2022 Volume 10 Pages 56-67
Recently, metabolomics has grown rapidly in the fields of food and agriculture. Complex physiological changes after harvest prompted the development of a new metabolomic analytical method offering more profound insight into these changes. This review presents the feasibility of a metabolomics approach to elucidate physiological changes during ripening, senescence, and disorders. Additionally, we introduce metabolomics for the authentication of agricultural products. Confirmation of species, varieties, and geographical origin via metabolomics can be useful to tackle adulteration and certify quality.
Metabolomics studies biological systems, focusing on metabolites, a broad range of small molecules [1]. This study can provide information about the cellular state of living things. Generally, metabolomics elucidates chemical and biochemical processes in biological samples and discovers biomarkers through profiling metabolites. Recently, agricultural research has utilized metabolomics to understand the physiological and biochemical changes under abiotic or biotic stress conditions as introduced in many review articles [2, 3, 4]. Food research also uses it to assure food quality, monitor food safety, and traceability [5, 6, 7].
“Postharvest” is the handling of agricultural products such as fruits, vegetables, legumes, spices, and nuts after they are separated from the parent plant. During this stage, they continue metabolic activity associated with physiological and biochemical changes. To evaluate these changes, respiration, ethylene production, transpiration, total soluble solids, soluble sugars, starch, and pigments have been measured. However, a conventional analysis provides little information on the physiological status of postharvest agricultural products because it covers only limited, specific parameters.
Alternatively, metabolomics produces multitudes of data regarding the physiological and biochemical changes during the lifespan of agricultural products. Understanding postharvest physiological changes leads to improved postharvest technology such as maintaining and assuring the quality of agricultural products [8, 9]. Additionally, discrimination of species, cultivar, and geographic origin verified via metabolic biomarkers will detect counterfeit products and certify quality [10, 11]. In recent years, several researchers have tried to apply the metabolomics approach to clarify the mechanism of ripening, senescence, and disorders [12, 13]. Moreover, it has been used to discern species or cultivars [9] and the geographic origin of the product [14, 15].
Figure 1 summarizes the general workflow and the representative outcomes of metabolomics for postharvest research. The metabolomic routine analysis consists of several steps from the sample extraction to obtain the metabolite of interest from the food matrix, followed by data acquisition using suitable analytical platform [16, 17]. Multivariate data analysis is usually used to better understand the objective of the study. Commonly, mass spectrometry (MS) is coupled to a liquid or gas chromatography for better compound separation. In the postharvest metabolomics the MS system is used as an analytical platform for profiling a large number of metabolites or performing global metabolites fingerprint. Furthermore, it can be used to find metabolic biomarkers and understand the metabolic pathway. The role of the multivariate analysis in metabolomics is to assist to reach the goal of the study, whether to explore the natural pattern in the spectral biochemical features that may lead to understand the physiological alteration, or to find the significant feature among the variables which is important for marker selection [18].
Here, we aim to provide information about the latest application of metabolic profiling and biomarker discovery in postharvest research. This review presents basic knowledge regarding the benefits of metabolomics in understanding physiological changes and authentication of several agricultural products.
Figure 1: The workflow of metabolomics in postharvest research
After harvesting, fruits and vegetables continue a biological activity. These metabolic changes affect their quality. Especially in climacteric fruits, ripening is important to achieve optimum quality. Because fruits are not of optimum quality at harvesting time, ripening continues in postharvest until fruits are ready to be eaten. In ripening, drastic metabolic changes occur, leading to quality improvement. Senescence is also an important physiological phenomenon affecting quality. Complex metabolic shifting occurs in this stage. Understanding ripening and senescence mechanisms is needed for maintaining and assuring the quality of fresh produce. Metabolomics is one solution and has recently become popular for understanding physiological changes and disorders during ripening and senescence in fresh produce as shown in Table 1.
