was observed in 2008, but was not significant. There was no effect of curing temperature on ZR concentration, but ZR decreased during curing, and then increased towards the end of storage. The concentration of IPA did not vary according to cultivar or curing temperature. There was no change in IPA during curing, but the concentration increased during cold storage, reaching a peak at around the time of onset of sprouting (Supplementary Data Tables S1, S4).

PCA reveals consistent changes in metabolites

Principal component analysis was used to identify key variables within the large physiological and metabolomic profiling set for each experimental year. The PCA for 2007 showed that the samples could be grouped according to time, but did not separate according to cultivar or curing treatment (Fig. 1). Samples before and after cold storage were separated on PC1 (47·04 % of variation), while harvest and sprouting samples were separated from cured and pre-sprouting samples by PC2 (14·32 % of variation). This trend was consistent for the 2008 (Fig. 2) data, and similar for the 2009 (Fig. 3) data, although the 2009 data set did not include any PGR data. These analyses show that samples taken at harvest were characterized by high fructan and ABA concentrations, and cold-stored samples contain more fructose, sucrose and IPA. Sprouting samples were characterized by higher ZR and a higher ratio of monosaccharides to disaccharide.

Fig. 1.

Principle component analysis biplot of A. cepa ‘Wellington’ and ‘Sherpa’ bulbs cured at 20 or 28 °C and stored at 1 °C; after harvest (0), after curing (1), and before (3) and after (4) sprouting in cold storage in 2007.

Fig. 2.

Principle component analysis biplot of A. cepa ‘Wellington’ and ‘Sherpa’ bulbs cured at 20 or 28 °C and stored at 1, 3 or 6 °C; after harvest (0), after curing (1), and before (3) and after (4) sprouting in cold storage in 2008.

Fig. 3.

Principle component analysis biplot of A. cepa ‘Red Baron’, ‘Wellington’ and ‘Sherpa’ bulbs cured at 20 or 28 °C and stored at 1, 3 or 6 → 1 °C; after harvest (0), after curing (1), and before (2) and after (3) sprouting in cold storage in 2009.

Effect of post-harvest handling on onion gene expression

Transcriptional profiles of onion bulbs during storage were determined to provide additional evidence for the metabolomic changes observed. Samples of RNA were prepared from onions of two cultivars at various physiological ages, i.e. freshly harvested, cured, pre-sprouting and sprouting. Three biological replicates were hybridized for each cultivar and time combination. To test the sensitivity and consistency of the microarray analysis, the profiling obtained for three probes using quantitative real-time PCR was compared with the profiling for the same probes using the microarray. There was a qualitative similarity in the data, with a correlation coefficient (R²) of 0·71 between the microarray and quantitative real-time PCR expression values. Curing temperature did not have an effect on the gene expression profile. In total, 3111 and 1406 probes were identified as being significantly (BH-corrected P < 0·05) and differentially expressed at one or more time points during curing and storage in onions ‘Wellington’ and ‘Sherpa’, respectively. Of these, 1050 were common between the two cultivars, and 1036 remained after filtering for a fold change cut-off >2·0. These entities were subjected to K-means clustering analysis, which revealed eight clusters of temporal gene expression profiles, four representing upregulated genes and four downregulated genes (Fig. 4).

Fig. 4.

Cluster analysis for the time course of gene expression. The altered gene expression was grouped by the K-means clustering algorithm. The expression pattern over time can be linked with specific genes detailed in Tables 3 and 4. The y-axis indicates gene expression fold change and the x-axis represents the sampling time (0, at harvest; 1, after curing; 3, dormancy break; 4, sprouting).

