Comprehensive Two-Dimensional Gas Chromatography–Mass Spectrometry as a Tool for the Untargeted Study of Hop and Their Metabolites

Since hop secondary metabolites have a direct correlation with the quality of beer and other hop-based beverages, and the volatile fraction of hop has a complex composition, requiring effective separation, here we explore the application of headspace solid-phase microextraction as a sample preparation method, coupled with comprehensive two-dimensional gas chromatography–mass spectrometry (GC×GC–MS) analysis. The methodology involved the use of a DVB/PDMS fibre with 500 mg of hop cone powder, extracted for 40 min at 50 °C, for both GC–MS and GC×GC–MS. The varieties Azacca, Cascade, Enigma, Loral, and Zappa were studied comprehensively. The results demonstrate that GC×GC–MS increases the number of peaks by over 300% compared to classical GC–MS. Overall, 137 compounds were identified or tentatively identified and categorised into 10 classes, representing between 87.6% and 96.9% of the total peak area. The composition revealed the highest concentration of sesquiterpene hydrocarbons for Enigma, whilst Zappa showed a relatively significant concentration of monoterpene hydrocarbons. Principal component analysis for all compounds and classes, along with hierarchical cluster analysis, indicated similarities between Zappa and Cascade, and Azacca and Loral. In conclusion, this method presents an optimistic advancement in hop metabolite studies with a simple and established sample preparation procedure in combination with an effective separation technique.


