Tracking the development of the superficial scald disorder and effects of treatments with diphenylamine and 1-MCP using an untargeted metabolomic approach in apple fruit.

Superficial scald is a physiological storage disorder that significantly reduces the marketability of apple fruit. To gain fundamental knowledge about the biochemical pathways leading to the development of the disorder and mechanisms of treatments for prevention, an untargeted metabolomics experiment employing liquid chromatography and mass spectrometry with data independent acquisition was performed. Metabolomic changes of two apple cultivars 'Cortland' and 'Red Delicious' with scald development and scald control treatments, using diphenylamine and 1-MCP, at 0-1 °C for up to 7 months was investigated. In total, 833 features/compounds were analyzed, and among them 59 were found to change significantly in controls involved in scald development, and in response to DPA and 1-MCP treatments. Our results provide new evidence that metabolites in association with phenylpropanoid metabolism, antioxidant and redox systems, and amino acid metabolism are related closely to scald development and response to potential treatments.

Introduction

Superficial scald is a physiological disorder of apples characterized by brown coloring of fruit surface during cold storage (Lurie & Watkins, 2012). As some cultivars of apple and pear are sensitive to this disorder, the development of scald results in significant losses to the tree fruit industry. Since the late 1960’s, there have been a large number of publications reporting efforts to define the metabolic events leading to superficial scald and its mechanism of action (Lurie & Watkins, 2012). In general, it is accepted that chilling injury and antioxidant imbalances in the peel tissues may be responsible for the development of disorder. One of the hypotheses that has been suggested is that oxidation products of α-farnesene cause cell damage resulting in peel browning (Rowan et al., 1995, Rowan et al., 2001). A close relationship between increased oxidation products of α-farnesene, conjugated trienols (CTs) and scald disorder in apple fruit has been found (Whitaker et al., 2004). In summary, it is understood that superficial scald is a low temperature disorder, that ensues as a consequence of chill-induced impairment of the cytochrome pathway of electron transport, leading to accumulations of superoxide free radicals and hydrogen peroxide, and that these accumulated radicals interact to form hydroxyl radicals which are highly reactive and cause the oxidation of farnesene, leading to cellular damage and the consequent expression of the disorder (Pechous et al., 2005, Whitaker, 2004).

Certain apple fruit cultivars, such as ‘Granny Smith’, ‘Red Delicious’, ‘Cortland’ and ‘Law Rome’ seem to be more sensitive than other cultivars to develop this disorder. Development of superficial scald can be controlled by various postharvest treatments. DPA (diphenylamine) or ethoxyquin (6-ethoxy-2,2,4-trimethyl-1,2-dihydroquinoline) have been applied worldwide to prevent the disorder in apples and pears by commercial industry for more than 50 years (Huelin & Coggiola, 1968). Low oxygen storage can also reduce the development of apple scald (Sabban-Amin, Feygenberg, Belausov, & Pesis, 2011). Application of 1-methylcyclopropene (1-MCP), an inhibitor of ethylene production, can also significantly reduce apple scald (Lurie and Watkins, 2012). The fact that treatment with DPA, low oxygen storage or 1-MCP results in reductions of the disorder severity suggests both oxidative stress and ethylene may be responsible for scald development. Intermittent warming or temperature conditioning was also reported as causing reversal of abiotic stress with a significant reduction of scald disorder (Honaas et al., 2019, Lurie and Watkins, 2012).

Various ‘omics’ approaches have been applied to uncover the biochemistry of scald development of apples and to understand the effects of DPA and 1-MCP treatments. Untargeted metabolic profiling using GC/MS identified some triterpenoids similar to ursolic acid and ß-sitosterol in apple fruit peel tissue and their accumulation was linked with scald development (Rudell, Mattheis, & Hertog, 2009). In another metabolomic study, accumulations of α-farnesene oxidation products, methyl esters, phytosterols, and some compounds related to loss of chloroplast integrity and oxidative stress response were identified (Leisso, Buchanan, Lee, Mattheis, & Rudell, 2013). In another targeted metabolomic study, accumulation of five polyphenol compounds were quantified (Busatto et al., 2014). Among them, chlorogenic acid accumulated in the peel tissue and this was coincident with the rapid development of scald. A corresponding upregulation of gene expression relating to the chlorogenic acid pathway, namely phenylalanine ammonia lyase (MdPAL), p-coumarate 3 hydroxylase (MdC3H) and polyphenol oxidase (MdPPOm) was also found in that study. Therefore, it was proposed that α-farnesene and its oxidative product (CTols) may trigger a signal for browning in apple scald (Busatto et al., 2014). A quantitative proteomic study revealed that a group of 14 proteins, which are involved in several metabolism pathways, such as antioxidant and redox, phenylpropanoid, ethylene biosynthesis, allergens, sulfur amino acid-containing proteins as well as programed cell death, are closely linked to the scald development (Du, Song, Campbell Palmer, Fillmore, & Zhang, 2017). Further studies on characterization of laccase in apples suggested that the metabolism of phenolic compounds may be responsible for scald development, but processes other than polyphenol oxidase mediated process may be responsible for the development of browning symptom. This hypothesis is supported by the fact that neither DPA or 1-MCP reduce the accumulation of total phenolic compounds or individual phenolic compounds such as chlorogenic acid (Gong et al., 2018). A browning reaction mediated by laccase with catechin or epi-catechin as substrates was also proposed to be responsible for scald development. Research on ‘Granny Smith’ apple demonstrated the accumulation of chlorogenic acid and other phenolic compounds with antioxidant and scavenging activity (Busatto et al., 2018). Polyphenol oxidase (PPO) controls and maintains the level of chlorogenic acid. It was shown that 1-MCP treatment repressed PPO gene expression and upregulated a complex of enzymes involved in metabolism of unsaturated and long chain fatty acids and sorbitol which are improved for tolerance of low temperatures. This response was interpreted as being important cold acclimation and freezing tolerance (Busatto et al., 2018). Recently, transcriptome analysis on ‘Granny Smith’ apples suggested that fatty acid metabolism, organellar modifications and antioxidant biosynthesis are involved in recovery from oxidative stress. In addition, an energetically active process involving proteasome mediated protein turnover as well as new protein synthesis are also contributed to the recovery of the injury (Honaas et al., 2019). Several genes which are involved in photosynthesis, stress response, flavonoid biosynthesis and ethylene signaling, were induced by 1-MCP treatment resulted in scald tolerance (Karagiannis et al., 2020).

