- Email: [email protected]
Discrimination of Brazilian lager beer by 1H NMR spectroscopy combined with chemometrics
Accepted Manuscript Short communication Discrimination of Brazilian lager beer by 1H NMR spectroscopy combined with chemometrics Luis Augusto da Silva, Danilo Luiz Flumignan, Aristeu Gomes Tininis, Helena Redigolo Pezza, Leonardo Pezza PII: DOI: Reference:
S0308-8146(18)31481-X https://doi.org/10.1016/j.foodchem.2018.08.077 FOCH 23418
To appear in:
Food Chemistry
Received Date: Revised Date: Accepted Date:
7 May 2018 16 August 2018 19 August 2018
Please cite this article as: da Silva, L.A., Flumignan, D.L., Tininis, A.G., Pezza, H.R., Pezza, L., Discrimination of Brazilian lager beer by 1H NMR spectroscopy combined with chemometrics, Food Chemistry (2018), doi: https:// doi.org/10.1016/j.foodchem.2018.08.077
This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Discrimination of Brazilian lager beer by 1H NMR spectroscopy combined with chemometrics
Luis Augusto da Silvaa, Danilo Luiz Flumignanb, Aristeu Gomes Tininisb, Helena Redigolo Pezzaa, Leonardo Pezzaa,* a
Institute of Chemistry, São Paulo State University (UNESP), Rua Prof. Francisco Degni 55, Araraquara, Brazil
b
São Paulo Federal Institute of Education, Science and Technology (IFSP), Rua Stefano D'avassi 625, Matão, Brazil
Abstract 1
H NMR spectroscopy combined with chemometrics was employed to discriminate lager
beer samples from two different classes, according to their style and information provided on the label. Partial replacement of barley malt by adjuncts is a common practice adopted by large breweries, which can lead to a decrease in diastatic power, requiring the use of exogenous enzymes. For this reason, small variations in the spectral profile can occur in the carbohydrates region. Many studies have focused on differentiating beers according to type and brewing process. However, there have no studies concerning the discrimination of beers of the same type that differ only in style, using 1H NMR spectroscopy. In this study PCA (first three components explained 81.5% of the dataset variability), PLS-DA and SIMCA models proved to be powerful tool with predict power higher than 90% for distinguishing lager beers based on the raw materials employed in the brewing process. Keywords: Brazilian lager beer, malt barley and adjunct, NMR, PCA, PLS-DA, SIMCA
1. Introduction Beer, the world’s most popular fermented alcoholic beverage, is made using water, barley malt, hops, and yeast (Silva, Augusto, & Poppi, 2008). Malt is the main source of ____________________________________________________________________________________ *Corresponding author Email address: [email protected] (Leonardo Pezza)
fermentable sugars consumed by the yeast during the fermentation (Marcone et al., 2013). In this process, sugars are converted into ethanol, carbon dioxide, and other secondary metabolites such as acetic, malic, succinic, lactic, and pyruvic acids (Duarte et al., 2002). The crucial step in the brewing process is the choice of malt, because sensory characteristics such as taste, color, and flavor are influenced by the types and proportions of the carbohydrate sources. These attributes are decisive in influencing the consumer’s selection of beer. Lager-type beers dominate in Brazil and worldwide, with Standard American Lager and Premium American Lager being among the most consumed styles, differing mainly in terms of the types of carbohydrate sources used during brewing. Premium American Lagers are brewed with pure barley malt and also with special malts that impart specific colors and flavors. In the case of Standard American Lager, 30-50% of the base malt is replaced by alternative carbohydrate sources, known as adjuncts (Cooper, Evans, Yousif, Metz, & Koutoulis, 2016; Poreda, Czarnik, Zdaniewicz, Jakubowski, & Antkiewicz, 2014), such as corn, rice, unmalted barley, oats, sorghum, malt extracts, and sucrose-based syrups. The adjuncts may be solids, in the form of unmalted cereals, or liquids, in the form of malt extract or sucrose-based syrup (Bogdan & Kordialik-Bogacka, 2017). The use of adjuncts is common in large breweries, where the practice can reduce costs, compared to beers brewed with only barley malt, as well as assist in the production of a high quality beer that meets the requirements of the consumer (Kunz, Müller, Mato-Gonzales, & Methner, 2012). Partial replacement of barley malt by adjuncts contributes to the reduction of the diastatic potential, making it indispensable to use exogenous microbial enzymes, such as α and βamylase, responsible for promoting the cleavage of internal bonds of the starch molecules enabling the release of monosaccharides as glucose and fructose (Holmes, Casey, & Cook, 2017). Differences in the sugar composition of the beer arise from the incomplete
2
conversion of starch, due to the malting conditions, use of adjuncts, and addition of exogenous enzymes. These characteristics are important in the development of classification models based on differences among beer styles (Petersen, Nilsson, BǾjstrup, Ole Hindsgaul, 2014). However, little effort has been dedicated to the quality control and authentication of Brazilian beers, or to the differentiation of styles based on the carbohydrate sources used in the brewing process. There have been many previous studies concerning the classification and differentiation of beers using techniques such as chromatography (Clara Pérez-Ràfols, 2015; Spreng & Hofmann, 2018) and mass spectroscopy (Gallart-Ayala, Kamleh, Hernández-Cassou, Saurina, & Checa, 2016). These analytical techniques usually require some form of sample pretreatment, such as extraction, concentration, and dilution. As an alternative technique, 1H NMR spectroscopy is reproducible, nondestructive, precise, and capable of providing information about a wide range of chemical compounds in beer samples, in a single analysis and without any significant sample preparation (it is only necessary to degas the sample) (Kirtil, Cikrikci, McCarthy, & Oztop, 2017; Kuballa, Brunner, Thongpanchang, Walch, & Lachenmeier, 2018). The main advantages of 1H NMR, compared to other methods, are its simplicity and the speed of analysis (Petersen, Nilsson, BǾjstrup, Ole Hindsgaul, 2014; Bogdan & Kordialik-Bogacka, 2017). In this work, Principal Component Analysis (PCA), Partial Least Squared- Discriminant Analysis (PLS-DA) and Soft Independent Modeling of Class Analogies (SIMCA) were applied to 1H NMR spectra in the carbohydrates region, obtained for a set of forty lager-type beers, in order to investigate the relationship between variability in the spectral profile and the raw material used in the brewing process. To the best of our knowledge, there are no studies that aim to differentiate Brazilian lager beers by styles employing NMR spectroscopy combined with chemometric approach.