Introduction
Hop (Humulus lupulus L.) is a dioecious and diploid plant from the family Cannabaceae and is an extremely important raw material in drinks and beverages, such as sparkling teas and soft drinks and with major use in brewing.The use of hop as an ingredient was a significant turning point in the history of beer, increasing the beverage's popularity and changing the physical/chemical properties of the beverage because of the secondary metabolites present in hop [1,2].Nowadays, hop is the most expensive ingredient in brewing, with a market valued at USD 7.8 billion in 2022 and a growth prospect of more than USD 5.5 billion by 2030 [3].Hop flavour is associated with the female flowers or strobili (termed cones) having a yellow powder named lupulin, which is the major source of hop aroma and is composed of acids, essential oils, and resins.Hop improves the aroma, stability properties, and bittering characteristics of the final product [2,4].
The composition of hop is complex and has an intrinsic characteristic based on properties such as variety/cultivar, geographic origin, environmental factors, and stability, among others [4].The flavour, odour, and other physicochemical properties of hop are generally associated with five key categories of components: alpha acids, beta acids, polyphenols, terpenoids, and thiols.Alpha acids are responsible for the sensation of bitterness and foam stability, representing a mass fraction of 2 to 20% of the hop composition [5,6].Beta acids influence the aromatic profile of the beer and are present in lower concentrations, due to their biochemical transformation during processing of the beverage.Polyphenols, such as tannins and flavonoids, act as stabilising agents in beer, influencing turbidity and flavour [6].Finally, hydrocarbon and oxygenated terpenoids, as well as thiols, are some of the classes of components found in hop essential oil, which reportedly include more than 300 aroma compounds [4,7].Essential oil compounds undergo biotransformation by yeast fermentation, such as geraniol generating beta-citronellol and, finally, the esterification product citronellyl acetate [8].
Numerous hop varieties are currently available in the market, with more than 100 cultivars being patented in the last decade and new products being continually developed and reported [9].There are 363 results for Humulus lupulus species-metabolite relationships in the KNApSAcK Core System database [10], which aids in searching metabolomics and analytical studies.Metabolomics is a defined research field that has developed over the last 25 years, which involves the comprehensive analysis of small molecules (metabolites) present in complex biological matrices.Within this broad field, there are a number of different sub-disciplines that study the behaviour of matrices such as hop proteomics [11], genomics [12], beeromics [13], hopomics [14], and volatilomics [15].
Regarding the analysis of volatile organic compounds (VOCs) and volatilomics, gas chromatography (GC) is a standardised technique that has been applied to the determination of plant metabolites for more than 50 years [16].Multiple analytical studies have targeted key aroma compounds in hop essential oil, such as myrcene (related to fresh odour), and linalool and geraniol (both related to floral notes) [17].GC has also been applied in numerous analytical approaches for hop studies, such as chemical profiling [18][19][20], comparison of different types of sample preparation [21][22][23], and the study of the effect of hop in the brewing process [24,25].
According to the level of complexity of the hop matrix, the full resolution and identification of compounds can be difficult to achieve.Comprehensive two-dimensional gas chromatography (GC×GC) is designed for such multi-component samples [22].Advantages brought by GC×GC compared with conventional GC include enhancement of separation power, increased peak capacity, and increase in sensitivity [32].
The use of GC×GC in the metabolomics field with respect to performance attributes was shown by Wong and coworkers [33].In that study, GC×GC coupled with highresolution quadrupole time-of-flight mass spectrometry (qTOFMS) was applied to samples of Eucalyptus spp.leaf oils, characterising four different species and their untargeted metabolic profiles [33].The expression of metabolomic profiles was also extended to hop for the study of new genotypes, chemotyped by Yan and coworkers [34].
The application of GC×GC to hop analysis (Table 1) has been reported by thirteen research studies between 2003 and 2019, according to the Web of Science database, and data are summarised here as illustrative of capabilities for untargeted metabolomics.The first study was published 20 years ago and shows the potential of the technique for the analysis of hop essential oil [35].With the development of the GC×GC, different setups were described, such as the parallel comprehensive two-dimensional gas chromatography (2GC×2GC) [36], and a range of detectors and modulators.Greater resolution power was achieved more recently, with the largest number of compounds reported by Yan and collaborators [34], who separated 306 and identified 99 compounds in their samples.Table 1 shows sample preparation and the column set applied in each study.Not only were comprehensive analyses of hop presented, but also the detection of compounds such as 4-mercapto-4-methylpentan-2-one (4MMP) (IUPAC: 4-methyl-4-sulfanylpentan-2-one (4MSP)).4MSP is a black-currant-like odorant with an odour detection threshold as low as 0.00055 µg L −1 in beers [37], and these studies explore the advantage of the improved sensitivity of the GC×GC system.Odorants in hop samples were also reported in other studies such as that of Eyres, Marriott, and Dufour [38], which were concerned with the detection and possible identification of the aroma-active compounds/regions present in the samples.The aim of the present study was the analysis of different commercial hop samples, applying HS-SPME and GC×GC techniques to obtain the chemical profile and chemotyping of five different hop varieties, to highlight the general value of GC×GC to generate valuable information for characterisation and metabolomics studies.Thus, the primary objective was to demonstrate the applicability of this new methodology for volatile metabolite profiling of hops and associated benefits.From the accompanying literature review (Table 1), we observed that comprehensive analyses combining SPME and GC×GC techniques for hop analysis were not previously reported in the literature.This paper presents the first report of HS-SPME-GC×GC-MS for hop samples, establishing its capacity and presenting new data on the volatile profile of different species of hop.
The hop samples Azacca (AZAC) and Loral (LORA) were kindly donated by Carlton & United Breweries (Asahi-CUB, Abbotsford, Australia) and the samples Cascade (CASC), Enigma (ENIG), and Zappa (ZAPP) were obtained from various other local suppliers.All samples were obtained in the format of pellet type 90, stored in the presence of nitrogen, and kept in a freezer until ready for analysis.More information about the hop samples is presented in Table S1 of the Supplementary Materials.

Sample Preparation
The sample preparation and analysis procedure (Figure 1) was adapted and optimised based on previous studies [20,22,46].The data regarding this initial examination of various experimental parameters, including the choice of the SPME fibres, among 65 µm DVB/PDMS (pink), 50/30 µm DVB/CAR/PDMS (grey), 100 µm PDMS (red), and 85 µm CAR/PDMS (blue), will be reported elsewhere.The following conditions were selected for the present study.HS-SPME was performed with a manual holder and a 65 µm DVB/PDMS (pink) fibre from Supelco (Castle Hill, Australia), conditioned in accordance with the manufacturer's guidelines.In the first step, the pellets were ground in a mortar and pestle with 500 mg taken into a 20 mL screw top vial with clear glass.The vials were heated at 50 • C using a hot plate, with an equilibrium time of 10 min and 30 min for sorption.After extraction, the DVB/PDMS (pink) fibre was introduced into the GC injector and desorbed for 3 min.where "Ix" is the retention index, "n" is the number of carbons in the alkane prior to the analyte being determined, "tRx" is the retention time of the analyte, "tRn" is the retention time of the alkane prior to the analyte, and "tRn+1" is the retention time of the alkane after the analyte.

