Keywords

5.1 Introduction

Metabolomics deals with the systematic analysis of endogenous and exogenous small molecules (<1kD) involved in primary or secondary metabolic processes. Metabolomics is fascinating due to the vast diversity of the metabolites classified into amino acids, lipids, nucleotides, carbohydrates, organic acids, etc. The organism’s metabolic repertoire changes consistently and significantly during growth and development and interactions with environmental factors. The metabolic shifts thus represent the physiological state of the organism and have a strong correlation to the phenotype. Metabolomics is applied to all arena of biological sciences ranging from human health to agriculture (Kim et al. 2016; Tian et al. 2016). Assessing food quality is a prime area of the food industry today since there is increasing awareness among consumers regarding food safety and composition. Food metabolomics is an emerging area in metabolomics, enabling food processors and scientists to understand the biochemistry and composition of food more quickly and efficiently. With the ever-increasing population, reduction in agricultural lands, climate change, and environmental pollutants, food metabolomics analysis is imperative for food security and human health. This chapter focuses on metabolomics workflows, applications, and challenges in agriculture, livestock, processed food, human nutrition, and plant resource food. New advances such as food metabolomics databases, the application of metabolomics to screen genetically modified (GM) crops, and organic food are also discussed. This chapter also highlights the challenges in the food metabolomics research, such as sample preparation, data analysis, identification of unknown compounds.

5.2 Metabolomics Analysis

5.2.1 Study Design

The study design of metabolomics analysis should be based on the biological question. Two different metabolomics approaches are employed based on the analysis requirement: untargeted and targeted (Fig. 5.1). If the research is hypothesis-driven, primarily to recognize the prior characterized and biochemically annotated metabolites, then a targeted approach is preferred. The analysis could be undertaken quantitatively or semi-quantitatively based on the standards. Untargeted metabolomics is ideal for hypothesis development, as it simultaneously identifies and quantifies several unknown/known metabolites. The identification of unknown features is performed by matching the metabolic MS/MS fingerprints with the public spectral library repository or standards, which leads to limited metabolite identification. Therefore, many potentially useful information in MS/MS data sets remains uncurated. Several in silico tools like Global Natural Product Social Molecular Networking (GNPS) could catalog the uncurated MS/MS data sets via a spectral correlation and visualization approach (Wang et al. 2016). Both methods have their advantages and limitations. There is no universal metabolomics workflow that is one-for-all due to vast metabolites complexity.

Fig. 5.1
A flow diagram indicates the metabolomic workflow. It includes the following. Sample preparation, L C, data alignment and analysis, validation, L C M S or M S, and match with a spectral library. Another branch from validation contains known standards, L C M S run optimization, run of samples, and quantification.

Metabolomics workflows (a) targeted and (b) untargeted

Full size image

5.2.2 Sample Collection, Quenching, and Storage

Sample collection and storage are very critical. The metabolites are very dynamic; hence sampling time becomes essential, and samples should be collected systematically. Consistency is the key to sample collection, particularly for long-term experiments. Food habits, age, sex, social-economic status, geographic location, etc., should be considered for animal models or human subjects. Likewise, species, environmental factors like watering patterns, nutritional content, light, moisture, day-night cycles, and development stage should be accounted in plant study. The tissue/bacterial samples should be washed with buffer or water before storage to remove any media/external components. Likewise, for exometabolomics research, the samples should be filtered to remove any cells in the media. Further, samples may be spiked with the known concentration of metabolite “standard” to evaluate any degradation/change in the metabolite during storage. As soon as the samples are collected, they should be quenched with liquid nitrogen or solvents like methanol and stored at −80 °C until processed further.