Every probe was compared with other known genes in GenBank non-redundant databases (Altshul et al., 1997), and categorized into functional categories based on their homology to known genes. Three of the eight clusters contained probes with GO terms that were significantly (BH-corrected P < 0·05) over-represented. In cluster 1, GO terms for photosynthesis, thylakoid, plastid and cellular processes were significantly over-represented, and, similarly, the GO term for thylakoid was significant in cluster 4, and RNA binding in cluster 7. These findings reflect the functional classification of the probes (Table 2). Overall, approx. 34 % of the probes were designated as unclassified, and approx. 11 % as housekeeping. The top 30 most down- and upregulated genes are listed in Tables 3 and 4. In cluster 1, many of the most highly downregulated probes were annotated as having high homology to alliin lyase, which may suggest that this is the same gene; however, these probes were all designed to ESTs that map to different tentative consensus sequences. In addition, a probe with homology to lachrymatory factor synthase, a gene involved in onion flavour metabolism, is also represented in this category. These probes account for the high representation of secondary metabolism in the functional categories (Table 2). In cluster 0, the largest functional category contained probes with homology to genes involved in stress and defence response. Many clusters contain probes with high homology to PGRs, including cluster 2 (ethylene-responsive transcription factor, cyokinin-O-glucosyl transferase 1 and gibberellin receptor GID1L2), cluster 3 [ethylene receptor, S-adenosyl methionine (SAM) synthase, auxin efflux carrier component], cluster 4 (auxin-responsive gene family member), cluster 5 (gibberellin 20 oxidase 2), cluster 6 (auxin-responsive gene family member, ethylene insensitive) and cluster 7 (auxin-induced protein). Probes with homologies to genes involved in cell wall organization were represented in clusters 2, 3 and 5, and probes with homologies to genes involved in the regulation of the cell cycle are most highly represented in the upregulated clusters 2, 4 and 5.

Table 2.

Functional grouping of genes differentially expressed during onion curing and storage as a percentage of the probes in each K-means cluster

	Cluster
Functional category	0	1	2	3	4	5	6	7	Total upregulated	Total downregulated
Housekeeping	9·4	5·4	15·1	9·2	11·8	17·4	9·8	5·5	12·6	8·8
Stress/defence	21·9	1·8	1·0	11·2	2·6	7·6	4·5	10·3	4·8	8·5
Chaperones	–	7·1	0·5	0·7	1·0	0·7	0·9	0·7	0·7	1·7
Photosynthesis	–	19·6	1·5	0·7	0·5	2·1	0·9	2·1	1·5	3·7
Cell wall	–	–	2·5	5·9	0·5	2·1	–	1·4	1·6	2·6
Secondary metabolism	12·5	37·5	3·5	3·9	1·5	2·8	5·4	6·8	3·5	10·5
Cell death	–	1·8	0·5	0·7	–	–	–	–	0·1	0·6
Peptidase/kinase	9·4	–	7·5	5·9	7·2	2·8	7·1	6·8	6·3	5·7
Transport/secretion	6·3	–	8·0	8·6	9·7	9·0	7·1	8·9	8·9	6·5
Signalling	–	–	3·0	0·7	5·1	2·8	3·6	4·1	3·8	1·4
Metabolism	6·3	10·7	10·6	11·2	6·2	4·9	8·9	11·6	8·3	9·9
PGR-related	–	–	2·0	2·0	0·5	0·7	1·8	0·7	1·0	1·4
Cell cycle	–	1·8	5·0	0·7	3·6	4·2	1·8	0·7	3·5	1·1
Transcription factor	–	1·8	2·5	2·6	9·2	2·8	9·8	15·1	7·2	4·5
Phosphatase	–	–	1·0	–	1·0	–	0·9	1·4	0·9	0·3
Unclassified	34·4	12·5	35·7	36·2	39·5	40·3	37·5	24·0	35·2	32·7

Table 3.

The 30 most downregulated genes post-harvest in A. cepa ‘Sherpa’ and ‘Wellington’ sampled at harvest, after curing at 20 or 28 °C for 6 weeks, and after 15 (non-sprouting) and 25 (sprouting) weeks storage at 1 °C