Gas Chromatography (GC-MS and GC×GC-MS)
Chromatographic analyses were performed using two different systems.The GC-MS system was an Agilent 7890A GC with a 7000 triple quadrupole MS (Agilent Technologies, Mulgrave, Australia), with the first quadrupole operated with total ion transfer.The GC×GC-MS system was an Agilent 7890A GC with a 5975C single quadrupole MS (Agilent Technologies) and included an SSM1800 solid-state modulator (SSM) system from J&X (J&X Technologies, Nanjing, China).
The GC-MS method was adapted from hop studies in the literature [22] and included a DB-5ms UI column (30 m × 0.25 mm I.D. × 0.25 μm df) connected to the MS transfer line where "I x " is the retention index, "n" is the number of carbons in the alkane prior to the analyte being determined, "t Rx " is the retention time of the analyte, "t Rn " is the retention time of the alkane prior to the analyte, and "t Rn+1 " is the retention time of the alkane after the analyte.

Gas Chromatography (GC-MS and GC×GC-MS)
Chromatographic analyses were performed using two different systems.The GC-MS system was an Agilent 7890A GC with a 7000 triple quadrupole MS (Agilent Technologies, Mulgrave, Australia), with the first quadrupole operated with total ion transfer.The GC×GC-MS system was an Agilent 7890A GC with a 5975C single quadrupole MS (Agilent Technologies) and included an SSM1800 solid-state modulator (SSM) system from J&X (J&X Technologies, Nanjing, China).
The GC-MS method was adapted from hop studies in the literature [22] and included a DB-5ms UI column (30 m × 0.25 mm I.D. × 0.25 µm d f ) connected to the MS transfer line by a deactivated fused silica (DFS, 1.0 m × 0.18 mm I.D.) and a glass press fit.Helium, grade 99.99%, was used as a carrier gas in constant flow (1.0 mL min −1 ).The injector was set in split mode ( 50 The GC×GC-MS method including the solid-state modulator operation was also based on previous research [40,47] and included a DB-5ms UI 1  ), and were connected to the main columns using glass press fit connectors.The GC temperature programme was the same as used in GC-MS analyses, except for the MS source 230 • C, quadrupole 150 • C, threshold 80, 12500 scans (u/s), and 22.8 scan/s.The solid-state modulator configuration was adapted from the literature [47] and included the same temperature program of the GC oven for the modulator entry oven and a 20 • C offset above the GC oven for the exit oven.The trap temperature program was −50 • C (8 min) and 2 • C min −1 to −20 • C (37 min).

Data Analysis
The GC-MS and GC×GC-MS data processing and chromatographic analyses were performed using Agilent MassHunter workstation Qualitative Analysis version 10.0 (Agilent Technologies, Santa Clara, CA, USA) and J&X Canvas W1.5.14.30115 (J&X Technologies, Shanghai, China) software.The MS identification was performed using the NIST 11 mass spectrometry library.
Statistical analysis was performed by Excel ® (Microsoft, Redmond, WA, USA), MATLAB ® software, version 7.13 (MathWorks, Natick, MA, USA), including the PLS Toolbox, version 6.5 (Eigenvector Technologies, Manson, WA, USA).The multivariate statistics analysis included principal component analysis (PCA) and hierarchical cluster analysis (HCA).This analysis included pre-processing autoscaling and removal of outliers in the graphs and Hotelling's T2 versus Q residuals.Two types of PCA were performed, viz.by the areas (%) of compounds found in the samples, and by the sum of areas (%) of the classes of compounds.For HCA for the classes of compounds, Ward's method was applied and PCA with 3 components.