Liquid chromatography-mass spectrometry (LC–MS) has become the primary platform for metabolomics studies due to rapid separation and broad range detection of small molecules (Alseekh & Fernie, 2018). Research has been conducted using LC–MS untargeted metabolomics to investigate stress response in legume crops (Jorge et al., 2015, Rathahao-Paris et al., 2016). In these studies, various compounds in different classes, including primary metabolites (e.g. sugars, fatty acids and amino acids) and secondary metabolites (flavonols and phenolic acids) were separated for identification using LC–MS. In addition to legume plants, LC–MS based metabolomics has been applied to investigate the metabolomics changes in Brassicaceae (De Vos, Schipper, & Hall, 2012a), tomato (Rogachev & Aharoni, 2012) and wine grape (Ghan et al., 2015). A group of A-Type procyanidin dimers in yellow raspberries fruit were also detected and quantified (Carvalho et al., 2016). Research on potato leaf revealed that modulation of the steroidal glycoalkaloids biosynthetic pathway at a single enzyme site leads to wider metabolic changes, and the knockdown of the glycoalkaloid metabolism gene expression results in a differential response to Colorado potato beetle (Paudel, Davidson, Song, Maxim, Aharoni, & Tai, 2017).

The biochemical mechanism and molecular bases of apple scald development as well as control mechanisms remain unclear. Untargeted (global) metabolomics employing LC/MS to perform comprehensive comparative analysis of relatively low molecular weight metabolites may provide insights into the interaction of the development of this physiological disorder and the effects of treatments that can significantly inhibit the scald development of apples. The objectives of this study were to determine the metabolomics changes in association with scald development and its response to DPA and 1-MCP treatments, to reveal metabolomics processes responsible for both scald development and control, and to gain novel insight into apple scald disorder.

Materials and methods

Apple fruit and treatments

Apple fruit (Malus × domestica Borkh.) ‘Cortland’ and ‘Red Delicious’ were harvested before the climacteric stage in 2013 from a commercial orchard in Berwick, NS, Canada. Two biological replicates per cultivar of two tree blocks were selected and harvested. Fruit samples and treatment have been published previously (Du, Song, Campbell Palmer, Fillmore, & Zhang, 2017). Briefly, fruit were treated with DPA (Decco No Scald® DPA aerosol) and 1-MCP (EthylBloc, 0.14%, Rohm and Haas Company, Philadelphia, PA. USA) separately before storage. Control fruit were kept at 22 °C without any treatment. Fruit were then stored under controlled atmosphere storage (CA) (3.0 kPa O2 + 1.0 kPa CO2) at 0–1 °C for 7 months. Quality assessment and sampling were taken place and evaluated after 4 and 7 months storage (Du, Song, Campbell Palmer, Fillmore, & Zhang, 2017).

Scald development

Scald development was evaluated at the same day after removal from storage, and after an additional 7 days at 22 °C. Scald rating was conducted based on 1 to 5 scale previously reported (Watkins et al., 2000). Twelve (12) apples were used for each evaluation. Peels with ca. 0.5 cm flesh tissue were collected, pooled and frozen in liquid N2, and stored at −86 °C for further analysis (Du, Song, Campbell Palmer, Fillmore, & Zhang, 2017).

Metabolomics analysis

Sample extraction and preparation

The extraction of metabolites from apple tissue was modified from a protocol developed for broccoli tissues that had been reported previously (De Vos et al., 2012b, Gong et al., 2018). Briefly, 0.5 g of frozen, ground apple peel tissue was extracted twice with 0.7 mL of extraction solvent (80:20 methanol: water, V/V, 0.1% formic acid). The samples were mixed and sonicated for 20 min, followed with centrifugation at 10,000×g for 10 min at room temperature (Thermal ICE Microlite, Fisher Scientific Company, Ottawa, Ontario). The supernatants from the extractions were combined and transferred to a new, pre-weighed microcentrifuge tubes and dried in a vacuum centrifuge (Thermo Fisher). The dried extracts were re-dissolved in 1 mL 10% methanol, 0.1% formic acid and mixed via sonication for 10–15 s and then vortexed for 10 s. After another centrifugation at 10,000×g for 10 min, the supernatants were transferred to HPLC vials for injection.

LC separation was conducted using a Waters NanoAcquity UPLC (Waters Corporation , Milford MA, USA) equipped with a BEH C18 1.7 µm 1.0 × 100 mm column (Waters, part number 186002346). The UPLC mobile phase was comprised of mobile phase A with 0.1% formic acid in 99.9% water, and mobile phase B with 0.1% formic acid in 99.9% acetonitrile. The UPLC gradient program was run over 25 min at a flow rate of 45 µL min−1, with initial conditions of 95% mobile phase A over 13.5 min and dropped to 50%. Over the next minute, the percentage of mobile phase A dropped further to 5% and was held for 3.5 min. Then, the percentage of mobile phase A returned to the initial percentage (95%) and remained there for 7.0 min. Column temperature was held at 35 °C and samples were kept in the auto sampler sample compartment at 4 °C. Injection volume was 1.0 µL using the partial loop mode.

Mass spectrometry

Apple compound analysis by LC/MS was conducted on a Synapt XS HDMS mass spectrometer (Waters Corperation, Milford MA, USA) with electrospray ionization source (ESI) using the Waters MSE acquisition (data independent acquisition) method in the continuum mode. The mass detector was operated in negative mode at high resolution, having a mass resolution of 40,000. The Masslynx (version 4.2) for MSE acquisition parameters were as follows: capillary voltage was 2.0 eV, in negative mode, m/z range detected was 75–1,000 amu using a scan speed of 0.4 s with an inter scan delay of 0.02 s. The collision energy was set at 6 eV for the low energy and the high energy ramp was 15–35 eV. The cone voltage was 20 V. Argon was used as a collision gas. Leucine enkephalin ([M−H]− m/z: 554.2615) was used as a lock mass solution. The lock mass was applied during the sample acquisition in 30 s intervals. Sodium formate was used as a calibrant over the mass range of 100–1500 amu. A standard mixture of 12 compounds was also prepared and run as QC and to assist with identification based on retention time, mass accuracy and mass fragmentation.

