Carotenoid retention during post-harvest storage of Capsicum annuum: the role of the fruit surface structure

Abstract

In this study, a chilli pepper (Capsicum annuum) panel for post-harvest carotenoid retention was studied to elucidate underlying mechanisms associated with this commercial trait of interest. Following drying and storage, some lines within the panel had an increase in carotenoids approaching 50% compared with the initial content at the fresh fruit stage. Other lines displayed a 25% loss of carotenoids. The quantitative determination of carotenoid pigments with concurrent cellular analysis indicated that in most cases, pepper fruit with thicker (up to 4-fold) lipid exocarp layers and smooth surfaces exhibit improved carotenoid retention properties. Total cutin monomer content increased in medium/high carotenoid retention fruits and subepidermal cutin deposits were responsible for the difference in exocarp thickness. Cutin biosynthesis and cuticle precursor transport genes were differentially expressed between medium/high and low carotenoid retention genotypes, and this supports the hypothesis that the fruit cuticle can contribute to carotenoid retention. Enzymatic degradation of the cuticle and cell wall suggests that in Capsicum the carotenoids (capsanthin and its esters) are embedded in the lipidic exocarp layer. This was not the case in tomato. Collectively, the data suggest that the fruit cuticle could provide an exploitable resource for the enhancement of fruit quality.

Carotenoid pigments in chilli pepper confer post-harvest colour and nutritional quality. Analysis of diverse commercial genotypes indicates the involvement of the fruit surface in carotenoid retention.

Introduction

In chilli peppers (Capsicum annuum), carotenoid pigments are responsible for the different yellow, orange, and red colouration found in ripe fruit. In addition to carotenoids conferring aesthetic and nutritional properties to the fruit, the post-harvest retention of colour is also an important fruit quality trait.

The pepper crop is in demand from consumers all year round. However, due to environmental conditions in pepper-growing regions, the crop cannot be grown and harvested throughout the year. Therefore, storage of dry peppers following harvest is essential. Pepper fruits must retain quality, including colour, during post-harvest storage. Post-harvest storage losses are observed not only in fruit crops, but also in other important crops, including rice, wheat, and maize (Kumar and Kalita, 2017), and this affects colour and nutritional quality. Furthermore, whilst increased provitamin A capacity has been engineered in rice, resulting in a carotenoid-enriched variety known as ‘Golden rice’ (Beyer et al., 2002), maintaining such increased nutritional properties has proved challenging, as carotenoid degradation occurs upon post-harvest storage (Gayen et al., 2015).

Capsanthin and capsorubin are the two major carotenoids found in red pepper fruits, which are almost unique to the ripe fruit of peppers. Capsanthin and capsorubin are synthesized from antheraxanthin and violaxanthin, respectively, by the action of CAPSANTHIN/CAPSORUBIN SYNTHASE (CCS) (Bouvier et al., 1994). Pepper carotenoids are commonly found to be esterified (Biacs et al., 1989; Minguez-Mosquera and Hornero-Mendez, 1994b), making them more stable than free carotenoids (Biacs et al., 1989). Carotenoids not only contribute to the colour properties to pepper fruits, but further have high antioxidant capacity and are therefore beneficial to human health. This antioxidant capacity is a result of their ability to scavenge reactive oxygen species (ROS), due to their conjugated double bond structure. Upon interaction of carotenoids with singlet oxygen (¹O₂), physical quenching occurs, in which energy is transferred between the two molecules. The excess energy of ¹O₂ is transferred to the carotenoid and yields ground state oxygen and a triplet excited carotene (Stahl and Sies, 2003). Oxidation of carotenoids results in the formation of cleavage products, collectively referred to as apocarotenoids. Excessive carotenoid degradation results in colour loss. In a recent study using an identical chilli pepper panel, it was found that increased unsaturated fatty acids along with altered lipid and carotenoid-derived volatiles are associated with genotypes displaying altered colour/carotenoid retention properties (Berry et al., 2021). These findings have led to the hypothesis that post-harvest loss of colour/carotenoids in chilli fruit derives from lipid peroxidation and the engagement of carotenoids to scavenge and dissipate ROS, while being degraded in the process. The question arising now is: how are these ROS initiated?

Waxy cuticles are an essential part of a plant’s physiology and play a fundamental role in the plant’s interaction with the environment. (Domínguez et al., 2011). The cuticle has been further suggested to have a role in the post-harvest quality of fruit (Lara et al., 2014, 2019). Plant cuticles are comprised of two components: cutin monomers and waxy, or lipophilic, components, which are embedded within the cutin matrix, as reviewed in Yeats and Rose (2013). The cuticle provides a barrier to control the entrance and exit of gases in fruits, which lack stomata. Intact pepper fruit cuticle has been shown to be permeable to a small amount of carbon dioxide and oxygen, though this permeability increases significantly upon wounding of the cuticle (Banks and Nicholson, 2000). Whilst gaseous exchange is essential, negative effects may be observed as a result. The role of antioxidants is evident in protecting against the harmful effects of ROS, which may be formed during the process of gaseous exchange.

There is some dispute over whether this ‘cuticular’ layer, which penetrates several cell layers deep, can be termed the ‘cuticle’. For this reason, the ‘cuticular layer’ spanning several cell layers has been termed the ‘wax exocarp’, which is defined as the outermost lipophilic layer of the pericarp of an angiosperm fruit, external to the mesocarp (Martin and Rose, 2014). Pepper fruit cuticular components, and associated cuticle morphology quantitative trait loci have been previously identified (Popovsky-Sarid et al., 2017).

In the present study, the association of post-harvest carotenoid retention traits in chilli with altered fruit surface structure has been investigated.

Materials and methods

Materials

All chemicals were purchased from Sigma-Aldrich, UK, unless stated otherwise. A commercial panel of 12 chilli pepper genotypes displaying diversity in colour intensity and retention phenotypes was provided by Syngenta. Capsicum annuum L. cv. CM334 (Criollo de Morelos 334) was also included in the diversity panel. Material was grown in glasshouses (25°C, 16/8 h light/dark).

Storage conditions

Pepper fruits were harvested when ripe and dried in an oven (30–40 °C) under a 12/12 h light cycle. Pepper fruits were dried for 2 weeks before being stored in hessian bags (4 °C) for a period of between 2 and 12 weeks. These conditions were selected in order to replicate the industrial drying and storage process that occurs during commercial growing and storage of chilli peppers.