Probe	Tentative annotation	Fold change*	Onion sequence ID	Cluster
CUST_97_PI404013531	Chlorophyll A/B-binding protein	322·9	X95687·1	1
CUST_10495_PI403527117	Alliin lyase precursor	266·5	BQ580283	1
CUST_2252_PI403527117	Starch synthase	256·8	TC6090	1
CUST_2105_PI403527117	Alliin lyase precursor	218·2	TC5943	1
CUST_9364_PI403527117	Alliin lyase precursor	197·6	BQ580052	1
CUST_312_PI403527117	Chlorophyll A/B-binding protein	181·7	TC4150	1
CUST_3578_PI403527117	BURP domain-containing protein	173·2	TC7416	1
CUST_10965_PI403527117	Starch synthase	169·5	BE205631	1
CUST_8035_PI404013528	Lipoxygenase	144·0	BI095683	1
CUST_11714_PI403527117	Alliin lyase precursor	138·6	BQ579865	1
CUST_6269_PI404013528	Alliin lyase precursor	137·7	BE205573	1
CUST_2477_PI403527117	Chlorophyll A/B-binding protein	128·2	TC6315	1
CUST_1366_PI403527117	Chlorophyll A/B-binding protein	118·8	TC5204	1
CUST_2054_PI403527117	3-Hydroxy-3-methylglutaryl-coenzyme A reductase	116·2	TC5892	1
CUST_5317_PI403527117	Alliin lyase precursor	114·5	BQ579913	1
CUST_7489_PI403527117	Glutamate-cysteine ligase, chloroplast precursor	111·9	CF450735	1
CUST_6781_PI403527117	POEI3 – pollen Ole e I allergen and extensin family protein precursor	102·3	CF435504	1
CUST_2601_PI403527117	POEI1 – pollen Ole e I allergen and extensin family protein precursor	93·2	TC6439	1
CUST_5265_PI404013528	Adenine phosphoribosyltransferase	92·4	BQ580038	1
CUST_8043_PI404013528	Pherophorin-C2 protein precursor	90·3	BI095694	1
CUST_915_PI403527117	Annexin	85·8	TC4753	1
CUST_379_PI403527117	Chlorophyll A/B-binding protein	85·0	TC4217	1
CUST_14_PI404013531	Alliinase	70·2	Z12620·1	1
CUST_1269_PI403527117	MYB family transcription factor	69·3	TC5107	1
CUST_5044_PI403527117	Alliin lyase precursor	69·3	BQ580118	1
CUST_8097_PI403527117	3-Hydroxy-3-methylglutaryl-coenzyme A reductase	65·4	CF449184	1
CUST_7_PI404013531	Alliinase	64·8	Z12621·1	1
CUST_42_PI404013531	Lachrymatory factor synthase	64·5	AB089203·1	1
CUST_342_PI403527117	Chlorophyll A/B-binding protein	62·5	TC4180	1
CUST_102_PI404013531	Alliinase	61·9	M98267·1	1

The cluster numbers relate to Fig. 4.

* Fold change compared with expression at harvest, calculated as 2x, where x = absolute value of (normalized condition 1 – normalized condition 2).

Table 4.

The 30 most upregulated genes post-harvest in A. cepa ‘Sherpa’ and ‘Wellington’ sampled at harvest, after curing at 20 or 28 °C for 6 weeks, and after 15 (non-sprouting) and 25 (sprouting) weeks storage at 1 °C

Probe	Tentative annotation	Fold change*	Onion sequence ID	Cluster
CUST_7598_PI403527117	No significant match	8611·4	CF449784	0
CUST_1748_PI404013528	OsRCI2-6 – hydrophobic protein LTI6B	434·9	AA508914	0
CUST_3047_PI403527117	LTPL129 – protease inhibitor/seed storage/LTP family protein precursor	265·6	TC6885	0
CUST_2917_PI404013528	Alanine precursor	167·1	CF451215	0
CUST_1734_PI404013528	Glycine/proline-rich protein	139·1	AA451576	0
CUST_2079_PI403527117	Glycine-rich cell wall structural protein precursor	118·4	TC5917	0
CUST_1639_PI403527117	Caffeoyl-CoA O-methyltransferase	112·4	TC5477	0
CUST_5851_PI403527117	OsPOP4 – putative prolyl oligopeptidase homologue	106·4	CF438948	0
CUST_11680_PI403527117	Integral membrane protein	105·7	CF443686	0
CUST_11157_PI403527117	Caffeoyl-CoA O-methyltransferase	104·6	CF443345	0
CUST_8031_PI404013528	Glutathione S-transferase	100·2	BI095676	0
CUST_7555_PI403527117	No significant match	96·5	CF450666	0
CUST_12036_PI403527117	Hypoxia-responsive family protein	88·0	CF441799	0
CUST_565_PI403527117	OsAPx2 – cytosolic ascorbate peroxidase-encoding gene 4,5,6,8	87·5	TC4403	0
CUST_3316_PI403527117	Glutathione S-transferase	82·1	TC7154	0
CUST_579_PI403527117	Stem-specific protein TSJT1	81·3	TC4417	0
CUST_10249_PI403527117	NADH-ubiquinone oxidoreductase chain 3	66·5	CF442073	0
CUST_11370_PI403527117	DUF538 domain-containing protein	57·5	CF443772	0
CUST_9938_PI403527117	Endoribonuclease	57·2	CF445119	0
CUST_1014_PI403527117	Peroxidase precursor	50·4	TC4852	5
CUST_10976_PI403527117	Cysteine proteinase inhibitor 8 precursor	50·0	BE205657	0
CUST_10222_PI403527117	Pherophorin-C2 protein precursor	44·4	CF446675	2
CUST_995_PI403527117	Phosphate-induced protein 1 conserved region domain-containing protein	44·3	TC4833	5
CUST_5445_PI403527117	HD domain-containing protein 2	43·5	CF435292	5
CUST_4440_PI403527117	Multicopper oxidase domain-containing protein	41·5	CF437212	0
CUST_2280_PI404013528	Gibberellin 20 oxidase 2	41·2	CF449051	5
CUST_7158_PI403527117	Kinase, pfkB family	39·6	CF437420	0
CUST_12100_PI403527117	Glycosyl hydrolase	36·6	CF446683	0
CUST_10751_PI403527117	Ubiquitin-conjugating enzyme	36·4	CF444379	0
CUST_917_PI403527117	Ribonucleoside-diphosphate reductase small chain	35·9	TC4755	5