Metabolites detection and identification

Data processing and analysis

Two separate approaches were conducted to performed the data analysis. The first approach was to use Progenesis QI (version 2.4, Nonlinear Dynamics, Waters Corporation, Milford MA, USA) to detect masses of peak groups/clusters (i.e. chromatographic peaks) of raw data files to generate a result table including all detected mass traces, retention time and integrated intensity values with normalization. Initial PCA analysis was performed using Progensisi QI package. The data were also exported to Excel for further statistical analysis with Genstat (16th edition, VSN International, Hemel Hempstead, UK). The second approach was to convert raw data files from the LC/MS of each sample and replicates to NetCDF files using DataBridge software included in Masslynx 4.2. Mass peak detection and retention time correction was accomplished using the XCMS package (Smith, Want, & O’Maille, 2006). Peak detection, clustering and pathway plot were performed with XCMS (Tautenhahn, Patti, Rinehart, & Siuzdak, 2012).

Metabolites identification

For identification of metabolites, Progenesis QI was employed to de-convolute the low energy molecular ions and high energy spectra and to obtain the mass fragments of corresponding molecular ions. Both information on molecular ion and mass spectra were processed and used for search against available databases such as Chemspider and FoodDB (http://foodb.ca/spectra/ms/search). The search parameters in both databases were set for MS and MS/MS mass tolerance as ±5.0 ppm respectively. Combining efforts from MS/MS fragmentation spectra, accurate mass, isotope analysis, and searching of publicly available databases were applied. The elemental composition calculator of MassLynx 4.2 was also used to determine the elemental composition of compounds. Molecular formulas were determined using accurate mass and isotopic pattern. After manual inspection, only peaks with a Progenesis QI score >39 and fragment score and variation error below ±5.0 ppm were considered as tentative identifications. Authentic standards were used to verify some of the identified features and retention times. The metabolites were annotated according to the Metabolomics Standard Initiative guidelines (Sumner, Amberg, & Barrett, 2007).

Statistical analysis

The output of the Progenesis QI results were exported to excel spreadsheets and further analyzed through GenStat with p value at 0.05. The trees were used as random effects while the Cv (Red Delicious, Cortland) × Set (Harv, Storage) × Treatment (Ctrl, DPA and 1-MCP) × Store (Feb, May) were used as the fixed effects for the ANOVA. A corresponding PCA analysis was also conducted using Progenesis QI. Pathway analysis was conducted using default settings in XCMC online (Gowda et al., 2014, Tautenhahn et al., 2012).

Results

Development of superficial scald disorder in apples and their response to DPA and 1-MCP treatments

Development of superficial scald was evaluated at day 0 and 7 after 4 and 7 months storage and reported previously (Du, Song, Campbell Palmer, Fillmore, & Zhang, 2017). Both cultivars, ‘Cortland’ and ‘Red Delicious’ showed significant scald development in control fruit after 4 and 7 month storage. Development of scald was reduced by DPA and 1-MCP treatments with scald index below 0.2 for both cultivars. No significant difference on scald index between 1-MCP and DPA was found (Supplementary information Fig. S1 with permission license no. 4725350449434).

Identification of the major phenolic compounds in ‘Cortland’ and ‘Red Delicious’

Using an untargeted LC/MS analysis as an unbiased procedure, the relative abundance of hundreds of both known and unknown metabolites in apple fruit were obtained. Quality control of the analysis was conducted using the 12 reference (known) compounds as reference standards with 7 concentrations, which presented the product ions and corresponding mass spectra and retention time with good reproducibility (data not shown). In this present study, total of 833 features/compounds with molecular ions and corresponding MS/MS fragments with retention time were obtained from negative mode (Supplementary Table S1). The present chromatographic condition resolves good quality ion chromatography of all major feature compounds and provides quantitative information. Among the 833 total features, 69 of them were tentatively identified (Table 1).

Table 1

Significantly changed phenolic compounds putatively identified in ‘Cortland’ and ‘Red delicious’ during 7 months storage and in response to DPA and 1-MCP treatments employing untargeted qTOF LC/MS.