Results and Discussion
One of the crucial steps in the analysis of hop varieties and their volatile compounds is to achieve an acceptable separation of peaks, leading to the increase in selectivity and improvement in the identification of substances due to reduced overlapping interferences.The complexity of hop samples can be represented by the single separation dimension for the Cascade hop GC-MS result shown in Figure 2A, for which the given GC column is apparently of conventional separation quality with a peak capacity for the 32 min analysis of approximately 300; this indicates responses that have considerable peak overlap.For example, "d" (15.4 min) is a linalool with a cluster of poorly resolved peaks, and peak "e" at 22.0 min (geraniol) has an evident shoulder before a peak on the geraniol tail.The molecular composition of such regions will be indeterminate.The complexity becomes immediately clear when GC-MS (Figure 2A) is compared with GC×GC-MS (Figure 2B) with retention times on the DB-5ms UI columns arranged to show the same peaks aligned vertically in each panel.The existence of significantly more compounds is confirmed in the latter GC×GC result, simply through the addition of the second dimension ( 2 D) separation.Being a more polar column (wax-type), the 2 D retention proceeds from less to more polar along this axis, and so this indicates the relative polarity of compounds that coeluted on the 1 D column, and can serve as a check for possible chemical class.This leads to the notion of molecular structure-retention relationships in GC×GC [48].It is evident that the 2D presentation of GC×GC with most peaks resolved covers a large proportion of the volatile (headspace) composition of the sample, leading to the suggestion that this embodies the requirements for comprehensive metabolite profiling [49] and is a quintessential untargeted method for volatile compounds [50].
The GC×GC-MS configuration scenario for metabolite analysis, generating a greater number of peaks, should also include better identification by the MS through more intense peaks that improve minor constituent detectability, fewer matrix interferences, and less phase bleeding.Here GC-MS analysis was similar to that in the literature [22].For GC×GC-MS, 1 D was a nonpolar DB-5ms UI phase column (5% phenyl methylpolysiloxane), and two 2 D phase columns were tested: an intermediate polarity BPX50 phase (50% phenyl polysilphenylene-siloxane) and a polar phase (SUPELCOWAX 10; polyethylene glycol), both of the same dimensions.The results (Figure S1) displayed an enhanced separation with the polar 2 D phase, better resolution of peaks, and a larger number of recorded metabolites.The modulation period (P M ) used for GC×GC corresponds to the time available for completion of the 2 D analysis before the wrap-around might occur.Times of 4, 5, 6, and 7 s were tested, with the best separation for 6 s, and a little wrap-around.As for the choice of the SPME fibre, the higher number of peaks, which we correlate with better coverage of total metabolites, was the pink fibre (Figure S2).
The SPME method presented in this work is advantageous for the study of volatiles in hop.Compared to other methods previously applied (Table 1), SPME has the advantages of being easier, requiring fewer resources, being solventless, and using mild temperatures and small amounts of sample.This limits compound degradation, while being environmentally friendly with less energy requirements and generating less waste.This sample preparation method is established in the literature, the fibres are commercially available, and automation is possible, which can facilitate and reduce costs in operation.
The intra-day and inter-day precision were evaluated using Cascade hop with HS-SPME-GC×GC-MS selected for 22 compounds and tabulated in Table S2.The intra-day precision (n = 5 injections on the same day) displayed RSD % of retention time of 0.00-0.23%for 1 t R , 0.05-4.30%for 2 t R , and between 1.51% and 9.63% for peak areas.The inter-day precision was analysed with n = 9, where three replicates were injected daily over 3 different days with results of 0.22-0.58%for 1 t R , 1.21-8.77%for 2 t R , and 1.65-12.81%for the area.These results indicate an acceptable repeatability for the proposed method, as compared with the values of Yan et al. [34], which means a low value of RSD even when considering the SPME reproducibility for peak area.The GC×GC-MS configuration scenario for metabolite analysis, generating a greater number of peaks, should also include better identification by the MS through more intense peaks that improve minor constituent detectability, fewer matrix interferences, and less phase bleeding.Here GC-MS analysis was similar to that in the literature [22].For GC×GC-MS, 1 D was a nonpolar DB-5ms UI phase column (5% phenyl methylpolysiloxane), and two 2 D phase columns were tested: an intermediate polarity BPX50 phase (50% phenyl polysilphenylene-siloxane) and a polar phase (SUPELCOWAX 10; polyethylene glycol), both of the same dimensions.The results (Figure S1) displayed an enhanced separation with the polar 2 D phase, better resolution of peaks, and a larger Compared to previous studies (Table 1), the present methodology resulted in the highest number of peaks tentatively identified in the hop by using GC×GC-MS.The profile of the metabolites present in the five different hops identified numerous peaks by GC×GC-MS, totalling 205 for AZAC, 258 for CASC, 421 for LORA, 472 for ZAPP, and 413 for ENIG.Comparing the number of compounds reported with the GC-MS for the samples (Table 2) proved the improvement in metabolic coverage by GC×GC-MS with increases of 140.2% for CASC, 273.5% for ENIG, and 316.8% for ZAPP. Figure 2 (see Figures S6 and S7 for the other samples) represents aligned peaks with seven metabolites that have significant characteristics for the flavour and odour of hop: (a) β-pinene, (b) β-myrcene, (c) 2-methylbutyl isobutyrate, (d) linalool, (e) geraniol, (f) caryophyllene, and (g) humulene.All these components are now apparently free from interference, although the suite of non-polar compounds could potentially lead to overlapping compounds.Better separation should correspond to improved MS matching with databases.The overall results demonstrate evidence of a better separation, clear distinction of compounds and, consequently, an increase in identified peaks.The tentative identification was performed using the NIST 11 database, and the highest probability compound identity was determined, considering the MS RMatch (>700) and retention indices (±20 units) based on the van den Dool and Kratz equation (Table S3 and Figures S3-S5).A total of 137 compounds were identified with GC×GC-MS analysis of five hops and a total of 73 compounds for GC-MS analysis of three hops (n = 3, Table S4).