The cluster numbers relate to Fig. 4.

* Fold change compared with expression at harvest, calculated as 2x, where x = absolute value of (normalized condition 1 – normalized condition 2).

DISCUSSION

Biochemical and physiological analyses of onions of different cultivars (‘Wellington’, ‘Sherpa’ and ‘Red Baron’, not treated with maleic hydrazide) grown at different sites over 3 years, cured at different temperatures (20, 24 and 28 °C) and stored under different regimes (1, 3, 6 and 6 → 1 °C) were carried out in order to study the mechanisms regulating post-harvest onion sprout development. The underlying targeted metabolomic and transcriptomic changes that occurred during curing and storage were determined. The first onion oligonucleotide microarray was constructed and used to determine differential gene expression in selected samples during onion curing and storage.

Biochemical and physiological analyses were supported by transcriptional changes; however, direct alignments between transcripts and metabolites were not possible due to the lack of comprehensive sequence data for onion. There were greater transcriptional differences between samples at harvest and before sprouting than between the samples taken before and after sprouting, with some significant changes occurring during the relatively short curing period. In addition, and in agreement with the biochemical and physiological analysis, no significant transcriptional differences could be found between the curing temperatures. This suggests that there is no detrimental effect of reducing the curing temperature from 28 to 20 °C. The similar response of ‘Sherpa’ and ‘Wellington’ with storage time was also confirmed by the sprouting behaviour, although these varieties are reputed to have different commercial storage properties (Nigel Kingston, Syngenta, pers. comm.). It is suggested that the reason for this may be that in this experiment, small changes in growth of the internal sprout leaves were monitored to give a true indication of the physiological state. If it is assumed that the onion bulb is in a dormant state at harvest, then the changes in gene expression that occur after this time are likely to represent the transition from endo-dormancy to sprout suppression, and suggest that endo-dormancy may be relatively short – ending just after curing.

The bulb mono- to disaccharide ratio is a potential biochemical marker for sprouting