| Purpose | Fresh produce | Analytical platform | Multivariate Analysis | Biomarkers | Biomarker’s function | Ref. |
|---|---|---|---|---|---|---|
| Ripening stages | Mangosteen | GC-Q | PCA, PLS, HCA | psicose, fructose, xylose, galacturonic acid, and glucose (pericarp) xylose, xylulose, ribulose, glucuronate, 2-aminoisobutyric, ryptophan (flesh) phenylalanine, valine, isoleucine, serine, tyrosine (seeds) |
Distinguishing the ripening level 0 to 6 | [19] |
| Avocado | LC-QTOF GC-TOF |
PCA | glutamic acids, aspartic acids, alanine, galacturonic acid | Characterization of the fast and slow ripening | [20] | |
| Kiwifruit | GC-Q | PCA,PLS-DA | sucrose, myo-inositol, citric acid, malic acid, fructose, glucose, quinic acid | Characterization of natural and artificial ripening | [21] | |
| Pineapple | GC-Q | PCA, OPLS-DA | melezitose, inositol, xylonic acid, gluconic acid, raffinose (flesh) inositol, mannose, galactose, sucrose, asparatic acid (peel) |
Distinguishing the early and late ripening | [22] | |
| Senescence stages | Pear | LC-QqQ | PLS-DA, OPLS-DA | lysoPC, 16:0, 18:1, 18:2, 18:3, lysoPE 16:0, 18:2, MAG 18:2, 18:3, 18:4, punicic acid, 9-hydroxy-(10E, 12Z, 15Z)-octadecatrienoic acid, 4-hydroxysphinganine | Understanding the contribution of metabolites in postharvest softening during senescence | [13] |
| Lulo | GC-Q | PLS-DA | methyl (E)-2-butenoate, 4-heptanone, o-xylene, (Z)-3-hexenyl acetate, hexyl acetate, 3,7-dimethyl-1,6-octadien-3-ol, methyl ester, the alcohols 1-penten-3-ol, (E)-3-hexen-1-ol, pentanal aldehyde, 1,7,7-trumethyl-bicyclo[2.1.1]-heptan-2-one ketone | Evaluating the changes of the VOCs during senescence contributed to lulo aroma | [27] | |
| Disorders during storage | Basil | GC-Q | PCA, PLS | 1,8-cineole, (Z)-β-ocimene, 1-hexanol, (E)-3-hexen-1-ol, 1-octanol, α-guaiene, α-Terpineol, bicyclogermacrene, hexanal, (E)-2-hexanal | Understanding the trend of the volatile compound under CI and early diagnose CI | [32] |
| Mango | LC-Q GC-QqQ |
PCA | galloylquinic acid, gallic acid esters, gallotannins, quercetin 3-O-rhammoside, mangiferin, myositol, linoleic acid, sugar | Characterization of mangoes under the CI condition | [33] |
GC-Q: gas chromatography-quadrupole; LC-Q: liquid chromatography-quadrupole; LC-QTOF: liquid chromatography-quadrupole time-of-flight; GC-TOF: gas chromatography time-of-flight; LC-QqQ: liquid chromatography-triple quadrupole; PCA: principal component analysis; PLS: partial least square; PLS-DA: partial least squares-discriminant analysis; OPLS-DA: orthogonal projections to latent structures modelling discriminant analysis
Uneven ripening at harvest causes delivery of inconsistent quality to consumers. Uniformity attracts consumers and guarantees high quality. To address this problem, researchers have applied metabolomics to understand the ripening mechanism in fruits. Parijadi et al. [19] found that psicose, fructose, xylose, and other 13 compounds become metabolic markers enabling description of the degree of mangosteen ripening by using GC-MS metabolomics. Similarly, Pedreschi et al. [20] applied untargeted multiplatform metabolomics using both LC-MS and GC-MS to reveal the heterogeneity of ripening stages of ‘Hass’ avocado. In their experiment, avocado samples were separated into five clusters based on time to ripeness (9 days, 13 days, 17 days, 20 days, and >22 days). Amino acids such as glutamic acids, aspartic acids, alanine, and galacturonic acid contributed to the ripening of ‘Hass’ avocado. Particularly, glutamic acids, aspartic acids, and alanine were detected between fast and slow ripening clusters while galacturonic acid was only detected in the fastest ripening clusters. This observation indicates that metabolic profiling can elucidate the heterogeneity of ‘Hass’ avocado ripening stages.