5.2.3 Workflow

Typical metabolomics workflow (Fig. 5.2) includes(i) Study designing, the most critical factor for any metabolomics study. Study design comprises but is not limited to asking the right question or hypothesis based on which sample size, controls, sampling time, metabolite extractions solvents/methods, analytical tools, data analysis, etc., are decided.(ii) Metabolite extractions: the intracellular metabolite extraction goals include separation of the small molecules from cell debris or other cellular biomolecules like protein, nucleic acids, etc. For the exometabolites, like the extraction of metabolites from the soil, the aim is to separate metabolites from the complex matrix. The metabolite extraction approaches should be consistent and minimize the losses due to biochemical/photochemical conversion or degradation. However, the biases are inevitable due to the wide dynamic range of metabolites and varying solubility quotient (Phapale et al. 2020). A wide range of solvent choices is available based on metabolite interest and chemistry. Apart from the solvent, temperature, pH, and desired molecular weight of the metabolite of interest should be considered. Monophasic and biphasic extraction approaches are employed based on the metabolites of interest. In monophasic extraction, only one solvent is used, while for biphasic extraction, a combination of polar and non-polar solvents are used to achieve a comprehensive metabolite coverage. Samples may be spiked with the known concentrations of standards to normalize for metabolite losses during extraction. Again, the standard spiked could be stable isotope-labeled (absolute standards) or pseudo standards that are primary or secondary metabolite not present in the experimental sample. For example, reserpine, curcumin (found exclusively in plants) could be spiked in serum/blood samples. Likewise, process control or negative control is vital to eliminate any contamination from leached pipette tips, centrifuge tubes or mass spec peak tubings, etc. Last but not least, sample homogenization should be considered; based on our information on the spatial localization of the metabolites, which suggests different regions of the tissue might have different concentrations or compositions of metabolites. The extracted metabolites should be aliquoted to prevent any metabolite degradation due to freeze-thaw.(iii) Sample complexity could be reduced on-line or off-line. For on-line metabolite separation, liquid/gas chromatography (LC/GC), capillary electrophoresis (CE), and ion-mobility spectrometry (IMS) (Fig. 5.3) are attached to the mass spectrometry. The most common off-line metabolite separation includes solid-phase extraction and LC-based fractionation based on the retention times. This step is essential in complex samples, where several masses co-elutes and make the data analysis challenging.(iv) Analytical instruments: are selected based on experimental needs. For example, Gas-chromatography is employed to analyze the volatiles, or direct infusion mass spectrometry is used to analyze less complex metabolite extracts. GC uses gas (usually an inert gas or an unreactive gas) as a mobile phase and solid or liquid stationary phases. GC separation is highly robust as it separates metabolites based on vapourization temperatures. Thus, the technique is limited to the analysis (identification and semi-quantification) of molecules that are vaporized below 350–400 °C without decomposing or reacting with the GC components. Primarily to lower the vapor pressure, the metabolites are derivatized. LC uses liquid (acid, base, or neutral solvents) as a mobile phase and solid stationary phase. LC separation is highly variable and depends on the combination of mobile and solid phase, solvents, column properties like column material, length, diameter, pore size, temperature, etc. LC offers versatility and no or limited sample pre-processing requirement (Fig. 5.3a). CE separates ionized molecules in the liquid phase based on their electrophoretic mobility. The greater the electric field applied, the faster the mobility. Thus, the approach is inclusive for the analysis of ions (charged molecules) but not neutral species (Fig. 5.3b). IMS separates ionized molecules in the gas phase based on their mobility in a carrier buffer gas (Fig. 5.3c). Mass spectrometry (MS) measures the mass-to-charge ratio (m/z) of the ionized molecules. MS finds application in absolute or relative quantification of the metabolites, identify unknown molecules based on molecular weight and fragmentation patterns, a structure prediction. MS comprises three major components: ionization source, a mass analyzer, and an ion detector. The ionization source converts molecules to gas-phase ions, which could be negatively or positively charged. Hard and soft ionizations could be employed. Hard ionization like Electron Impact ionization (EI) causes extensive fragmentation of the ions (ambiguous identification of the molecule weight) and is incompatible with LC, hence a method of choice for GC-MS. While soft ionization like electrospray ionization (ESI), atmospheric pressure chemical ionization (APCI), and matrix-assisted laser desorption ionization (MALDI) has gained popularity and increased the MS applications in metabolomics analysis. A mass analyzer sorts and separates the ionized molecules according to mass-to-charge (m/z) ratios. Quadrupole (Q), Time-of-flight (TOF), Ion trap, Orbitrap, etc., are few examples are mass analyzers that are interchangeably used for metabolomics analysis based on the prerequisites for sensitivity, precision, accuracy, or resolution. A detector detects the sorted ions, and a mass spectrum/chromatogram is generated representing the m/z ratio against intensity/relative abundance (Fig. 5.3d). Each technique has advantages and limitations (Johnson and Gonzalez 2012). To reduce the complexity of the analysis, lower the sample volume and minimize analytical variation, several separation-free MS techniques like direct infusion-MS, MALDI-MS, mass spectrometry imaging (MSI), and direct analysis in real-time (DART)-MS are gaining popularity. Mass spectrometry imaging (MSI) has revolutionized MS-based metabolomics by providing spatial resolution. MSI is operated in two modes: imaging (Stoeckli et al. 2001) to correlate with histology and profiling to know the overall metabolites (Cornett et al. 2006). MSI technique is applied in broad areas of plant biology, including development, defense, and responses to abiotic and biotic stress, and the developing field of spatial-temporal metabolomics. Nuclear Magnetic Resonance (NMR) relies on detecting the electromagnetic signals generated due to the perturbation of the nuclei in a weak oscillating magnetic field (Fig. 5.3e). NMR finds application in quantifying the known compounds or identifies unknown compounds based on the match against the spectral libraries, or infer the basic structures. Chromatography or reducing the metabolite complexity is essential to reduce the matrix effect, identify reteition time that adds third dimention to the MS data, and enhance the MS/MS data quality.(v) Data analysis: the aim of the data analysis is to get rid of the aritifacts, contanimants and redundant peaks, identify/quantify biologically relevant peak and statistical and pathway analysis. Metabolomics data analyis could be performed using propriety tools like mass propessional profiler (MPP), compound discoverer, etc., or open source tools like mzMine or XCMS. Metabolomics data analysis pipeline is shown in Fig. 5.4.

