Fermented Pepper Paste)

molecules

Article Metabolomics Provides Quality Characterization of Commercial Gochujang (Fermented Pepper Paste)

Gyu Min Lee, Dong Ho Suh, Eun Sung Jung and Choong Hwan Lee *

Department of Bioscience and Biotechnology, Konkuk University, Seoul 05029, Korea; [email protected] (G.M.L.); [email protected] (D.H.S.); [email protected] (E.S.J.) * Correspondence: [email protected]; Tel.: +82-2-2049-6177; Fax: +82-2-455-4291

Academic Editor: Marcello Iriti Received: 3 June 2016; Accepted: 8 July 2016; Published: 15 July 2016

Abstract: To identify the major factors contributing to the quality of commercial gochujang (fermented red pepper paste), metabolites were proﬁled by mass spectrometry. In principal component analysis, cereal type (wheat, brown rice, and white rice) and species of hot pepper (Capsicum annuum, C. annuum cv. Chung-yang, and C. frutescens) affected clustering patterns. Relative amino acid and citric acid levels were signiﬁcantly higher in wheat gochujang than in rice gochujang. Sucrose, linoleic acid, oleic acid, and lysophospholipid levels were high in brown-rice gochujang, whereas glucose, maltose, and γ-aminobutyric acid levels were high in white-rice gochujang. The relative capsaicinoid and luteolin derivative contents in gochujang were affected by the hot pepper species used. Gochujang containing C. annuum cv. Chung-yang and C. frutescens showed high capsaicinoid levels. The luteolin derivative level was high in gochujang containing C. frutescens. These metabolite variations in commercial gochujang may be related to different physicochemical phenotypes and antioxidant activity.

Keywords: gochujang; quality characterization; metabolomics; GC-TOF-MS; UHPLC-LTQ-ESI-IT- MS/MS; antioxidant activity

1. Introduction Food is essential to sustain life and has undergone continuous development worldwide for a long time. As the industrialization of food products has progressed over the past few decades, the public interest in health and well-being has also increased [1]. To meet these demands, application of scientific knowledge to the food industry has become necessary to control quality and safety as well as to maintain the beneficial components in commercial food products [2]. However, various factors contribute to food quality, such as biological origin, ecological conditions, microbiome, and processing methods. Particularly, fermented foods contain more complex factors and mechanisms than other food types. Gochujang (fermented red pepper paste) is a traditional food that has unique properties, including spicy flavor, pungency, sweet taste, and distinctive texture. It is rich in nutrient components, such as essential amino acids, fatty acids, organic acids, and sugars, which are generated from raw materials during fermentation [3]. Furthermore, gochujang has various functional and biological activities, such as anti-obesity [4,5], anti-tumor [6], and anti-cancer [7] effects. The general manufacturing process for commercial gochujang is complex because of the combination of various raw materials and the diverse conditions required for production, which are summarized in Figure1. Because of these complexities, it is difficult to control the quality of commercial gochujang. Thus, it is important to identify the major factors, as well as their mechanisms of action, affecting the quality of commercial gochujang.

Molecules 2016, 21, 921; doi:10.3390/molecules21070921 www.mdpi.com/journal/molecules Molecules 2016, 21, 921 2 of 14

Molecules 2016, 21, 921 2 of 14

Figure 1. The schematic diagram of the general manufacturing process and the major raw ingredients Figure 1. The schematic diagram of the general manufacturing process and the major raw ingredients for commercial gochujang. * Additives; salt, starch syrup, garlic, onion, and soy powder are typically for commercialon the ingredient gochujang. label *of Additives; the products. salt, starch syrup, garlic, onion, and soy powder are typically on the ingredient label of the products. Metabolomics have been developed in numerous fields, including pharmacology, toxicology, plant physiology, food science, and nutrition [8–11]. Metabolomics techniques are very important Metabolomics have been developed in numerous fields, including pharmacology, toxicology, tools in food science in particular to ensure food safety, quality, and traceability [12]. For example, plant physiology,these methods food can science, be applied and to nutrition detect contaminants, [8–11]. Metabolomics pathogens, and techniques toxins in food are [13]; very to importantevaluate tools in foodthe science quality in of particular fruits or beverages to ensure based food on metabolite safety, quality, profiles and [14]; traceabilityand to monitor [12 metabolic]. For example, changes these methodsduring can bestorage applied and fermentation to detect contaminants,[15,16]. Metabolomi pathogens,cs approaches and are toxins actively in used food to [investigate13]; to evaluate the qualityfermented of fruits foods orsuch beverages as cheeses, basedwines, fermented on metabolite soy products, profiles and [probiotics14]; and [17]. to Metabolomics monitor metabolic approaches have been used to evaluate gochujang to confirm the bioactivities of extracts or specific changes during storage and fermentation [15,16]. Metabolomics approaches are actively used to compounds and to investigate the bioconversion of compounds during fermentation [3–7]. However, investigatefew studies fermented have examined foods such differences as cheeses, in metabolites wines, to fermented characterize soy commercial products, gochujang. and probiotics [17]. MetabolomicsIn this approaches study, the haveuntargeted been metabolite used to evaluate profiling gochujangof various commercial to confirm gochujang the bioactivities preparations of extracts or specificwas compounds performed using and GC-TOF-MS to investigate and the ultrahigh bioconversion performance of compoundsliquid chromatography-linear during fermentation trap [3–7]. However,quadrupole-ESI-ion few studies have trap-MS/MS examined (UHPLC-LTQ-ESI-IT- differences in metabolitesMS/MS) with to characterize multivariate statistical commercial analysis gochujang. In thisto investigate study, the the untargetedmajor factors metabolitecontributing profilingto the metabolite of various state and commercial physicochemical gochujang quality preparationsof the final commercial form of gochujang. In addition, we propose specific metabolites as quality markers for was performed using GC-TOF-MS and ultrahigh performance liquid chromatography-linear trap commercial gochujang. quadrupole-ESI-ion trap-MS/MS (UHPLC-LTQ-ESI-IT-MS/MS) with multivariate statistical analysis to investigate2. Results the major factors contributing to the metabolite state and physicochemical quality of the final commercial form of gochujang. In addition, we propose specific metabolites as quality markers 2.1. Metabolite Profiling of Commercial Gochujang Samples for commercial gochujang. To investigate the factors affecting the differences among the final products of a diverse 2. Resultsgroup of gochujang samples (Figure 1 and Table 1) and to characterize the metabolites affecting the physicochemical quality of commercial gochujang, multivariate statistical analysis was performed using 2.1. Metabolitethe GC-TOF-MS Profiling and of CommercialUHPLC-LTQ-ESI-IT Gochujang-MS/MS Samples data sets (Figure 2). The variance of each PCs of the three-dimensional (3D) PCA score plots were PC1 (12.3%), PC2 (11.9%), and PC3 (7.6%) (Figure 2A) and ToPC1 investigate (9.9%), PC2 the (6.7%), factors and affectingPC3 (6.3%) the(Figure differences 2B). Although among the total the variances final products are low, the of aexplained diverse group of gochujangvariances samples (R2) values (Figure of PCA models1 and were Table over1) 0.4 and (R2X to(cum)characterize = 0.582 and 0.477), the reasonably. metabolites And the affecting distinct the physicochemical quality of commercial gochujang, multivariate statistical analysis was performed using the GC-TOF-MS and UHPLC-LTQ-ESI-IT-MS/MS data sets (Figure2). The variance of each PCs of the three-dimensional (3D) PCA score plots were PC1 (12.3%), PC2 (11.9%), and PC3 (7.6%) (Figure2A) and PC1 (9.9%), PC2 (6.7%), and PC3 (6.3%) (Figure2B). Although the total variances are 2 2 low, the explained variances (R ) values of PCA models were over 0.4 (R X(cum) = 0.582 and 0.477), reasonably. And the distinct patterns for commercial gochujang are shown based on their raw materials Molecules 2016, 21, 921 3 of 14

Moleculespatterns2016, 21, 921for commercial gochujang are shown based on their raw materials in the 3D PCA 3score of 14 plots. As shown in Figure 2, the samples were mainly divided into two groups by cereal type, gochujang Molecules 2016, 21, 921 3 of 14 made with wheat (WG) and gochujang made with rice (RG). Within the cluster of RG, there was in the 3D PCA score plots. As shown in Figure2, the samples were mainly divided into two groups patternsdifferentiation for commercial into groups gochujang of goarechujang shown based prepared on their using raw materials brown rice in the (RbG) 3D PCA and score white plots. rice (RwG). In by cereal type, gochujang made with wheat (WG) and gochujang made with rice (RG). Within the Asthe shown expanded in Figure figures, 2, the samplesthe two were clusters mainly of dividedRbG and into RwG two groups were separatedby cereal type, according gochujang to hot pepper cluster of RG, there was differentiation into groups of gochujang prepared using brown rice (RbG) and madespecies with used wheat to produce (WG) and gochujang. gochujang Themade RbG with cluster rice (RG). was Within separated the cluster into two of RG,groups there based was on the PCA white rice (RwG). In the expanded ﬁgures, the two clusters of RbG and RwG were separated according differentiationscore plot, including into groups one of withgochujang C. annuum prepared (CA) using and brown another rice (RbG) mixed and with white C. riceannuum (RwG). cv. In Chung-yang to hotthe pepper expanded species figures, used the to two produce clusters gochujang. of RbG and TheRwG RbG were cluster separated was according separated to intohot pepper two groups (CAY). RwG was also divided into two clusters as follows: one prepared with CA and the other basedspecies on the used PCA toscore produce plot, gochujang. including The one RbG with clusterC. was annuum separated(CA) into and two another groups mixedbased on with the C.PCA annuum containing C. frutescens (CF). In summary, 18 commercial gochujang samples were separated and cv. Chung-yangscore plot, including (CAY). RwG one with was C. also annuum divided (CA) into and two another clusters mixed as with follows: C. annuum one prepared cv. Chung-yang with CA and clustered based on cereal grain type (WG, RbG, and RwG) and secondarily on hot pepper species the other(CAY). containing RwG wasC. also frutescens divided(CF). into Intwo summary, clusters as 18 follows: commercial one prepared gochujang with samples CA and werethe other separated containing(RbG-CA, C. RbG-CAY, frutescens (CF). RwG-CA, In summary, and RwG-CF) 18 commercial in the gochujang PCA score samples plots (Figurewere separated 2). and and clustered based on cereal grain type (WG, RbG, and RwG) and secondarily on hot pepper species clustered based on cereal grain type (WG, RbG, and RwG) and secondarily on hot pepper species (RbG-CA,(RbG-CA, RbG-CAY, RbG-CAY, RwG-CA, RwG-CA, and and RwG-CF) RwG-CF) in in thethe PCAPCA score score plots plots (Figure (Figure 2).2 ).