In kiwifruit, exogenous ethylene treatment is commonly used as artificial ripening to accelerate the process because these fruits need a longer ripening time. Qualitative metabolite analysis is needed to investigate the differences of quality between normal ripening (NR) and exogenous ripening (ER). This work noted that concentrations of sucrose, myo-inosistol, citric acid, and malic acid were higher in NR than ER fruits and fructose, glucose, and quinic acid were higher in ER fruits. This observation indicated that ethylene treatment in kiwifruit during storage can help the ripening and result in good quality [21]. The potential of metabolomics has also been exhibited in non-climacteric products such as capsicum [12], pineapple [22], and cherry [23]. The common non-climacteric fruits are not ripe after harvesting, but several commodities such as capsicum have a unique ripening behavior. For instance, the field ripening stages of capsicum are classified into seven groups, i.e., green, deep green, breaker, breaker red, bright red, deep red, and deep red + dried. Harvesting at the green and deep green stages results in failure of full ripening. Breaker red stages are the best time to harvest. Capsicum fruits harvested in this stage can develop to the fully red stages during storage [24]. In this regard, Aizat et al. [12] utilized GC-MS and LC-MS to elucidate the unique ripening mechanism in capsicum. GC-MS was used for screening potential markers among the ripening stages, and then, LC-MS was applied to enhance the characterization of selected metabolites. From the metabolic profiling in capsicum, the modification of sugar, amino acids, organic acids, and polyamines was found during ripening. These results highlighted the fundamental role of metabolites in renewing the grading method in non-climacteric products.
Flavor is an essential attribute to evaluate senescence and is a combination of aroma and taste. The shifting of flavor depends on fruit type. Sugars, acids, and volatiles produce and alter fruit flavor [25, 26]. Iguaran et al. [27] conducted a comprehensive volatile analysis to clarify the causes of off flavor during senescence in lulo fruits. In this study, headspace-solid phase microextraction–gas chromatography–mass spectrometry (HS-SPME-GC-MS) was applied to detect the volatile compound contributed to the off flavor of lulo fruits, and 11 volatile compounds were found to be potential markers of off flavor and particularly, methyl ester was dominant in lulo fruit senescence. In citrus fruits, flavor is also used as a reference to assess senescence. Sun et al. [28] applied metabolomic profiling using GC-MS to understand the metabolic role in organic acid changes. Profiling highlighted that succinic acids, γ-aminobutyric acid (GABA), and glutamine increased, whereas 2-oxoglutaric acid decreased during orange senescence. The postharvest phenomenon from ripening to senescence in pear was captured via metabolic profiling. The degradation of 18 species of sugars such as D-fructose-6-phosphate, arabinose, and trehalose 6-phosphate did not affect pear softening during senescence, whereas lysoPC 16:0, 18:1, 18:2, 18:3, lysoPE 16:0, 18:2, MAG 18:2, 18:3, 18:4, punicic acid, 9-hydroxy-(10E, 12Z, 15Z)-octadecatrienoic acid, and 4-hydroxysphinganine involved the softening [13].
Storage temperature affects enzyme activity, which induces metabolite changes in fresh produce. Inadequate postharvest storage conditions not only influence ripening and senescence but also result in physiological disorders. Chilling injury (CI) occurs in chilling sensitive products under low temperatures. CI is caused by membrane damage due to lipid peroxidation. The imbalance of respiratory and other metabolic processes accelerates the occurrence of lipid peroxidation [29, 30]. The common symptoms of CI are pitting, failure of ripening, internal discoloration, water soaking, and browning [31]. Metabolomics gave a satisfying result for investigating the metabolic alteration caused by CI. Cozzolino et al. [32] used GC-MS with HS-SPME to investigate the profile of volatile organic compounds (VOCs) in basil leaves. Principal component analysis (PCA) was utilized for analyzing the dominant effect of the cultivar on VOCs profile and partial least squares regression (PLSR) was applied to identify the marker which had strong relationship to the storage duration. This study found ten potential markers, especially, 1,8-cineole was the most abundant volatile which was considered as a marker to diagnose the early symptoms of CI. Identification of metabolomic changes also provided an understanding of the CI tolerances in mangoes. The galloylquinic acids, gallic acid ester, gallotannins, quercetin 3-O-rhammoside, mangiferin, myositol, linoleic acid, and sugar increased during cold storage which reflect the CI tolerance in mangoes. This understanding provided knowledge about the function of antioxidant and phenolic levels in suppressing cell damage by attacking reactive oxygen species [33]. Apple (var. domestica Borkh. Mansf.) is susceptible to necrosis characterized by superficial scald. This physiological disorder is induced by cold storage. A metabolomics approach identified phytosterol metabolism as responsible when superficial scald occurred [34].