Fig. 5.2
A workflow with labels study design, sample preparation, reduce sample complexity, analytical instruments, data processing from left to right and corresponding attributes below each label.

Typical metabolomics workflow

Full size image
Fig. 5.3
Five illustrations. A. Chromatography. Carrier gas, G C column, buffer, and H P L C column. B. Capillary electrophoresis. High voltage power supply. C. I M S. Ion source and ion mobility cell. D. Mass spectrometry. Ion source, mass analyzer, and ion detector. E. N M R. Sample, magnetic field, and ion detector.

(a) Chromatography, (b) capillary electrophoresis, (c) ion mobility spectrometry, (d) mass spectrometry, and nuclear magnetic resonance working principle

Full size image
Fig. 5.4
A workflow. 1. Raw data processing. 2. Data processing. 3. M S polarity combine. 4. Data cleaning. 5. Feature prioritization. 6. Feature identification. The flow diagram at the top includes align, profile, annotate, M S polarity combine, feature prioritize, and metabolite identified.

Untargeted metabolomics data analysis workflow

Full size image

5.3 Food Metabolomics

‘Food metabolomics‘is the application of metabolomics in food systems, including processes from agriculture-to-human nutrition: (a) soil quality, (b) food resource, (c) food processing, and (d) human nutrition (Fig. 5.5).

Fig. 5.5
Four applications. 1. Soil metabolomics. Pesticide, soil nutrients, and fertilizer. 2. Food resources. Livestock, genetic modifications, chemicals, and plant produce. 3. Food processing. Fermentation, baking, and preservation. 4. Human nutrition. Bio-fluid, human diet, tissue or organs, disease and culture, and microbiota.

Applications of food metabolomics

Full size image

Considering the constant change and evolution in the food supply and the ever-increasing and continuous launch of new products and formulations into the marketplace, there is a need to keep track of the vital food metabolites. Moreover, raw agricultural products, their composition like nutritional value, assimilated environmental contaminants, pesticides, etc., are also continually changing, leading researchers, nutritionists, and food policymakers to emphasize beyond single nutrients to develop holistic nutrition plans foods, food groups, and dietary patterns. Also, there is a growing urge for transparent and easily accessible information about nutrients and other food components. Typical workflow and challenges encountered to perform food metabolomics experiments are shown in Fig. 5.6.

Fig. 5.6
An illustration of the workflow and challenges. 1. Sample preparation. The workflow indicates blood, urine, and food products. 2. Data acquisition. N M R, L C M S, and G C M S. 3. Data analysis. Statistics and pathway analysis. 4. Identification of the unknown. 5. Translation.

Food metabolomics workflow and challanges

Full size image

Several natural and non-natural food metabolomics annotated databases are made available in the public domain (Table 5.1). Briefly, food databases are in place to get the list of possible metabolites, chemical and biological properties, structures, spectral data, related metabolic pathways, and their presence or concentration in food products based on the query m/z or elemental formula. Recently USDA launched an integrated, research-focused data system, ‘FoodData Central,’ that provides expanded data on nutrients and other food components and links to sources of related agricultural, food, dietary supplement, and additional information (https://fdc.nal.usda.gov/). The data in FoodData Central includes five data types, including Foundation Foods and Experimental Foods. In a similar effect, Dr. David Wishart Research Group at the University of Alberta annotated a comprehensive 28,000 food metabolite ‘FooDB’ (http://www.foodb.ca/) that comprises food constituents flavor, color, taste, texture, and aroma. Various European research groups have also taken the initiative to develop databases with tailored phytochemical components of the food metabolome. One such example is Phenol Explorer, which comprises 500 dietary polyphenols and their known human metabolites in over 400 foods (Rothwell et al. 2012). A similar ongoing project is PhytoHUB that will contain a comprehensive inventory of dietary phytochemicals and their human metabolites and structures either obtained from previous publications or predicted in silico (http://phytohub.eu/).

Table 5.1 List of food metabolomics databases
Full size table

5.3.1 Soil Quality

Soil quality is one of the most critical factors determining plant health. The typical soil quality indicators are organic matter content, salinity, tilth, compaction, available nutrients, rooting depth, and microbial populations. Untargeted soil metabolomics approaches are used to analyze extractable organic matter using LCMS (Swenson et al. 2015) and NMR (Johns et al. 2017) or discriminate the suppressive and non-suppressive soils for disease resistance (Rochfort et al. 2015). Soil metabolomics is also employed to assess the microbial (Boiteau et al. 2019; Rai et al. 2020) and plant metabolites (Petriacq et al. 2017; van Dam and Bouwmeester 2016) in soil.