3
2. Materials and Methods 2.1. Samples A total of forty beer samples (20 Premium American Lager and 20 Standard American Lager) were purchased in local supermarkets in Araraquara (São Paulo State, Brazil). All the beers had approximately the same remaining shelf life. The beer samples were degassed in an ultrasonic bath at room temperature for 10 min. Subsequently, 540 μL of the beer sample, 60 μL of D2O and 0.05% sodium 3-(trimethylsilyl) propionate (TSP, as a chemical shift reference) were placed into a 5 mm NMR tube.
2.2. Measurement of pH The pH was measured (in triplicate) in 50 mL volumes of the decarbonated beers, using a pH meter (Model 692 pH/Ion meter, Metrohm-Herisau, Switzerland) fitted with a combination electrode and a temperature probe, following the standard method (AOAC, 2016). Table 1 provides the average pH values of the beer samples.
2.3. NMR Measurements 1
H NMR spectra of the beer samples were acquired using a Bruker Avance III HD 600
spectrometer (Bruker Biospin, Rheinstetten, Germany) operating at 14.1 Tesla (600.13 MHz for 1H). The instrument was equipped with a Triple Inverse TCI Cryo-probehead, automated tuning, a BCU (Bruker Cooler Unit I) temperature control accessory, and a Sample Express autochanger. Icon NMR software (Bruker) was used to control the entire process of tuning the equipment and acquiring the data. Samples were analyzed at 303.1 K after allowing 5 min for temperature equilibration. For all the 1H NMR spectra, water peak suppression was performed using a pulse sequence “zgcppr”. The numbers of scans and dummy scans were
4
set to 64 and 4, respectively. The free induction decays (FIDs) were collected into 32768 (32 k) data points using a 90◦ pulse, spectral width of 12.02 ppm, acquisition time of 2.3 s, and transformation with line broadening (LB = 0.3 Hz). The spectra were automatically phased and baseline corrected using the TopSpin 3.5 software package (Bruker Biospin, Rheinstetten, Germany). The analysis time for each sample was 15 min (Kuballa, Brunner, Thongpanchang, Walch, & Lachenmeier, 2018).
2.4. Chemometric methods PCA is a statistical tool that is widely used for exploratory data analysis. It reduces the dimensionality of the data, finding linear combinations of the original independent variables that represent the maximum amounts of variation. Each variable can be considered as a different dimension (Wold, Esbensen, & Geladi, 1987). PCA changes the original data matrix, X, into a product of two smaller matrices, as shown in Eq. 1: X=TPT+E
(1)
where T is the scores, P is the loadings, and E is an error matrix. The beer 1H NMR spectra were organized in a matrix X (40, 5805). Supplementary material illustrates the matrix, where the lines and columns corresponding to the samples and the carbohydrates region chemical shifts, respectively. The signals for ethanol at 3.65 ppm and residual water at 4.76 ppm were removed. This data set was preprocessed by mean centering, standardized by normalization at 1.00, and analyzed by unsupervised PCA. Then, PLS-DA model was built by correlating the matrix X of class variables (lager beer style) with a vector y of dependent variables (1H NMR chemical shifts). In this work, we labeled the classes as 1 (Standard American Lager) and 2 (Premium American Lager). SIMCA model, class modeling was used to build a separate model for each category (similar to class in PLS-DA). In SIMCA, the similarity among individuals of a particular
5
category is obtained by means of a PCA model. To validate the pattern recognition analyses (PLS-DA and SIMCA), the samples were split into 28 and 12 for the training and prediction sets, respectively, by using the Kennard-Stone algorithm. Moreover, the data set was preprocessed and mathematical transformed by mean-centering, normalization and smooth (Gondim, Junqueira, Souza, Ruisánchez, & Callao, 2017; Granato et al., 2018). All analyzes were performed using the Pirouette v. 4.5 (rev. 1) software package (Infometrix, Bothel, WA, USA).
3. Results and Discussion Beer samples are fairly complex matrices, with the major components being water, ethanol, and sugars. Fig. 1a shows an overlay of the collection of the 1H NMR spectra in the range 0-10 ppm for the set of forty lager-type beers. The signals for ethanol (1.17 and 3.65 ppm), residual water (4.76 ppm), and the TSP chemical shift reference (0.00 ppm) are indicated. The small variations observed in the chemical shifts could be caused by pH differences or intermolecular interactions between the chemical compounds contained in the beer (Liu, Dong, Yin, Li, & Gu, 2012; Monakhova et al., 2011), as can be seen for the 2.04, 2.35, and 2.57 ppm signals corresponding to acetic, pyruvic, and succinic acids, respectively. This causes small spectral misalignment. The strategy adopted to overcome this difficulty was simpler than using preprocessing algorithms or binning operations. The spectra were manually aligned using the ethanol δ(CH2) signal at 3.65 ppm for a reference sample. Sample 17 (Table 1) was selected as the reference, because the pH value and alcohol strength were equal to the mean values for the sample set. This procedure avoided the possibility that small variations in chemical shift, especially in the carbohydrates region, might contribute to fake groupings and hinder application of the exploratory analysis. The alignment of the spectra was fundamental in order to be able to compare the data for the
6
beer samples and identify the main sources of variability, so the carbohydrates region (3.06.0 ppm) was selected for the application of PCA, because in this region the signals were perfectly aligned. Fig. 1b displays the anomeric region (at 4.95 ppm) associated with maltooligosaccharides and α-limit dextrins containing α-1,6 glycosidic linkages that cannot be cleaved by β-amylase enzymes, which could be indicative of incomplete conversion of the starch to fermentable sugars (Boulton, 2013). The assignment of the anomeric proton signals was based on NMR experiments and information described elsewhere (Bent. O. Petersen, Mathias Nilsson, Marie BǾjstrup, Ole Hindsgaul, 2014; Dal Poggetto, Castañar, Adams, Morris, & Nilsson, 2017; Duarte et al., 2002; Petersen, Meier, & Duus, 2012; Schievano, Tonoli, & Rastrelli, 2017). The PCA scores plot for the first two components (PC1 and PC2), shown in Fig. 2a, revealed two clusters, corresponding to Premium American Lager (red circles) and Standard American Lager (blue circles), together with a region where the areas for the Premium and Standard American Lager samples overlapped. The first two PCs accounted for 72.3% of the data variability (PC1 = 62.2%; PC2 = 10.1). Fig. 2b shows the scores plot for the first three components, which together explained 81.5% of the dataset variability (PC3 = 9.2%). Greater dispersion can be seen for the Premium American Lager samples, compared to the Standard American Lager samples. This was probably due to the technology employed by large breweries in the production of Standard American Lager, with similar beers being brewed, regardless of the brand or label. Large breweries have recently invested heavily in the production and marketing of Premium American Lager, although small breweries are responsible for producing the best Premium American Lager beers and have gained the preference of the Brazilian consumer. For this reason, multinational breweries have sought to acquire promising national breweries, in order to increase the variety of beers available to consumers and expand the field of their beer businesses.