Food adulteration is illegal worldwide but continues even now. Adulteration occurs when the authentic substance is replaced with a cheaper one to increase the volume and weight of the product for financial gain [35]. Because it threatens food reliability, researchers have worked to develop methods to detect food adulteration, especially for expensive products like honey and meat. Honey is popular among consumers because of its health benefits but is expensive, making it a prime target for adulteration. Adding a cheaper sweetener like corn syrup or inverted sugar syrup is a common trick to increase profits [36]. Similarly, minced beef adulterated with other meat, like pork, is often found in the market because of high beef demand [37]. In this case, metabolomics has been used to identify food adulteration [36, 37]
Additionally, the metabolomics method efficiently verifies the true composition from different species, varieties, and origins. It is essential to certify authenticity to protect economic value [11, 38]. Nowadays, metabolomics authentication has been utilized in agricultural products to ensure quality, establish the brand, and avoid false claims of origin, species, and variety. Table 2 and 3 show representative examples of metabolomics used to discriminate or distinguish the agricultural product of different species, varieties, and origins.
| Fresh produce | Analytical platform | Multivariate Analysis | Biomarkers | Biomarker’s function | Ref. |
|---|---|---|---|---|---|
| Wild mushrooms | GC-IT | PCA | 3-octanol, 3-octanone, linalool, 1-octanol, 1-pentanol, (E)-2-octen-1-ol, hexanol, hexanal, (E)-2-octenal, ρ-anisaldehyde, sesquiterpene-like compound. | Distinguishing the wild mushroom from six species | [9] |
| Grape | GC-TOF | PCA, PLS-DA, | serine, phenylalanine, L-homoserine, glutamic acid | Distinguishing the four grape varieties from 120 commercial brands. | [14] |
| Pineapple | GC-Q | PCA, PLSROPLS-DA, | GABA, valine, alanine, sucrose, threonic acid, 5-hydroxytryptamine, serine, methionine | Distinguishing the taste of pineapple from three cultivars | [38] |
| White tea | LC-Q-TOF | PCA, PLS-DA, HCA | theanine, aspartic acid, asparagine, AMP, flavan-3-ols, theasinensisns, procyanidin B3, theobromine. | Characterizing the taste of white tea from three subtypes | [40] |
| Peach | GC-Q | PCA | fructose, glucose, malic acid, citric acid | Distinguishing 15 peach varieties | [41] |
| Mango | GC-Q | PCA, OPLS-DA | nicotinic acid, glutamic acid, aspartic acid, glycine, ribose | Distinguishing five mango varieties | [42] |
| Honey | GC-Q LC-QqQ |
PCA, HCA | leptosin, acetyl-2-hydroxy-4-2-(2-methoxyphenyl)-4-oxobutanate, 3-hydroxy-1-(2-methoxyphenyl)-penta-1,4-dione, kojic acid, 5-methyl-3furancaerboxylic acid | Discrimination of Manuka, Kanuka, and Jelly bush honey for tackling the adulteration | [43] |
GC-IT: gas chromatography-ion-trap; GC-TOF: gas chromatography time-of-flight; LC-Q-TOF: liquid chromatography-quadrupole time-of-flight; GC-Q: gas chromatography-quadrupole; LC-QqQ: liquid chromatography-triple quadrupole; PCA: principal component analysis; PLS: partial least square; PLS-DA: partial least squares-discriminant analysis; OPLS-DA: orthogonal projections to latent structures modelling discriminant analysis; HCA: hierarchical cluster analysis; PLSR: partial least squares regression.