Figure 2. Principal component analysis (PCA) score plots of commercial gochujang derived from the Figure 2. Principal component analysis (PCA) score plots of commercial gochujang derived from the GC-TOF-MSFigure 2. Principal(A) and UHPLC-LTQ-IT-MS/MS component analysis ( B(PCA)) data sets; score ● QC, plot qualitys of commercial control samples; gochuj ● WG,ang wheat derived from the GC-TOF-MS (A) and UHPLC-LTQ-IT-MS/MS (B) data sets; ‚ QC, quality control samples; ‚ WG, wheat gochujang;GC-TOF-MS □ RG (A, rice) and gochujang UHPLC-LTQ-IT-MS/MS (including RbG and (B RwG);) data ●sets; RbG, ● QC,brown quality rice gochujang; control samples; ● RwG, ● WG, wheat gochujang;white rice˝ RG gochujang;, rice gochujang ● RbG-CA, (including brown rice gochujang RbG and containing RwG); ‚ C.RbG, annuum brown; ● RbG-CAY, rice gochujang; brown rice‚ RwG, gochujang; □ RG, rice gochujang (including RbG and RwG); ● RbG, brown rice gochujang; ● RwG, whitegochujang rice gochujang; containing‚ C.RbG-CA, annuum cv. brown Chung-yang; rice gochujang ● RwG-CA, containingwhite rice gochujangC. annuum containing; ‚ RbG-CAY, C. annuum; brown white rice gochujang; ● RbG-CA, brown rice gochujang containing C. annuum; ● RbG-CAY, brown rice rice gochujang● RwG-CF, containingwhite rice gochujangC. annuum containingcv. Chung-yang; C. frutescens. Explained‚ RwG-CA, variances white (R rice2X(cum) gochujang) and predictive containing gochujang2 containing C. annuum cv. Chung-yang; ● RwG-CA, white rice gochujang containing C. annuum; C. annuumabilities; ‚ (QRwG-CF,(cum)) of PCA white models rice gochujangwere 0.582 and containing 0.342 in (AC.) and frutescens 0.477 and. Explained 0.186 in (B), variances respectively. (R2 X ) 2 (cum) ● RwG-CF, white rice gochujang containing C. frutescens. Explained variances (R2X(cum)) and predictive and predictive abilities (Q2 ) of PCA models were 0.582 and 0.342 in (A) and 0.477 and 0.186 in 2 (cum) abilities (Q2(cum)) ofTable PCA 1. models Information were for 0.582 commercial and 0.342 gochujang in (A) andsamples. 0.477 and 0.186 in (B), respectively. (B), respectively. Sample Species of Type of Cereal Pungency b Abbreviation c Table 1. Information fora commercial gochujang samples. No. Table 1. InformationHot for Pepper commercial gochujang samples. 1 Sample Species of 2 WheatType gochujang of Cereal Pungency b Abbreviation c Capsicum annuum a a b WG-CA c Sample No.No. Type of Cereal SpeciesHot of Hot Pepper Pepper Abbreviation 3No. (WG) Hot Pepper Pungency 1 4 1 5 2 Wheat gochujang 2 Wheat gochujang Capsicum annuum WG-CA Capsicum annuum WG-CA Capsicum annuum 6 Capsicum annuum RbG-CAWG-CA 3 (WG) 3 3 Brown(WG) (WG)rice 7 4 4 gochujang (RbG) 8 Capsicum annuum 5 RbG-CAY 5 9 cv. Chung-yang 6 Capsicum annuum RbG-CA 6 Capsicum annuum RbG-CA 6 Capsicum annuum BrownBrown rice rice RbG-CA 7 7 gochujanggochujang (RbG) (RbG) 8 8 CapsicumCapsicum annuum annuumcv. RbG-CAY RbG-CAYRbG-CAY 9 9 Chung-yangcv. Chung-yang

Molecules 2016, 21, 921 3 of 14 Molecules 2016, 21, 921 3 of 14 patterns for commercial gochujang are shown based on their raw materials in the 3D PCA score plots. As shown in Figurepatterns 2, the samples for commercial were mainly gochujang divided are into shown two based groups on bytheir cereal raw materialstype, gochujang in the 3D PCA score plots. made with wheat (WG)As shown and gochujangin Figure 2, made the samples with rice were (RG). mainly Within divided the cluster into two of RG,groups there by wascereal type, gochujang differentiation into madegroups with of go wheatchujang (WG) prepared and gochujang using brown made rice with (RbG) rice and (RG). white Within rice (RwG).the cluster In of RG, there was the expanded figures,differentiation the two clusters into groups of RbG of and gochujang RwG were prepared separated using according brown rice to (RbG) hot pepper and white rice (RwG). In species used to producethe expanded gochujang. figures, The RbG the cluster two clusters was separated of RbG into and two RwG groups were based separated on the according PCA to hot pepper score plot, includingspecies one with used C. to annuum produce (CA) gochujang. and another The RbG mixed cluster with was C. separatedannuum cv. into Chung-yang two groups based on the PCA (CAY). RwG was alsoscore divided plot, including into two one clusters with C.as annuumfollows: (CA)one preparedand another with mixed CA andwith the C. otherannuum cv. Chung-yang containing C. frutescens(CAY). (CF). RwG In wassummary, also divided 18 commercial into two gochujang clusters as samples follows: were one preparedseparated withand CA and the other clustered based on containingcereal grain C. type frutescens (WG, (CF).RbG, Inand summary, RwG) and 18 secondarily commercial on gochujang hot pepper samples species were separated and (RbG-CA, RbG-CAY,clustered RwG-CA, based and on RwG-CF) cereal grain in the type PCA (WG, score RbG, plots and (Figure RwG) 2). and secondarily on hot pepper species (RbG-CA, RbG-CAY, RwG-CA, and RwG-CF) in the PCA score plots (Figure 2).