Rai et al. has recently developed a high throughput method for the extraction and quantification of siderophores, highly specialized iron-chelating secondary metabolite, in the soil (Rai et al. 2020). Siderophores have ecological significance and find application as a biomarker for soil quality, the remedy of polluted sites, and improving nutrient metal uptake by crops or other plants. The detection and quantification of soil metabolites are challenging due to the complexity of soil matrix and metabolites’ structural diversity. The workflow has the potentials to identify the soil biomarkers of plant health, as both plants and microbes secrete siderophores in response to iron limitation or cataloging soil-borne plant pathogens, a prerequisite for food security. On the other hand, identifying siderophores from beneficial rhizosphere bacteria can protect plants from pathogens and their virulence factors and keep them healthy. A yet another potential is discovering novel siderophores in the soil, which is obscure due to our inability to culture microbes in laboratory conditions. Microbes’ critical need to sequester essential iron provides an achilles heel for new antibiotic development by utilizing the siderophore-based molecular recognition covalently attached to antimicrobial peptides. The unique metabolites of the soil-borne pathogens are used as biomarkers for plant diseases such as macrocarpa for Rhizoctonia solani-suppressive soil (Hayden et al. 2019) or siderophore of beneficial bacteria that promotes or inhibits Ralstonia solanacearum infections (Pollak and Cordero 2020). Metabolomics is also employed to investigate the impact of soil microbial population on the leaf metabolome and herbivore feeding behavior (Badri et al. 2013).

5.3.2 Food Resources

Environmental factors such as geographical origins, application of chemicals (fertilizers, pesticides, industrial wastes), stress (abiotic and biotic), and genetic factors, including genetically modified (GM) crops in agriculture, significantly influence food resource production (Kim et al. 2016). Advancements in the analytical tools and ease to analyze a wide variety of plant samples ranging with varying culture conditions, geographic locations, growing seasons, exposure to stress, etc., could lead to the assessment of food resource quality changes due to environmental and genetic factors.

5.3.2.1 Metabolomics to Screen Geographic Origin

Geographical origins of food resources affect the quality and the prices, and hence is becoming an important issue for consumers and producers due to increasing international trade. It is almost impossible to determine the geographical origins of food resources by their phenotypes. Metabolomics is thus employed as a tool to assess the legitimacy and source of specific food resources such as hazelnuts (Klockmann et al. 2017), tobacco (Zhao et al. 2015), coffee (Arana et al. 2015), green tea (Lee et al. 2015), Spanish Extra Virgin Olive Oils (Gil-Solsona et al. 2016), grape (Teixeira et al. 2014), wine (Amargianitaki and Spyros 2017), milk and dairy products (Brescia 2005; Renou et al. 2004), honey (Zhou et al. 2018b), fish, and seafood (Cubero-Leon et al. 2014), beef (Man et al. 2021) since the metabolite profiles differ depending on their geographical origins. NMR fingerprinting is used to differentiate coffee beans based on country of origin (Arana et al. 2015). Zhao et al. correlated metabolite changes of carbon and nitrogen pools in tobacco plants from two different locations due to climatic factors such as rainfall and temperature (Arana et al. 2015).

5.3.2.2 Metabolomics for Organic Food Authentication

There is an increasing demand for organic products, and hence organic food items are expensive. The lack of reliable chemical markers to discriminate between organic and conventional products makes them an attractive target for fraudulent malpractices. Metabolomics could solve this problem by precisely detecting and quantifying the chemicals like pesticides or fertilizers in food products. The ability of metabolomics to detect unknown targets is particularly beneficial to identify/detect new adulterants. A non-targeted metabolomics approach identifies food markers that discriminate between organic and conventional tomato crops (Martínez Bueno et al. 2018). Also, the presence of pesticides is determined in the amphibian liver (Van Meter et al. 2018), urine samples of healthy individuals (Reisdorph et al. 2020), and pregnant women (Sem et al. 2013). Growers sometimes apply growth promoter substances like testosterone, progesterone, auxin, and gibberellins beyond permissible limits to improve food resources quality and production yield. Metabolomics has recently been introduced as a new tool to detect illegal and excessive uses of growth promoters (Rodriguez-Celma et al. 2011; Šimura et al. 2018; Stephany 2010).

Organic agricultural systems rely on non-conventional soil fertilization techniques like the application of organic manures, biological pest controls, and multiannual crop rotation, including legumes and other green manure crops. In contrast, chemical plant protection products, including pesticides, are applied to protect plants in conventional agriculture. Organic production systems increase the susceptibility of the plants to pathogens, hence resulting in the accumulation of inducible protective secondary metabolites such as phenolics; chlorogenic acid (Malik et al. 2009; Novotná et al. 2012; Young et al. 2005), and flavonoids (Mitchell et al. 2007). Metabolomics studies distinguish conventionally and organically grown produces (Martínez Bueno et al. 2018; Novotná et al. 2012; Vallverdu-Queralt et al. 2011). Factors such as plant response, rhizosphere microbiome are accountable (Bradi et al. 2013).