Carotenoid analysis

Carotenoids were extracted from lyophilized and homogenized chilli pepper powder. A 10 mg aliquot of tissue was used in the extraction process, employing chloroform (500 µl) and HPLC-grade methanol (250 µl). The suspension was incubated on ice, in the dark, before HPLC-grade water (250 µl) was added. A phase separation was created, and the organic layer was collected. Chloroform (500 µl) was again added to the material for extraction, a phase separation was created, and the organic layer was pooled with the initial organic phase. Separation and detection of carotenoids was performed by HPLC with photodiode array (PDA) detection, using a C30 reverse-phase column (250 × 4.6 mm), purchased from YMC, Wilmington, NC, USA. The solvent system used has been detailed previously (Fraser et al., 2000). Carotenoids isolated from pepper fruit exocarp discs (1 cm) were extracted by washing discs in chloroform and methanol (1:1 ratio; 10 ml) in dark conditions, on a rotator. Wash solution was replaced every 24 h, and all solvent used for extraction were pooled and evaporated. Exocarp carotenoids were analysed by HPLC-PDA, as described.

Cuticle component analysis

Cutin monomer analysis

Discs (1 cm) of fruit pericarp tissue were dissected, and cuticle tissue was isolated using pectinase (1.5% w/v) and cellulase (0.1% w/v) in citrate buffer (0.2 mM, pH 3.7), with sodium azide (1 mM). Samples were incubated with shaking in dark conditions (35 °C; 4 d, 100 rpm). Cutin monomer extraction was performed as previously described (Cohen et al., 2019). Analysis of extracted cutin monomers was performed using a GC-MS system (Agilent 7683 autosampler, 7890A gas chromatograph, and 5975C mass spectrometer).

Cuticle wax analysis

Fruit discs of 1 cm were dissected and dipped in chloroform (10 ml; 10 s), with the cuticle side facing downwards, into solvent to extract cuticle-bound waxes. Deuterated triacontane (C₃₀, 10 µg) was added as an internal standard. Ten discs per biological sample were dipped. Following cuticle wax extraction, the extract was transferred to a cleaned glass vial and dried under nitrogen. Extracts were resuspended in fresh chloroform (500 µl) and transferred to a glass GC-MS vial. Extracts were dried under nitrogen, and derivatized for analysis using pyridine (30 µl) and N-methyl-N-(trimethylsilyl) trifluoroacetamide (MSTFA; 70 µl). Samples were incubated (40 °C; 1 h) before analysis. Analysis of cuticle waxes was performed using a GC-MS system (Agilent 7820A gas chromatograph and 5977B mass spectrometer).

For both cutin monomer and cuticle wax analysis, a DB-1HT capillary column (Agilent, J+W 122-1131; 30 m×250 µm×0.1 µm) was used, with a flow rate of 1.2 ml min^–1. Samples were injected in splitless mode at 200 °C, with helium as the carrier gas. The oven temperature was held at 70 °C for 2 min, before ramping at 10°C min^–1 to 150 °C, and then ramping at 3 °C min^–1 to 310 °C. The final temperature was held for a further 20 min. MS was performed in full scan mode. Identification of chromatogram components was performed by comparing mass spectra with literature-reported spectra, in-house libraries, and the NIST 2.0 MS library. Quantification was performed by calculating peak areas relative to the internal standard.

Fruit microscopy

Light microscopy

Fresh fruit were harvested, and hand sectioning was used to dissect cross-sections from the middle of the fruit. Sections were collected in distilled water. Nile Red (0.5 mg ml^–1) in acetone was used to stain lipids, and Fast Green (0.5% w/v) in ethanol (90%) was used to stain subepidermal cell layers. Sections were washed in distilled water, before mounting on glass slides. A Leica DM 500 compound light microscope was used for visualization, with ×10 and ×20 objective lenses. IMS (Imagic, Imaging Ltd) software was used to capture images.

Exocarp thickness was measured using ImageJ software employing light microscopy images. Exocarp was defined as the area stained pink following Nile Red staining. Three images per genotype were measured, and six measurements per image were recorded.

Cryo-scanning electron microscopy

Cross-sections of fruit were sectioned, frozen in liquid nitrogen, and then sublimed (2 min) and coated with platinum. Cryo-SEM was performed using a Quorum Technologies PP3010T cryo-SEM system.

Scanning electron microscopy

Fruit samples were mounted on aluminium pin stubs (SEM sample supports), using adhesive carbon tabs. SEM was carried out using a Zeiss Evo LS15 scanning electron microscope. Sample imaging was carried out using variable pressure mode, and images were captured via a C2D detector.

Transcriptomic analysis

Three independent fruits at each of the designated developmental stages from one plant were used, with each plant representing a biological replicate. The transcriptomic analysis focused on line R8 because it showed medium carotenoid retention and high colour retention, compared with other lines such as R5 and R6 that had high carotenoid retention but medium colour retention. In addition, R8 was one of the most amenable lines to the extraction of RNA with good integrity profiles. Abrasive paper was used to isolate fruit epidermal cells, and total RNA was extracted and purified using TRIzol (Invitrogen, UK) and the PureLink RNA Minikit (Thermo Fisher Scientific, UK) according to the manufacturer’s instructions. RNA was treated with TURBO DNase (ThermoFisher Scientific) to remove contaminating genomic DNA. Following quality control, an mRNA library was prepared using NEBNext Ultra II RNA Library Prep with sample purification beads, and in-house (Syngenta Ltd, Research Triangle Park, NC, USA) 8 bp indexes. Libraries were sequenced using a Hi-Seq system (Illumina). Gene count was normalized using the R package EDASeq v2.16.3 (Risso, 2013), and filtered. Reads were mapped to the published CM334 chilli pepper genome (Kim et al., 2014). The unprocessed transcriptomic data have been deposited with the NCBI under BioProject PRJNA640935.

Hydrogen peroxide fruit dip

Fruits were harvested and dried as previously described. Fruits were treated with varying concentrations of hydrogen peroxide (H₂O₂), either before or after fruit drying, and then stored (4 °C) for 1 month. Carotenoid content was measured using HPLC-PDA analysis.

Data processing and statistical treatment

All experiments typically used 3–6 biological replicates, unless stated otherwise. This refers to one plant from which three fruit are used/harvested and analysed as an independent biological replicate. Graphs were compiled using GraphPad Prism 8 software, and means, SD, and SE were calculated using Excel (Microsoft). Significance testing, including Student’s t-tests and ANOVA, were carried out using XLstat software (Addinsoft). Student’s t-test significances were represented as: *P<0.05; **P<0.01.