Figure 2. Principal component analysis (PCA) score plots of commercial gochujang derived from the GC-TOF-MS (A) and UHPLC-LTQ-IT-MS/MSFigure 2. Principal component (B) data analysis sets; ● QC, (PCA) quality score control plots samples;of commercial ● WG, gochujwheat ang derived from the gochujang; □ RG, riceGC-TOF-MS gochujang (including(A) and UHPLC-LTQ-IT-MS/MS RbG and RwG); ● RbG, (B) browndata sets; rice ● QC,gochujang; quality control● RwG, samples; ● WG, wheat white rice gochujang; gochujang;● RbG-CA, brown□ RG, rice gochujang containing(including C.RbG annuum and ;RwG); ● RbG-CAY, ● RbG, brown brown rice rice gochujang; ● RwG, gochujang containing C. annuum cv. Chung-yang; ● RwG-CA, white rice gochujang containing C. annuum; Molecules 2016, 21, 921 white rice gochujang; ● RbG-CA, brown rice gochujang containing C. annuum; ● 4RbG-CAY, of 14 brown rice Molecules 2016●, 21 RwG-CF,, 921 white rice gochujanggochujang containing containing C. C. annuum frutescens cv.. Chung-yang;Explained variances ● RwG-CA, (R2X white(cum)) and rice predictivegochujang4 of 14 containing C. annuum; 2 abilities (Q (cum)) of PCA● RwG-CF, models were white 0.582 rice gochujangand 0.342 in containing (A) and 0.477 C. frutescens and 0.186. Explained in (B), respectively. variances (R 2X(cum)) and predictive Table 1. Cont. 2 Table 1.abilitiesInformation (Q (cum) for) of commercial PCA models gochujang were 0.582 samples. and 0.342 in (A) and 0.477 and 0.186 in (B), respectively. Sample Table 1. InformationSpecies for commercial of gochujang samples. Type of Cereal Pungency b Abbreviation c No. Hot TablePepper 1. aInformation for commercial gochujang samples. SampleSample No. Type of Cereal Species ofSpecies Hot Pepper of a Pungency b Abbreviation c Type of Cereal Pungency b Abbreviation c No.10 Hot Pepper a 10 Sample Species of 11 Type of Cereal Pungency b Abbreviation c 11 1 No. Hot Pepper a 12 12 2 Wheat gochujang Capsicum annuum RwG-CA 1 Capsicum annuum WG-CA 13 Capsicum annuum RwG-CA 13 3 White(WG) rice 2 Wheat gochujang 14 White rice Capsicum annuum WG-CA 14 4 gochujang (RwG) 15 gochujang (RwG)3 (WG) 15 5 16 4 16 6 Capsicum annuum RbG-CA 17 Brown rice5 CapsicumCapsicum frutescens frutescens RwG-CF 17 RwG-CF 7 6 Capsicum annuum RbG-CA 18 18 gochujang (RbG) 8 CapsicumBrown rice annuum a a 7 b RbG-CAYb Main speciesMain of species pepper of powder pepper were powder indicated gochujangwere by indica the ingredients (RbG)ted by the label ingredients of products. labelRelative of products. pungency Relative 9 8 cv. Chung-yang Capsicum annuum RbG-CAY pungency levels of hot peppers classified by Scoville heat units (SHU); mildly pungent levels of hot peppers classified by Scoville9 heat units (SHU); mildlycv. pungentChung-yang (700–3000 SHU); (700–3000 SHU), moderately pungent (3000–25,000 SHU), and highly pungent c moderately(25,000–70,000 pungent (3000–25,000 SHU) [18]. SHU); c Abbreviation and highly of total pungent samples (25,000–70,000 used in the SHU) text;[ 18WG,]. wheatAbbreviation gochujang; of total samples used in the text; WG, wheat gochujang; RG, rice gochujang; RbG, brown rice gochujang; RwG, white riceRG, gochujang; rice gochujang; RbG-CA, brownRbG, brown rice gochujang rice gochujan containingg; RwG,C. annuum white; RbG-CAY, rice gochujang; brown rice RbG-CA, gochujang brown containingriceC. gochujang annuum cv. Chung-yang;containing C. RwG-CA, annuum white; RbG-CAY, rice gochujang brown containing rice gochujangC. annuum containing; RwG-CF, whiteC. annuum C. frutescens rice gochujangcv. Chung-yang; containing RwG-CA,. white rice gochujang containing C. annuum; RwG-CF, white rice gochujang containing C. frutescens. 2.2. Comparisons of Metabolites, Physicochemical Characteristics, and Antioxidant Activity According to Different Cereal Grains Used in Gochujang 2.2. Comparisons of Metabolites, Physicochemical Characteristics, and Antioxidant Activity According to ToDifferent compare Cereal the Grains differences Used in in Gochujang metabolites of gochujang according to the type of cereal grain used for preparation, PLS-DA was conducted using the GC-TOF-MS (Figure3A) and To compare the differences in metabolites of gochujang according to the type of cereal grain used for UHPLC-LTQ-ESI-IT-MS/MSpreparation, PLS-DA was data conducted (Figure using3B). The the threeGC-TOF-MS groups (Figure of WG, 3A) RbG, and and UHPLC-LTQ-ESI-IT-MS/MS RwG were clearly separateddata from(Figure each 3B). other The basedthree groups on the of PLS-DA WG, RbG, score and plots. RwG The were WG clearly samples separated were separatedfrom each fromother based both theon RbGthe PLS-DA and RwG score groups plots. byThe PLS1 WG samples (12.9% inwere Figure separated3A and from 7.1% both in Figurethe RbG3B), and while RwG the groups RbG by PLS1 group(12.9% was distinct in Figure from 3A the and WG 7.1% and in RwGFigure groups 3B), wh accordingile the RbG to group PLS2 (10.8%was distinct in Figure from3A the and WG 6.6% and RwG in Figuregroups3B). Significantlyaccording to PLS2 different (10.8% variables in Figure were 3A selectedand 6.6% by in PLS1 Figure and 3B). PLS2 Significantly in the PLS-DA different score variables p plots basedwere onselected variable by PLS1 importance and PLS2 in projection in the PLS-DA (VIP) score values plots > 0.7 based and on-values variable < 0.05. importance Next, primary in projection and secondary(VIP) values metabolites > 0.7 and (based p-values on < GC-MS 0.05. Next, and LC-MS/MSprimary and data second sets,ary respectively) metabolites were(based tentatively on GC-MS and λ identifiedLC-MS/MS by retention data sets, time, respectively) molecular weight,were tentatively ultraviolet identified light lambda-max by retention ( maxtime,, nm), molecular MS/MS weight, fragment patterns, in-house library, library of National Institute of Standards and Technology (NIST ultraviolet light lambda-max (λmax, nm), MS/MS fragment patterns, in-house library, library of National MS SearchInstitute program, of Standards version and 2.0, Technology Gaithersburg, (NIST MD,MS Search USA), program, standard version compounds, 2.0, Gaithersburg, and references MD, USA), (Tablesstandard S1 and S2).compounds, and references (Tables S1 and S2). SignificantlySignificantly different different metabolites, metabolites, including including amino amino acids, organicacids, organic acids, fattyacids, acids, fatty sugarsacids, sugars and and sugar alcohols,sugar alcohols, flavonoids, flavonoids, capsaicinoids capsaicinoids and capsinoids, and capsinoids, and lipids, and lipids, according according to the to type the of type cereal of cereal grainsgrains used inused gochujang in gochujang were identified,were identified, and their and relativetheir relative contents contents were determinedwere determined (Table (Table S3). S3). AmongAmong them, metabolitesthem, metabolites with significantly with significantly distinct distinct patterns patterns (p-value (p-value < 0.05) < such 0.05) as such unique as unique high or high or low contentslow contents in specific in specific gochujang gochujang groups groups (WG, RbG,(WG, andRbG, RwG) and RwG) were visualizedwere visualized with a with heat a map heat map (Figure(Figure3C). The 3C). heat The map heat was map prepared was usingprepared the fold-change using the fold-change values normalized values to normalized the average to relative the average contentsrelative of each contents metabolite of each in metabolite all gochujang in all samples. gochujang The samples. levels ofThe most levels amino of most acids amino [leucine acids (1 [leucine), valine( (12),), valine isoleucine (2), isoleucine (3), proline (3), (4proline), glycine (4), (glycine5), serine (5), ( 6serine), threonine (6), threonine (7), pyroglutamic (7), pyroglutamic acid ( 9acid), (9), glutamicglutamic acid (11 acid), and (11), phenylalanine and phenylalanine (12)], citric(12)], acidcitric (16 acid), xylose (16), xylose (23), and (23 3), secondaryand 3 secondary metabolites metabolites [apigenin-[apigenin-C-hexoside-C-hexoside-C-pentosideC-pentoside (32), dihydrocapsiate(32), dihydrocapsiate (49), and(49), linoleic and linoleic ethanolamide ethanolamide (68)] were (68)] were distinctivelydistinctively higher higher in the WGin the group WG group than in than the otherin the samples.other samples. Lipid-related Lipid-related molecules molecules such as such fatty as fatty acids [linoleicacids [linoleic acid (19 acid) and (19 oleic) and acid oleic (20 acid)] and (20)] lysophospholipids and lysophospholipids [lysoPC [lysoPC 14:0 (55 14:0), lysoPC (55), lysoPC 18:2 (60 18:2), (60), lysoPClysoPC 16:0 ( 6316:0), lysoPC(63), lysoPC 18:1 ( 6518:1), and(65), lysoPE and lysoPE 18:2 ( 5818:2)], sucrose(58)], sucrose (29), and(29), two and other two secondaryother secondary metabolitesmetabolites including including luteolin- luteolin-C-hexosideC-hexoside (34) and (34) quercetin- and quercetin-O-rhamnosideO-rhamnoside (37) were (37) presentwere present at at significantlysignificantly higher higher levels inlevels the RbGin the group RbG thangroup in thethan other in the samples. other samples. For RwG, For some RwG, sugars some such sugars as such glucoseas ( 26glucose) and maltose(26) and ( 31maltose) were ( present31) were at present higher levelsat higher than levels in the than other in samples.the other Insamples. addition, In theaddition, γ relativethe contents relative of contents two other of primarytwo other metabolites primary metabolites ( -aminobutyric (γ-aminobutyric acid (GABA) acid (10 (GABA)) and palmitic (10) and acid palmitic

Molecules 2016, 21, 921 5 of 14

Molecules 2016, 21, 921 5 of 14 (18)) and two secondary metabolites (genistein-O-dihexoside (40) and dihydrocapsaicin (52)) were also highacid in the (18 RwG)) and group. two secondary metabolites (genistein-O-dihexoside (40) and dihydrocapsaicin (52)) Thewere results also high for in antioxidant the RwG group. activity, TPC, TFC, and physicochemical characteristic assays (amino The results for antioxidant activity, TPC, TFC, and physicochemical characteristic assays (amino type nitrogen, acidity, pH, salinity, and reducing sugar content) for the commercial gochujang samples type nitrogen, acidity, pH, salinity, and reducing sugar content) for the commercial gochujang are presented according to the type of cereal grain in Figure3 and Figure S1. The values for amino type samples are presented according to the type of cereal grain in Figure 3 and Figure S1. The values for nitrogenamino and type TPC nitrogen were and the highestTPC were in the WG highest (Figure in 3WGD,G), (Figure as were 3D,G), the as values were the for values acidity for and acidity salinity (Figureand S1). salinity Antioxidant (Figure S1). activity, Antioxidant TFC, andactivity, pH TFC, were and higher pH were in WG higher and in RbG WG thanand RbG in RwG than (Figurein RwG 3E,F and Figure(Figure S1), 3E,F whereas and Figure reducing S1), whereas sugar contentreducing was sugar signiﬁcantly content was higher significan in WGtly higher and RwG in WG than and in RbG (FigureRwG S1). than in RbG (Figure S1).