7
The loading plots (Fig. 3) show the contributions of the carbohydrates to the separation of the beer samples shown in the score plots. The PC1 loadings revealed positive values for signals corresponding to maltooligosaccharides and maltose (5.22, 4.63, 3.41 and 3.27 ppm), indicating that these compounds were present at higher relative concentrations in the Premium American Lager style beers. On the other hand, the PC1 loadings showed negative values at the positions corresponding to the anomeric proton of dextrins (5.31, 5.34, and 5.37 ppm), indicating that these sugars were present at higher relative concentrations in the Standard American Lager style beers. A possible explanation for this could be the use of exogenous enzymes during the brewing process, which acts to substantially increase the fermentable sugar content. The PC2 loadings could be used to clarify the overlapping of samples in the positive region. Significant contributions of the anomeric proton signals of different sugars (5.37 and 4.95 ppm) could have been due to inefficient activity of endogenous enzymes during the mashing process (Petersen et al., 2014), especially βamylase, which only hydrolyzes α-1,4 bonds and is unable to cleave the α-1,6 bonds represented by the signal at 4.95 ppm (Boulton, 2013). PLS-DA model was built with the two classes: 1 (Standard American Lager) and 2 (Premium American Lager). The discriminant analyses model was validated using leaveone-out cross validation and the number of latent variable (3), preprocessed and mathematical transformed by mean-centering, normalization and smooth. Using these model was possible to discriminate styles beers with powerful of 100% in the calibration and prediction sets (Table 2). Results of PLS-DA analysis (Fig. 4) revealed that no sample from the training and prediction set was misclassified. The analysis was repeated using a modeling approach by means of the SIMCA algorithm, where 4 PCs were selected for the model and the corresponding results are reported in Table 2. Using SIMCA model was
8
possible to discriminate styles beers with powerful of 96.4% in the calibration and 91.6% in the prediction sets. This work specifically focused on the possibility of discriminating beers by style, according to the label information and 1H NMR spectra. The results obtained indicated that chemometric analysis using exploratory (PCA) and discriminant (PLS-DA) analyses can be employed in a first step to build a full classification procedure, based on the raw material used in the brewing process. Whereas, all pattern recognition chemometric methods were successful in the step for training and predicting the beer classes, and the models produced can be used as a screening used for authentication of the beer style. These results can be interesting to the Ministry of Agriculture Livestock and Food Supply (MAPA), government agency that are tasked with the verification of authenticity and quality of animal source food and alcoholic, non-alcoholic and fermented beverages. In the literature, various studies shown that the evaluation of the authenticity of animal source foods, such as milk, dairy, honey, salmon, olive oil and sausages using chemometric approaches were successful (Cruz et al., 2013; Matera et al., 2014; Souza et al., 2011). In the same way, Granato et. al. (2011) obtained excellent discrimination of Brazilian lager and brown ale beers based on chemical composition, color and antioxidant activity. Quality and authenticity control of the beer are of great importance to both the brewery industry, whose responsibility it is to provide clear and accurate labeling of their products, and consumers (Oliveri & Simonetti, 2016). It is important to emphasize that evaluation of the aforementioned items is fundamental to the breweries can meet changing trade and consumer requirements and making its brands more profitable. However, severer laws should be created to oblige breweries to provide complete information concerning the ingredients used in the manufacture of the product. The descriptions of the ingredients on the label (water, barley malt, unmalted cereals or carbohydrates, and hops) are insufficient
9
and do not allow the consumer to obtain adequate information concerning specification of the amounts of the ingredients and the characteristics of the product.
4. Conclusions In summary, the findings demonstrated that the approach adopted, involving NMR analysis combined with chemometrics, can be very useful when applied to a suitable set of samples, providing fast and reliable information on beer composition, considering the main carbohydrates and adjuncts used in brewing processes. The results showed that it was possible to discriminate between beers of the same type, but different style, using a simple approach without any need for preprocessing algorithms or binning operations, focusing attention only on the carbohydrate region in the 1H NMR spectrum. This approach could be used in other screening studies, for example to identify and quantify sugars derived from adjuncts, and could be extended to other compounds in different regions of the spectrum.
Acknowledgments The authors thank the Brazilian National Research Council (CNPq) and the Coordination for the Improvement of Higher Level Personnel (CAPES) for financial support.
Conflict of interest The authors declare that there are no conflicts of interest.