Peak No.	Compounds Rt_m/z#	Adducts	Formula	Score	Mass Error (ppm)	Description	P1*	P2**	P3***
1	2.23_295.0663	M + Na-2H	C9H14N4O6	47.6	1.22	N4-carboxy-1-(beta-d-ribofuranosyl)-1H-imidazole-4,5-diamine	0.001	ns	<0.001
2	2.67_329.0871	M−H	C14H18O9	44.7	−2.28	1-O-vanilloyl-beta-d-glucose	0.007	ns	0.002
3	3.00_467.1192	M + Na-2H	C19H26O12	53.1	4.69	gaultherin	0.017	0.008	ns
4	3.03_137.0235	M−H	C5H4N3O2-	53.4	−0.54	2-amino-4-pyrimidinecarboxylate	0.004	ns	0.002
5	3.03_299.0765	M−H	C13H16O8	42.3	−2.47	4-hydroxybenzoyl glucose	0.002	ns	<0.001
6	3.29_575.1190	M−H2O−H	C30H26O13	49.9	−0.74	(+)-gallocatechin-(4alpha->8)-(−)-epicatechin	0.014	0.004	ns
7	3.54_368.0978	M−H	C16H19NO9	39.0	−2.43	[3-(hexopyranosyloxy)-2-oxo-2,3-dihydro-1H-indol-3-yl]acetic acid	0.009	ns	0.002
8	3.58_329.0870	M−H	C14H18O9	43.0	−2.47	vanillic Acid 4-β-d-glucopyranoside	<0.001	ns	<0.001
9	3.58_167.0341	M−H2O−H	C8H10O5	46.3	−4.84	2-hydroxy-2-(2-methylenecyclopropyl)succinic acid	<0.001	0.049	<0.001
10	3.65_162.0520n	M−H2O−H, M−H	C6H10O5	47.0	−4.98	l-fucono-1,5-lactone	0.004	0.031	0.009
11	3.66_220.0606	M−H	C11H11NO4	43.6	−4.21	2,2-dimethyl-5-(2(1H)-pyridinylidene)-1,3-dioxane-4,6-dione	<0.001	<0.001	<0.001
12	3.99_299.0765	M−H	C13H16O8	47.1	−2.56	1-O-(2-Hydroxybenzoyl)-beta-d-glucopyranose	0.002	0.001	0.092
13	4.16_451.1237	M−H	C21H24O11	45.4	−2.01	aspalathin	0.008	ns	0.006
14	4.26_449.1084	M−H	C21H22O11	43.9	−1.21	dihydroquercetin 3-rhamnoside	0.009	0.004	ns
15	4.50_577.1417	M−H	C30H26O12	52.5	−1.21	procyanidin B1	ns	ns	ns
16	4.61_481.1344	M + Na-2H	C20H28O12	44.1	3.57	paeonolide	0.045	0.015	ns
17	4.73_355.1024	M−H	C16H20O9	40.5	−2.96	1-O-feruloyl-beta-d-glucose	<0.001	ns	<0.001
18	4.75_466.1108n	M−H2O−H, M−H, M + Na-2H	C21H22O12	54.0	−0.64	plantagoside	0.005	0.001	ns
19	4.91_327.1077	M−H2O−H	C15H22O9	43.8	−2.29	deutzioside	0.062	ns	0.020
20	4.99_307.1026	M−H2O−H	C12H22O10	47.4	−2.68	neohesperidose	<0.001	<0.001	<0.001
21	5.15_354.0942n	M−H, M + Na-2H	C16H18O9	46.7	−2.42	chlorogenic acid	ns	ns	ns
22	5.16_245.0811	M−H	C14H14O4	42.6	−3.48	benzyl 1-hydroxy-6-oxo-2-cyclohexene-1-carboxylate	0.058	ns	0.031
23	5.16_289.0710	M−H	C15H14O6	41.4	−2.63	catechin	ns	ns	0.048
24	5.41_465.1033	M−H	C21H22O12	46.8	−1.09	epi-catechin-3′-glucoronide	<0.001	<0.001	ns
25	5.44_865.1977	M−H	C45H38O18	52.6	−0.98	arecatannin B1	0.023	ns	0.011
26	5.51_578.1418n	M−H, M + Na-2H	C30H26O12	53.5	−1.05	procyanidin B2	0.002	ns	<0.001
27	5.58_263.1128	M + Na-2H	C10H18N4O3	39.4	1.08	N′-[(2Z,3E)-3-(Hydroxyimino)-2-butanylidene]-2-(4-morpholinyl)acetohydrazide	0.015	0.040	0.034
28	5.67_436.0999n	M−H2O−H, M−H	C20H20O11	39.8	−1.60	Swertianolin	<0.001	<0.001	ns
29	5.74_425.1658	M + Na-2H	C16H28N4O8	40.9	0.94	1,4,13,16-tetraoxa-7,10,19,22-tetraazacyclotetracosane-6,9,20,23-tetrone	0.018	0.027	0.066
30	5.77_193.0497	M−H2O−H	C10H12O5	52.7	−4.17	3,5-dimeoxy-4-dimethoxy-4-hydroxyphenylacetic acid	0.046	ns	0.014
31	5.77_355.1026	M−H2O−H	C16H20O9	49.2	−2.43	gardoside	0.021	ns	0.006
32	5.94_320.0525n	M−H2O−H M−H	C15H12O8	41.0	−2.33	(2R)-2-(3,4-Dihydroxyphenyl)-3,5,7,8-tetrahydroxy-2,3-dihydro-4H-chromen-4-one	0.002	<0.001	ns
33	5.96_179.0345	M−H	C9H8O4	41.2	−2.76	caffeic acid	ns	ns	ns
34	5.96_245.0811	M−H	C14H14O4	47.4	−3.54	marmesin	0.024	ns	0.020
35	5.96_290.0783n	M−H, M + Na-2H	C15H14O6	42.0	−2.59	(+)-epicatechin	0.020	0.068	0.029
36	5.96_579.1502	M−H	C30H28O12	52.9	−1.12	5,7-dihydroxy-2-(4-hydroxy-2,5-cyclohexadien-1-ylidene)-2H-chromen-3-yl 6-O-[(2E)-3-(4-hydroxyphenyl)-2-propenoyl]-d-glucopyranoside	0.040	0.057	0.085
37	6.08_381.1758	M−H	C16H30O10	42.2	−2.12	dimethyl 3,6,9,12,15,18-hexaoxaicosane-1,20-dioate	0.007	ns	0.002
38	6.13_866.2057n	M−H, M + Na-2H	C45H38O18	50.1	−0.18	procyanidin C1	0.024	ns	0.010
39	6.16_173.0446	M−H2O−H	C7H12O6	45.2	−4.99	quinic acid	0.033	ns	0.010
40	6.16_337.0921	M−H	C16H18O8	50.8	−2.23	3-O-p-Coumaroylquinic acid	0.047	ns	0.014
41	6.40_381.1758	M−H	C16H30O10	40.6	−2.12	dimethyl 3,6,9,12,15,18-hexaoxaicosane-1,20-dioate	0.010	ns	0.004
42	6.57_329.0870	M−H	C14H18O9	39.0	−2.52	hydroxytyrosol 3-O-β-d-glucuronide	0.079	0.034	ns
43	6.69_395.1915	M + Na-2H	C16H30N4O6	46.2	0.78	glu-val-lys	0.011	ns	0.005
44	6.69_441.1971	M + Na-2H	C18H28N8O4	49.4	−2.16	his-his-lys	0.005	ns	0.002
45	6.88_319.0451	M−H	C15H12O8	47.0	−2.69	dihydromyricetin	0.036	0.011	ns
46	7.12_609.1458	M−H	C27H30O16	39.6	−0.56	rutin	<0.001	<0.001	<0.001
47	7.13_189.0760	M−H	C8H14O5	39.5	−4.23	1,4,7,10-Tetraoxacyclododecan-2-one	0.037	ns	0.041
48	7.16_475.1816	M−H	C21H32O12	43.7	−0.98	kanokoside A	0.040	ns	0.013
49	7.29_385.1133	M−H2O−H	C17H24O11	40.1	−1.88	gardenoside	0.004	ns	<0.001
50	7.36_463.0877	M−H	C21H20O12	49.2	−1.14	isoquercetin	ns	ns	ns
51	7.90_547.2392	M + Na-2H	C23H42O13	41.7	3.79	methyl 2,3,4-tri-O-methyl-beta-d-xylopyranosyl-(1->4)-2,3-di-O-methyl-beta-d-xylopyranosyl-(1->4)-2,3-di-O-methyl-beta-d-xylopyranoside	0.033	ns	0.016
52	7.90_623.1614	M−H	C28H32O16	43.2	−0.58	isorhamnetin 3-O-rutinoside	<0.001	0.078	<0.001
53	8.02_301.0342	M−H	C20H18O11	51.1	−4.01	quercetin	0.01	0.003	ns
54	8.09_353.1809	M + Na-2H	C14H28N4O5	41.4	0.81	ser-val-lys	0.002	ns	<0.001
55	8.18_289.0710	M−H	C15H14O6	42.1	−2.63	(+)-leucopelargonidin	0.028	0.015	ns
56	8.18_452.1307	M−H	C21H24O11	40.5	−2.67	3-hydroxyphloretin 2′-O-glucoside	0.045	0.029	ns
57	8.20_447.0928	M−H	C21H20O11	40.1	−1.15	quercitrin	ns	ns	ns
58	8.25_567.1716	M−H	C26H32O14	45.5	−0.58	phloretin xylosyl-galactoside	0.026	ns	0.009
59	8.27_477.1035	M−H	C22H22O12	52.7	−0.72	isorhamnetin 3-glucoside	0.003	ns	0.001
60	8.41_333.0609	M−H	C16H14O8	46.5	−2.09	2-(3,4-dihydroxy-5-methoxyphenyl)-3,5,7-trihydroxy-2,3-dihydro-4H-chromen-4-one	0.012	0.003	ns
61	8.52_131.0704	M−H	C4H10N3O2-	41.1	−0.97	1,1-diethyl-3-oxo-2-triazanolate	<0.001	<0.001	ns
62	8.82_317.1962	M−H	C16H30O6	39.8	−2.37	(+)-neomenthyl beta-d-glucoside	<0.001	0.076	<0.001
63	8.93_373.1284	M + Na-2H	C18H24O7	44.1	4.47	3-hydroxy-1-(8-hydroxy-6-methoxy-3-methyl-1-oxo-3,4-dihydro-1H-isochromen-7-yl)-3-methyl-2-butanyl acetate	0.066	ns	0.024
64	9.00_435.1291	M−H	C21H24O10	55.3	−1.17	phloridzin	ns	ns	ns
65	9.02_167.0341	M−H2O−H	C8H10O5	46.0	−4.69	7-oxabicyclo[2.2.1]heptane-2,3-dicarboxylic acid	0.006	ns	0.002
66	9.14_413.1447	M−H2O−H	C19H28O11	41.7	−1.48	zizybeoside I	<0.001	ns	<0.001
67	9.22_566.1631n	M−H, M + Na-2H	C26H30O14	42.4	−0.79	3-(5-hydroxy-3,6,7-trimethoxy-4-oxo-4H-chromen-2-yl)-2,6-dimethoxyphenyl beta-d-glucopyranoside	<0.001	<0.001	0.002
68	10.58_214.1439	M−H	C11H21NO3	49.7	−4.32	propyl N-methyl-N-pentanoylglycinate	0.023	0.015	ns
69	10.83_409.1498	M + Na-2H	C18H28O9	42.1	4.55	12-hydroxyjasmonic acid glucoside	0.026	ns	0.025