FigureFigure 3. Partial 3. Partial least least square-discriminate square-discriminate analysisanalysis (PLS-DA) (PLS-DA) score score plots plots using using the theGC-TOF-MS GC-TOF-MS (A); (A); UHPLC-LTQ-ESI-IT-MS/MS (B) data sets, heat map of specific metabolites (C); amino type nitrogen UHPLC-LTQ-ESI-IT-MS/MS (B) data sets, heat map of specific metabolites (C); amino type nitrogen contents (D); antioxidant activity tested by ABTS (E); total flavonoid contents (TFC) (F); and total contents (D); antioxidant activity tested by ABTS (E); total flavonoid contents (TFC) (F); and total polyphenolic contents (TPC) (G) according to the type of cereal grain used in commercial gochujang; polyphenolic▼ WG, wheat contents gochujang; (TPC) ♦ ( GRbG,) according brown rice to gochujang; the type of ●cereal RwG, white grain rice used gochujang. in commercial The numbers gochujang; İ WG,in wheatthe heat gochujang; map correspond RbG, to the brown compound rice gochujang; numbers presented‚ RwG, in white Table rice S3. gochujang.Different letters The in numbers the in thebar heat graph map indicate correspond significant to the difference compound by ANOVA numbers followed presented by Duncan’s in Table S3.multiple-range Different letterstest in the bar(p-value graph < 0.05). indicate significant difference by ANOVA followed by Duncan’s multiple-range test (p-value < 0.05). 2.3. Metabolite Variations and Antioxidant Activity According to Species of Hot Pepper Used in Gochujang 2.3. MetaboliteAccording Variations to the andPCA Antioxidant models, which Activity showed According different to patterns Species according of Hot Pepper to minor Used factors in Gochujang and the ingredient labels on the products, the gochujang samples were categorized based on the different According to the PCA models, which showed different patterns according to minor factors species of hot pepper present, including RbG-CA, RbG-CAY, RwG-CA, and RwG-CF (Table 1 and and the ingredient labels on the products, the gochujang samples were categorized based on the Figure 2). To investigate the metabolites contributing to differentiation according to species of hot differentpepper species in RbG of or hot RwG, pepper OPLS-DA present, was including conducted RbG-CA, using the MS RbG-CAY, spectrum RwG-CA, data set from and UHPLC-LTQ- RwG-CF (Table1 and FigureESI-IT-MS/MS2). To (Figure investigate 4A,D), the and metabolites variables were contributing selected using to VIP differentiation values (>1) and according p-values (<0.05). to species of hotThese pepper metabolites in RbG are or presented RwG, OPLS-DA in Table S2. was Amon conductedg them, specific using thepepper-related MS spectrum variables data were set from UHPLC-LTQ-ESI-IT-MS/MShighlighted in the loading (FigureS-plots (Figure4A,D), 4B,F), and variables and their wererelative selected contents using were VIPvisualized values using (>1) and p-valuesbox-and-whisker (<0.05). These plots metabolites (Figure 4D,H). are presentedThe antioxidant in Table activity, S2. AmongTFC, and them, TPC results specific are pepper-related presented variablesin Figure were 4C,G. highlighted In a comparison in the of loading RbG-CA S-plots and RbG-CAY (Figure 4(FigureB,F), and4B–D), their a total relative of 19 metabolites contents were visualizedwere identified using box-and-whisker as differential metabolites plots (Figure (Table4D,H). S2), among The antioxidantwhich the relative activity, levels TFC, of capsaicinoids and TPC results and capsinoids, including nordihydrocapsaicin (50), capsaicin (51), dihydrocapsaicin (52), and are presented in Figure4C,G. In a comparison of RbG-CA and RbG-CAY (Figure4B–D), a total of dihydrocapsiate (49), were significantly higher in RbG-CAY than in RbG-CA. These patterns were 19 metabolites were identified as differential metabolites (Table S2), among which the relative levels of capsaicinoids and capsinoids, including nordihydrocapsaicin (50), capsaicin (51), dihydrocapsaicin (52), and dihydrocapsiate (49), were significantly higher in RbG-CAY than in RbG-CA. These patterns were Molecules 2016, 21, 921 3 of 14

patterns for commercial gochujang are shown based on their raw materials in the 3D PCA score plots. As shown in Figure 2, the samples were mainly divided into two groups by cereal type, gochujang made withMolecules wheat 2016 (WG), 21, 921and gochujang made with rice (RG). Within the cluster of RG, there was 6 of 14 differentiation into groups of gochujang prepared using brown rice (RbG) and white rice (RwG). In Molecules 2016, 21, 921 similar to the results for ABTS, TFC, and TPC, which3 of 14 were significantly higher in RbG-CAY than in the expanded figures, the two clusters of RbG and RwG were separated according to hot pepper MoleculesRbG-CA.2016 In, 21a, comparison 921 of RwG-CA and RwG-CF (Figure 4F–H), a total of 30 metabolites 6were of 14 species used to produce gochujang. The RbG cluster was separated into two groups based on the PCA patterns for commercial gochujang are shown basedidentified on their (Table raw S2), materials among in which the 3D the PCA relative score levelsplots. of luteolin-O-apiosyl-glucoside (33), luteolin- score plot, includingMolecules one2016 , with21, 921 C. annuum (CA) and another mixed with C. annuum cv. Chung-yang6 of 14 As shown in Figure 2, the samples were mainlyC-hexoside divided into(34), two luteolin groups (43 by), dihydrocapsiatecereal type, gochujang (49), nordihydrocapsaicin (50), capsaicin (51), and (CAY). RwGsimilar was to also the divided results forinto ABTS, two clusters TFC, and as TPC,follows: which one were prepared signiﬁcantly with CA higher and the in RbG-CAYother than made with wheat (WG) and gochujang madeindihydrocapsaicin with RbG-CA.similar rice (RG).to In the a resultsWithin( comparison52) were for the ABTS, higher cluster of TFC, RwG-CA in of RwG-CF and RG, TPC, andthere thanwhich RwG-CF was in were RwG-CA significantly (Figure (Figure4F–H), higher 4H). a total in RbG-CAY of 30 metabolites than in containing C. frutescensRbG-CA. In (CF). a comparison In summary, of RwG-CA 18 commercial and RwG-CF gochujang (Figure 4F–H),samples a totalwere of separated 30 metabolites and were differentiation into groups of gochujang preparedwere using identiﬁed brown (Table rice (RbG) S2), among and white which rice the(RwG). relative In levels of luteolin-O-apiosyl-glucoside (33), clustered basedidentified on cereal (Table grain S2), type among (WG, which RbG, the and relative RwG) levels and of secondarily luteolin-O-apiosyl-glucoside on hot pepper (species33), luteolin- the expanded figures, the two clusters of RbGluteolin- and RwGC-hexoside were separated (34), luteolin according (43), dihydrocapsiate to hot pepper (49), nordihydrocapsaicin (50), capsaicin (51), (RbG-CA, RbG-CAY,C-hexoside RwG-CA, (34), luteolin and RwG-CF) (43), dihydrocapsiate in the PCA score(49), nordihydrocapsaicin plots (Figure 2). (50), capsaicin (51), and species used to produce gochujang. The RbG clusterand dihydrocapsaicin was separated into (52 two) were groups higher based in RwG-CF on the PCA than in RwG-CA (Figure4H). score plot, including one with C. annuum (CA) anddihydrocapsaicin another mixed (with52) were C. annuumhigher in cv.RwG-CF Chung-yang than in RwG-CA (Figure 4H). (CAY). RwG was also divided into two clusters as follows: one prepared with CA and the other containing C. frutescens (CF). In summary, 18 commercial gochujang samples were separated and clustered based on cereal grain type (WG, RbG, and RwG) and secondarily on hot pepper species (RbG-CA, RbG-CAY, RwG-CA, and RwG-CF) in the PCA score plots (Figure 2).

Figure 2. Principal component analysis (PCA) score plots of commercial gochujang derived from the GC-TOF-MS (A) and UHPLC-LTQ-IT-MS/MS (B) data sets; ● QC, quality control samples; ● WG, wheat gochujang; □ RG, rice gochujang (including RbG and RwG); ● RbG, brown rice gochujang; ● RwG, white rice gochujang; ● RbG-CA, brown rice gochujang containing C. annuum; ● RbG-CAY, brown rice gochujang containing C. annuum cv. Chung-yang; ● RwG-CA, white rice gochujang containing C. annuum; ● RwG-CF, white rice gochujang containing C. frutescens. Explained variances (R2X(cum)) and predictive Figure 2. Principal component analysis (PCA) score plots of commercial gochujang derived from the abilities (Q2(cum)) of PCA models were 0.582 and 0.342 in (A) and 0.477 and 0.186 in (B), respectively. GC-TOF-MS (A) and UHPLC-LTQ-IT-MS/MS (B) data sets; ● QC, quality control samples; ● WG, wheat Molecules 2016, 21gochujang;, 921 □ RG, rice gochujang (including RbG and RwG); ● RbG, brown rice gochujang;4 ●of RwG,14 Table 1. Information for commercial gochujang samples. white rice gochujang; ● RbG-CA, brown rice gochujang containing C. annuum; ● RbG-CAY, brown rice Sample Species of gochujang containing C. annuum cv. TableChung-yang; 1. Cont ●. RwG-CA,Type white of Cereal rice gochujang containing C. annuum; Pungency b Abbreviation c a ● RwG-CF, white rice gochujang containing No.C. frutescens . Explained variances (R2HotX(cum) Pepper) and predictive Sample 2 Species of 1 FigureFigure 4. Orthogonal 4. Orthogonal partial partial least least squares squares discriminant discriminant analysis analysis (OPLS-DA) (OPLS-DA) score score plots plots (A,E (),A loading,E), loading abilitiesType (Q (cum) of Cereal) of PCA models were 0.582 and 0.342FigurePungency in (A 4.) andOrthogonal b 0.477 and partialAbbreviation 0.186 leastin (B squares), respectively. c discriminant analysis (OPLS-DA) score plots (A,E), loading No. Hot Pepper a 2 S-plotsWheatS-plots (B ,gochujangF (),B ,box-and-whiskerF), box-and-whisker plots plots ( D(D,H,H)) ofof significantly different different metabolites metabolites derived derived from from the the S-plots (B,F), box-and-whiskerCapsicum plots annuum (D,H) of significantly different metabolitesWG-CA derived from the UHPLC-LTQ-IT-MS/MS data set, and the results for ABTS, TFC, and TPC (C,G) for brown rice (A–D) 10 Table 1. Information for3 commercialUHPLC-LTQ-IT-MS/MS gochujang(WG) samples. data data set, and th thee results for ABTS, TFC, and TPC (C,G) for brown rice (A–D) 11 4 and white rice (E–H) commercial gochujang according to species of hot pepper present; RbG-CA, and whitewhite ricerice ((EE––HH)) commercialcommercial gochujanggochujang accordingaccording to speciesspecies ofof hothot pepperpepper present;present; RbG-CA,RbG-CA, 12 Sample Species5 of brown rice gochujang containing C. annuum; RbG-CAY, brown rice gochujang containing C. annuum Type of CerealCapsicum annuum brown ricePungency gochujang b containing RwG-CAAbbreviation C. annuum ;; c RbG-CAY,brown brownrice rice gochujanggochujang containingcontaining C.C. annuum a cv. Chung-yang; ● RwG-CA, white rice gochujang containing C. annuum; ● RwG-CF, white rice 13 No. Hot Pepper6 Capsicum annuum RbG-CA White rice cv.cv. Chung-yang; Chung-yang;gochujangBrown rice containing ●‚ RwG-CA, C. frutescens. white Relative rice gochujang pungency levelscontaining of hot peppersC. annuum classified;; ●‚ RwG-CF, by Scoville whitewhite heat ricerice 14 1 7 gochujang (RwG) gochujanggochujangunits (SHU); containing (RbG) C.C. mildly frutescens.frutescens. pungent Relative (700–3000 pungency SHU), levels moderately of hot peppers pungent classifiedclassified (3000–25,000 by Scoville SHU), heat 15 2 Wheat gochujang 8 Capsicum annuum Capsicum annuum WG-CA unitsand (SHU); highly mildly pungent pungent (25,000–70,000 (700–3000 SHU) SH [18].U), The metabolite moderately numbers pungentRbG-CAY are identical (3000–25,000 to those SHU), 16 3 (WG) 9 units (SHU); mildlycv. pungent Chung-yang (700–3000 SHU), moderately pungent (3000–25,000 SHU), and in Table highly S2. Different pungent letters (25,000–70,000 in the bar graph SHU) indi cate[18]. significant The metabolite differences numbers by ANOVA are identical followed to by those 17 4 Capsicum frutescens RwG-CF and Duncan’s highly multiple-range pungent (25,000–70,000test (p-value < 0.05). SHU) The [ 18red]. pepper The metabolite diagram is numbersthe relative are pungency identical levels to those 18 5 in Table S2. Different letters in the bar graph indicate significant differences by ANOVA followed by in Tableof hot S2. pepper. Different letters in the bar graph indicate significant differences by ANOVA followed by a 6 Capsicum annuumDuncan’s multiple-range testb (p-valueRbG-CA < 0.05). The red pepper diagram is the relative pungency levels Main species of pepper powder were indicated by the ingredients label of products. Relative Brown rice Duncan’s multiple-range test (p-value < 0.05). The red pepper diagram is the relative pungency levels 7 of hot pepper. pungency levels of hotgochujang peppers (RbG) classified by Scoville heat units (SHU); mildly pungent 8 Capsicum annuumof hot pepper. (700–3000 SHU), moderately pungent (3000–25,000 SHU), and highly pungentRbG-CAY