The final list of compounds tentatively identified by GC×GC-MS (Table 3) includes their molecular formula, CAS identifier, retention time, retention indices from literature and experiment, the relative chromatographic area of the peaks in each sample, and compound chemical class.Analysis of selected standards was performed for specific compound identification confirmation.The main classes of volatiles identified and the number of compounds were as follows: alcohols (3), aldehydes (4), esters (47), hydrocarbons (7), ketones (9), monoterpene hydrocarbons (17), oxygenated monoterpenes (12), sesquiterpene hydrocarbons (32), and oxygenated sesquiterpenes (3).The total amount of volatiles is expressed in the last line of the table, demonstrating that there remains a small fraction of unknown compounds, from 3.1% to 12.4% of the total area, which could not be identified using the reported criteria.However, this is an improvement in the identified or tentatively identified molecules over previous studies of essential oils since studies such as Wong et al. [33] reported 50.8-90.0%of the total area of the sample by GC×GC.The most abundant compounds detected using the SPME sampling method were related to the sesquiterpene hydrocarbons humulene (108) and caryophyllene (103), and the monoterpene hydrocarbon β-myrcene (22).These three compounds are reported in the literature as the major components present in hop, being responsible for up to 90% of the composition of hop essential oils [4].All these substances are formed from precursors obtained through the methyl-D-erythritol 4-phosphate (MEP) pathway followed by the transformation by prenyltransferases into geranyl diphosphate (GPP) [51,52].β-Myrcene is formed by GPP during hop growth by a monoterpene synthase (MTS2) [52,53] and has the odour characteristics of peppery, spicy, balsam, plastic, and terpene [24].For the samples studied, the highest area of β-myrcene was reported for the hop ENIG (37.82%), followed by ZAPP (17.06%) and CASC (5.93%).Caryophyllene, or β-caryophyllene, is an isomer of humulene, also identified as α-humulene or α-caryophyllene, and both compounds are produced by sesquiterpene synthase 1 (HlSTS1) from the precursor β-farnesene formed after GPP [52,53].Humulene is a metabolite that originated in the final stages of hop cone maturation and was reported in the samples with the highest area of 54.07% for the sample LORA, 41.36% for AZAC, and 36.44% for CASC; lowest humulene abundances, although still with a considerable percentage, were 16.04% for ENIG and 10.05% for ZAPP.Caryophyllene showed lower concentrations than humulene in the samples, except for ZAPP, which had twice the caryophyllene area than humulene.Table 3 lists the major compounds in the five hop samples.Another observation noted here and confirmed from the literature was the relation between β-myrcene and humulene, where β-myrcene has an inverse trend in concentration compared to humulene, described by a common intermediate in their biosynthesis via α-acids and β-acids [4,54].
Some compounds have been identified in only one of the samples (i.e., unique to a single hop), with possible use as chemical markers including those in Table 4.  * Even though an enantioselective column was not used, and so no chirality of a compound can be assessed, this was the entry returned by the database search.
The elucidation of the GC×GC distribution of compounds tentatively identified can be expressed by the apex plot in Figure 3a, which was constructed considering all the identified molecules present in Table 3.The classes can be seen as different markers, where the location of peaks shows a region for various classes.For instance, esters (orange circles) persist over the full range of 1 t R but cluster between 2 t R = 1.0 and 2.0 s, and are shown above the hydrocarbons (dark blue circle).The terpenoid compounds, Figure 3b, demonstrate the power of separation of GC×GC through a clear distinction between the classes of MHyd, OM, SHyd, and OS.
The classes of compounds represented in this study may be considered descriptors of the composition of the hop, displaying the metabolite formation and the identity of the cone.Figure 4 represents the compositional profile of each hop according to the percentage obtained by our methodology (subject to the limitations of reporting peak areas by using SPME), by a clear distinction of the samples.According to Figure 4, all the hops expressed the highest concentration of sesquiterpene hydrocarbons as the major class present, as expected based on the presence of caryophyllene (103) and humulene (108), with the largest amounts for AZAC, LORA, and CASC, respectively.The hop ENIG and ZAPP expressed a considerable concentration of monoterpene hydrocarbons, which differentiates these from the other hop; this characteristic can be related to the high concentration of β-myrcene (22).Furthermore, β-myrcene has a direct relationship with the amount of essential oil produced by the hop with regards to its biosynthetic pathway, giving a different complexity for the hop [4].The amount of data provided by GC×GC illustrates the ease in profiling and describing the sample composition, which should be translated into multivariate statistics, a powerful tool to aid this data interpretation.Considering the data in Table 3, a principal component analysis (PCA) interpretation was applied to the compounds present.For the PCA, the model was built with the relative peak area (%) related to each compound with the selection of autoscale as preprocessing and 3 PCs representing 81.70% of the variance with 34.19% for PC1, 24.24% for PC2, and 23.27% for PC3.The biplot graphs, shown in Figure 5, represent (A) PC1×PC2 and (B) PC1×PC3, where the samples (scores) are represented by red triangles and the compounds (loadings) by blue squares.To improve the view of the compounds related to each sample, coloured regions were drawn for AZAC (blue), CASC (pink), ENIG (yellow), LORA (green), and ZAPP (grey).Regarding this result, a diversity of substances is related to the samples and, as expected, the complexity of compounds formed during hop metabolism.The distribution of compounds can be seen with the ZAPP hop related to the positive part of PC1, while ENIG is present in the negative part of PC2; moreover, AZAC is present in the positive part of PC1 and the negative part of PC2, while PC3 shows LORA in the positive section and CASC in the negative.The amount of data provided by GC×GC illustrates the ease in profiling and describing the sample composition, which should be translated