5.3.2.3 Metabolomics to Screen Genetically Modified (GM) Crops

Genetically modified (GM) food crops are resistant to diseases, pests, and unfavorable environmental conditions, produce high-quality foods with less effort and lower costs; however, GM food resources continue to be a controversial issue due to their safety and relation to human health and the environment (Simo et al. 2014). Organization for Economic Co-operation and Development (OECD), comprising 37 member countries, is established to globally deal with GM food resources safety issues (Kearns et al. 2021). OECD and European Food Safety Authority (EFSA) has developed guidance and regulations for GM crops to evaluate their safety and nutritional value (Bedair and Glenn 2020). Metabolomics provides comprehensive information about GM food composition compared to their corresponding non-GM counterparts (Simo et al. 2014). Metabolomics is effectively employed to assess the safety of widely consumed GM maize by comparing plasma metabolome and fecal microbiota in GM maize and non-GM near-isogenic maize-fed rats (Mesnage et al. 2019). Metabolomics is also employed to evaluate and assess GM food resources such as soybean (Garcia-Villalba et al. 2008; Inaba et al. 2007), rice (Jiao et al. 2010; Zhou et al. 2009), maize (Frank et al. 2012; Piccioni et al. 2009), wheat (Baker et al. 2006), tomato (Kusano et al. 2011b; Noteborn et al. 2000), potato (Catchpole et al. 2005), poplar (Srivastava et al. 2013), carrot (Cubero-Leon et al. 2018), and barley (Kogel et al. 2010). A recent study of GM rice with Cr1C gene transformation showed no significant difference in metabolic profile than the parent line (Chang et al. 2021). Another study reported differential metabolite profiles between wild-type and cryIAc and sck genes; for improving insect resistance in GM rice (Zhou et al. 2009). Targeted quantitative metabolomics could provide additional information for safety and nutritional assessment for GM crops with traits known to modify metabolic pathways. Integrating metabolomics with other omics data provides more comprehensive knowledge about risk assessment of GM crops (Kok et al. 2019). Metabolomics application for the safety assessment of GM crops, which is ever-growing and introduced to the global market, provides relevant information regarding the associated metabolite alteration. The challenges faced in such metabolomics applications include chemical complexity, identification of unknown metabolites, matrix effect as each plant is different, and dynamic concentration range. One key hurdle in using data from omics studies with GM crops, including metabolomics, is the difficulty to assess whether there is any impact on safety in the observed differences amongst the 1000s of signals characterized by the untargeted profiling method(s) (Bedair and Glenn 2020). Application of metabolomics to screen GM crops is infancy, and to get a comprehensive understanding of the detected metabolite changes in a biological context, big-data generated needs to be analyzed together with other ‘omics’ data such as proteomics and transcriptomics as proposed by the new Foodomics strategy (Ibanez et al. 2012). The development of advanced tools and databases is essential for metabolomics studies.

5.3.2.4 Metabolomics to Elucidate Environmental Stress Feedback in Food Resources

Environmental stress (biotic/abiotic) is a major limiting factor of agriculture production, affecting both yields and nutritional content. The early detection of stress symptoms could help reduce loss. Plant metabolomics has emerged as the most promising tool to decipher the metabolic changes caused by (a) climatic and seasonal variations, (b) biotic factors including pathogens and beneficial/symbiotic associations (Alseekh and Fernie 2018) for high-quality food resource production since the last decade.