Results

Carotenoid retention classification of the commercial study panel

Previous studies (Berry et al., 2021) have classified the colour retention properties of the commercial panel used, on visual colour intensity, image analysis, and end-point carotenoid content, following post-harvest storage (Table 1). To increase the robustness of this classification, the present study measured carotenoid contents of the pepper fruit before oven drying and throughout post-harvest storage (Table 1). Over this period, the 13 genotypes displayed varying levels of carotenoids. The genotype R3 contained only 2.5 mg g^–1 DW of carotenoids at the fresh fruit stage, whereas genotype R7 contained >18 mg g^–1 DW of carotenoids at the same fruit stage (Supplementary Table S1). Furthermore, the change in carotenoid content during the 12 week post-harvest storage period differed between the 13 pepper varieties analysed. Carotenoid retention was calculated as the change in carotenoid content between the fresh fruit time point and following 12 weeks of post-harvest storage. Some genotypes displayed very little change in total carotenoid content during post-harvest 4 °C storage, such as lines R1 and R12, whereas other lines showed significant increases (R5) or decreases (R7) in total carotenoid content. Effectively, this approach took into consideration initial carotenoid intensity, biosynthesis, and/or degradation during drying and throughout storage, all contributing to a measure of retention. However, other factors augmenting colour retention should not be ignored and it should be stated that carotenoid retention and colour retention are not the same parameter.

Table 1.

Change in carotenoid content following post-harvest storage of diverse pepper genotypes

Pepper genotype	Change in carotenoid content (%)	Carotenoid retention	Colour retentiona
R1	-0.6	Medium	High
R2	2.7	Medium	Low
R3	-18.3	Low	Low
R4	-20.9	Low	Low
R5	48.3	High	Medium
R6	24.2	High	Medium
R7	-25.7	Low	Medium
R8	-0.6	Medium	High
R9	-17.3	Low	Low
R10	8.5	Medium	Medium
R11	5.7	Medium	High
R12	-1.0	Medium	Low
CM334	-18.0	Low	ND

Carotenoid content was measured before and after post-harvest storage and the change in carotenoid content calculated. Carotenoid retention classification is allocated based on change in carotenoid content (see the Materials and methods). Data represent the mean values of six biological replications, where a biological replicate is a plant from which three fruit are analysed. Carotenoid and colour retention are linked but are not the same parameter.

^a Colour retention is based on image analysis as described in Berry et al. (2021).

Pepper genotypes were characterized as low, medium, or high carotenoid retention dependent on the change in carotenoid content during post-harvest storage. Arbitrary values were used to determine carotenoid retention phenotypes: lines which decreased in carotenoid content by >10% were deemed to be low retention (R3, R4), lines which showed a change in carotenoid content between –10% and 10% were deemed to be medium carotenoid retention (R1, R2), and lines which increased in carotenoid content by >10% were characterized as high retention (R5, R6). However, it should be stressed that a medium classified genotype that shows negligible loss in carotenoids over storage by definition has a colour retention trait, as is the case with the R8 line. In this case, the line has been referred to as a medium/high genotype to take into consideration the multiple classification parameters used. The R8 line is also ideal for this study as, besides having carotenoid retention, it has high colour retention, meaning that other parameters besides carotenoids could contribute to the trait of interest, in comparison with lines R5 and R6 that may have high carotenoid retention but display medium colour retention. Thus, there is a greater opportunity to elucidate parameters beyond carotenoids that contribute to colour.

Fruit surface texture is associated with carotenoid retention following post-harvest storage

Post-harvest drying of pepper fruits resulted in vastly different fruit surface textures when comparing the 13 pepper genotypes. Following oven drying, some pepper fruits, such as genotypes R1, R6, and R8, retained the smooth, waxy surface texture that all genotypes had at the fresh time point, whilst other genotypes, such as R3, R4, and R7, displayed significant surface ‘wrinkling’. Therefore, all pepper genotypes were characterized as having either a smooth cuticle surface or a cracked cuticle surface (Supplementary Fig. S1). Genotypes characterized as low carotenoid retention tended to display a wrinkled surface texture following drying, whilst medium and high carotenoid retention fruits tended to display a smooth surface texture following drying (Fig. 1A).

Fig. 1.

Fruit surface texture is associated with the carotenoid retention trait following post-harvest storage. (A) Pepper fruits following oven drying. R3 and R4 display a ‘wrinkled’ surface texture after drying; R6 and R8 display a ‘smooth’ surface texture after drying. (B) SEM images of dried pepper fruit surface. R3 and R4 display surface cracks; R6 and R8 display smooth surfaces. Magnifications of ×300 and ×3000 were used. Three biological replicates were used as described in the Materials and methods.

SEM further supported these differences between medium/high and low carotenoid retention pepper genotypes in their fruit surface texture. Low carotenoid retention genotypes R3 and R4 both had a ‘wrinkled’ surface texture, and showed evidence of surface cracks (Fig. 1B), whilst the medium/high carotenoid retention genotypes R6 and R8 both had smooth surface textures with no cracks present on the surface (Fig. 2A, B).

Fig. 2.

Exocarp thickness is associated with the carotenoid retention phenotype. (A) Light microscopy images of pepper fruit exocarp. (B) Cryo-SEM images of pepper fruit exocarp. (C) Exocarp thickness of the pepper diversity panel measured using light microscopy images. Means ±SE bars are shown (n=3). Three biological replicates (per genotype) were used with six measurements per image.

Fruit exocarp thickness is associated with the carotenoid retention phenotype

Due to the differences observed between medium/high and low carotenoid retention genotypes in fruit surface structure following post-harvest drying, further analysis of the fruit surface structure was carried out.

The outer surfaces of the pepper fruit, defined as the fruit exocarp, of fruits within the pepper diversity panel were observed using light microscopy (Fig. 2A) to determine whether variation in the fruit exocarp structure is associated with the carotenoid retention phenotype. Nile Red was used to stain the wax exocarp and Fast Green was used to stain pericarp, to clearly differentiate between these two tissue types. This method demonstrated that there was a clear difference in exocarp thickness within the fruits of the 13 pepper genotypes studied (Fig. 2A; Supplementary Fig. S2). Staining showed that epidermal cells were embedded within the wax exocarp layer and, in some cases, the wax exocarp layer penetrated several cell layers deep into the fruit. Whilst some genotypes had a very thin exocarp layer, for example R3 and R7, which showed just one layer of cells embedded within the wax exocarp layer, other genotypes, such as R5 and R6, displayed up to four or five layers of cells embedded within the wax exocarp. This difference in exocarp thickness was further supported following cryo-SEM, which demonstrated that low carotenoid retention lines R3 and R4 both had just one layer of cells embedded within the outer layer, whilst lines R6 and R8 both displayed several cell layers embedded within the outer layer (Fig. 2B).