into multivariate statistics, a powerful tool to aid this data interpretation.Considering the data in Table 3, a principal component analysis (PCA) interpretation was applied to the compounds present.For the PCA, the model was built with the relative peak area (%) related to each compound with the selection of autoscale as preprocessing and 3 PCs representing 81.70% of the variance with 34.19% for PC1, 24.24% for PC2, and 23.27% for PC3.The biplot graphs, shown in Figure 5, represent (A) PC1×PC2 and (B) PC1×PC3, where the samples (scores) are represented by red triangles and the compounds (loadings) by blue squares.To improve the view of the compounds related to each sample, coloured regions were drawn for AZAC (blue), CASC (pink), ENIG (yellow), LORA (green), and ZAPP (grey).Regarding this result, a diversity of substances is related to the samples and, as expected, the complexity of compounds formed during hop metabolism.The distribution of compounds can be seen with the ZAPP hop related to the positive part of PC1, while ENIG is present in the negative part of PC2; moreover, AZAC is present in the positive part of PC1 and the negative part of PC2, while PC3 shows LORA in the positive section and CASC in the negative.
Characterising some of the compounds, attention may be called to the unique compounds in each of the samples with the presence of clusters close to their respective red triangles shown in Figure 5. Characterising each hop variety, AZAC shows the influence of compounds hexanal (1), p-cymene (33), the sesquiterpene hydrocarbons δ-cadinene (124), calamenene (125) and α-calacorene (130), and the ketones 2-nonanone (48), 2-decanone (65), 2-undecanone (79), 2-dodecanone (96) and 2-tridecanone (116).This shows a significant connection between AZAC and the ketone formed in its metabolism.Regarding reports from the literature [55], the hop AZAC was studied by GC-MS and a method for the fraction of hydrodistilled essential oil followed by HS-SPME.This study also showed a high concentration of caryophyllene and a high percentage of humulene as reported by results in Table 3.
For the hop CASC, the monoterpene hydrocarbons camphene (14) and perillene (52) and the oxygenated monoterpenes geraniol (73), cis-linalool oxide (44), and trans-linalool oxide (47) were related by PCA analysis.This shows that CASC has a relation with the auto-oxidation of myrcene [56,57] by the formation of geraniol, an important floral odorant in the hop essential oil, as the presence of camphene and linalool oxide in the two isomeric forms, related to the "European hop aroma" [17,57].The hop CASC is widely studied in the literature, as in the GC×GC-TOFMS study from Yan et al. [34], and by GC-MS from other studies [55].In the results of Yan et al., the essential oil fraction of CASC was hydrodistilled and as a result also displayed the presence of geraniol and perillene, proving certain similarities.When compared to the composition of humulene (36.44%) and β-myrcene (5.93%) (noting limitations due to the SPME sampling), this paper shows a contrary behaviour compared to the literature [34,55,58], where the concentration of βmyrcene is higher than humulene, which is explained as a distinct sample with a probable cause of low content of β-myrcene and high content of humulene related to the ripening period, more specifically an early harvest or a not-well-ripe hop [4,54].Characterising some of the compounds, attention may be called to the unique compounds in each of the samples with the presence of clusters close to their respective red triangles shown in Figure 5. Characterising each hop variety, AZAC shows the influence of compounds hexanal (1), p-cymene (33), the sesquiterpene hydrocarbons δcadinene (124), calamenene (125) and α-calacorene (130), and the ketones 2-nonanone (48), 2-decanone (65), 2-undecanone (79), 2-dodecanone (96) and 2-tridecanone (116).This shows a significant connection between AZAC and the ketone formed in its metabolism.Regarding reports from the literature [55], the hop AZAC was studied by GC-MS and a method for the fraction of hydrodistilled essential oil followed by HS-SPME.This study Regarding ENIG, three classes were significant: monoterpene hydrocarbons comprising β-myrcene (22), β-ocimene (35), and trans-β-ocimene (36), the two sesquiterpene hydrocarbons (E)-β-farnesene (107) and β-eudesmene (117), and the esters isobutyl 2-methylbutanoate (25) and 2-methylbutyl octanoate (106).Low concentrations of βfarnesene has been reported in low concentrations in hop [4,54], and this can be one of the differentiating components of ENIG hop, while the presence of the monoterpene β-myrcene showed the highest concentration, followed by the two forms of β-ocimene.
Based on all the information described above and noting some patterns in the PCA (Figure 5), the classes of compounds were also studied according to PCA and HCA.For the PCA, the model was developed with the sum of the relative chromatographic areas (%) of each class using the selection of autoscale as preprocessing and 3 PCs representing 90.58% of variance displaying the variance of 39.78% for PC1, 36.72% for PC2, and 14.08% for PC3.Regarding HCA, autoscale was used as preprocessing, Ward's method was applied as an algorithm, and PCA was used to choose three PCs.The PCA graphs are expressed as biplot graphs, as shown in Figure 6A (PC1×PC2) and Figure 6B (PC1×PC3), while HCA is represented in Figure 6C.The PCA for the classes of compounds separates the ZAPP sample by PC2, LORA by PC3, and ENIG by PC1.As explained previously, this represents that the sample ZAPP was most influenced by esters, LORA by alcohols, and ENIG by monoterpene hydrocarbons, with the influence of ketones over the AZAC hop.HCA gave the highest similarity between classes of ZAPP and CASC, followed by LORA and AZAC, which means that compared to the results shown by PCA, the first two classes were largely influenced by esters, aldehydes, oxygenated monoterpenes, and sesquiterpenes, while the second two were more affected by sesquiterpene hydrocarbons and ketones.The PCA for the classes of compounds separates the ZAPP sample by PC2, LORA by PC3, and ENIG by PC1.As explained previously, this represents that the sample ZAPP was most influenced by esters, LORA by alcohols, and ENIG by monoterpene hydrocarbons, with the influence of ketones over the AZAC hop.HCA gave the highest similarity between classes of ZAPP and CASC, followed by LORA and AZAC, which means that compared to the results shown by PCA, the first two classes were largely influenced by esters, aldehydes, oxygenated monoterpenes, and sesquiterpenes, while the second two were more affected by sesquiterpene hydrocarbons and ketones.