(25,000–70,0009 SHU) [18]. c Abbreviation of totalcv. Chung-yangsamples used in the text; WG, wheat gochujang; RG, rice gochujang; RbG, brown rice gochujang; RwG, white rice gochujang; RbG-CA, brown rice gochujang containing C. annuum; RbG-CAY, brown rice gochujang containing C. annuum cv. Chung-yang; RwG-CA, white rice gochujang containing C. annuum; RwG-CF, white rice gochujang containing C. frutescens.

2.2. Comparisons of Metabolites, Physicochemical Characteristics, and Antioxidant Activity According to Different Cereal Grains Used in Gochujang To compare the differences in metabolites of gochujang according to the type of cereal grain used for preparation, PLS-DA was conducted using the GC-TOF-MS (Figure 3A) and UHPLC-LTQ-ESI-IT-MS/MS data (Figure 3B). The three groups of WG, RbG, and RwG were clearly separated from each other based on the PLS-DA score plots. The WG samples were separated from both the RbG and RwG groups by PLS1 (12.9% in Figure 3A and 7.1% in Figure 3B), while the RbG group was distinct from the WG and RwG groups according to PLS2 (10.8% in Figure 3A and 6.6% in Figure 3B). Significantly different variables were selected by PLS1 and PLS2 in the PLS-DA score plots based on variable importance in projection (VIP) values > 0.7 and p-values < 0.05. Next, primary and secondary metabolites (based on GC-MS and LC-MS/MS data sets, respectively) were tentatively identified by retention time, molecular weight, ultraviolet light lambda-max (λmax, nm), MS/MS fragment patterns, in-house library, library of National Institute of Standards and Technology (NIST MS Search program, version 2.0, Gaithersburg, MD, USA), standard compounds, and references (Tables S1 and S2). Significantly different metabolites, including amino acids, organic acids, fatty acids, sugars and sugar alcohols, flavonoids, capsaicinoids and capsinoids, and lipids, according to the type of cereal grains used in gochujang were identified, and their relative contents were determined (Table S3). Among them, metabolites with significantly distinct patterns (p-value < 0.05) such as unique high or low contents in specific gochujang groups (WG, RbG, and RwG) were visualized with a heat map (Figure 3C). The heat map was prepared using the fold-change values normalized to the average relative contents of each metabolite in all gochujang samples. The levels of most amino acids [leucine (1), valine (2), isoleucine (3), proline (4), glycine (5), serine (6), threonine (7), pyroglutamic acid (9), glutamic acid (11), and phenylalanine (12)], citric acid (16), xylose (23), and 3 secondary metabolites [apigenin-C-hexoside-C-pentoside (32), dihydrocapsiate (49), and linoleic ethanolamide (68)] were distinctively higher in the WG group than in the other samples. Lipid-related molecules such as fatty acids [linoleic acid (19) and oleic acid (20)] and lysophospholipids [lysoPC 14:0 (55), lysoPC 18:2 (60), lysoPC 16:0 (63), lysoPC 18:1 (65), and lysoPE 18:2 (58)], sucrose (29), and two other secondary metabolites including luteolin-C-hexoside (34) and quercetin-O-rhamnoside (37) were present at significantly higher levels in the RbG group than in the other samples. For RwG, some sugars such as glucose (26) and maltose (31) were present at higher levels than in the other samples. In addition, the relative contents of two other primary metabolites (γ-aminobutyric acid (GABA) (10) and palmitic

Molecules 2016, 21, 921 7 of 14

Speciﬁcally, the levels of luteolin and luteolin derivatives were signiﬁcantly different only in the comparison between RwG-CA and RwG-CF. These metabolite variations were similar to the patterns for ABTS, TFC, and TPC, which showed higher levels in RwG-CF than in RwG-CA.

3. Discussion The quality of fermented foods is affected by numerous factors such as the type of raw materials, diversity of microbes used in fermentation, and other fermentation conditions [16,17,19]. The quality of gochujang, a fermented red pepper paste, is affected by diverse factors derived from the complex manufacturing process, including the conditions used for the steaming, fermentation, and aging procedures; numerous raw materials used such as cereal grains, fermented soy powder, hot pepper powder, salt, and other additives; and their combinations (Figure1). In this study, we conducted the untargeted metabolite profiling with several physicochemical assays on commercial gochujang samples to identify the factors important for the quality characterization of commercial gochujang. Based on the results of metabolite profiling, clustering patterns were determined using PCA models, with different cereal grains as main factors and the species of hot pepper as minor factors (Figure2). In WG, most amino acids were present in relatively high levels (Figure3C). In general, amino acids play important roles in the aroma, flavor, and nutritional quality of food products. Aspartic acid (8), glutamic acid (11), and phenylalanine (12) are related to the umami flavor [20]; proline (4) and threonine (7) contribute to the sweet flavor; and leucine (1), valine (2), and isoleucine (3) are related to the bitter flavor [17,19]. The other amino acids such as glycine (5), serine (6), and pyroglutamic acid (9) were also high in WG. Aspartic acid (8) and glutamic acid (11) are acidic amino acids and likely contributed to the high acidity of WG (Figure S1). Amino acids have been shown to contribute to antioxidant activity [21], and the high levels of amino acids in WG might have affected the patterns of ABTS and amino type nitrogen content (Figure3D,E). Furthermore, these specific metabolite states may be related to the natural characteristics of the grain storage proteins known as wheat prolamins [22]. These proteins are generally present in cereal grains containing starch and some lipids, and the proportion of these proteins in wheat grains is higher than in rice seeds [22,23]. These common structural features related to specific amino acid residues lead to high proportions of glutamine, proline, and other amino acids (e.g., glycine, methionine, and phenylalanine) [22]. In addition, the levels of organic acids such as fumaric acid (14), malic acid (15), citric acid (16), and gluconic acid (17) were also high in WG (Table S3 and Figure3). Citric acid is widely used as an additive in the food industry and contributes to a sour taste along with other organic acids [24]. Based on our results, the general properties of wheat grains affected taste-related components such as amino acids and organic acids and the quality characteristics of the final product, WG. In the RG (RbG and RwG), lipid-related components and some sugars were present at significantly higher levels than in WG (Table S3 and Figure3). The average values of ABTS, TFC, and TPC were higher in RbG than in RwG. First, brown rice (Oryzae sativa) possesses bran layers containing abundant phenolic compounds and other nutritional components such as essential fatty acids and various lipids [25]. In RbG, the relative levels of most fatty acids; lipids such as linoleic acid (19), oleic acid (20), lysoPC 14:0 (55), lysoPE 18:2 (58), lysoPC 18:2 (60), lysoPC 16:0 (63), and lysoPC 18:1 (65); and sucrose (29) were higher than in the other gochujang samples. The high levels of these specific metabolites in RbG were attributed to the nutritional properties of brown rice. Rice bran phospholipids consist of palmitic (16:0), oleic (18:1), and linoleic (18:2) acids, as well as lysoPC 16:0, lysoPC 18:1, and lysoPC 18:2, in the outer section of the rice grain [26,27]. Ninety percent of the free sugar in rice grain is sucrose, and more than 60% of this sugar is present in the outer layer of the rice grain. Thus, the sucrose content of brown rice is generally higher than that of milled (white) rice [28]. Furthermore, various health-promoting effects of phospholipids such as lowering cholesterol and cardiovascular risk, inhibiting metastasis, regulating inflammatory reactions, and improving immunological functions have been reported [27]. Therefore, brown rice may be useful for producing fermented foods that contain more lipids than products made using wheat or white rice as raw materials. Second, white Molecules 2016, 21, 921 8 of 14 rice is composed of 80%–90% starch and is milled more than brown rice. In RwG, the relative levels of glucose (26) and maltose (31) were higher than in RbG and WG (Table S3 and Figure3). The levels of these metabolites in RwG were affected by the properties of the white rice grains. The starch level in the endosperm is higher than on the grain surface [27,29] and is composed of amylose and amylopectin. During digestion or fermentation, glucose and maltose are produced through hydrolysis [29]. According to Shu et al., the level of amylose-lipid complexes is lower in white rice than in brown rice. A high ratio of amylose-lipid complexes reduces the decomposition sensitivity of rice starch by enzymes such as α-amylase [30]. Amylose hydrolysis of white rice occurs at a faster rate than for brown rice. Thus, fermented foods prepared with white rice may be sweeter than those prepared with brown rice without any other additives. As shown in Figure4, the significantly different metabolite levels and the results of three assays were visualized according to the hot pepper species used in gochujang. The values for antioxidant activity, TFC, and TPC were higher in gochujang containing CAY or CF (RbG-CAY and RwG-CF) than in gochujang containing CA (RbG-CA and RwG-CA) (Figure4C,G). These patterns closely corresponded to the variation in capsaicinoids and capsinoids such as dihydrocapsiate (49), nordihydrocapsaicin (50), capsaicin (51), and dihydrocapsaicin (52), which are known to have antioxidant activity (Figure4D,H) [ 31]. Most capsaicinoids are pungent, and the pungency level of each compound is described in Scoville heat units (SHU). The reported pungency levels of capsaicin (51) and dihydrocapsaicin (52) are 1.5 ˆ 107 SHU, whereas that of nordihydrocapsaicin (50) is 9.1 ˆ 106 SHU. Capsinoids, which are capsaicinoid-like substances, such as capsiate and dihydrocapsiate (49) are not pungent [32]. These pungency-related metabolites present at high levels in RbG-CAY and RwG-CF resulted from the general characteristics of the hot peppers. According to Choi et al. CAY pepper was found to contain the highest levels of total capsaicinoids among 11 Korean red peppers [33]. The CF pepper, known as “Bird chili” or “Thai chili”, is widely used in Thailand. This hot pepper has a tiny cone shape and is highly pungent, with SHU values of 5 ˆ 104–1 ˆ 105 SHU [32]. In addition, some flavonoids, including luteolin-O-apiosyl-glucoside (33), luteolin-C-hexoside (34), and luteolin (43), were selected as metabolites to differentiate between RwG-CA and RwG-CF (Figure4F,H), which showed similar patterns for TFC (Figure4G). The levels of these metabolites were higher in RwG-CF than in RwG-CA. Our findings are consistent with those of Wahyuni et al., who evaluated the differences in 32 Capsicum spp. pepper fruits in terms of flavonoids such as glycosyl derivatives of quercetin, luteolin, and apigenin and their relative levels [34]. The relative levels of luteolin derivatives are high in C. frutescens, which is the species used to prepare RwG-CF; other samples used different species such as C. annuum. Thus, our data suggest that CAY and CF peppers can be used to control the level of pungency and alter the flavonoid contents in gochujang.