Abiotic stress is responsible for global crop yield reduction ranging from 50% to 70% (Boyer 1982). Climate change and population growth have worsened the situation (Raza et al. 2019). Understanding plants’ responses to such stressors to determine methods for improving crops quantitatively and qualitatively is inevitable. Abiotic stress in plants leads to the synthesis of phytohormones to impart stress resistance (Han et al. 2012; Rai et al. 2016); the oxidative stress disturbs the stomatal conductance and activates several signaling mechanisms and the dysregulated gene expression profiles (Rai et al. 2016). In particular, all essential plants’ mechanisms from germination to maturity are severely affected by abiotic stresses. The major abiotic plant health stressors include drought, salinity, temperature extremes, waterlogging, heavy metal, and chilling. Metabolomics has been applied for probing unique metabolites that regulate the abiotic stress tolerance mechanism in crops with two prime objectives: (i) identification of biomarkers for abiotic stress and (ii) investigate metabolic variations under abiotic stresses to detect different metabolites that permit restoration of plant homeostasis and normalize metabolic modifications (Arbona et al. 2013). Water deficit (drought) caused by global climate changes seriously endangers plant survival and crop productivity (Lesk et al. 2016). Metabolic profiling of drought-stressed wheat (Michaletti et al. 2018), barley (Chmielewska et al. 2016), rice (Lawas et al. 2019), and soybean (Das et al. 2017) are carried out to elucidate vital metabolites/biomarkers for drought tolerance. Soil Salinity is rapidly increasing, and about 20% of irrigated land is salt-affected. Salinity stress causes >20% losses in crop plants (Food and Agriculture Organization 2015). The GC-MS-based metabolic profiling of two salt-sensitive (Sujala and MTU 7029) and tolerant varieties (Bhutnath, and Nonabokra) of indica rice showed accumulation of two signaling molecules, serotonin, and gentisic acid, which may serve as a biomarker to produced salt-tolerant rice varieties (Gupta and De 2017). Metabolomics is employed to study metabolic remodeling due to salinity stress in several other crop plants like tomato (Rouphael et al. 2018), millet (Pan et al. 2020), strawberry (Antunes et al. 2019). Temperature stress, including heat (Abdelrahman et al. 2020; Escandon et al. 2018; Raza 2020) or cold (Furtauer et al. 2019; Xu et al. 2020), disturbs the homeostasis and physiological mechanisms. Metabolomics analysis of temperature-stressed wheat (Qi et al. 2017; Thomason et al. 2018), tomato (Almeida et al. 2020; Paupiere et al. 2017; Zhang et al. 2019), and maize (Obata et al. 2015; Sun et al. 2016; Urrutia et al. 2021) are studied to identify the effect of temperature stress. Heavy metal stress has become a significant concern on various terrestrial ecosystems due to extensive industrialization (Guerrero et al. 2019; Shahid et al. 2015). Suboptimal concentrations of trace metals such as Zn, Cu, Mo, Mn, Co, Ni, As, Pb, Cd, Hg, Cr, and Al reduce crop metabolism, growth, and productivity (Tiwari and Lata 2018). Metabolomics analysis of high Zn and Cu treated beans (Jahangir et al. 2008), Zn-deficient tea plants (Zhang et al. 2017), heavy metal stressed (Cu, Fe, and Mn) Brassica rapa (Jahangir et al. 2008), Cr-toxicated sunflower (Gonzalez Ibarra et al. 2017) and Fe-toxicated rice (Turhadi et al. 2019) are performed. A better understanding of the nutrient-limitation, including macro- and micro-nutrients, would enhance the food/fodder nutritional contents. Metabolic changes due to macro-nutrient: nitrogen (Kusano et al. 2011a; Rai et al. 2017), phosphorus (Jones et al. 2018; Vance et al. 2003), and potassium (Zeng et al. 2018), deficiency has been studied in plants/microalgae using HPLC and enzymatic activities (Scheible et al. 2004; Tschoep et al. 2009), FT-ICR-MS (Hirai et al. 2004), LC-MS (Peng et al. 2008), GC-MS (Heyneke et al. 2017; Urbanczyk-Wochniak and Fernie 2005), CE-MS (Takahara et al. 2010; Takahashi et al. 2009), 1H-NMR (Broyart et al. 2010).

Plant-microbe (pathogenic/beneficial) interactions (Biotic factors) trigger a plethora of primary and secondary metabolites changes, which could be easily detected by metabolomics (Allwood et al. 2008; Castro-Moretti et al. 2020). A wide range of phytopathogens, including fungi, bacteria, viruses, viroids, mollicutes, parasitic higher plants, and protozoa, are known to cause plant disease. Metabolomics is an emerging tool to study plant-pathogen interactions (Castro-Moretti et al. 2020). An integrated metabolo-proteomic approach showed induced phenolic acid and phenylpropanoids in Fusarium graminearum infected wheat (Gunnaiah et al. 2012). NMR analysis indicated an increased accumulation of disease-resistant biomarkers (Sarrocco et al. 2016). Metabolomic analysis of the susceptible and resistant wheat cultivars infected with the fungal pathogen Zymoseptoria tritici showed that immune and defense-related metabolites in resistant and susceptible wheat cultivars using FT-ICR-MS (Seybold et al. 2020). Likewise, the metabolomics responses in wheat against viral infection; wheat streak mosaic virus (Farahbakhsh et al. 2019), rice against fungal infection; Rhizoctonia solani (Suharti et al. 2016), Magnaporthe grisea (Jones et al. 2010), insect attack; gall midge (Agarrwal et al. 2014), bacterial infection; Xanthomonas oryzae pv. oryzae (Sana et al. 2010), maize against fungal infection; Fusarium graminearum (Zhou et al. 2018a), southern corn leaf blight (Vasmatkar et al. 2019), and insect attack; Ostrinia furnacalis (Guo et al. 2019) are studied. Adverse environmental conditions, including temperature, soil fertility, light, water deficit, give an edge to the pathogens and increase disease severity. However, we have very little knowledge about these tripartite interactions, suggesting future investigations towards understanding the multi-dimensional nature of plant-pathogen interactions in changing climate conditions (Velásquez et al. 2018). Weeds are yet another problem that impacts crop productivity at multiple levels, such as competing for nutrients, reducing crop yields and nutritional content, interfering with harvest efficiency, and recurrence. Metabolomics analysis of the canola plant extracts detected allelopathic metabolites (3,5,6,7,8-pentahydroxy flavones, p-hydroxybenzoic acid, and sinapyl alcohol) that inhibit the rye root and shoot development (Asaduzzaman et al. 2015). Similarly, weed-suppressing metabolites in wheat and legume are investigated (Latif et al. 2019).