* P1 value (Cv × Set × Treat × Store).

**P2 value (PfCv.Set.Ctrl vs D,M.fStore).

***P3 value (DPA vs.1-MCP).

Compounds in bold are confirmed with reference standards.

Quantitative changes of the major phenolic compounds in ‘Cortland’ and ‘Red Delicious’

Based on the ANOVA, among the 833 features monitored, a group of 160 features were found to be significantly changed during this study with p value <0.05 (Supplementary Table S1). Among them, 59 were tentatively identified (Table 1). Results indicate that most of them (20) were changed during storage and also either increased or decreased in response to DPA and 1-MCP treatments. Among the changed features, thirty five features changed significantly during storage but showed no significant changes in response to DPA and 1-MCP treatment. In another group, ten features showed significant changes during storage with significant response to DPA and 1-MCP treatment, as well as different response to DPA and 1-MCP treatment (Table 1).

Quantitative changes of the major phenolic compounds in ‘Cortland’ and ‘Red Delicious’ in response to DPA and 1-MCP treatment

A corresponding PCA analysis on the compounds that changed significantly revealed an important relationship among the compounds during storage from ‘Cortland’ and ‘Red Delicious’ as well as in response to DPA and 1-MCP treatments (Fig. 1A). Most of them are shown on the upper part of the plot and closely related to scald at day 0 and day 7. Features shown at the bottom of the plot are closely related to initial stage and in response to 1-MCP treatment. A subset of PCA plot from Fig. 1A provides more detail on some of the putatively identified features that have a close relationship with scald development (Fig. 1B).

Fig. 1A

PCA plot of significantly changed metabolites in ‘Cortland’ and ‘Red Delicious’ apples during storage and in relation to scald development as well as response to DPA and 1-MCP treatments. Data Cort for ‘Cortland’; RDel for ‘Red Delicious’; Ctrl for control, Init for initial at harvest; Feb for at 4 month and May for 7 month storage.

Fig. 1B

PCA plot (zoomed from Fig. 1A) of significantly changed metabolites in ‘Cortland’ and ‘Red Delicious’ apples during storage and in relation to scald development as well as response to DPA and 1-MCP treatments. Features were tentatively identified with numbers indicated in Table 1. 2: 1-O-vanilloyl-beta-d-glucose; 4: 2-Amino-4-pyrimidinecarboxylate; 5:1-O-(4-Hydroxybenzoyl)-beta-D-glucopyranose; 8: 2-Hydroxy-2-(2-methylenecyclopropyl)succinic acid; 9: Vanillic Acid 4-β-d-glucopyranoside; 11: 2,2-Dimethyl-5-(2(1H)-pyridinylidene)-1,3-dioxane-4,6-dione; .12: 1-O-(2-Hydroxybenzoyl)-beta-d-glucopyranose; 26: Procyanidin B2; 41: Dimethyl 3,6,9,12,15,18-hexaoxaicosane-1,20-dioate; 43: glu-val-lys; 44: His-his-lys; 46: Rutin; 47: 1,4,7,10-Tetraoxacyclododecan-2-one; 58: Phloretin xylosyl-galactoside; 62: (+)-Neomenthyl beta-d-glucoside. Data Cort for ‘Cortland’; ‘RedDel for Red Delicious’; Init for initial at harvest; Feb for at 4 month and May for 7 month storage.