4. Materials and Methods

4.1. Chemicals and Reagents Methanol, acetonitrile, and water were purchased from Fisher Scientiﬁc (Pittsburgh, PA, USA). Formic acid, methoxyamine hydrochloride, pyridine, N-methyl-N-(trimethylsilyl)triﬂuoroacetamide (MSTFA), diethylene glycol, gallic acid, naringin, 6-hydroxy-2,5,7,8-tetramethylchroman-2-carboxylic acid (Trolox), hydrochloric acid, potassium persulfate, 2,21-azinobis(3-ethylbenzothiazoline-6-sulfonic acid) diammonium salt (ABTS), Folin-Ciocalteu1s phenol reagent, 37% formaldehyde solution, sodium hydroxide solution, dinitrosalicylic acid (DNS), formononetin, norvaline, and other analytical standard compounds were purchased from Sigma-Aldrich (St. Louis, MO, USA).

4.2. Sample Preparation

4.2.1. Sample Information Eighteen commercial gochujang samples were purchased from a local market (Seoul, Korea) and were stored at 4 ˝C until analysis. Based on the ingredient labels of the gochujang samples, the raw material information is summarized in Table1. In addition, the general manufacturing process Molecules 2016, 21, 921 9 of 14 for commercial gochujang with the raw material information for the samples used in this study are schematized in Figure1. As shown in Figure1, diverse raw materials were used to produce the commercial gochujang samples using a highly complex process. After the cereal grains were steamed, a pure strain of Aspergillus or Bacillus spp. was inoculated into the steamed cereals and fermented to form a koji. Next, the koji was mixed with fermented soybean powder and aged for approximately 20 days. Hot pepper powder and other additives such as salt, starch syrup, garlic, and onion were mixed, and the product was aged and sterilized. Finally, the commercial gochujang was packed. In this study, a total of 18 commercial gochujang samples prepared from three types of cereal grains, including wheat, brown rice, and white rice, and different species of hot pepper powder, including Capsicum annuum (CA), C. annuum cv. Chung-yang (CAY), and Caspsicum frutescens (CF), were used; the combinations of the major raw materials are shown in Figure1. The general relative pungency levels of hot pepper fruits [18] are described using red pepper diagrams with other information for each sample in Table1.

4.2.2. Sample Extraction and Derivatization for Metabolite Profiling Each commercial gochujang sample (300 mg) was extracted using 80% methanol (800 µL) with formononetin (10 µL, 600 ppm) as an internal standard. The samples were sonicated for 15 min and homogenized using a Retsch MM400 Mixer Mill (Retsch GmbH & Co., Haan, Germany) at 30 Hz for 3 min. After centrifugation for 5 min at 11,000ˆ g and 4 ˝C, the supernatants were collected, and the residues were extracted again using same procedure. The supernatants were filtered through a 0.2-µm polytetrafluoroethylene (PTFE) syringe filter and completely concentrated using a speed-vacuum concentrator (Biotron, Seoul, Korea). Dried extracts were re-suspended in 80% methanol to a final concentration of 100,000 ppm (100 mg/mL) and filtered. For the UHPLC-LTQ-ESI-IT-MS/MS analysis, the resolved extracts were diluted to 50,000 ppm (50 mg/mL). For the GC-TOF-MS analysis, 5 µL of norvaline (20 mg/mL) was added to 50 µL (100 mg/mL) of the extracts as an internal standard and concentrated again. Derivatized extracts were obtained by oximation and silylation. First, dried extracts were oximated with 100 µL of methoxyamine hydrochloride in pyridine (20 mg/mL) at 30 ˝C for 90 min. Next, the extracts were silylated with 100 µL of MSTFA at 37 ˝C for 30 min. The final concentration of the derivatized extracts was set to 25,000 ppm (25 mg/mL). To evaluate the technical variables including extraction, sample analysis, and data processing, quality control (QC) samples were also prepared by pooling 10 µL of each sample for both GC-MS and LC-MS analyses. The extracts were prepared at a final concentration of 10,000 ppm (10 mg/mL) for the antioxidant activity tests.

4.3. GC-TOF-MS Analysis GC-TOF-MS analysis was performed using an Agilent 7890A GC system coupled with an Agilent 7693 autosampler (Agilent, Santa Clara, CA, USA) equipped with a Pegasus HT TOF-MS (LECO Corp., St. Joseph, MI, USA) system. An Rtx-5MS capillary column (30 m ˆ 0.25 mm i.d.; 0.25 µm particle size; Restek Corp., Bellefonte, PA, USA) was used with helium gas and a ﬂow rate of 1.5 mL/min. The front inlet and transfer line temperatures were 250 ˝C and 240 ˝C, respectively. The starting oven temperature was maintained for 2 min at 75 ˝C, increased to 300 ˝C for 15 min at a rate of 15 ˝C/min, and then maintained at 300 ˝C for 3 min. The ion source temperature and electron energy were set to 230 ˝C and ´70 eV, respectively. The detector voltage was 1600 V, and the scanning mass range was 45–800 m/z. One microliter of derivatized sample was injected into the GC with a split ratio of 10:1. Following a randomized injection order, three analytical replications were analyzed for each gochujang sample. To evaluate the technical variables, we analyzed spiked QC samples for each 15 samples analyzed.

4.4. UHPLC-LTQ-ESI-IT-MS/MS Analysis To analyze the gochujang extracts, UHPLC-LTQ-ESI-IT-MS/MS was used. The LTQ XL linear ion trap mass spectrometer is consist of an electrospray interface (Thermo Fisher Scientiﬁc, Inc., Waltham, Molecules 2016, 21, 921 10 of 14

MA, USA) coupled to a DIONEX UltiMate 3000 RS Pump, RS Autosampler, RS Column Compartment, and RS Diode Array Detector (Dionex Corporation, Sunnyvale, CA, USA). Ten microliters of each sample was injected and separated using a Thermo Scientific Syncronis C18 UHPLC column (100 mm ˆ 2.1 mm i.d.; 1.7 µm particle size). The mobile phase consisted of A (0.1% formic acid in water, v/v) and B (0.1% formic acid in acetonitrile, v/v). The flow rate was 0.3 mL/min. The initial condition was 10% of solvent B for 1 min. Next, the gradient was linearly increased to 100% of solvent B over 14 min and maintained for 3 min. After then, the gradient was reduced to 10% for 1 min, followed by re-equilibration of the column to the initial conditions for 3 min. The photodiode array was set to 200–600 nm for detection and managed by a 3D field. Ion trapping was performed in positive, negative, and full-scan ion modes within a range of 150–1000 m/z. The operating parameters were set as follows: source voltage, ˘5 kV; capillary voltage, 39 V; and capillary temperature, 275 ˝C. After the full-scan MS analysis, tandem MS (MS/MS) analysis was carried out to collect MSn information by scan-type turbo data-dependent scanning (DDS), under the same conditions in both positive and negative ion modes. Among ions separated by mass-to-charge (m/z) ratio in the first stage of mass spectrometry (MS1), the first and second most intense ion (precursor ion) are selected and subjected to collision-induced dissociation (CID) to create fragment ions. The product ions are separated and detected in the MS2, and so on. The normalized collision energy was set to 35% and the dwell times in the trap were set to 25 ms. Three analytical replications were analyzed for each gochujang sample. To evaluate the technical variables, we spiked QC samples for each 15 samples analyzed.