References AOAC (2016). Official Methods of Analysis of AOAC International 20th. Washington DC. Petersen, B. O., Nilsson, M., BǾjstrup, M., Ole Hindsgaul, S. M. (2014). 1H NMR spectroscopy for profiling complex carbohydrate mistures in non-fractionated beer. Food Chemistry, 150, 65–72. Bogdan, P., & Kordialik-Bogacka, E. (2017). Alternatives to malt in brewing. Trends in Food Science and
10
Technology, 65, 1–9. Boulton, C. (2013). Encyclopedia of Brewing. Chichester: John Wiley & Sons, Ltd. Clara Pérez-Ràfols, J. S. (2015). Liquid chromatographic fingerprints and profiles of polyphenolic compounds applied to the chemometric characterization and classification of beers. Analytical Methods, 7, 8733– 8739. Cooper, C. M., Evans, D. E., Yousif, A., Metz, N., & Koutoulis, A. (2016). Comparison of the impact on the performance of small-scale mashing with different proportions of unmalted barley, Ondea Pro®, malt and rice. Journal of the Institute of Brewing, 122(2), 218–227. Cruz, A. G., Cadena, R. S., Alvaro, M. B. V. B., Sant’Ana, A. S., Oliveira, C. A. F., Faria, J. A. F.,Ferreira, M. M. C. (2013). Assessing the use of different chemometric techniques to discriminate low-fat and full-fat yogurts. LWT - Food Science and Technology, 50(1), 210–214. Dal Poggetto, G., Castañar, L., Adams, R. W., Morris, G. A., & Nilsson, M. (2017). Relaxation-encoded NMR experiments for mixture analysis: REST and beer. Chem. Commun., 53(54), 7461–7464. Duarte, I., Barros, A., Belton, P. S., Righelato, R., Spraul, M., Humpfer, E., & Gil, A. M. (2002). Highresolution nuclear magnetic resonance spectroscopy and multivariate analysis for the characterization of beer. Journal of Agricultural and Food Chemistry, 50(9), 2475–2481. Fan, S., Zhong, Q., Fauhl-Hassek, C., Pfister, M. K. H., Horn, B., & Huang, Z. (2018). Classification of Chinese wine varieties using 1H NMR spectroscopy combined with multivariate statistical analysis. Food Control, 88, 113–122. Gallart-Ayala, H., Kamleh, M. A., Hernández-Cassou, S., Saurina, J., & Checa, A. (2016). Ultra-highperformance liquid chromatography–high-resolution mass spectrometry based metabolomics as a strategy for beer characterization. Journal of the Institute of Brewing, 122(3), 430–436. Gondim, C. de S., Junqueira, R. G., Souza, S. V. C. de, Ruisánchez, I., & Callao, M. P. (2017). Detection of several common adulterants in raw milk by MID-infrared spectroscopy and one-class and multi-class multivariate strategies. Food Chemistry, 230, 68–75. Granato, D., Branco, G. F., Faria, J. D. A. F., & Cruz, A. G. (2011). Characterization of Brazilian lager and brown ale beers based on color, phenolic compounds, and antioxidant activity using chemometrics. Journal of the Science of Food and Agriculture, 91(3), 563–571. Granato, D., Putnik, P., Kovačević, D. B., Santos, J. S., Calado, V., Rocha, R. S., Pomerantsev, A. (2018). Trends in Chemometrics: Food Authentication, Microbiology, and Effects of Processing. Comprehensive
11
Reviews in Food Science and Food Safety, 17 (3), 663–677. Holmes, C. P., Casey, J., & Cook, D. J. (2017). Mashing with unmalted sorghum using a novel low temperature enzyme system: Impacts of sorghum grain composition and microstructure. Food Chemistry, 221, 324–334. Kirtil, E., Cikrikci, S., McCarthy, M. J., & Oztop, M. H. (2017). Recent advances in time domain NMR & MRI sensors and their food applications. Current Opinion in Food Science, 17, 9–15 Kuballa, T., Brunner, T. S., Thongpanchang, T., Walch, S. G., & Lachenmeier, D. W. (2018). Application of NMR for authentication of honey, beer and spices. Current Opinion in Food Science, 19, 57–62. Kunz, T., Müller, C., Mato-Gonzales, D., & Methner, F. J. (2012). The influence of unmalted barley on the: Oxidative stability of wort and beer. Journal of the Institute of Brewing, 118(1), 32–39. Liu, C., Dong, J., Yin, X., Li, Q., & Gu, G. (2012). Influencing factors of hydrogen bonding intensity in beer. Journal of Food Science and Technology, 51(11), 2964–2976. Marcone, M. F., Wang, S., Albabish, W., Nie, S., Somnarain, D., & Hill, A. (2013). Diverse food-based applications of nuclear magnetic resonance (NMR) technology. Food Research International, 51(2), 729–747. Matera, J. A., Cruz, A. G., Raices, R. S. L., Silva, M. C., Nogueira, L. C., Quitério, S. L., Conte Júnior, C. A. (2014). Discrimination of Brazilian artisanal and inspected pork sausages: Application of unsupervised, linear and non-linear supervised chemometric methods. Food Research International, 64, 380–386. Monakhova, Y. B., Schäfer, H., Humpfer, E., Spraul, M., Kuballa, T., & Lachenmeier, D. W. (2011). Application of automated eightfold suppression of water and ethanol signals in 1H NMR to provide sensitivity for analyzing alcoholic beverages. Magnetic Resonance in Chemistry, 49(11), 734–739. Oliveri, P., & Simonetti, R. (2016). Chemometrics for Food Authenticity Applications. In Advances in Food Authenticity Testing (pp. 701–728). Elsevier Ltd. Petersen, B. O., Meier, S., & Duus, J. O. (2012). NMR assignment of structural motifs in intact β-limit dextrin and its α-amylase degradation products in situ. Carbohydrate Research, 359, 76–80. Poreda, A., Czarnik, A., Zdaniewicz, M., Jakubowski, M., & Antkiewicz, P. (2014). Corn grist adjunct application and influence on the brewing process and beer quality. Journal of the Institute of Brewing, 120(1), 77–81. Schievano, E., Tonoli, M., & Rastrelli, F. (2017). NMR Quantification of Carbohydrates in Complex Mixtures. A Challenge on Honey. Analytical Chemistry, 89(24), 13405–13414.
12
Silva, G. A. da, Augusto, F., & Poppi, R. J. (2008). Exploratory analysis of the volatile profile of beers by HSSPME-GC. Food Chemistry, 111(4), 1057–1063. Souza, S. S., Cruz, A. G., Walter, E. H. M., Faria, J. A. F., Celeghini, R. M. S., Ferreira, M. M. C., Sant’Ana, A. de S. (2011). Monitoring the authenticity of Brazilian UHT milk: A chemometric approach. Food Chemistry, 124(2), 692–695. Spreng, S., & Hofmann, T. (2018). Activity-Guided Identification of in Vitro Antioxidants in Beer. Journal of Agricultural and Food Chemistry, 66(3), 720–731. Vogels, J. T. W. E., Terwel, L., Tas, A. C., van den Berg, F., Dukel, F., & van der Greef, J. (1996). Detection of Adulteration in Orange Juices by a New Screening Method Using Proton NMR Spectroscopy in Combination with Pattern Recognition Techniques. Journal of Agricultural and Food Chemistry, 44(1), 175–180. Wold, S., Esbensen, K., & Geladi, P. (1987). Principal component analysis. Chemometrics and Intelligent Laboratory Systems, 2(1–3), 37–52.
13
FIGURE CAPTIONS Figure 1: Overlay of 1H NMR spectra. (a) Full spectral range (0-10 ppm), ethanol signals, residual water and sodium 3-(trimethylsilyl)propionate (TSP) are indicated, (b) anomeric proton signal α(1-6). Figure 2: PCA of sugar region (3.2-5.5 ppm). (a) Scores scatter plot of PC1 = 62.2% and PC2 = 10.1%, (b) Scores scatter plot of PC1 = 62.2%, PC2 = 10.1% and PC3 = 9.2%, blue circles; Standard American Lager, red circles; Premium American Lager. Figure 3: The loadings bi-plot in the sugar region. Loadings plot (PC1 black line and PC2 blue line). Figure 4: PLS-DA predictions. Calibration, training samples (a) and Prediction, test samples (b). Standard American Lager (blue diamond) and Premium American Lager (red diamond)
14
15
16
17
18
Table 1: Physicochemical attributes of Brazilian beers of different styles analyzed in this study (n=40).