Among the compounds which significantly increased during storage, further analysis demonstrated that a group of features were significantly reduced by DPA and 1-MCP treatments (Table 1). Our analysis identified tentatively 17 features significantly changed during storage and in response to treatments of DPA and 1-MCP, while no difference between DPA and 1-MCP was found (Table 1 and Figs. 2, 3). A group of 8 compounds which were not only significantly changed during storage, but also increased in response to DPA and 1-MCP treatments were identified tentatively as compounds: gaultherin; (+)-gallocatechin-(4alpha->8)-(−)-epicatechin, 1-O-(2-hydroxybenzoyl)-beta-d-glucopyranose, paeonolide, epicatechin-3′-glucuronide, (2R)-2-(3,4-dihydroxyphenyl)-3,5,7,8-tetrahydroxy-2,3-dihydro-4 h-chromen-4-one, 3-hydroxyphloretin 2′-O-glucoside and Propyl N-methyl-N-pentanoylglycinate (Fig. 2). A second group of compounds, which were significantly decreased in response to DPA and 1-MCP treatments (Fig. 3) were tentatively identified as: dihydroquercetin 3-rhamnoside, plantagoside, swertianolin, 1,4,13,16-Tetraoxa-7,10,19,22-tetraazacyclotetracosane-6,9,20,23-tetrone, 5,7-dihydroxy-2-(4-hydroxy-2,5-cyclohexadien-1-ylidene)-2H-chromen-3-yl 6-O-[(2E)-3-(4-hydroxyphenyl)-2-propenoyl]-d-glucopyranoside, dihydromyricetin, (+)-leucopelargonidin, 2-(3,4-Dihydroxy-5-methoxyphenyl)-3,5,7-trihydroxy-2,3-dihydro-4H-chromen-4-one and 1,1-diethyl-3-oxo-2-triazanolate.

Fig. 2

Significantly changed metabolites during 7 month storage and increased in response to DPA and 1-MCP treatments. Features were tentatively identified and listed in Table 1. A: Gaultherin; B: (+)-gallocatechin-(4alpha->8)-(−)-epicatechin; C: 1-O-(2-Hydroxybenzoyl)-beta-d-glucopyranos; D: Paeonolide; E: Epi-catechin-3′-glucuronide; F: (2R)-2-(3,4-Dihydroxyphenyl)-3,5,7,8-tetrahydroxy-2,3-dihydro-4H-chromen-4-one. Error bars indicate standard error of means.

Fig. 3

Significantly changed metabolites during 7 month storage and decreased in response to DPA and 1-MCP treatments. Features were tentatively identified and listed in Table 1. A: Dihydroquercetin 3-rhamnoside; B: Plantagoside; C: Swertianolin; D: 1,4,13,16-Tetraoxa-7,10,19,22-tetraazacyclotetracosane-6,9,20,23-tetrone; E: 5,7-Dihydroxy-2-(4-hydroxy-2,5-cyclohexadien-1-ylidene)-2H-chromen-3-yl6-O-[(2E)-3-(4-hydroxyphenyl)-2-propenoyl]-d-glucopyranoside; F: Dihydromyricetin; G: (+)-Leucopelargonidin; H: 2-(3,4-Dihydroxy-5-methoxyphenyl)-3,5,7-trihydroxy-2,3-dihydro-4H-chromen-4-one; I: 1,1-Diethyl-3-oxo-2-triazanolate. Error bars indicate standard error of means.

Another group of features, which were significantly changed during storage, showed significant responses to DPA and 1-MCP treatment as well as differential response to DPA and 1-MCP treatment, and were identified as: 2-hydroxy-2-(2-methylenecyclopropyl)succinic acid; l-fucono-1,5-lactone; neohesperidose; beta-d-galactofuranose; rutin; 3-(5-hydroxy-3,6,7-trimethoxy-4-oxo-4H-chromen-2-yl)-2,6-dimethoxyphenyl beta-d-glucopyranoside, as shown in Table 1 and Fig. 4.

Fig. 4

Significantly changed metabolites during 7 month storage and in response to DPA and 1-MCP treatments. A: 2-Hydroxy-2-(2-methylenecyclopropyl)succinic acid; B: l-Fucono-1,5-lactone; C: Neohesperidose; D: beta-d-Galactofuranose; E: Rutin; F: 3-(5-Hydroxy-3,6,7-trimethoxy-4-oxo-4H-chromen-2-yl)-2,6-dimethoxyphenyl beta-d-glucopyranoside. Error bars indicate standard error of means.

Considering the scald development in control fruit after 7 months storage as compared with DPA and 1-MCP treatment, those compounds that changed significantly during storage and were reduced or increased by DPA and 1-MCP treatments, would be relevant to reveal the scald development and control mechanisms of control treatments.

Biological pathways classifications of metabolomics changes

Based on the analysis with XCMS using Arabidopsis as a model plant (Rinehart et al., 2014), compounds significantly changed and identified as potential metabolites associated with scald development and in response to DPA and 1-MCP treatments were found. They are involved in metabolism of sorbitol biosynthesis II, free phenylpropanoid acid biosynthesis, salicylate glucosides biosynthesis III (signaling compounds), coumarin biosynthesis (via 2-coumarate), leucopelargonidin and leucocyanidin biosynthesis, suberin monomers biosynthesis and gallate biosynthesis. These are therefore potential metabolic pathways contributing to the development of scald and its control mechanism (Fig. 5).