4.5. Data Processing and Multivariate Analysis After the GC-TOF-MS and UHPLC-LTQ-ESI-IT-MS/MS analyses, raw data files were converted into a computable document format (*.cdf) using LECO Chroma TOF software (Version 4.4, LECO Corp.) and using the thermo file converter program in Thermo Xcalibur software (version 2.1, Thermo Fisher Scientific). The acquired NetCDF format (*.cdf) files were processed to determine retention time, baseline correction, peak detection, and alignment using the MetAlign software package [35]. After alignment, the resulting data file (*.csv) including the corrected peak retention times, peak areas, and corresponding mass (m/z) data was exported to Microsoft Excel (Microsoft, Redmond, WA, USA) for further analysis. Multivariate statistical analysis using SIMCA-P+ 12.0 software (version 12.0, Umerics, Umea, Sweden) was performed to compare metabolite differences between commercial gochujang samples by unsupervised principal component analysis (PCA), supervised partial least squares discriminant analysis (PLS-DA), and orthogonal PLS-DA (OPLS-DA). Significant differences (p-value < 0.05) between selected metabolites, their relative contents, and the values from the physicochemical characteristics assays were evaluated by one-way analysis of variance using SPSS 18.0 (SPSS Inc., Chicago, IL, USA). The differences between metabolites were visualized by heat map analysis using MEV software [36].

4.6. Determination of Antioxidant Activity (by ABTS), Total Polyphenol Content (TPC), and Total Flavonoid Content (TFC) Three assays—ABTS, TPC, and TFC—were conducted using procedures described by Jung et al. (2013) with slight modifications [10]. For the ABTS test, a dark-blue-colored stock solution was prepared by dissolving 7 mM ABTS in a 2.45 mM potassium persulfate solution, after which it was maintained in the dark for 24 h. The solution was diluted to reach an absorbance 0.7 ˘ 0.03 at 750 nm using a microplate reader (Spectronic Genesys 6, Thermo Electron, Madison, WI, USA). Then, 10 µL of the extracted sample was added to the diluted ABTS solution (190 µL) in a 96-well plate. After 6 min in dark, absorbance was measured at 750 nm using a microplate reader. The result was presented as Trolox equivalent activity concentration (TEAC), with the standard solution concentration curve ranging from 0.015 mM to 1.000 mM. For the analysis of total polyphenolic content (TPC), the Folin-Ciocalteu colorimetric method was used. In brief, 0.2 N Folin-Ciocalteu’s phenol reagent (100 µL) was added to 20 µL of each sample in Molecules 2016, 21, 921 11 of 14 a 96-well plate, followed by incubation for 5 min in the dark. Then, a 7.5% sodium carbonate solution (80 µL) was added. The mixture was allowed to react for 60 min, and the absorbance was measured at 750 nm using a microplate reader. The result was calculated as gallic acid equivalent concentration (GAE), and the standard solution concentration curve ranged from 3.906 ppm to 500 ppm. The total flavonoid content (TFC) was determined by measuring the absorbance of the reacted samples placed in 96-well plates. Briefly, 90% diethylene glycol (180 µL), 1 N NaOH (20 µL), and each extracted sample (20 µL) were mixed and reacted in the dark. After 60 min, the absorbance was recorded at 405 nm using a microplate reader. The result was expressed as naringin equivalent concentration (NE). The range of the standard solution concentration curve was from 1.562 ppm to 200 ppm. All experiments were performed in triplicate.

4.7. Analysis of Salinity, pH, Total Acidity, Amino Type Nitrogen, and Reducing Sugar Contents To evaluate physicochemical characteristics, each gochujang sample (10 g) was mixed with 100 mL of distilled water, homogenized on a vortexer for 2 min, and sonicated for 1 min. After the mixed solutions were centrifuged for 10 min at 4 ˝C and 4000ˆ g, the supernatants were collected and divided into three 100-mL beakers (20 mL each) to determine pH, total acidity, and amino type nitrogen. The pH was determined using a pH meter (Orion 3 Star pH benchtop, Thermo Fisher Scientific, Inc.). Total acid contents and amino type nitrogen were calculated by the formol titration method as described by Kim et al. with some modifications [37]. Total acidity was calculated by titration of the sample solution to pH 8.4 with a 0.1 N sodium hydroxide solution. After the titration, the consumed amounts of the sodium hydroxide solution (V1) was converted into percent lactic acid using the following formula: total acidity (%) = ((0.009 ˆ V1 ˆ D ˆ F)/S) ˆ 100, where 0.009 is the conversion factor for lactic acid. V1 is the first consumption volume of the sodium hydroxide solution (mL), and D, F, and S are the dilution rate (10), factor of the 0.1 N sodium hydroxide solution (1.0028), and the sample amount (10 g), respectively. To determine the amino type nitrogen contents, 20 mL of neutralized formaldehyde (37%, pH 8.4) was added to the above titrated sample solutions. After 1 min, the sample solutions were re-titrated to pH 8.4 with a 0.1 N sodium hydroxide solution (V2). The amount of the 0.1 N sodium hydroxide solution consumed (V2) was used to calculate the amino type nitrogen value. The milligram percentage of amino type nitrogen was expressed using the following formula: amino type nitrogen (mg%) = ((V2 ˆ 1.4 ˆ D ˆ F)/S) ˆ 100, where V2 is the second consumption volume of the sodium hydroxide solution (mL) and 1.4 is the nitrogen equivalent amount in 1 mL of the 0.1 N sodium hydroxide solution. Sample salinity (200 µL per each) was determined using a portable refractometer for salt measurements (HANNA Instruments, Inc., Padova, Italy). The reducing sugar contents of each gochujang sample were estimated using the dinitrosalicylic acid (DNS) method with some modifications; glucose was used as a standard [38]. To detect reducing sugars, a mixture of DNS solution (400 µL) and the prepared samples (200 µL) was boiled for 5 min and then placed on ice for 10 min. Next, the mixtures were diluted with distilled water to a measureable concentration range. The absorbance of the diluted mixtures was measured using a spectrophotometer (Genesys 6, Thermo Electron Co., Waltham, MA, USA) at 550 nm. All assays were conducted in duplicate.

5. Conclusions Through metabolite proﬁling, we found that metabolites related to cereal grain types and hot pepper species were major factors that distinguished different commercial gochujang samples. Characteristic metabolites such as amino acids, citric acid, fatty acids, and some sugars showed signiﬁcant patterns depending on cereal grain type. The relative contents of capsaicinoids, luteolin, and luteolin derivatives were affected by the species of hot pepper used to prepare the gochujang. These metabolites may be useful as quality markers for commercial gochujang using a metabolomics approach. Furthermore, metabolomics approaches may also be suitable for characterization based on differential factors in other fermented foods. Molecules 2016, 21, 921 12 of 14

Supplementary Materials: Supplementary materials can be accessed at: http://www.mdpi.com/1420-3049/21/ 7/921/s1. Acknowledgments: This research was supported by the Basic Science Research Program through the National Research Foundation of Korea (NRF) grant funded by the Korea government (MSIP) (No. NRF-2014R1A2A1A11050884) and was funded by the Strategic Initiative for Microbiomes in Agriculture and Food, Ministry of Agriculture, Food and Rural Affairs, Republic of Korea (as part of the (multi-ministerial) Genome Technology to Business Translation Program) (No. 916005-2). Author Contributions: C.H.L. conceived and designed the experiments; G.M.L. performed the experiments and analyzed the data; G.M.L., D.H.S., and E.S.J. conducted the data interpretation; and G.M.L. wrote the paper. All authors approved the final manuscript. Conflicts of Interest: The authors declare no conflict of interest.

Abbreviations The following abbreviations are used in this manuscript: PCA principal component analysis PLS-DA partial least squares discriminant analysis OPLS-DA orthogonal partial least squares discriminant analysis VIP variable importance in projection MS mass spectrometry GC-TOF-MS gas chromatography-time of flight-mass spectrometry UHPLC-LTQ-ESI-IT-MS/MS ultrahigh performance liquid chromatography-linear trap quadrupole-electrospray ionization-ion trap-MS/MS MSTFA N-methyl-N-(trimethylsilyl) trifluoroacetamide DNS dinitrosalicylic acid Trolox 6-hydroxy-2,5,7,8-tetramethylchroman-2-carboxylic acid ABTS 2,21-azinobis(3-ethylbenzothiazoline-6-sulfonic acid)diammonium salt TFC total flavonoid content TPC total polyphenolic content TEAC Trolox equivalent activity concentration NE naringin equivalent GAE gallic acid equivalent CA C. annuum CAY C. annuum cv. Chung-yang CF C. frutescens WG wheat gochujang RG rice gochujang RbG brown rice gochujang RwG white rice gochujang RbG-CA brown rice gochujang containing C. annuum RbG-CAY brown rice gochujang containing C. annuum cv. Chung-yang RwG-CA white rice gochujang containing C. annuum RwG-CF white rice gochujang containing C. frutescens