Clear differences were observed in the thickness of the wax exocarp layer between different genotypes within the pepper panel. Whilst lines R1, R5, R6, and R8 all had an exocarp thickness ranging from 90 µm to 120 µm, lines R3 and R7 both had a thinner exocarp, with a thickness between 30 µm and 40 µm. The general trend appeared in which genotypes characterized as high or medium carotenoid retention also tended to have a thicker exocarp (Fig. 2C).

Biochemical profiling of pepper fruit cuticle

The carotenoid retention phenotype is only observed upon ripening of the pepper fruit, thus the cuticle composition was analysed in ripe fruit using an established GC-MS-based method (see the Materials and methods), as this was deemed to be the most biologically relevant developmental stage for this trait. As a comparator, the cuticle of mature green fruit was also profiled (Supplementary Table S2).

Significantly lower levels of 10,16-dihydroxyhexadecanoic acid were observed in both low carotenoid retention genotypes, R3 and R4, when compared with the medium/high carotenoid retention genotype, R8, but not, however, compared with R6 (Fig. 3E). This compound accounts for ~90% of the total content of the cutin polymer, suggesting that low carotenoid retention is associated with lower total cutin content. In ripe fruit, R3 (low retention) had significantly lower levels of ferulic acid compared with genotypes R4 (low retention) and R6 (high retention) (Fig. 3A), and of octadeca-9,12-dienoate compared with genotypes R4, R6 (high retention), and R8 (medium/high retention) (Fig. 3C). The low carotenoid retention genotype, R3, displayed significantly lower total cutin monomer content when compared with the medium/high carotenoid retention genotype, R8, but no significant difference was observed between other genotypes (Fig. 3G). Overall, the differences in cutin monomer content reflect the differences observed in pepper exocarp thickness. When the cuticle components are represented as percentage composition per genotype (Supplementary Table S3), the data showed good corroboration. Only minor compositional changes are evident, the most notable being increased ferulic acid in R4, R6, and R8 (ranging from a 2- to 10-fold) increase. There were also minor compositional increases in the total cutin monomer component of R6 (high) and R8 (medium/high). A caveat that should be stated is the large dynamic range in the amounts of cuticle components experienced, and domination by total cutin monomers makes compositional changes difficult to interpret.

Fig. 3.

Cuticle components of ripe fruit associated with the carotenoid retention phenotype. Cutin monomer components and cuticular waxes were measured in ripe fruits displaying variation in carotenoid retention phenotype: R3, low carotenoid retention; R4, low carotenoid retention; R6, high carotenoid retention; R8, medium/high carotenoid retention. (A–G) Cutin monomer components. (H and I) Cuticular lipophilic components. Only compounds in which significant differences were reported between pepper genotypes are reported here. Means ±SE (n=3) bars are shown. C₃₂ alkane (100 µg) was utilized as an internal standard in cutin monomer extraction; deuterated traiacontane (C₃₀) (10 µg) was utilized as an internal standard in cuticle lipophilic component extraction. Amounts were calculated as relative to the respective internal standards. There were three biological replicates per genotype, and seven cuticle discs per line were analysed.

The low carotenoid retention genotype, R3, displayed a significantly thinner exocarp compared with the medium/high carotenoid retention genotype, R8, and this significant difference between these two genotypes is further supported by the differences in their cutin compositions. The high carotenoid retention genotype, R6, also showed a thicker exocarp compared with R3 (low retention). However, no difference in total cutin monomer was reported between these two varieties. Thus, although there was no absolute correlation with the panel studied, the data suggested a trend between increased cutin monomer and carotenoid retention.

Cuticular lipophilic components, which are embedded within the cutin matrix, were analysed in the four pepper fruit genotypes (R3, R4, R6, and R8). Whilst β-sitosterol levels were decreased in the low carotenoid retention genotype, R3 (Fig. 3I), no other significant differences, with which cuticular lipophilic composition and cuticle thickness were associated, were detected (Supplementary Table S2). The low carotenoid retention genotype, R4, displayed significantly higher levels of hentriacontane compared with the other genotypes analysed (Fig. 3H); however, this did not follow the emerging trend in which exocarp thickness, and hence cuticle composition, associates with the carotenoid retention phenotype. Consequently, cuticle lipophilic components did not appear to be associated with the carotenoid retention phenotype, or with exocarp thickness.

Cuticle biosynthesis gene expression displays an association with changes observed in cuticle structure between high and low carotenoid retention pepper genotypes