The changes in dry weight, pungency and concentrations of fructans, simple sugars and flavonols are consistent with those previously reported (Chope et al., 2006, 2007a; Downes et al., 2010) and consistent between storage years despite the differences in the degree of fall-down at harvest. Changes in fructose, glucose and sucrose in stored onion bulbs are well reported and have been reviewed by Chope et al. (2007a). During storage in all years, the concentration of fructans decreased as they were enzymatically hydrolysed to form increasing amounts of fructose, as has previously been shown by others (Suzuki and Cutliffe, 1989; Salama et al., 1990; Pak et al., 1995; Ernst et al., 1998). It has been hypothesized that carbohydrate content is correlated with storage life. Suzuki and Cutliffe (1989) found a significant positive correlation between fructan content and percentage marketable bulbs in eight onion cultivars stored at 6–10 °C for 4 months. Higher fructose content at harvest was correlated with extended storage life in ‘Robusta’ onion bulbs stored at 4 °C for 3 months (Rutherford and Whittle, 1982). In this experiment, there was no significant effect of storage temperature on changes in sugars. Similarly, Benkeblia et al. (2002) found that the pattern of changes in total soluble sugar content of ‘Rouge Amposta’ onion bulbs was similar at 4, 10 and 20 °C, suggesting that the catabolism of carbohydrates is more dependent on physiological stage than temperature. In contrast, fructose concentration was higher in ‘Sentinel’ onion bulbs stored at 0 and 15 °C than in those stored at 30 °C, suggesting that hydrolysis of fructans increased at low temperatures (Salama et al., 1990). It is likely that the range of storage temperatures tested in the present study (1–6 °C) was too narrow to show any difference in sugar metabolism. It has previously been suggested that a peak in glucose (Benkeblia and Selselet-Attou, 1999) or sucrose (Benkeblia et al., 2005) precedes sprouting. However, these phenomena seem to be cultivar specific, and a universal biochemical marker of sprouting is yet to be identified.

Using PCA, the ratio of monosaccharides (fructose and glucose) to disaccharide (sucrose), along with the concentration of ZR have been identified as important factors in discriminating between sprouting and pre-sprouting samples. The mono- to disaccharide ratio has the advantage that it is easier and more cost-effective than ZR to measure. With the screening of a detailed time course during storage of a number of samples from a range of different cultivars it may be possible that this relatively simple parameter could give important information on the dormancy/sprouting status of a stored onion bulb.

Association of gene expression profiling with dormancy and sprout suppression

Onions bulbs have evolved as a storage organ to allow the plant to over-winter. During the transition from dormancy (endo-dormancy) to sprout suppression (eco-dormancy) and subsequent growth, the bulb undergoes the transition from sink organ to source, to sustain cell division in the meristematic tissue. The mechanisms controlling these processes are yet to be elucidated.

Probes with tentative annotations relating to defence/stress formed the second largest functional category (following unclassified) in the highly upregulated cluster 0. Similarly, a high proportion of defence-related transcript-derived fragments were differentially expressed during the potato tuber life cyle (Trindade et al., 2004) and during endo-dormancy release in raspberry buds (Mazzitelli et al., 2007). In the highly downregulated cluster 1, approx. 20 % of the probes were placed in the functional category related to photosynthesis. Bulb formation is well known to be influenced by photoperiod and, as light is perceived by the leaves and the signals transmitted to other plant parts, then the downregulation of these light-responsive elements could follow the dieback of onions and subsequent removal of the leaves at harvest known as topping.

It would be expected that expression of genes related to cell cycle regulation would increase during storage, as the meristematic cells begin to divide and elongate to form new sprout tissue. Indeed, upregulated clusters 2, 4 and 5 contained ten, seven and six probes, respectively, with tentative annotation for genes including histones and tubulins. It is also likely that the transition from dormancy to sprout suppression will also involve changes in the cell walls, as modifications are made to control the transport of reserve compounds and cell division resumes at the meristem. In general, cellulose synthase decreased post-harvest (six probes with tentative annotation for cellulose synthesis downregulated), and, indeed, cellulose concentration has been shown to decrease in ‘MBL87-WOPL’ onions stored at 6·6 °C for 12 weeks (Coolong et al., 2008). Polygalacturonase and pectinesterase have been associated with a decrease in onion firmness caused by degradation of the middle lamella (Coolong et al., 2008). Our transcriptional profiling identified probes with homology to genes involved in cell wall modification, including polygalacturonase, pectinesterase, β-galactosidase and pectate lyase, as being upregulated after harvest (eight probes in clusters 2, 4 and 5). Taken together, this suggests that the cell walls are being degraded and the bulb is becoming softer; however, an increase in expression of the transcript does not necessarily translate to an increase in enzyme activity.