| Fresh produce | Analytical platform | Multivariate Analysis | Biomarkers | Biomarker’s function | Ref. |
|---|---|---|---|---|---|
| Adzuki beans | GC-TOF | OPLS-DA | citric acid, malic acid | Distinguishing adzuki beans from Korea and China | [10] |
| Asian palm civet coffee | GC-Q | PCA, OPLS-DA | citric acid, malic acid, inositol | Differentiating the original Civet coffee and three cultivars of coffee from three region | [44] |
| Coffee beans | GC-Q | PCA | glycerol, glucono-1,5-lactone, gluconic acid, sorbitol, galactitol, galactinol | Distinguishing coffee beans from three regions of Indonesia | [45] |
| Brewed Arabica coffee |
GC-Q | PCA | (4´-hydroxyphenyl)-2-butanone | Distinguishing the aroma of coffee from Ethiopia, Tanzania, and Guatemala | [48] |
| Garlic | GC-Q LC-Q-TOF |
PCA, PLS-DA | (E)-1-Allyl-2-(prop-1-en-1yl) disulfane,1-(2-Methyl-1-cyclopenten-1-yl)-ethanone, 3,4-Dimethylthiophene-2,5-dione, mequinol, 2-Methoxyphenol,1,2-dimethoxybenzene | Distinguishing the Chinese garlic from four origins | [49] |
| Hazelnut | LC-QqQ | PCA-LDA | PC 18:2/18:2, 16:0/18:2, 18:1/18:2, 16:0/18:3, 18:2/18:2, PE 18:2/18:2, 16:0/18:2, 18:1/18:1, DG 18:1/18:1, 18:2/18:2, 16:0/18:1, 16:0/16:1, TG 2:0/18:1/18:2, 2:0/18:2/18:2/18:2, 14:0/16:0/18:1, 15:0/16:0/18:1, 16:0/16:1/18:1, 17:1/18:1/18:2, 18:2/18:2/18:3 | Distinguishing the hazelnut from six countries (Turkey, Italy, Georgia, Spain, France, and Germany) | [50] |
| Black pepper | LC-Q-Orbitrap | PCA, OPLS-DA | reynosin, artabsinolide D, tatridin B | Differentiating black pepper from Brazil, Vietnam, and Sri Lanka | [51] |
| Dates fruit | LC-FT GC-TOF |
PCA,OPLS-DA | ethanolamine, GABA, serotonin, tyramine, tryptamine, phenethylamine, serine, glutamate, tyrosine, tryptophan, phenylalanine, riboflavin, niacin, pyridoxine, nicotinate | Distinguishing dates fruit from 14 countries | [52] |
GC-TOF: gas chromatography time-of-flight; LC-Q-Orbitrap: liquid chromatography couple-quadrupole-Orbitrap; LC-FT: liquid chromatography-Fourier transform; GC-TOF: gas chromatography-time-of-flight; GC-Q: gas chromatography-quadrupole; LC-Q-TOF: liquid chromatography-quadrupole time-of-flight; LC-QqQ: liquid chromatography-triple quadrupole; PCA: principal component analysis; PCA-LDA: linear discriminant analysis based on PCA scores; PLS-DA: partial least squares-discriminant analysis; OPLS-DA: orthogonal projections to latent structures modelling discriminant analysis.
Discrimination by appearance among different species of agricultural products like tea leaves, coffee, and some fruits and vegetables is often challenging. This difficulty leads to adulteration and threatens the validity of a quality guarantee. The quality of several products is characterized by their unique aroma and taste. Commonly, these aromas and tastes are assessed by sensory evaluation [39]. However, this evaluation is inefficient because evaluator training is costly, and an untrained organoleptic evaluator produces subjective results. Metabolomics can overcome this problem by detection and quantification of specific biomarker metabolites that accurately discriminate between species and varieties. Moreover, the metabolic profile aids understanding of product characteristics that are useful to enhance the product brand.
For instance, mushroom is a multipurpose fresh product used for food and pharmaceuticals. Aroma is an essential attribute in mushrooms as each cultivar has a unique aroma correlated with quality. Malheiro et al. [9] used HS-SPME-GC-MS metabolomics to define six species of wild mushrooms on the basis of their volatile compounds. Untargeted analysis was followed by targeting 46 volatiles to discover the discriminant markers. PCA was employed to elucidate the biomarkers that contributed to discriminate the wild mushroom species. From this report, the volatile profile accurately described each species on the basis of their unique aroma. Eleven volatile compounds were fundamental to distinguish mushroom species. For example, 3-octanol and 1-octanol were important for L.nuda species, linalool for T.fracticum, and 3-octanone for H.crustuliniforme.