Conclusions
In this paper, a method for HS-SPME-GC×GC-MS using hop samples was developed, applied, and demonstrated to be a powerful technique to identify metabolites in the samples.By suitable experimental design of the method, we believe this to be a "hoptimal" analysis strategy for total VOC composition in the headspace of the hop cone.The identification of compounds as an untargeted study through this methodology provides considerable coverage of volatile metabolite expression, and it is possible to describe both major and minor compounds, including those most likely not readily measured by single-dimension GC separations since substantially fewer compounds are reported in GC-MS analysis.Thus, the identification of 137 substances in five diverse hop samples is described, with the potential of separating 471 peaks.This of course also highlights that in terms of metabolite identification, just having the ability to separate individual compounds is not the same as being able to unambiguously identify them.In this regard, in comparison with the conventional technique of GC-MS, GC×GC-MS improves the separation through an increase of over 300% in the number of peaks recognised as discrete components.This study reports the comprehensive study of hop through detailed chemical class assignment compounds, which can largely be separated in the 2D space of GC×GC analysis.Multivariate statistics analysis proved a similarity between the samples ZAPP and CASC, and the samples AZAC and LORA.The use of HS-SPME-GC×GC-MS is a bright light in understanding the metabolites present in hop.

Supplementary Materials:
The following supporting information (abbreviated titles here) can be downloaded at: https://www.mdpi.com/article/10.3390/metabo14040237/s1. Table S1: Hop samples information; Table S2: Intra-day and inter-day precision of selected peaks; Figure S1: A comparison between two chromatographic settings for GC×GC; Table S3: n-Alkanes the series C8-C21 and their respective retention times; Figure S2