RNA-seq was used to determine gene expression patterns in pepper fruit epidermal cells during cuticle development. From the R3 (low) and R8 (medium/high) retention genotypes, total RNA was isolated from pepper fruit epidermal tissue, in order to provide spatiotemporal resolution to gene expression analyses. This comparison was made based on the classification of retention provided in Table 1, exocarp thickness, fruit surface texture, biochemical changes in the cuticle composition, and logistical issues in obtaining RNA of sufficient quality from epidermal tissues. Genes involved in cuticle biosynthesis were identified in pepper, based on homology with orthologues in other species, previously reported to be involved in cuticle biosynthesis. Differentially expressed transcripts for gene ontologies associated with exocarp, epidermal/subepidermal cell layer, and development/differentiation only reveal one gene transcript throughout fruit development and ripening; this gene was Ca03g05300, cell differential protein rcd1 putative (Supplementary Dataset S1). A comparison of differentially expressed gene transcripts between R3 (low) and R8 (medium/high) retention are reported in Table 2. The pepper orthologue of the ABC transporter PERMEABLE CUTICLE1/ABCG32 (PEC1), which functions in the transfer of cutin components over the plasma membrane in Arabidopsis thaliana (Bessire et al., 2011), was identified as significantly up-regulated in the medium/high carotenoid retention genotype compared with the low carotenoid retention genotype. Lipid transport proteins (LTPs) are considered to be involved in the transport of cutin monomers to the site of cuticle synthesis (Yeats and Rose, 2008). Several LTPs were identified as differentially expressed between the medium/high and low carotenoid retention genotypes. Bodyguard (BDG) has been identified as playing a major role in cutin biosynthesis (Kurdyukov et al., 2006), and an orthologue of this gene was found to be up-regulated in the medium/high carotenoid retention genotype compared with the low retention genotype. Interestingly, a MYB16 transcription factor was found to be down-regulated in the medium/high carotenoid retention genotype during early fruit development. The orthologue of this gene has been shown to be involved in the regulation of cuticle development, along with WAX INDUCER/SHINE1 (WIN1/SHN1) in A. thaliana (Oshima et al., 2013). Thus, it could be expected that this gene would be up-regulated in the medium/high carotenoid retention genotype, as these fruits had a thicker fruit cuticle compared with the low carotenoid retention genotype. However, WIN1/SHN1 orthologues were not found to be differentially expressed in pepper. Two genes were annotated as ECERIFERUM1 CER1 in the pepper Sol Genomics Network and, interestingly, they displayed differences in expression between the medium/high and low retention genotypes. CA09G18740 was shown to be up-regulated in the medium/high retention line at anthesis +45 d, whilst CA12G22670, also annotated as CER1, was shown to be down-regulated in the medium/high retention genotype in ripe fruits. CER1 is involved in reduction and decarbonylation of the very long chain fatty acids to cuticular alkanes (Bernard et al., 2012). A cytochrome P450 enzyme designated midchain alkane hydroxylase (MAH1) is involved in the oxidation of alkanes to secondary alcohols and ketones in Arabidopsis (Greer et al., 2007). MAH1 genes were found to be up-regulated in the medium/high carotenoid retention genotype compared with the low carotenoid retention genotype throughout fruit development (Table 2).

Table 2.

Differentially expressed genes associated with fruit cuticle biosynthesis

Pepper gene	SGN annotation	Description	A+20	A+30	A+45	Ripe	References	Homologue	Homologue function
CA06G14420	ABC transporter	Putative ABC transporter					Bessire et al. (2011)	At2G26910	ABCG32 full transporter
CA06G14430	Unknown function						Bessire et al., (2011)	At2G26910	ABCG32 full transporter
CA01G00270	Valacyclovir hydrolase	Putative					Kurdyukov et al. (2006)	At4G24140	BDG3 BODYGUARD3
CA01G19070	CD2	Cutin deficient 2					Isaacson et al. (2009)	Solyc01G091630	CD2 CUTIN DEFICIENT2
CA05G18530	Unknown function						Gable et al. (2004)	At3G55360	CER10 ECERIFERUM10 Enoyl-CoA reductase
CA09G18740	CER3	Predicted protein ECERIFERUM 3-like					Rowland et al. (2007)	At5G57800	CER3 ECERIFERUM3 involved in alkane formation
CA11G06990	CER3	Predicted protein ECERIFERUM 3-like					Rowland et al. (2007)	At5G57800	CER3 ECERIFERUM3 involved in alkane formation
CA02G23370	3-ketoacyl-CoA synthase	3-ketoacyl-CoA synthase					Vogg et al. (2004)	At1G68530	CER6 ECERIFERUM6 b-Ketoacyl-CoA synthase
CA05G19790	Cytochrome P450	Cytochrome P450					Li-Beisson et al. (2009)	At3G10570	CYP77A6 CYP77A subfamily
CA08G18140	Predicted long chain acyl-CoA synthetase 2-like	Predicted long chain acyl-CoA synthetase 2-like					Schnurr et al. (2004)	At1G49430	Long chain acyl-CoA synthase
CA08G08360	Predicted long chain acyl-CoA synthetase 4-like	Predicted long chain acyl-CoA synthetase 4-like					Fulda et al. (2002)	At1G64400	Long chain acyl-CoA synthase
CA03G30090	Unknown function						Kranz et al. (1998)	At3G28910	MYB30 MYB transcription factor
CA02G28250	Transcription factor	Transcription factor					Kranz et al. (1998)	At3G28910	MYB30 MYB transcription factor
CA03G30090	Unknown function						Stracke et al. (2001)	At5G62470	MYB96 MYB transcription factor
CA02G28250	Transcription factor	Transcription factor					Stracke et al. (2001)	At5G62470	MYB96 MYB transcription factor
CA10G12030	LTP	Lipid transfer protein					Popovsky-Sarid et al. (2017)
CA10G12060	LTP	Lipid transfer protein
CA09G18740	CER1	Fatty acid hydroxylase
CA12G22670	CER1	Fatty acid hydroxylase
CA11G14620	MYB16	MYB-related transcription factor
CA02G20380	LCFL	Long chain fatty acid CoA ligase
CA03G16520	ERF	Ethylene responsive factor 2A
CA02G17970	PE	Pectinesterase
CA07G11250	ACCOXIDASE	Fruit ripening
CA10G16170	ACCOXIDASE	Fruit ripening
CA10G10710	LTP	Lipid transfer protein
CA10G10770	LTP	Lipid transfer protein
CA10G18900	CYP96A/MAH1	Cuticle development
CA10G18910	CYP96A/MAH1	Cuticle development
CA10G08490	LTP	Lipid transfer protein
CA10G18310	CYTB5	Cuticle development

Red, down-regulated R3; blue, up-regulated R3. Transcript number was assessed at four fruit development stages: anthesis +20 d, anthesis +30 d, anthesis +45 d, and ripe. Genes were identified based on literature searches and using BLAST to identify gene orthologues. Sol Genomic Network gene descriptions are noted, if provided. Genes were considered to be significantly differentially expressed if P<0.05, FDR<0.05. The biological replication is described in the Materials and methods.

The differentially expressed genes involved in cuticle biosynthesis between the medium/high and low carotenoid retention genotypes provide further evidence that the carotenoid retention phenotype is controlled, in part, by the structure of the fruit surface.

Carotenoids associated with the fruit exocarp may influence the carotenoid retention trait

Upon enzymatic isolation of exocarp discs from pepper fruit, it was observed that the red pigment was retained within the wax exocarp tissue, despite the removal of all other pericarp tissue (Fig. 4A). It was consequently hypothesized that carotenoids may be located within the epidermal cells found in the isolated exocarp layer. The waxy exocarp layer penetrated several cell layers deep within the fruit, particularly in genotypes R6 and R8; these carotenoids may remain within the exocarp due to the protective nature of this layer. A comparison was made with tomato fruit exocarp discs, but the red pigment observed in pepper exocarp discs was not present in tomato discs (Supplementary Fig. S3).