Fig. 5

Metabolic pathways cloud plot for significantly changed metabolites (p < 0.01) in ‘Cortland’ and ‘Red Delicious’ apples during 7 month storage and in relation to scald development as well as in response to DPA and 1-MCP treatments. The radius of each circle represents the number of metabolites relative to the number of metabolites represented by other circles. Darker circles mean more pathways are represented. A corresponding table is also provided as Supplementary data (Supporting data) for the name of the information (Gowda et al., 2014, Tautenhahn et al., 2012).

Discussion

Metabolomics is a comprehensive study of small molecules which undergo chemical transformation during metabolism of biological samples. Liquid chromatography–mass spectrometry (LC/MS) has become one of the more successful methods suitable for cell metabolomics studies due to high sensitivity, specificity, and reproducibility. Two major data acquisition strategies, namely data dependent acquisition (DDA) and data independent acquisition (DIA) have been widely developed. Data dependent acquisition (DDA) obtains fragmenting ions based on the priority of ion intensity. The drawback with DDA is that not all low intensity ions can be fragmented and generated into a feature to be quantified, despite that they may be biologically important (Schrimpe-Rutledge, Codreanu, Sherrod, & McLean, 2016). Unlike data dependent acquisition (DDA), data independent acquisitions (DIA) operates and measures fragment masses for all detected precursor ions independent of intensity, resulting in improved detection of features. Therefore, LC/MS metabolomics with DIA provides a better and unbiased way to identify metabolites in any biological sample (Li, Vissers, Silva, Golick, Gorenstein, & Geromanos, 2009). Different manufacturers have developed different DIA strategies. MSE is one of the examples of DIA, which ramps collision energy to fragment all ions to obtain a convoluted product ion spectrum that includes all initial ions. Specific software algorithms have to used to process these data and de-convolute the spectra into individual features (Li et al., 2009).

In present study, we applied the MSE (DIA) approach and investigated as well as quantified group of metabolites under the extraction and chromatographic conditions in apple fruit. Among the features detected, 69 were tentatively identified. The most changed metabolites which increased significantly in control after 4 and 7 months storage and were reduced by DPA and 1-MCP respectively, revealed potential correlations with scald development and control mechanism.

With the aid of reference standards, features such as caffeic acid, phloretin, catechin, epicatechin, chlorogenic acid, procyanidin B2 and C1, phlorizin, quercetin, quercitrin, rutin and avicularin (quercetin-3-o-arabinoside) were identified and quantified. Quality control of reference compounds of precursor ions, fragments and retention time indicated that the present LC/MS protocol confirms and validates the current study dataset with good matching and reproducibility. Among them, epicatechin, procyanidin B2 and C1, quercetin and rutin showed significant changes in both ‘Cortland’ and ‘Red Delicious’ apples. Amongst other significantly changed compounds, 30 identified and quantified compounds were generally reduced by DPA and 1-MCP, plus another 6 compounds showed a differential response between DPA and 1-MCP treatment. These compounds were mainly phenolics, amino acids and short-chain terpenes (Table 1).

Employing the DIA approach in this study, additional groups of compounds were also quantified and tentatively identified to be linked with scald development and negatively regulated by DPA and 1-MCP treatment (Figs. 1–3 and Table 1). From the identified compounds, seventeen compounds were not only significantly changed during storage, but also increased or reduced differently in response to DPA and 1-MCP treatments (Fig. 2, Fig. 3, Fig. 4). Most of these compounds are associated with metabolism of phenolic compounds in apple peel tissue. Those identified compounds are shown in Table 1.

Our study also revealed a group of compounds that were increased during storage and were further enhanced by DPA and 1-MCP. The biological functions of these compounds are not well known, but this group of compounds seems to serve as some protection to prevent the tissue damages. For example, gaultherin is known for a natural salicylate derivative and together with other compounds like (+)-gallocatechin-(4alpha->8)-(−)-epicatechin, belongs to the class of organic compounds known as biflavonoids and polyflavonoids. 1-O-(2-hydroxybenzoyl)-beta-d-glucopyranos, paeonolide, epi-catechin-3′-glucuronide, a potent antioxidant, and (2R)-2-(3,4-dihydroxyphenyl)-3,5,7,8-tetrahydroxy-2,3-dihydro-4H-chromen-4-one may serve as antioxidants to reduce the incidence or severity of scald.

In contrast, another group of compounds which increased during storage and reduced by DPA and 1-MCP treatment were also identified (Fig. 3A–C). Most of these compounds are involved various stages of metabolism of phenolic compounds. Astilbin is a flavanonol, a type of flavonoid compound found in grapes, which has a role as a radical scavenger, an anti-inflammatory agent. Paeonolide (astragalin 7-glucoside) is a plant glycoside that contains a non-reducing end α-l-arabinopyranoside and is found in broccoli. Plantagoside also known as (2″,3″-di-O-p-coumaroylafzelin) is a phenolic glycosides found in many plant species (http://foodb.ca/spectra/ms/search). Quinic acid and shikimic acid are key intermediates in the biosynthesis of aromatic compounds in living systems. Procyanidin B2 is a proanthocyanidin consisting of two molecules of (−)-epicatechin joined by a bond between positions 4 and 8′ in a β-configuration. Procyanidin C1, also known as cinnamtannin A1 or procyanidol C1, as well as (±)-dihydromyricetin belong to the class of organic compounds known as biflavonoids and polyflavonoids. Leucoanthocyanidins is a colourless chemical compound belonging to the family of leucoanthocyanidins. Although it is not possible to link any of these compounds yet with browning color formation, these results support the possibility that the phenolic compounds may contribute to the browning color formation.