Figure 1 .
Figure 1.Schematic representation of the HS-SPME and GC×GC-MS analyses of hop.

Figure 2 .
Figure 2. A comparison between two chromatograms of (A) GC-MS and (B) GC×GC-MS applying HS-SPME for Cascade hop, where the GC-MS and 1 D column of GC×GC-MS is a DB-5ms UI (30 m × 0.25 mm I.D. × 0.25 μm df), the 2 D column is SUPELCOWAX 10 (1.0 m × 0.10 mm I.D. × 0.10 μm df), and the structure of the selected compounds (C) is represented by a-g indicated in (A,B).

Figure 2 .
Figure 2. A comparison between two chromatograms of (A) GC-MS and (B) GC×GC-MS applying HS-SPME for Cascade hop, where the GC-MS and 1 D column of GC×GC-MS is a DB-5ms UI (30 m × 0.25 mm I.D. × 0.25 µm d f ), the 2 D column is SUPELCOWAX 10 (1.0 m × 0.10 mm I.D. × 0.10 µm d f ), and the structure of the selected compounds (C) is represented by a-g indicated in (A,B).

Metabolites 2024 , 25 Figure 3 .
Figure 3. (A) Apex plot of the compounds present in the hop samples AZAC, CASC, ENIG, LORA, and ZAPP showing the classes of compounds (refer to Table 3 for abbreviations).(B) Apex plot showing clustering of MHyd, OM, SHyd, and OS classes.

Figure 3 .
Figure 3. (A) Apex plot of the compounds present in the hop samples AZAC, CASC, ENIG, LORA, and ZAPP showing the classes of compounds (refer to Table 3 for abbreviations).(B) Apex plot showing clustering of MHyd, OM, SHyd, and OS classes.

Figure 4 .
Figure 4. Compositional plots for the hop samples AZAC, CASC, ENIG, LORA, and ZAPP based on the classes of compounds expressed in area percentage (%), according to the SPME methodology employed.

Figure 4 .
Figure 4. Compositional plots for the hop samples AZAC, CASC, ENIG, LORA, and ZAPP based on the classes of compounds expressed in area percentage (%), according to the SPME methodology employed.

Figure 6 .
Figure 6.Multivariate statistical analysis of PCA expressed by the biplot graph of (A) PC1×PC2 and (B) PC1×PC3, and (C) HCA for the hops AZAC, CASC, ENIG, LORA, and ZAPP regarding the classes of compounds.

Table 1 .
Review table for literature studies of hop by GC×GC.

Table 2 .
Comparison of peaks from 3 hop samples by GC-MS and GC×GC-MS.

Table 3 .
The composition of hop samples determined by using HS-SPME-GC×GC-MS.