Fig. 4.

Carotenoids are associated with pepper fruit exocarp. (A) Pepper fruit exocarp discs (1 cm diameter). (B) Carotenoid content of pepper exocarp discs, expressed as exocarp to whole fruit ratio. Means ±SE bars are shown (n=3). Ten discs per biological replicate were analysed together with three biological replicates.

Carotenoids were isolated from exocarp discs by sequential chloroform and methanol washes over a period of 3 d. HPLC analysis was used to quantify carotenoids extracted from exocarp discs. These values were then compared with the carotenoid amounts extracted from whole fruit discs, and results were expressed as a ratio (Fig. 4B). This approach allowed the quantity of exocarp-bound carotenoids to be compared in a relative manner with the whole-fruit carotenoid content, regardless of variations in total fruit carotenoid content between the genotypes.

Interestingly, significant differences were only observed in the ratio of capsanthin and capsanthin esters between the genotypes analysed. This suggests that these carotenoids may be responsible for causing the observed red colour of exocarp discs in these genotypes. A significantly greater exocarp to whole fruit ratio for capsanthin diesters was observed in the medium/high carotenoid retention genotype, R8, relative to the three other genotypes analysed (Fig. 4B). This trend was further reflected in total carotenoid content when comparing R8 (medium/high retention) with the other genotypes analysed (Fig. 4B).

Initiation of carotenoid degradation reveals the crucial protective role of the exocarp

The fruit surface structure has been demonstrated to be associated with the carotenoid retention phenotype; it was hypothesized that the fruit surface may be crucial in protecting against carotenoid degradation. Cracks were not observed on the surface of medium/high carotenoid retention fruits. Therefore, it could be postulated that the fruit surface may provide a protective barrier against oxidative degradation of carotenoids.

Pepper fruits were treated with H₂O₂, as an oxidative agent, to determine the role of the fruit surface in protecting against carotenoid degradation. Fruits were harvested when ripe; control fruits were treated with varying concentrations of H₂O₂ immediately following harvest, and were oven dried and stored for 4 weeks.

Following post-harvest storage, a decrease in total carotenoid content was observed in the low carotenoid retention genotype, R3, peppers treated with 2 mM H₂O₂ before drying when compared with peppers treated with 0 mM H₂O₂ (Fig. 5A), at which point the cuticle remained smooth and exhibited no cracks. No decrease was observed in peppers treated with 0.2 mM H₂O₂ before drying. In contrast, R3 peppers treated with both 0.2 mM and 2 mM H₂O₂ after fruit drying displayed decreases in total carotenoid content (Fig. 5B) compared with peppers treated with 0 mM H₂O₂ at this time point. Fruits treated with H₂O₂ following drying were more susceptible to carotenoid degradation when a lower concentration (0.2 mM) H₂O₂ was used. After drying, the fruit surface displayed cracking; this structural alteration could have facilitated the entrance of oxidative species into the fruit and promoted carotenoid degradation.

Fig. 5.

Carotenoid content of pepper fruits after post-harvest storage, following treatment with hydrogen peroxide. (A) Total carotenoid content (µg g^–1 DW) after post-harvest storage, following hydrogen peroxide treatment before fruit drying. (B) Total carotenoid content (µg g^–1 DW) after post-harvest storage, following hydrogen peroxide treatment after fruit drying. Student’s t-test was used to determine significant differences between the control condition (0 mM H₂O₂) and test concentrations. Means ±SE bars (n=3) are shown. Three biological replicates were used as described in the Materials and methods.

The medium/high carotenoid retention genotype, R8, displayed no difference in total carotenoid content, regardless of the H₂O₂ concentration used (Fig. 5A, B). This was observed in fruits treated both before and after fruit drying. The fruit surface remained smooth in the R8 genotype (medium/high retention) and intact following drying, suggesting a protective barrier was in place to limit the entry of oxidative species into the fruit.

Discussion

Carotenoid retention as a key quality trait of pepper fruits

The retention of colour during post-harvest storage is an important quality trait for ripe chilli pepper fruit.

The major carotenoid conferring red colour in chilli peppers is capsanthin, whilst other xanthophylls such as violaxanthin, neoxanthin, and antheraxanthin, along with β-carotene, are also present in fruits (Minguez-Mosquera and Hornero-Mendez, 1994b; Hornero-Méndez et al., 2000; Berry et al., 2019). Whilst carotenoid content in harvested pepper fruit has been studied previously (Berry et al., 2019), the change in carotenoid content during post-harvest storage, defined as carotenoid retention, has been understudied to date. Identification of genotypes which retain high levels of carotenoids is crucial to elucidate the mechanisms underlying this trait. The study presented here has demonstrated that pepper genotypes can retain carotenoids to different extents following post-harvest storage (Table 1). Interestingly, some pepper fruit genotypes showed an increase in carotenoid content during post-harvest storage, including genotypes R5 and R6, which showed carotenoid increases of 48% and 24%, respectively, following oven drying. This increase in carotenoids during post-harvest storage has previously been observed in pepper varieties, and it has been postulated that biosynthesis can proceed during post-harvest storage (Park and Lee, 1975; Minguez-Mosquera and Hornero-Mendez, 1994a; Mínguez-Mosquera et al., 2000). In addition to carotenoid pigments, the measure of colour can encompass several other parameters (Berry et al., 2021). The inclusion of image analysis data in the classification of genotypes has added to the robustness and contextual interpretation of the present data.

Carotenoid retention is a key quality trait not only in pepper, but also in other crop species. Carotenoid losses during post-harvest storage are well reported in a variety of crop species (Burt et al., 2010; Bechoff et al., 2011), and this affects both the colour quality and nutritional quality. Elevated carotenoid content has been engineered in a variety of crop plants, including rice (Beyer et al., 2002), maize (Zhu et al., 2008), and cassava (Welsch et al., 2010). Whilst these studies have been successful in engineering elevated carotenoid content and, consequently, increased provitamin A capacity in these crops, significant challenges have been encountered due to the fact that these crops do not retain these carotenoids during post-harvest storage (Hidalgo and Brandolini, 2008; Ortiz et al., 2016).