It was reported by Busatto et al. (2014) that a significant accumulation of three specific phenolic compounds in ‘Granny Smith’ apple: vanillin, trans-piceid and catechin were dominant in control fruit after 2 months storage as compared to initial harvest, while a further 7 days of control resulted in further accumulation of chlorogenic acid and isomers, cis-piceid, epicatechin, procyanidin (B1, B2 and B4) and quercetin-3-rhamnose, with a close relationship to the severity of scald symptoms. In addition, control at day 7 plus 1-MCP showed much lower levels of phenols characterized by phlorizin, rutin and quercetin-3-glc (Busatto et al., 2018). All these changes result in destabilization of plastid membrane by oxidative stress that lead to tissue browning, while phenolic antioxidants may quenched that process. Our current results showed neither significant increase of chlorogenic acid or decrease of phlorizin and quercetin-3-glucoside. Rather some significant changes in epicatechin, procyanidin B2 and procyanidin C1 were found to be increased during storage, but not affected by DPA and 1-MCP treatments, although they responded to treatments differently (Table 1). The biosynthesis of chlorogenic acid and isomers requires p-coumaric acid as precursor. Gene expression study indicated the increasing transcripts of p-cinnamate 3-hydrolase (C3H) and 4-coumarate-CoA ligase (4CL) are associated with biosynthesis of chlorogenic acid and isomers (Hoffmann et al., 2004 ). Our previous proteomic study reported that the increase of 4-coumarate-CoA ligase-like (4CL) protein was associated with scald development and is reduced by DPA and 1-MCP treatments (Du et al., 2017). Since 4-coumarate-CoA ligase-like (4CL) converts not only 4-courmarate into 4-courmaroyl CoA but also vanillic acid into vanilloyl acid as well as ferulic acid into feruly CoA, it is possible that 4CL could regulates the substrates used in downstream browning color development in the scald tissue (Du et al., 2017). The identification and quantification of 1-O-vanilloyl-beta-d-glucose and its isomers as well as 1-O-feruloyl-beta-d-glucose supports this possibility. Further more, a temporal increase of neohesperidose, beta-d-galactofuranose, rutin, and phloretin xylosyl-galactoside during storage at 4 month and decrease in response to the treatment of 1-MCP was also found (Fig. 4).

The difference between our studies and Busatto’s study seemed to be in differences in cultivars as well as storage time. In addition, we also investigated the effect of treatments of DPA and 1-MCP, while Busatto’s work applied 1-MCP only. Mixed results have also been reported so far on the consequence of 1-MCP treatment on metabolism of phenolic compounds in apples during storage. For example, no significant changes in the individual phenolic compound was found in ‘Cripps Pink’ apple after 1-MCP treatment. Among the some phenolic compounds identified, 1-MCP treatment seemed to increase only gallic acid in the peel tissue (Hoang, Golding, & Wilkes, 2011). No significant change in phenolic compounds, including epi-catechin, catechin, chlorogenic acid, and phloridzin, was found during storage with 1-MCP treatment in ‘Empire’ apple (Lee, Rudell, Davies, & Watkins, 2012). The concentration of phenolic compounds in apple peel from different varieties also showed the significant difference among the varieties. ‘Granny Smith’ showed a lower level in chlorogenic acid than ‘Pink Lady’. ‘Fuji’ and Golden (Spain), also showed the least amount of phloridzin (Alarcón-Flores, Romero-González, Martínez Vidal, & Garrido Frenich, 2015). The concentration of polyphenolics varied among the apple cultivars and growing regions can also cause significant differences as well (McGhie, Hunt, & Barnett, 2005). Furthermore, when apple cultivars were compared, ‘Granny Smith’, as a white flesh and green peel cultivar, has lower phenolic compound content overall, but higher total flavan-3-ols content in flesh and total flavonols in peel (Alarcón-Flores et al., 2015, Bars-Cortina et al., 2017). Epicatechin and its polymerized forms are the main flavan-3-ols detected in apple peels (Bars-Cortina, Macià, Iglesias, Romero, & Motilva, 2017). It is well know that the competitive nature of substrates under the subclasses of flavonoids pathway can lead to the divergence of the metabolism of final subclass compounds (Henry-Kirk, McGhie, Andre, Hellens, & Allan, 2012). There has been no direct evidence yet to prove that flavan-3-ols compounds are responsible for brown color formation in apple scald. Recently, a study demonstrated that some of the scald associated metabolites such as ethylene, α-farnesene and 6-methy-5 heptene-2-one in ‘Granny Smith’ apples are is not directly related to scald severity rather, to onset and progression of scald severity (Mditshwa, Fawole, Vries, Van Der Merwe, Crouch, & Opara, 2016). It was reported that the increase of the concentration of caffeoyl-d-quinic acid and catechin in infected fruit and subsequent oxidation to ortho-quinones is the response of the plums to pathogen infection, the effectiveness of this defence response is highly dependent o-quionone formation respectively PPO activity (Fuchs & Spiteller, 1998). Therefore, the brown color formation in damaged peel in ‘Cortland’ and ‘Red Delicious’ apples may be involved by more complicated group of compounds which requires further investigations. The potential involvement of laccase and correlation of chlorogenic acid isomers rather than chlorogenic acid itself may also be important in scald development in apples was also hypothesized. Further evidence and investigation on the metabolism between p-coumarate and chlorogenic acid and isomers as well as the metabolism of the monomeric catechin and epicatechin to polymeric proanthocyanidins and their links to scald development seem to be necessary to reveal the mechanism of browning process of scald development (Gong et al., 2018).

Despite the robust analysis and wide coverage of metabolites of LC/MS analysis, one of the challenges that remain is that not all features/compounds can successfully be identified due to lack of reference standards and MS/MS library for metabolites in apples or fruit in general, which is a common bottleneck in untargeted metabolomics studies (Chaleckis et al., 2019, Schrimpe-Rutledge et al., 2016). Nevertheless, our study reveals new insights into biological development and control of apple scald. It has also made possible links in combination with orthogonal technologies, such as genomics and proteomics studies from previous reports. Thus, this study provides not only prospects of metabolomics information, but also its integration into systems biology on apple scald development and control.

Conclusion

To gain better insight on superficial scald development and control, an untargeted metabolomic analysis employing UPLC–LC/MS was conducted to quantify the metabolomic changes of apple scald tissues as well as tissues in response to DPA and1-MCP after storage. It is demonstrated that there are significant changes in various groups of metabolites such as terpenes and phenolic compounds during storage as well as in response to treatments. The changes in metabolomics composition observed between control and treatment could be used to suggest the involvement of metabolic pathways to the development of the disorder and the mechanism of inhibitory treatments by DPA and 1-MCP. Among the metabolites with significant changes, metabolites were identified could serve as targets for metabolic profiling or biological markers to be monitored over the storage. Findings from this study may lead to new insight into metabolic processes related to superficial scald, the development of physiological markers for monitoring and control.

Declaration of Competing 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.