In tea, taste is an essential attribute for consumers. Taste differences are characterized not only by varieties but also by subtypes. Metabolite diversity affects the specific taste of tea. Yang et al. [40] utilized ultrahigh performance liquid chromatography–time of flight/mass spectrometry (UHPLCQ-TOF/MS) to characterize the taste of different Chinese white tea subtypes, namely, Silver Needle, White Peony, and Shou Mei. Metabolomics identified 99 non-volatiles, and clarified that theanine, aspartic acid, asparagine, and AMP were responsible for the umami taste. Additionally, they found that flavan-3-ols, theasinensins, procyanidin B3, and theobromine were responsible for bitterness and astringency by applying PLSR. Monti et al. [41] conducted metabolic profiling to elucidate the chemical biodiversity of 15 peach varieties at different ripening stages. This study demonstrated that specific metabolites involving the organoleptic and nutritional properties depended on peach variety. Six varieties were characterized by fructose and glucose, and two varieties were characterized by malic and citric acid.
To several consumers, mango has an exotic taste. Worldwide, there are approximately 350 commercial cultivars, each with a unique taste. Sato et al. [42] investigated metabolites in mango characterizing the taste of five cultivars from Indonesia. In this report, GC-MS identified 95 metabolites of interest. Orthogonal projection to latent structure-discriminant analysis (OPLS-DA) separated these metabolites into three groups in the five cultivars. Nicotinic acid, glutamic acid, aspartic acid, glycine, and ribose contributed to the identification. These results highlighted the potential role of biomarker metabolites that could be commercially useful for sorting mango varieties. Similarly, taste is crucial in pineapple, and metabolic profiling by GC-MS revealed that GABA, valine, and alanine defined the ‘Red Spanish’ cultivar, sucrose, threonic acid, and 5-hydroxytryptamine (serotonin) defined ‘Smooth Cayenne’, and threonine, serine, and methionine defined ‘Queen’ [38].
Honey is produced by honeybees from flower nectar. In New Zealand, Manuka honey from the species Leptospermum scoparium J.R is popular among consumers because of its health benefits but is costly. The cost makes Manuka honey a target for adulteration. Mixing Manuka honey and Kanuka honey is a common way to increase profits. Metabolic profiling of honey using UHPLC-PDA-MS/MS for non-volatile and HS-SPME-GC/MS for volatiles discriminated clearly among Manuka, Kanuka, and Jelly Bush honey. PCA found that leptosin, acetyl-2-hydroxy-4-2-(2-methoxyphenyl)-4-oxobutanate, 3-hydroxy-1-(2-methoxyphenyl)-penta-1,4,-dione, kojic acid, and 5-methyl-3furancaerboxylic acid were biomarkers in Manuka honey [43]. Additionally, the utilization of biomarkers elucidated by metabolomics using GC-MS distinguished wine produced from four grape varieties. In this study, 120 commercial brands representing four varieties from six different countries were used for analysis. Organic acids and sugar are dominant metabolites in all wine; however, serine, phenylalanine, L-homoserine, and glutamic acid were found to be the best biomarkers in discriminating between the four varieties using PLS-DA [14].
3.2 Discrimination of geographical originConsumers’ concern over quality certification of agricultural products regarding the geographical origin is increasingly common for coffee and tea [15, 44, 45]. Traceability of geographical origin is usually used to protect the product from fraud. Recently, the European Union and the United Kingdom introduced protected designation of origin (PDO) and protected geographical indication (PGI) classifications. PDO or PGI aims to protect the original product from adulteration. PDO and PGI products are registered in a list of quality products to trace their specifications [46, 47]. Because of this trend, researchers are motivated to develop advanced technology for discerning geographical origins. The metabolomics approach using MS and NMR is a powerful tool for authenticating several agricultural commodities.
Coffee is a popular beverage brewed from roasted coffee beans. The most common coffee beans used all over the world are C. arabica and C. robusta. Recently, the global coffee market is more concerned with aroma and taste. Different coffee origins produce specific aromas and tastes. In this context, HS-SPME-GC-MS comprehensively detected metabolites from samples of different origins, namely, Ethiopia, Tanzania, and Guatemala. The metabolic profiles from different origins were successfully separated on PCA score plot. Ethiopian coffee aroma was characterized mainly by 4-(4′-hydroxyphenyl)-2-butanone [48]. In Indonesia, coffee production is spread over islands such as Java, Sumatra, Sulawesi, Bali, and Papua. Every island has several coffee production areas. Particularly, Mandheling is famous in Sumatera, Toraja in Sulawesi Island, and Kintamani in Bali. They are well known in the global market. Different coffee bean origins result in unique metabolites. The metabolomics approach is a valuable method to distinguish the origin of coffee beans. For instance, a comprehensive analysis for metabolic profiling of primary metabolites using non-targeted GC/MS successfully identified 64 compounds in coffee beans from nine areas in Indonesia. PCA separated the metabolic profiles of each coffee bean into three groups, i.e., western, central, and eastern Indonesia. Glycerol, glucuno-1,5-lactone, gluconic acid, sorbitol, galactitol, and galactinol were potential markers distinguishing different regions. As mentioned above, exhaustive profiling and specific biomarkers are more valuable in authenticating coffee bean origin than utilization of conventional analyses like the cupping test [45].