Sample no. Beer style 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40
Premium American Lager Premium American Lager Premium American Lager Premium American Lager Premium American Lager Standard American Lager Standard American Lager Standard American Lager Standard American Lager Standard American Lager Standard American Lager Standard American Lager Standard American Lager Standard American Lager Standard American Lager Premium American Lager Premium American Lager Premium American Lager Premium American Lager Premium American Lager Premium American Lager Premium American Lager Premium American Lager Premium American Lager Premium American Lager Standard American Lager Standard American Lager Standard American Lager Standard American Lager Standard American Lager Standard American Lager Standard American Lager Standard American Lager Standard American Lager Standard American Lager Premium American Lager Premium American Lager Premium American Lager Premium American Lager Premium American Lager
Carbohydrate sourcea
Ethanola (%, v/v)
Pure barley malt Pure barley malt Pure barley malt Pure barley malt Pure barley malt Barley malt and adjunct* Barley malt and adjunct* Barley malt and adjunct* Barley malt and adjunct* Barley malt and adjunct* Barley malt and adjunct* Barley malt and adjunct* Barley malt and adjunct* Barley malt and adjunct* Barley malt and adjunct* Pure barley malt Pure barley malt Pure barley malt Pure barley malt Pure barley malt Pure barley malt Pure barley malt Pure barley malt Pure barley malt Pure barley malt Barley malt and adjunct* Barley malt and adjunct* Barley malt and adjunct* Barley malt and adjunct* Barley malt and adjunct* Barley malt and adjunct* Barley malt and adjunct* Barley malt and adjunct* Barley malt and adjunct* Barley malt and adjunct* Pure barley malt Pure barley malt Pure barley malt Pure barley malt Pure barley malt
4.4 4.5 5.0 5.5 5.1 5.0 4.7 4.9 4.8 4.5 4.5 4.6 5.0 4.5 5.0 6.0 4.8 4.8 5.5 5.0 5.5 4.8 4.8 5.0 4.5 4.7 4.7 4.5 4.6 4.7 4.5 4.5 5.0 4.7 4.6 6.2 4.2 5.0 4.7 5.0
a
As reported on the label Mean coefficient of variation = 1.2% *There is no information of the type of adjunct employed in brewing b
19
pHb 4.40 4.51 4.56 4.48 4.40 4.06 4.10 4.07 4.09 3.96 4.19 4.16 4.03 3.98 4.03 4.59 4.35 4.45 4.21 4.45 4.69 4.80 4.43 4.39 4.59 4.42 4.20 4.24 4.15 4.30 3.91 4.27 4.27 4.01 4.27 4.61 4.91 4.66 4.28 4.38
Table 2: Results of PLS-DA and SIMCA Method Class Mathematical (Lager) transformation Premium American NNormalize and PLS-DA Standard American Smooth Overall
LV/PC* LLV 3 3
Correct classification rate (%) Training Prediction 100% (14/14) 100% (6/6) 100% (14/14) 100% (6/6) 100% (28/28) 100% (12/12)
Premium American NNormalize and 4 92.9% (13/14) 100% (6/6) Standard American Smooth 4 100% (14/14) 83.3% (5/6) Overall 96.4% (27/28) 91.6% (11/12) *LV: number of latent variable in the case of PLS-DA and PC: number of principal component in the case of SIMCA SIMCA
20
HIGHLIGHTS
Differentiation of Brazilian lager beers based on carbohydrate source.
Partial replacement of barley malt with adjuncts alters the NMR spectral profile.
NMR spectroscopy allied chemometrics was used to discriminate styles of beer.
Discriminant (PLS-DA) analysis was useful in separating beers of different styles.
21
S0308-8146(18)31481-X https://doi.org/10.1016/j.foodchem.2018.08.077 FOCH 23418
To appear in:
Food Chemistry
Received Date: Revised Date: Accepted Date:
7 May 2018 16 August 2018 19 August 2018
Please cite this article as: da Silva, L.A., Flumignan, D.L., Tininis, A.G., Pezza, H.R., Pezza, L., Discrimination of Brazilian lager beer by 1H NMR spectroscopy combined with chemometrics, Food Chemistry (2018), doi: https:// doi.org/10.1016/j.foodchem.2018.08.077
This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Discrimination of Brazilian lager beer by 1H NMR spectroscopy combined with chemometrics
Luis Augusto da Silvaa, Danilo Luiz Flumignanb, Aristeu Gomes Tininisb, Helena Redigolo Pezzaa, Leonardo Pezzaa,* a
Institute of Chemistry, São Paulo State University (UNESP), Rua Prof. Francisco Degni 55, Araraquara, Brazil
b
São Paulo Federal Institute of Education, Science and Technology (IFSP), Rua Stefano D'avassi 625, Matão, Brazil
Abstract 1
H NMR spectroscopy combined with chemometrics was employed to discriminate lager
beer samples from two different classes, according to their style and information provided on the label. Partial replacement of barley malt by adjuncts is a common practice adopted by large breweries, which can lead to a decrease in diastatic power, requiring the use of exogenous enzymes. For this reason, small variations in the spectral profile can occur in the carbohydrates region. Many studies have focused on differentiating beers according to type and brewing process. However, there have no studies concerning the discrimination of beers of the same type that differ only in style, using 1H NMR spectroscopy. In this study PCA (first three components explained 81.5% of the dataset variability), PLS-DA and SIMCA models proved to be powerful tool with predict power higher than 90% for distinguishing lager beers based on the raw materials employed in the brewing process. Keywords: Brazilian lager beer, malt barley and adjunct, NMR, PCA, PLS-DA, SIMCA
1. Introduction Beer, the world’s most popular fermented alcoholic beverage, is made using water, barley malt, hops, and yeast (Silva, Augusto, & Poppi, 2008). Malt is the main source of ____________________________________________________________________________________ *Corresponding author Email address: [email protected] (Leonardo Pezza)