The pepper fruit surface structure is associated with the carotenoid retention trait

The cuticle of fruit species has been acknowledged as a modulator of post-harvest quality (Lara et al., 2014), influencing the shelf-life potential of fruit crops (Lara et al., 2019). This study proposes the pepper fruit surface, and specifically the fruit cuticle, as a key structural feature in protecting the fruit against carotenoid losses during post-harvest storage. The carotenoid retention phenotype is associated with the fruit surface structure of pepper, which was noted because genotypes characterized as low carotenoid retention tended to show a ‘wrinkled’ or ‘cracked’ fruit surface upon fruit post-harvest drying, whilst genotypes characterized as high carotenoid retention tended to show a smooth fruit surface upon drying (Figs 1, 2). Ferulic acid has previously been demonstrated to be associated with suberin deposition in melon fruits (Cohen et al., 2019). Suberin deposition occurs in response to wounding, to seal the compromised tissue. Interestingly, high levels of ferulic acid were observed in genotype R4, which further displayed wrinkling upon drying. Consequently, the high levels of ferulic acid, which are precursors for suberin, could be associated with cracking of the fruit surface, which involves suberin deposition.

Upon further inspection of the fruit surface using light microscopy, it was noted that pepper genotypes with a smooth surface also possessed a thicker fruit exocarp. The exocarp described here consisted of a lipidic layer in which cells were embedded (Fig. 2). The fruit cuticle is normally localized only on the outer surface of epidermal cells (Martin and Rose, 2014); however, in some genotypes studied here, this lipid layer penetrates several cell layers deep within the fruit, and is consequently termed the ‘exocarp’. Microscopic analysis of the tomato cultivar M82 revealed a small amount of subepidermal cuticular material deposition (Buda et al., 2009; Yeats et al., 2012); however, this did not appear to penetrate multiple cell layers deep into the fruit as presented here in the case of chilli pepper fruits. The cuticular layer has been reported to surround more than a single cell layer below the outermost epidermis in Ailsa Craig tomatoes (Mintz-Oron et al., 2008); however, deposition of cuticular components penetrating below the epidermal cell layer has not been widely reported to the extent demonstrated in this study in pepper fruits. Genotypes possessing a smooth surface tended to have more layers of cells embedded within this lipid layer. A resistance to cuticle cracking has previously been demonstrated to correlate with a thicker fruit cuticle in the case of cherry tomato (Matas et al., 2004).

The cuticle is composed of two components: a cutin-rich section and embedded cuticular lipophilic components (Yeats and Rose, 2013). Here, increased total cutin monomer content, along with an increase in specific cutin monomers, for example 10,16-dihydroxyhexadecanoic acid, were shown to be correlated with the high carotenoid retention trait in pepper fruits (Fig. 3). In contrast, cuticular wax content was not shown to be associated with the carotenoid retention phenotype (Fig. 3). This indicates that the thicker exocarp consisted of a cutin matrix, which may be supported by the finding that the tomato cd1 mutant, which presents a deficiency in cutin, displayed a significantly thinner cuticular layer with less subepidermal deposits than the M82 control (Yeats et al., 2012). These findings suggest that the subepidermal deposits observed in the M82 tomato, and in the pepper genotypes presented here, are largely comprised of cutin. Cuticular waxes showed no association with exocarp thickness, or with the carotenoid retention trait, and this may be due to their localization to the outermost layer of the fruit surface. Further, total cuticular waxes have been shown not to correlate with water loss rate in a diverse pepper collection (Parsons et al., 2013). This indicates that cuticle waxes may not play a critical role in traits influenced by the fruit cuticle structure, as demonstrated by their lack of correlation to water loss rate, or to the carotenoid retention trait.

The cutin matrix penetrates several cell layers deep within the fruit pericarp in the form of subepidermal deposits; this suggests that subepidermal cell layers, along with epidermal cells, may synthesize cuticular components. Previous identification of attached and detached subepidermal globules in the tomato M82 cuticle raised questions regarding the mechanism responsible for depositing cuticular material in subepidermal layers (Buda et al., 2009). Two hypotheses were presented: (i) subepidermal cuticular material is derived from epidermal cells and trafficked into subepidermal walls; or (ii) subepidermal cells, along with epidermal cells, synthesize small amounts of cuticular components (Buda et al., 2009). The data presented here may support the second of these two hypotheses, in that subepidermal cells can synthesize cuticular components, as a significant trafficking network would be required to explain the depth to which cuticular components penetrate the fruit pericarp in some of the pepper genotypes described in this study. Cuticle precursor transport is not entirely understood; however, data presented here suggest that the claim that epidermal cells are responsible for cuticle synthesis and transport (Suh et al., 2005) may need to be reconsidered.

The differential expression of cuticle biosynthesis genes between the medium/high and low carotenoid retention pepper genotypes supports the finding that the fruit surface structure is associated with the carotenoid retention phenotype. Gene products identified previously as involved in pepper cuticle biosynthesis (Popovsky-Sarid et al., 2017) were amongst those identified as differentially expressed gene transcripts between the medium/high and low carotenoid retention genotypes. Several genes involved in the biosynthesis of A. thaliana cuticle (Mintz-Oron et al., 2008; Yeats and Rose, 2013) were found to have orthologues in pepper, which were differentially expressed between the medium/high carotenoid retention genotype, with a thick exocarp (Table 2), and the low carotenoid retention genotype, which possessed a thinner fruit exocarp. An orthologue of the BDG, known in A. thaliana to be involved in cutin biosynthesis (Kurdyukov et al., 2006), is significantly up-regulated in the medium/high retention genotype fruits. This supports the finding that cutin monomer content is also increased, and that the exocarp is thicker, in this medium/high carotenoid retention genotype (Fig. 2). However, the Cutin Deficient 2 (CD2) gene, which in tomato has been reported to regulate cutin monomer biosynthesis (Isaacson et al., 2009), is up-regulated in the low carotenoid retention pepper genotype. Despite the fact that a cd2 tomato mutant was found to have a 98% decrease in cutin content, suggesting the crucial role of CD2 in regulating cutin biosynthesis in tomato fruit (Isaacson et al., 2009), the same influence of this gene product may not be exerted in pepper. Differential expression of genes involved in cutin monomer transport, including PEC1 (Bessire et al., 2011) and various LTP genes (Yeats and Rose, 2008), may indicate that transport of these cuticular components is critical to determining the extent of subepidermal cutin deposits, and consequently exocarp thickness. Increases in these gene products could result in increased subepidermal cutin deposition. Therefore, it would seem that cuticle precursor transport is a critical step in determining cutin content. Several genes encoding components of cuticular wax biosynthesis, such as ECERIFERUM1 (CER1), ECERIFERUM3 (CER3), and the midchain alkane hydroxylase (MAH1), which are involved in alkane formation from very long chain and midchain fatty acid precursors, respectively (Greer et al., 2007; Bernard et al., 2012), were also shown to be differentially expressed. In these cases, differences in gene expression did not correlate with cuticular wax content, potentially suggesting that post-transcriptional regulatory mechanisms are operating. The data also show examples where gene transcript levels do not correlate with previously predicted functional studies; for example, MYB16 (Popovsky-Sarid et al., 2017) is known to be a positive regulator of cuticle development but is down-regulated in R8 (Natarajan et al., 2020). These studies were performed in hot pepper accessions of a different species, Capsicum chinese, and, in the case of Natarajan et al. (2020), low carotenoid (yellow) accessions were studied. Despite this difference, the studies were in agreement that 10,16-dihydroxyhexadecanoic acid is the dominant cutin monomer and is influential in altering total cutin. In such cases, the effect of the genetic background should be considered. The creation of isogenic lines harbouring the trait could assist in future studies.