The fact that many temporal changes in expression were detected in the probes on the onion microarray suggests that the possibility exists for generating a diagnostic microarray chip that could predict sprouting. In order to create this, a more detailed time course between curing and the onset of sprouting would be necessary, together with more sequence data for onions. Although many probes with different expression patterns were identified, the relatively small coverage of the onion genome on the microarray used in our experiments means that genes with still greater differences in expression may be yet to be discovered, and these in turn could provide breeders with targets to manipulate storage life. The samples in this study consisted of longitudinal wedges of bulb tissue, which gives information about the real concentrations of each substance in the bulb at that time, but does not provide information about transport within the bulb. Therefore, detailed spatial analysis will be necessary in the future, as certain parameters such as glucose and pyruvate have been shown to be differentially distributed in the bulb during storage (Abayomi and Terry, 2009).

PGRs mediated control of dormancy and sprouting

During post-harvest storage, a gradual change in the relative composition of PGRs occurs as the levels of growth inhibitors drop and the levels of growth promoters rise (Thomas, 1969; Thomas and Isenberg, 1972). The amount of ABA in the plant is a balance between synthesis and degradation. Plant development, environmental conditions such as drought stress, and other growth regulators affect these processes (Nambara and Marion-Poll, 2005). An effect of curing temperature on ABA concentration was only observed in bulb samples produced in 2007, suggesting that there may be an interaction with growing season. However, ABA concentration decreased during curing in both 2007 and 2008, and then increased after the onset of sprout growth, as was previously observed (Chope et al., 2006, 2007a, b). Chope et al. (2006, 2007a, b) found that a decline in endogenous ABA was correlated with storage life under both controlled atmosphere and regular atmosphere conditions. Yamazaki et al. (1999a, b) also demonstrated a functional role for ABA in maintaining bulb dormancy in A. wakegi. Although ABA can be considered as a growth promoter or inhibitor depending on circumstances (Sharp, 2002), it generally inhibits growth prior to plant establishment, such as in young shoots or seedlings (Lopez-Molina et al., 2001).

No probes with significant changes in expression pattern were annotated as being directly related to ABA; however, there were changes in aquaporins which facilitate movement of water and other uncharged small molecules across membranes, and are often regulated by ABA (Chaumont et al., 2005; Parent et al., 2009); probes with putative annotations as aquaporins were identified in cluster 7 (highly downregulated, n = 5) and cluster 4 (slightly upregulated, n = 2). Similarly, Mazzitelli et al. (2007) observed a downregulation of ESTs with similarity to an aquaporin gene in raspberry buds during the transition from endo-dormancy to para-dormancy. Here, two probes were also upregulated; however, aquaporins are part of a large and divergent gene family, which may be differentially regulated. Changes in aquaporin expression may be related to the need to control water movement between storage tissues and rapidly growing and expanding tissues, such as the sprout.

In this study, the cytokinins ZR, which decreased during curing, and IPA, which did not, were found to increase during storage and peaked with the onset of sprouting. Cytokinins generally stimulate cell division, and this increase in growth-promoting substances, which begins immediately after curing, suggests that there is a threshold above which sprouting occurs, or, more likely that the onset of sprouting is controlled by a delicate balance between the concentrations of growth promoters (such as cytokinins) and growth inhibitors (Galuszka et al., 2008). A putative cyokinin-O-glucosyl transferase 1 was upregulated in cluster 2. The glucosylated form of zeatin is thought to be important in transport, storage and protection against cytokinin oxidases, as O-glycosylation is reversible. This enzyme is likely to regulate the active and storage forms of zeatin. Therefore, an increase in expression of this probe could represent an increase in transport of cytokinins, particularly as the increase occurs mostly between samples after curing and before sprouting. Similarly, the greatest changes in ZR and IPA occurred between these times, strongly suggesting that an increase in cytokinins is necessary to stimulate cell division at the meristem.