fermentable sugars consumed by the yeast during the fermentation (Marcone et al., 2013). In this process, sugars are converted into ethanol, carbon dioxide, and other secondary metabolites such as acetic, malic, succinic, lactic, and pyruvic acids (Duarte et al., 2002). The crucial step in the brewing process is the choice of malt, because sensory characteristics such as taste, color, and flavor are influenced by the types and proportions of the carbohydrate sources. These attributes are decisive in influencing the consumer’s selection of beer. Lager-type beers dominate in Brazil and worldwide, with Standard American Lager and Premium American Lager being among the most consumed styles, differing mainly in terms of the types of carbohydrate sources used during brewing. Premium American Lagers are brewed with pure barley malt and also with special malts that impart specific colors and flavors. In the case of Standard American Lager, 30-50% of the base malt is replaced by alternative carbohydrate sources, known as adjuncts (Cooper, Evans, Yousif, Metz, & Koutoulis, 2016; Poreda, Czarnik, Zdaniewicz, Jakubowski, & Antkiewicz, 2014), such as corn, rice, unmalted barley, oats, sorghum, malt extracts, and sucrose-based syrups. The adjuncts may be solids, in the form of unmalted cereals, or liquids, in the form of malt extract or sucrose-based syrup (Bogdan & Kordialik-Bogacka, 2017). The use of adjuncts is common in large breweries, where the practice can reduce costs, compared to beers brewed with only barley malt, as well as assist in the production of a high quality beer that meets the requirements of the consumer (Kunz, Müller, Mato-Gonzales, & Methner, 2012). Partial replacement of barley malt by adjuncts contributes to the reduction of the diastatic potential, making it indispensable to use exogenous microbial enzymes, such as α and βamylase, responsible for promoting the cleavage of internal bonds of the starch molecules enabling the release of monosaccharides as glucose and fructose (Holmes, Casey, & Cook, 2017). Differences in the sugar composition of the beer arise from the incomplete
2
conversion of starch, due to the malting conditions, use of adjuncts, and addition of exogenous enzymes. These characteristics are important in the development of classification models based on differences among beer styles (Petersen, Nilsson, BǾjstrup, Ole Hindsgaul, 2014). However, little effort has been dedicated to the quality control and authentication of Brazilian beers, or to the differentiation of styles based on the carbohydrate sources used in the brewing process. There have been many previous studies concerning the classification and differentiation of beers using techniques such as chromatography (Clara Pérez-Ràfols, 2015; Spreng & Hofmann, 2018) and mass spectroscopy (Gallart-Ayala, Kamleh, Hernández-Cassou, Saurina, & Checa, 2016). These analytical techniques usually require some form of sample pretreatment, such as extraction, concentration, and dilution. As an alternative technique, 1H NMR spectroscopy is reproducible, nondestructive, precise, and capable of providing information about a wide range of chemical compounds in beer samples, in a single analysis and without any significant sample preparation (it is only necessary to degas the sample) (Kirtil, Cikrikci, McCarthy, & Oztop, 2017; Kuballa, Brunner, Thongpanchang, Walch, & Lachenmeier, 2018). The main advantages of 1H NMR, compared to other methods, are its simplicity and the speed of analysis (Petersen, Nilsson, BǾjstrup, Ole Hindsgaul, 2014; Bogdan & Kordialik-Bogacka, 2017). In this work, Principal Component Analysis (PCA), Partial Least Squared- Discriminant Analysis (PLS-DA) and Soft Independent Modeling of Class Analogies (SIMCA) were applied to 1H NMR spectra in the carbohydrates region, obtained for a set of forty lager-type beers, in order to investigate the relationship between variability in the spectral profile and the raw material used in the brewing process. To the best of our knowledge, there are no studies that aim to differentiate Brazilian lager beers by styles employing NMR spectroscopy combined with chemometric approach.
3
2. Materials and Methods 2.1. Samples A total of forty beer samples (20 Premium American Lager and 20 Standard American Lager) were purchased in local supermarkets in Araraquara (São Paulo State, Brazil). All the beers had approximately the same remaining shelf life. The beer samples were degassed in an ultrasonic bath at room temperature for 10 min. Subsequently, 540 μL of the beer sample, 60 μL of D2O and 0.05% sodium 3-(trimethylsilyl) propionate (TSP, as a chemical shift reference) were placed into a 5 mm NMR tube.
2.2. Measurement of pH The pH was measured (in triplicate) in 50 mL volumes of the decarbonated beers, using a pH meter (Model 692 pH/Ion meter, Metrohm-Herisau, Switzerland) fitted with a combination electrode and a temperature probe, following the standard method (AOAC, 2016). Table 1 provides the average pH values of the beer samples.
2.3. NMR Measurements 1
H NMR spectra of the beer samples were acquired using a Bruker Avance III HD 600
spectrometer (Bruker Biospin, Rheinstetten, Germany) operating at 14.1 Tesla (600.13 MHz for 1H). The instrument was equipped with a Triple Inverse TCI Cryo-probehead, automated tuning, a BCU (Bruker Cooler Unit I) temperature control accessory, and a Sample Express autochanger. Icon NMR software (Bruker) was used to control the entire process of tuning the equipment and acquiring the data. Samples were analyzed at 303.1 K after allowing 5 min for temperature equilibration. For all the 1H NMR spectra, water peak suppression was performed using a pulse sequence “zgcppr”. The numbers of scans and dummy scans were
4
set to 64 and 4, respectively. The free induction decays (FIDs) were collected into 32768 (32 k) data points using a 90◦ pulse, spectral width of 12.02 ppm, acquisition time of 2.3 s, and transformation with line broadening (LB = 0.3 Hz). The spectra were automatically phased and baseline corrected using the TopSpin 3.5 software package (Bruker Biospin, Rheinstetten, Germany). The analysis time for each sample was 15 min (Kuballa, Brunner, Thongpanchang, Walch, & Lachenmeier, 2018).