Similarly, Mi et al. [49] demonstrated the ability of metabolomics to distinguish the origin of garlic. They applied HS-SPME-GC-MS and UHPLC-Q-TOF/MS to identify both volatile and non-volatile compounds. GC-MS identified 68 volatile compounds, and LC/MS detected 854 non-volatile compounds. PCA and PLS-DA were applied to distinguish the garlic origin and identify the biomarker. From the multivariate statistical analysis, two compounds from ketones, i.e., 1-(2-methyl-1-cyclopenten-1-yl)-ethanone, 3,4-dimethylthiophene-2,5-dione, one compound from an alkane, i.e., (E)-1-allyl-2-(prop-1-en-1yl) disulfane, mequinol, 2-methoxyphenol, and 1,2-dimethoxybenzene contributed to the garlic origin discrimination. Klockmann et al. [50] also reported the ability of metabolomics to distinguish the origin of hazelnut. They utilized LC-QQQ-MS/MS to detect the metabolites, and linear discriminant analysis based on PCA scores (PCA-LDA) was applied to discriminate the hazelnut from six countries. Nineteen metabolites such as PC 18:2/18:2, DG 16:0/16:1, DG 18:1/18:1, etc., were found as potential markers responsible for distinguishing the origin of hazelnut.
Adzuki bean is usually used as an ingredient for traditional desserts in East Asia, such as Japan, Korea, and China. This bean is also famous in the global market and has become an export product. Metabolomics analysis is performed to discern the geographical origin of the adzuki bean to certify authenticity. GC-TOF/MS and OPLS-DA was utilized to investigate metabolite profile in adzuki beans from Korea and China. Malic acid and citric acid were important markers to distinguish adzuki bean production areas [10]. In black pepper, using high-resolution mass spectrometry (HRMS) using orbitrap mass analyzer followed by PLS-DA analysis, reynosin, artabsinolide D, tatridin B were found to be a strong marker to distinguish black pepper from Brazil, Vietnam, and Sri Lanka. In this study, the utilization of PLS-DA was more effective than PCA for discrimination purpose [51].
Increasing consumer concerns with quality and authenticity have put forward great challenges for scientists to grow the new analysis methods. The conventional methods focused on physicochemical changes cannot fulfill the new trend appropriately. They do not give sufficient information about the physiological changes and specific characteristics of agricultural products. Conversely, the advanced analytical method of metabolomics can provide a thorough understanding of metabolic changes through a comprehensive analysis of small molecules. This understanding can cope with the needs of consumers. As mentioned in this review, the metabolomics approach exhibited the advantage for maintaining and assuring product quality, traceability, and authenticity. However, the application of metabolomics in agricultural products is quite new compared with other fields like biomedical sciences. The standardization of the analytical methods such as sampling, extraction, and detection for agricultural products is necessary. Sample heterogeneity should be considered to get a robust result. The complex matrix from the plant sample also induces sensitivity fluctuation in mass detection due to the ion enhancement or suppression effect which occurs in the ion source.
As for the platform of the analytical method, GC-MS has been commonly used in agricultural products and primary metabolites such as amino acids and sugars were often targeted. In these studies, the number of detected metabolites was one hundred at most even though plants have thousands of metabolites. The number of detected metabolites should be increased to get more profound product identification information. Only a few trials using HRMS, where a broad range of metabolites can be detected based on untargeted metabolomics, have been conducted.
Additionally, despite many studies proposing metabolic markers for authentication, these results are still insufficient for practical use due to a shortage of quantitative data. First, the prediction model should be constructed based on quantitative data. Then, a simple and non-destructive method for detecting metabolite markers should be developed for real-time and on-site assurance and authentication.