Ethylene is a PGR that is clearly fundamental to the post-harvest physiology of many fresh produce types; however the literature on the role of ethylene in onion bulb dormancy and storage life is far from comprehensive (Chope and Terry, 2008). There are conflicting reports on how ethylene affects onion storage life. The observation that ‘Elba Globe’ onion bulbs produced ethylene in much greater amounts (actual amounts not specified) at the end of dormancy than at the beginning (Abdel-Rahman and Isenberg, 1974) suggests that ethylene may have a role in sprouting. In contrast, Benkeblia and Selselet-Attou (1999) found little variation in ethylene production (range of 4·4–4·6 nmol kg⁻¹ h⁻¹) of ‘Rouge Amposta’ onion bulbs during 6 months storage at 18 °C and 70 % relative humidity. The dichotomy between these findings implies that production of ethylene by onion bulbs could be cultivar dependent. An ethylene receptor, SAM synthase (cluster 3), and ethylene-insensitive 3 (a DNA-binding protein which regulates transcription in response to ethylene and is essential for ethylene-mediated responses) (cluster 6) were downregulated. This suggests that onions become less sensitive to ethylene, and also produce less ethylene with time in storage. However, one probe with homology to an ethylene-responsive transcription factor was upregulated (cluster 2). Onions are regarded as non-climacteric in their response to ethylene and, as such, have consistently low endogenous ethylene levels and lack the dramatic change in ethylene production found in climacteric fruit, which triggers ripening (Downes et al., 2010). There is a paucity of research on endogenous ethylene production of stored onions; however, the continuous application of ethylene gas has been shown to increase the storage life of onions (Bufler, 2009). More recently, a single 24 h treatment with ethylene before curing was shown to suppress sprouting in ‘Sherpa’ onions (Downes et al., 2010). Taken together, the efficacy of the single treatment and the potential reduction in sensitivity to ethylene throughout storage suggests that continuous ethylene treatment may not be necessary (Downes et al., 2010; Cools et al., 2011).

The concentrations of GAs and auxins were not measured; however, in cluster 2, a gibberellin receptor GID1L2 was upregulated. Gibberellins stimulate cell elongation, so an upregulation in its receptors could be explained by meristematic tissue preparing for growth. One of the most significantly upregulated probes in cluster 5 is annotated as a gibberellin 20 oxidase, which is involved in the production of active GAs. Taken together, this suggests a role for GAs in dormancy release. An auxin efflux carrier component (cluster 3) was downregulated along with an auxin-responsive gene family member (cluster 6) and an auxin-induced protein (cluster 7). However, an auxin-responsive gene family member was upregulated (cluster 4). This suggests that auxins may be more important in elongation of the sprout after initial growth has begun.

In conclusion, this study has provided a detailed analysis of the physiological, biochemical and transcriptional changes which occur in onion bulbs throughout storage using a newly developed onion microarray. These data suggest that, with the aim of reducing energy consumption, reducing curing temperatures from 28 to 20 °C is unlikely to result in any deleterious effects on onion bulb storage life and quality as long as the pathogen load is low. Lastly, results herein provide evidence for potential biomarkers (monosaccharide to disaccharide ratio and ZR) and the possibility of a diagnostic microarray chip which could be used to predict onion bulb sprouting.

SUPPLEMENTARY DATA

Supplementary data are available online at www.aob.oxfordjournals.org and consist of the following. Table S1: dry weight, pyruvate, non-structural carbohydrates and plant growth regulators in cultivars ‘Red Baron’, ‘Sherpa’ and ‘Wellington’ grown in 2007 at two sites and sampled at harvest, after curing at 20, 24 or 28 °C, and at intervals during cold storage at 1 °C. Table S2: fructans and flavonols in cultivars ‘Sherpa’ and ‘Wellington’ grown in 2007 at the Suffolk site and sampled at harvest, after curing at 20 or 28 °C, and at intervals during cold storage at 1 °C. Table S3: dry weight and non-structural carbohydrates in cultivars ‘Red Baron’, ‘Sherpa’ and ‘Wellington’ grown in 2008 at two sites and sampled at harvest, after curing at 20 or 28 °C, and at intervals during cold storage at 1, 3 or 6 °C. Table S4: dry weight and non-structural carbohydrates in cultivars ‘Red Baron’, ‘Sherpa’ and ‘Wellington’ grown in 2009 at the Bedfordshire site and sampled at harvest, after curing at 20 or 28 °C, and at intervals during cold storage at 1, 6 or 6 → 1 °C. Table S5: plant growth regulators in cultivars ‘Red Baron’, ‘Sherpa’ and ‘Wellington’ grown in 2008 at the Suffolk site and sampled at harvest, after curing at 20 or 28 °C, and at intervals during cold storage at 1, 3 or 6 °C. Table S6: multiple reaction monitoring transitions and conditions used for mass spectrometry.