2.4. Chemometric methods PCA is a statistical tool that is widely used for exploratory data analysis. It reduces the dimensionality of the data, finding linear combinations of the original independent variables that represent the maximum amounts of variation. Each variable can be considered as a different dimension (Wold, Esbensen, & Geladi, 1987). PCA changes the original data matrix, X, into a product of two smaller matrices, as shown in Eq. 1: X=TPT+E
(1)
where T is the scores, P is the loadings, and E is an error matrix. The beer 1H NMR spectra were organized in a matrix X (40, 5805). Supplementary material illustrates the matrix, where the lines and columns corresponding to the samples and the carbohydrates region chemical shifts, respectively. The signals for ethanol at 3.65 ppm and residual water at 4.76 ppm were removed. This data set was preprocessed by mean centering, standardized by normalization at 1.00, and analyzed by unsupervised PCA. Then, PLS-DA model was built by correlating the matrix X of class variables (lager beer style) with a vector y of dependent variables (1H NMR chemical shifts). In this work, we labeled the classes as 1 (Standard American Lager) and 2 (Premium American Lager). SIMCA model, class modeling was used to build a separate model for each category (similar to class in PLS-DA). In SIMCA, the similarity among individuals of a particular
5
category is obtained by means of a PCA model. To validate the pattern recognition analyses (PLS-DA and SIMCA), the samples were split into 28 and 12 for the training and prediction sets, respectively, by using the Kennard-Stone algorithm. Moreover, the data set was preprocessed and mathematical transformed by mean-centering, normalization and smooth (Gondim, Junqueira, Souza, Ruisánchez, & Callao, 2017; Granato et al., 2018). All analyzes were performed using the Pirouette v. 4.5 (rev. 1) software package (Infometrix, Bothel, WA, USA).
3. Results and Discussion Beer samples are fairly complex matrices, with the major components being water, ethanol, and sugars. Fig. 1a shows an overlay of the collection of the 1H NMR spectra in the range 0-10 ppm for the set of forty lager-type beers. The signals for ethanol (1.17 and 3.65 ppm), residual water (4.76 ppm), and the TSP chemical shift reference (0.00 ppm) are indicated. The small variations observed in the chemical shifts could be caused by pH differences or intermolecular interactions between the chemical compounds contained in the beer (Liu, Dong, Yin, Li, & Gu, 2012; Monakhova et al., 2011), as can be seen for the 2.04, 2.35, and 2.57 ppm signals corresponding to acetic, pyruvic, and succinic acids, respectively. This causes small spectral misalignment. The strategy adopted to overcome this difficulty was simpler than using preprocessing algorithms or binning operations. The spectra were manually aligned using the ethanol δ(CH2) signal at 3.65 ppm for a reference sample. Sample 17 (Table 1) was selected as the reference, because the pH value and alcohol strength were equal to the mean values for the sample set. This procedure avoided the possibility that small variations in chemical shift, especially in the carbohydrates region, might contribute to fake groupings and hinder application of the exploratory analysis. The alignment of the spectra was fundamental in order to be able to compare the data for the
6
beer samples and identify the main sources of variability, so the carbohydrates region (3.06.0 ppm) was selected for the application of PCA, because in this region the signals were perfectly aligned. Fig. 1b displays the anomeric region (at 4.95 ppm) associated with maltooligosaccharides and α-limit dextrins containing α-1,6 glycosidic linkages that cannot be cleaved by β-amylase enzymes, which could be indicative of incomplete conversion of the starch to fermentable sugars (Boulton, 2013). The assignment of the anomeric proton signals was based on NMR experiments and information described elsewhere (Bent. O. Petersen, Mathias Nilsson, Marie BǾjstrup, Ole Hindsgaul, 2014; Dal Poggetto, Castañar, Adams, Morris, & Nilsson, 2017; Duarte et al., 2002; Petersen, Meier, & Duus, 2012; Schievano, Tonoli, & Rastrelli, 2017). The PCA scores plot for the first two components (PC1 and PC2), shown in Fig. 2a, revealed two clusters, corresponding to Premium American Lager (red circles) and Standard American Lager (blue circles), together with a region where the areas for the Premium and Standard American Lager samples overlapped. The first two PCs accounted for 72.3% of the data variability (PC1 = 62.2%; PC2 = 10.1). Fig. 2b shows the scores plot for the first three components, which together explained 81.5% of the dataset variability (PC3 = 9.2%). Greater dispersion can be seen for the Premium American Lager samples, compared to the Standard American Lager samples. This was probably due to the technology employed by large breweries in the production of Standard American Lager, with similar beers being brewed, regardless of the brand or label. Large breweries have recently invested heavily in the production and marketing of Premium American Lager, although small breweries are responsible for producing the best Premium American Lager beers and have gained the preference of the Brazilian consumer. For this reason, multinational breweries have sought to acquire promising national breweries, in order to increase the variety of beers available to consumers and expand the field of their beer businesses.
7
The loading plots (Fig. 3) show the contributions of the carbohydrates to the separation of the beer samples shown in the score plots. The PC1 loadings revealed positive values for signals corresponding to maltooligosaccharides and maltose (5.22, 4.63, 3.41 and 3.27 ppm), indicating that these compounds were present at higher relative concentrations in the Premium American Lager style beers. On the other hand, the PC1 loadings showed negative values at the positions corresponding to the anomeric proton of dextrins (5.31, 5.34, and 5.37 ppm), indicating that these sugars were present at higher relative concentrations in the Standard American Lager style beers. A possible explanation for this could be the use of exogenous enzymes during the brewing process, which acts to substantially increase the fermentable sugar content. The PC2 loadings could be used to clarify the overlapping of samples in the positive region. Significant contributions of the anomeric proton signals of different sugars (5.37 and 4.95 ppm) could have been due to inefficient activity of endogenous enzymes during the mashing process (Petersen et al., 2014), especially βamylase, which only hydrolyzes α-1,4 bonds and is unable to cleave the α-1,6 bonds represented by the signal at 4.95 ppm (Boulton, 2013). PLS-DA model was built with the two classes: 1 (Standard American Lager) and 2 (Premium American Lager). The discriminant analyses model was validated using leaveone-out cross validation and the number of latent variable (3), preprocessed and mathematical transformed by mean-centering, normalization and smooth. Using these model was possible to discriminate styles beers with powerful of 100% in the calibration and prediction sets (Table 2). Results of PLS-DA analysis (Fig. 4) revealed that no sample from the training and prediction set was misclassified. The analysis was repeated using a modeling approach by means of the SIMCA algorithm, where 4 PCs were selected for the model and the corresponding results are reported in Table 2. Using SIMCA model was
8
possible to discriminate styles beers with powerful of 96.4% in the calibration and 91.6% in the prediction sets. This work specifically focused on the possibility of discriminating beers by style, according to the label information and 1H NMR spectra. The results obtained indicated that chemometric analysis using exploratory (PCA) and discriminant (PLS-DA) analyses can be employed in a first step to build a full classification procedure, based on the raw material used in the brewing process. Whereas, all pattern recognition chemometric methods were successful in the step for training and predicting the beer classes, and the models produced can be used as a screening used for authentication of the beer style. These results can be interesting to the Ministry of Agriculture Livestock and Food Supply (MAPA), government agency that are tasked with the verification of authenticity and quality of animal source food and alcoholic, non-alcoholic and fermented beverages. In the literature, various studies shown that the evaluation of the authenticity of animal source foods, such as milk, dairy, honey, salmon, olive oil and sausages using chemometric approaches were successful (Cruz et al., 2013; Matera et al., 2014; Souza et al., 2011). In the same way, Granato et. al. (2011) obtained excellent discrimination of Brazilian lager and brown ale beers based on chemical composition, color and antioxidant activity. Quality and authenticity control of the beer are of great importance to both the brewery industry, whose responsibility it is to provide clear and accurate labeling of their products, and consumers (Oliveri & Simonetti, 2016). It is important to emphasize that evaluation of the aforementioned items is fundamental to the breweries can meet changing trade and consumer requirements and making its brands more profitable. However, severer laws should be created to oblige breweries to provide complete information concerning the ingredients used in the manufacture of the product. The descriptions of the ingredients on the label (water, barley malt, unmalted cereals or carbohydrates, and hops) are insufficient
9
and do not allow the consumer to obtain adequate information concerning specification of the amounts of the ingredients and the characteristics of the product.