5.3.2.5 Metabolomics in Livestock Research

Traditional livestock analysis, such as feed consumption, is time-consuming, expensive, and requires specific equipment (Karisa et al. 2014). Reproductive trait measurements need animals to reach the maturity stage, while carcass trait evaluation requires animal slaughter which otherwise could have been used for breeding. Metabolomics has emerged as an efficient, cost-effective, non-invasive way to detect animal traits for livestock research and industry. Livestock metabolomics is instrumental in animal breeding farm trials for efficient and quicker quantitative phenotyping (Karisa et al. 2014; te Pas et al. 2017) and is primarily applied for biomarker identification for weight gain, milk quality (D’Auria et al. 2013), health (LeBlanc et al. 2005), fertility (Chapinal et al. 2012), etc. The metabolomic signatures associated with feed efficiency in beef cattle (Novais et al. 2019), dairy cow (Saleem et al. 2012) is studied. Also, metabolomics is employed to analyze antimicrobial resistance in livestock pathogens like E. coli (Campos and Zampieri 2019; Lin et al. 2019). Livestock metabolomics experimental design should take diet, diurnal variations, sex, and sampling time into consideration to reduce variability. Shortage of data resources makes data interpretation a challenge in livestock metabolomics Goldansaz et al. created a livestock metabolome database of more than 1000 metabolites detected in livestock metabolomic studies on cattle, sheep, goats, horses, and pigs (Goldansaz et al. 2017). Bovine Metabolome Database is a recent collection of more than 50,000 metabolites focusing on animal health which describes a healthy range of metabolites in bovine biofluids and tissues (Foroutan et al. 2020).

5.3.3 Food Processing

Our modern lifestyle and the ever-growing global population have caused increased food processing industry demands. Food processing can be defined as the physical and/or chemical manipulation of raw food to enhance nutritional and sensory quality and sustainability. Some of the food processes where metabolomics is applied include cheese (Afshari et al. 2020), tomato purees (Capanoglu et al. 2008), tinned vegetables, biscuits (Diez-Simon et al. 2019), alcoholic beverages (Álvarez-Fernández et al. 2015; Ichikawa et al. 2019), yogurt (Settachaimongkon et al. 2015) and milk (Rocchetti et al. 2020). Metabolomics finds its role in the food processing industry ranging from food preparation, packaging, and storage. For instance, food preparation processes like pasteurization, fermentation, etc., could affect the food’s nutritional and sensory quality either beneficially (improved digestibility, nutrients bioavailability, foodborne pathogens/toxins inactivation) or detrimentally (vitamins and nutrient loss, toxic compounds formation, conferring adverse effects on flavor, aroma, texture or color), monitored by metabolomics. Another exciting application of metabolomics is identifying the chemical species that contribute to flavor, texture, taste, and color quality, which could be used to develop meat using plants, considering the growing demand for plant-based foods (https://www.impossiblefoods.com/food). Metabolomics is applied in food safety for rapid and reliable monitoring of food contaminants (i) pesticides and other chemical residues like furans, dioxins, dioxin-like polychlorinated biphenyls (PCBs), non-dioxin-like PCBs (Tengstrand et al. 2012; Zainudin et al. 2015), (ii) foodborne pathogens like Salmonella sp., E. coli (Cevallos-Cevallos et al. 2011) and Listeria sp. (Jadhav et al. 2015), with short-term and long-term health risks (Pinu 2016). Predicting the end of shelf life before apparent spoilage (expiry date or best before) and determining effects of food processing on the shelf life is another application of metabolomics. For instance, metabolic profiling showed a correlation between diphenylamine oxidation treatment and extended shelf-life during storage in apples (Leisso et al. 2013). Likewise, the shelf life of meat at various storage temperatures was studied (Argyri et al. 2015). A comprehensive understanding of food metabolite composition during processing and storage will improve preservation methods. Metabolomic assessment of food packaging types is performed to detect contaminants and spoilage in packaging materials (Makkliang et al. 2015). Also, metabolomics could expand our understanding of the biodegradable materials used for packaging (Kleeberg et al. 1998) and microbial and plant strain improvement for efficient production of eco-friendly packaging materials.