Overexpression of the tomato MIXTA-like MYB transcription factor gene (SlMX1) has previously been shown to result in increased cuticle deposition in the peel, along with an increase in total fruit carotenoid content (Ewas et al., 2016). However, the present data showed that homologues of this gene transcript were not found to be differentially expressed between the medium/high and low carotenoid retention pepper genotypes.

Implications of the localization of carotenoids in cells within the fruit exocarp

Interestingly, carotenoids have been shown to be associated with the ripe pepper fruit exocarp. This phenomenon has not previously been observed in pepper, or in other similar fleshy fruit species, such as tomato. The observation of carotenoids within the lipidic exocarp layer following enzymatic degradation of the cell walls suggests that the lipid exocarp layer protects carotenoids from degradation. Consequently, the structure of the lipid exocarp directly influences the carotenoid retention phenotype.

Further to this, the ratio of exocarp to whole-fruit capsanthin diester content is significantly greater in the high carotenoid retention genotype, R8, compared with the low genotype, R3. This suggests that the waxy, lipophilic environment of the exocarp favours the storage of more non-polar carotenoids, specifically diesters in this case (Fig. 4). Esterified carotenoids have been shown to be more stable than their non-esterified counterparts (Schweiggert et al., 2007), and this may explain the increase in total carotenoid content of these cells in the high retention genotype, as esterified carotenoids are less susceptible to oxidative degradation. Whilst the mechanism by which increased esterified carotenoids are accumulated in these exocarp embedded cells is unknown, one explanation may be that these cells have an increased capacity for storage of carotenoids in subchromoplast organelles, specifically in fibrillar plastoglobuli, which are well documented regarding their role in the sequestration and storage of esterified carotenoids in pepper fruit (Deruère et al., 1994). An increase in fibrillar structures would facilitate the accumulation of increased pigment levels within the chromoplasts of high carotenoid retention genotypes, as has previously been demonstrated (Berry et al., 2019).

Carotenoids are localized to cells within exocarp. Spatial localization of carotenoids within outer fruit layers may influence the visual perception of the fruit, and their localization within cells embedded within a lipid layer appears to provide protection against degradation.

Pepper fruit surface protects carotenoids from degradation during post-harvest storage

Treatment of pepper fruits with a cracked surface structure (low carotenoid retention: R3, dried fruits) with H₂O₂ as an oxidative agent resulted in greater degradation of carotenoids than observed in smooth surface fruits (medium/high carotenoid retention: R8, fresh and dried fruits) (Fig. 5). This was presumably due to the cracked cuticle being more permeable to H₂O₂, therefore resulting in increased ROS within the fruit responsible for initiating endogenous lipid peroxidation and consequently increased carotenoid degradation. Capsorubin, capsanthin, and capsanthin diesters are well characterized as having high quenching capacity for singlet oxygen and hydroxyl free radicals (Nishino et al., 2016), but are degraded in the process. Collectively, these data provide further evidence that the fruit surface structure is critical in protecting the fruit from permeation of ROS or their precursors. Upon permeating the cuticle barrier, ROS can initiate lipid peroxidation. Carotenoids acting as cellular antioxidants are then degraded as they dissipate reactive molecular species associated with lipid peroxidation. These findings support those postulated by Berry et al. (2021), whereby altered carotenoid retention is associated with altered fatty acid content, lipid peroxidation, and the dissipation of ROS by carotenoids. The present data now demonstrate the involvement of the fruit surface structure as a means of either preventing or enabling the entry of pro-oxidants into the fruit tissue.

In summary, the present study has characterized a panel of chilli pepper accessions displaying altered post-harvest carotenoid/colour retention properties. Although not exclusive, the data imply a contributory role for the fruit surface structure in post-harvest retention of pepper colour. Further studies and the construction of biparental populations and/or genome-wide association studies (GWAS) have the potential to reveal robust correlations between metabolites, traits, and genetic components. These resources could in the future be used as a strategy to maintain colour/carotenoid retention in crops for improved quality and nutritional quality.

Supplementary data

The following supplementary data are available at JXB online.

Table S1. Pepper diversity panel storage experiment of carotenoid amounts.

Table S2. Cutin monomer and cuticular wax components of pepper fruit cuticles.

Fig. S1. Variation in pepper fruit surface appearance of the diversity panel genotypes following drying and post-harvest storage.

Fig. S2. Variation in exocarp structure in the pepper panel.

Fig. S3. Isolated exocarp discs of pepper and tomato fruits, before and after chloroform washing.

Dataset S1. Cuticle RNA-seq of differentially expressed genes, low versus high colour retention.

Author contributions

AH, DVR, and PDF: conceptualization; AH, PDF, AA, and HC: methodology; AH: formal analysis, investigation, visualization, and writing—original draft; PDF and AA: resources; AH and HMB: data curation; PDF, DVR, HMB, AA, and HC: supervision; AH, HC, DVR, HMB, AA, and PDF: writing—review and editing; PDF: project administration; PDF, DVR, and AA: funding acquisition.

Conflict of interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Funding

The work was supported through a Biotechnology and Biological Sciences Research Council iCASE with Syngenta Ltd, to ACH and PDF (Project BB/P001742/1), and Weizmann–UK making connections funding.

Data availability

Processed data are available in the manuscript and the supplementary data. The unprocessed transcriptomic data have been deposited with the NCBI under BioProject PRJNA640935.