References

1. Roberfroid, M.B. Concepts and strategy of functional food science: The European perspective. Am. J. Clin. Nutr. 2000, 71, 1660s–1664s. [PubMed] 2. Oms-Oliu, G.; Odriozola-Serrano, I.; Martín-Belloso, O. Metabolomics for assessing safety and quality of plant-derived food. Food Res. Int. 2013, 54, 1172–1183. [CrossRef] 3. Cho, J.Y.; Lee, H.J.; Shin, H.C.; Lee, J.M.; Park, K.H.; Moon, J.H. Behavior of ﬂavonoid glycosides contained in Korean red pepper paste (gochujang) during fermentation: Participation of a β-glucosidase inhibitor. Food Sci. Biotechnol. 2013, 22.[CrossRef] 4. Ahn, I.S.; Do, M.S.; Kim, S.O.; Jung, H.S.; Kim, Y.I.; Kim, H.J.; Park, K.Y. Antiobesity effect of Kochujang (Korean fermented red pepper paste) extract in 3T3-L1 adipocytes. J. Med. Food. 2006, 9, 15–21. [CrossRef] [PubMed] 5. Shin, H.W.; Jang, E.S.; Moon, B.S.; Lee, J.J.; Lee, D.E.; Lee, C.H.; Shin, C.S. Anti-obesity effects of gochujang products prepared using rice koji and soybean meju in rats. J. Food Sci. Technol. 2015.[CrossRef][PubMed] 6. Park, K.Y.; Kong, K.R.; Jung, K.O.; Rhee, S.H. Inhibitory effects of Kochujang extracts on the tumor formation and lung metastasis in mice. Prev. Nutr. Food Sci. 2001, 6, 187–191. Molecules 2016, 21, 921 13 of 14

7. Kim, J.Y.; Park, K.W.; Yang, H.S.; Cho, Y.S.; Jeong, C.H.; Shim, K.H.; Yee, S.T.; Seo, K.I. Anticancer and immuno-activity of methanol extract from onion Kochujang. Korean J. Food Preserv. 2005, 12, 173–178. 8. Beckonert, O.; Keun, H.C.; Ebbels, T.M.; Bundy, J.; Holmes, E.; Lindon, J.C.; Nicholson, J.K. Metabolic profiling, metabolomic and metabonomic procedures for NMR spectroscopy of urine, plasma, serum and tissue extracts. Nat. Protoc. 2007, 2, 2692–2703. [CrossRef][PubMed] 9. Nicholson, J.K.; Connelly, J.; Lindon, J.C.; Holmes, E. Metabonomics: A platform for studying drug toxicity and gene function. Nat. Rev. Drug Discov. 2002, 1, 153–161. [CrossRef][PubMed] 10. Jung, E.S.; Lee, S.; Lim, S.H.; Ha, S.H.; Liu, K.H.; Lee, C.H. Metabolite profiling of the short-term responses of rice leaves (Oryza sativa cv. Ilmi) cultivated under different LED lights and its correlations with antioxidant activities. Plant Sci. 2013, 210, 61–69. [CrossRef][PubMed] 11. Solanky, K.S.; Bailey, N.J.; Beckwith-Hall, B.M.; Bingham, S.; Davis, A.; Holmes, E.; Nicholson, J.K.; Cassidy, A. Biofluid 1H-NMR-based metabonomic techniques in nutrition research—Metabolic effects of dietary isoflavones in humans. J. Nutr. Biochem. 2005, 16, 236–244. [CrossRef][PubMed] 12. Castro-Puyana, M.; Herrero, M. Metabolomics approaches based on mass spectrometry for food safety, quality and traceability. TrAC Trends Anal. Chem. 2013, 52, 74–87. [CrossRef] 13. Warth, B.; Parich, A.; Atehnkeng, J.; Bandyopadhyay, R.; Schuhmacher, R.; Sulyok, M.; Krska, R. Quantitation of mycotoxins in food and feed from Burkina Faso and Mozambique using a modern LC-MS/MS multitoxin method. J. Agric. Food Chem. 2012, 60, 9352–9363. [CrossRef][PubMed] 14. Vrhovsek, U.; Masuero, D.; Gasperotti, M.; Franceschi, P.; Caputi, L.; Viola, R.; Mattivi, F. A versatile targeted metabolomics method for the rapid quantification of multiple classes of phenolics in fruits and beverages. J. Agric. Food Chem. 2012, 60, 8831–8840. [CrossRef][PubMed] 15. Summo, C.; Caponio, F.; Pasqualone, A.; Gomes, T. Vacuum-packed ripened sausages: Evolution of volatile compounds during storage. J. Sci. Food Agric. 2011, 91, 950–955. [CrossRef][PubMed] 16. Lee, S.Y.; Lee, S.; Lee, S.; Oh, J.Y.; Jeon, E.J.; Ryu, H.S.; Lee, C.H. Primary and secondary metabolite profiling of doenjang, a fermented soybean paste during industrial processing. Food Chem. 2014, 165, 157–166. [CrossRef] [PubMed] 17. Mozzi, F.; Ortiz, M.E.; Bleckwedel, J.; de Vuyst, L.; Pescuma, M. Metabolomics as a tool for the comprehensive understanding of fermented and functional foods with lactic acid bacteria. Food Res. Int. 2013, 54, 1152–1161. [CrossRef] 18. Weiss, E.A. Spice Crops; CABI: New York, NY, USA, 2002; p. 411. 19. Kang, H.J.; Yang, H.J.; Kim, M.J.; Han, E.S.; Kim, H.J.; Kwon, D.Y. Metabolomic analysis of meju during fermentation by ultra performance liquid chromatography-quadrupole-time of flight mass spectrometry (UPLC-Q-TOF MS). Food Chem. 2011, 127, 1056–1064. [CrossRef][PubMed] 20. Shiga, K.; Yamamoto, S.; Nakajima, A.; Kodama, Y.; Imamura, M.; Sato, T.; Uchida, R.; Obata, A.; Fukusaki, E. Metabolic profiling approach to explore compounds related to the umami intensity of soy sauce. J. Agric. Food Chem. 2014, 62, 7317–7322. [CrossRef][PubMed] 21. Martínez-Tomé, M.; García-Carmona, F.; Murcia, M.A. Comparison of the antioxidant and pro-oxidant activities of broccoli amino acids with those of common food additives. J. Sci. Food Agric. 2001, 81, 1019–1026. [CrossRef] 22. Shewry, P.R.; Halford, N.G. Cereal seed storage proteins: Structures, properties and role in grain utilization. J. Exp. Bot. 2002, 53, 947–958. [CrossRef][PubMed] 23. Duan, M.; Sun, S.S. Profiling the expression of genes controlling rice grain quality. Plant Mol. Biol. 2005, 59, 165–178. [CrossRef][PubMed] 24. Papagianni, M. Advances in citric acid fermentation by Aspergillus niger: Biochemical aspects, membrane transport and modeling. Biotechnol. Adv. 2007, 25, 244–263. [CrossRef][PubMed] 25. Tian, S.; Nakamura, K.; Kayahara, H. Analysis of phenolic compounds in white rice, brown rice, and germinated brown rice. J. Agric. Food Chem. 2004, 52, 4808–4813. [PubMed] 26. Zaima, N.; Yoshimura, Y.; Kawamura, Y.; Moriyama, T. Distribution of lysophosphatidylcholine in the endosperm of Oryza sativa rice. Rapid Commun. Mass Spectrom. 2014, 28, 1515–1520. [CrossRef][PubMed] 27. Liu, L.; Waters, D.L.; Rose, T.J.; Bao, J.; King, G.J. Phospholipids in rice: Significance in grain quality and health benefits: A review. Food Chem. 2013, 139, 1133–1145. [CrossRef][PubMed] 28. Tran, T.U.; Suzuki, K.; Okadome, H.; Homma, S.; Ohtsubo, K.I. Analysis of the tastes of brown rice and milled rice with different milling yields using a taste sensing system. Food Chem. 2004, 88, 557–566. [CrossRef] Molecules 2016, 21, 921 14 of 14

29. Kusano, M.; Fukushima, A.; Fujita, N.; Okazaki, Y.; Kobayashi, M.; Oitome, N.F.; Ebana, K.; Saito, K. Deciphering starch quality of rice kernels using metabolite proﬁling and pedigree network analysis. Mol. Plant. 2012, 5, 442–451. [CrossRef][PubMed] 30. Shu, X.; Jia, L.; Ye, H.; Li, C.; Wu, D. Slow digestion properties of rice different in resistant starch. J. Agric. Food Chem. 2009, 57, 7552–7559. [CrossRef][PubMed] 31. Rosa, A.; Deiana, M.; Casu, V.; Paccagnini, S.; Appendino, G.; Ballero, M.; Dessí, M.A. Antioxidant activity of capsinoids. J. Agric. Food Chem. 2002, 50, 7396–7401. [CrossRef][PubMed] 32. Wangcharoen, W.; Morasuk, W. Antioxidant capacity changes of bird chili (Capsicum frutescens Linn) during hot air drying. Kasetsart J. Nat. Sci. 2009, 43, 12–20. 33. Choi, S.H.; Suh, B.S.; Kozukue, E.; Kozukue, N.; Levin, C.E.; Friedman, M. Analysis of the contents of pungent compounds in fresh Korean red peppers and in pepper-containing foods. J. Agric. Food Chem. 2006, 54, 9024–9031. [CrossRef][PubMed] 34. Wahyuni, Y.; Ballester, A.R.; Sudarmonowati, E.; Bino, R.J.; Bovy, A.G. Metabolite biodiversity in pepper (Capsicum) fruits of thirty-two diverse accessions: Variation in health-related compounds and implications for breeding. Phytochem. 2011, 72, 1358–1370. [CrossRef][PubMed] 35. Wageningen. Available online: http://www.metalign.nl (accessed on 13 July 2016). 36. Microarray Software Suite. Available online: http://www.tm4.org (accessed on 13 July 2016). 37. Kim, H.E.; Han, S.Y.; Kim, Y.S. Quality characteristics of gochujang meju prepared with different fermentation tools and inoculation time of Aspergillus oryzae. Food Sci. Biotechnol. 2010, 19, 1579–1585. [CrossRef] 38. Miller, G.L. Use of dinitrosalicylic acid reagent for determination of reducing sugar. Anal. Chem. 1959, 31, 426–428. [CrossRef]

Sample Availability: Samples are available from the authors.

© 2016 by the authors; licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC-BY) license (http://creativecommons.org/licenses/by/4.0/).