5.3.4 Human Nutrition

Diet/nutrition is linked to metabolic disorders, including obesity, diabetes, cardiovascular disease, and aging (Shlisky et al. 2017). The traditional way to evaluate beneficial or detrimental effects of foods based on volunteers filling the questionnaires has several limitations: misreporting, bias and measurement error, high cost and time consumption, and unreliability for populations with cognitive impairment (Fallaize et al. 2014). Metabolomics is a powerful tool to overcome such limitations, and hence dietary biomarkers are emerging as an objective and accurate measure of dietary intake and nutrient status (O’Gorman et al. 2013). Novel metabolic signatures are associated with juice and fruits (Liu et al. 2015), grain, fish (Hanhineva et al. 2015), wine (Urpi-Sarda et al. 2015), and diet patterns (western vs. prudent) (Bouchard-Mercier et al. 2013). Also, metabolomics is used to monitor diet-related metabolic diseases (Sebedio 2017). LC/MS-based metabolic profiling suggested green tea mediated stimulation of hepatic lipid metabolism associated with obesity prevention (Lee et al. 2015). Likewise, the anti-hyperlipidemia effect of curcumin was demonstrated using NMR and MS-based metabolomics (Li et al. 2015). A yet another hot area of metabolomics research is the human gut-microbiota study. Recent studies have suggested that the gut microbiome secretes metabolites that impact human health (Dore and Blottiere 2015). Moreover, the gut microbiome varies between individuals and is greatly affected by diet. Interestingly, diet changes can manipulate the gut microbiome, currently being studied for use as a potential therapy (Shoaie et al. 2015). Selected food metabolomics studies are listed in Table 5.2.

Table 5.2 Comprehensive list of food metabolomics study
Full size table

5.4 Challenges in Food Metabolomics

Despite its enormous potential, there are several dark areas of food metabolomics: (i) lack of optimized workflow for sample processing, normalization, and data analysis, (ii) run-to-run variability, (iii) substantial matrix effects, (iv) limited spectral library coverage, (iv) availability of open-source data analysis tools and databases, and (v) lack of comprehensive guidelines for biomarker prediction and validation for food resources. To overcome the challenges of experimental design, power calculation should be performed to determine the sample size required for statistically significant results and reduce variability, particularly for field samples. Cultivated varieties, geographical location, and fertilization schemes should be taken into consideration. Quality control samples should be included to determine run-to-run variability; any data showing coefficient of variance higher than 20% should be discarded. The lack of external validation sets using samples, which are not part of the statistical models built for prediction, is a significant limitation in most studies published (Cubero-Leon et al. 2018). Matrix effects could to either eliminated by reducing the sample complexity or spiking the standards in the matrix. Absolute metabolite quantification workflows should be improved using a combination of different analytical tools. So far, targeted approaches are used for metabolite quantification, efforts for developing untargeted quantitative metabolomics methods are needed for efficient biomarker discovery. Further regulatory guidelines for biomarker discovery should be put in place for robust experimental design, data acquisition, validation, and translation. Extensive research is required for the instrument miniaturization, cost-effectiveness, accessibility, and ease-of-handing for the fieldable application of metabolomics in agriculture farms. This could be achieved by corroboration between farm workers, researchers, and engineers. Automated, user-friendly, and open-source metabolomics data analysis platforms should be developed for robust data interpretation.

5.5 Conclusion

With ever-increasing food demand, climate change, and the advent of food metabolomics in the past decade, the horizon of metabolomics application has increased from food resource production to food processing and human nutrition. Because of increased efficiency, cost-effectiveness, and accuracy, applications of food metabolomics are rapidly expanding, as discussed extensively in this chapter. Untargeted metabolomics is especially useful in detecting unknown adulterants or advancing food metabolomics research (Cubero-Leon et al. 2014). Food databases are handy in elucidating the unknown/novel metabolites (Table 5.1). Plant metabolite repertoire has a wide dynamic range and diverse chemical composition. With the current set-up, only ~10% of the metabolite are identified; hence there is enormous scope to explore the unknowns. Integrating the metabolomics information with other-omics (genomics, transcriptomics, and proteomics) and non-omics studies (physiological data), combined with reliable and broad-spectrum food metabolome databases and artificial intelligence technologies, could help identify and elucidate many ‘unknowns’ and monitor in real-time to predict dynamics and quality control for accelerating, automating and progressing production processing.

5.6 Perspective

Apart from the applications of food metabolomics discussed in this chapter, metabolomics could be employed in personalized nutrition, crop improvement, metabolic fingerprinting in livestock and plants. Metabolomics investigations have generated a plethora of information that will allow food manufacturers to develop nutritional and sensorial rich food. Metabolomics-assisted crop improvement could lead to high-yielding, stress-tolerant germplasm and create climate-smart crop varieties. However, deciphering a specific metabolite’s function (metabolite-phenotype) and decoding the structure of metabolic networks remains a major hurdle in the third decade of plant metabolomics. Identifying biomarkers related to plant biotic/abiotic stress, genetically modified organisms (GMOs), organic vs. conventional produce, and human nutrition using metabolomics and its translation in the agricultural fields, food industry, and clinics could be a future application. Also, metabolomics allows the determination of nutrient enhancement or reduction due to food processing via detection chemical alteration, which could significantly alter human health. Fingerprinting metabolic phenotypes of livestock and plants in response to defined feeding/fertilizing patterns and compositions is another emerging application. Finally, the food waste generated as a by-product of agriculture, the food processing industry, or household could be used to extract high-value bioactive compounds and nutraceuticals. In the years to come, an innovative sequential approach could simultaneously use food resources for food processing and simultaneously extract specific bioactive compounds and nutraceuticals to be used as functional foods.