4. Conclusions In summary, the findings demonstrated that the approach adopted, involving NMR analysis combined with chemometrics, can be very useful when applied to a suitable set of samples, providing fast and reliable information on beer composition, considering the main carbohydrates and adjuncts used in brewing processes. The results showed that it was possible to discriminate between beers of the same type, but different style, using a simple approach without any need for preprocessing algorithms or binning operations, focusing attention only on the carbohydrate region in the 1H NMR spectrum. This approach could be used in other screening studies, for example to identify and quantify sugars derived from adjuncts, and could be extended to other compounds in different regions of the spectrum.
Acknowledgments The authors thank the Brazilian National Research Council (CNPq) and the Coordination for the Improvement of Higher Level Personnel (CAPES) for financial support.
Conflict of interest The authors declare that there are no conflicts of interest.
References AOAC (2016). Official Methods of Analysis of AOAC International 20th. Washington DC. Petersen, B. O., Nilsson, M., BǾjstrup, M., Ole Hindsgaul, S. M. (2014). 1H NMR spectroscopy for profiling complex carbohydrate mistures in non-fractionated beer. Food Chemistry, 150, 65–72. Bogdan, P., & Kordialik-Bogacka, E. (2017). Alternatives to malt in brewing. Trends in Food Science and
10
Technology, 65, 1–9. Boulton, C. (2013). Encyclopedia of Brewing. Chichester: John Wiley & Sons, Ltd. Clara Pérez-Ràfols, J. S. (2015). Liquid chromatographic fingerprints and profiles of polyphenolic compounds applied to the chemometric characterization and classification of beers. Analytical Methods, 7, 8733– 8739. Cooper, C. M., Evans, D. E., Yousif, A., Metz, N., & Koutoulis, A. (2016). Comparison of the impact on the performance of small-scale mashing with different proportions of unmalted barley, Ondea Pro®, malt and rice. Journal of the Institute of Brewing, 122(2), 218–227. Cruz, A. G., Cadena, R. S., Alvaro, M. B. V. B., Sant’Ana, A. S., Oliveira, C. A. F., Faria, J. A. F.,Ferreira, M. M. C. (2013). Assessing the use of different chemometric techniques to discriminate low-fat and full-fat yogurts. LWT - Food Science and Technology, 50(1), 210–214. Dal Poggetto, G., Castañar, L., Adams, R. W., Morris, G. A., & Nilsson, M. (2017). Relaxation-encoded NMR experiments for mixture analysis: REST and beer. Chem. Commun., 53(54), 7461–7464. Duarte, I., Barros, A., Belton, P. S., Righelato, R., Spraul, M., Humpfer, E., & Gil, A. M. (2002). Highresolution nuclear magnetic resonance spectroscopy and multivariate analysis for the characterization of beer. Journal of Agricultural and Food Chemistry, 50(9), 2475–2481. Fan, S., Zhong, Q., Fauhl-Hassek, C., Pfister, M. K. H., Horn, B., & Huang, Z. (2018). Classification of Chinese wine varieties using 1H NMR spectroscopy combined with multivariate statistical analysis. Food Control, 88, 113–122. Gallart-Ayala, H., Kamleh, M. A., Hernández-Cassou, S., Saurina, J., & Checa, A. (2016). Ultra-highperformance liquid chromatography–high-resolution mass spectrometry based metabolomics as a strategy for beer characterization. Journal of the Institute of Brewing, 122(3), 430–436. Gondim, C. de S., Junqueira, R. G., Souza, S. V. C. de, Ruisánchez, I., & Callao, M. P. (2017). Detection of several common adulterants in raw milk by MID-infrared spectroscopy and one-class and multi-class multivariate strategies. Food Chemistry, 230, 68–75. Granato, D., Branco, G. F., Faria, J. D. A. F., & Cruz, A. G. (2011). Characterization of Brazilian lager and brown ale beers based on color, phenolic compounds, and antioxidant activity using chemometrics. Journal of the Science of Food and Agriculture, 91(3), 563–571. Granato, D., Putnik, P., Kovačević, D. B., Santos, J. S., Calado, V., Rocha, R. S., Pomerantsev, A. (2018). Trends in Chemometrics: Food Authentication, Microbiology, and Effects of Processing. Comprehensive
11
Reviews in Food Science and Food Safety, 17 (3), 663–677. Holmes, C. P., Casey, J., & Cook, D. J. (2017). Mashing with unmalted sorghum using a novel low temperature enzyme system: Impacts of sorghum grain composition and microstructure. Food Chemistry, 221, 324–334. Kirtil, E., Cikrikci, S., McCarthy, M. J., & Oztop, M. H. (2017). Recent advances in time domain NMR & MRI sensors and their food applications. Current Opinion in Food Science, 17, 9–15 Kuballa, T., Brunner, T. S., Thongpanchang, T., Walch, S. G., & Lachenmeier, D. W. (2018). Application of NMR for authentication of honey, beer and spices. Current Opinion in Food Science, 19, 57–62. Kunz, T., Müller, C., Mato-Gonzales, D., & Methner, F. J. (2012). The influence of unmalted barley on the: Oxidative stability of wort and beer. Journal of the Institute of Brewing, 118(1), 32–39. Liu, C., Dong, J., Yin, X., Li, Q., & Gu, G. (2012). Influencing factors of hydrogen bonding intensity in beer. Journal of Food Science and Technology, 51(11), 2964–2976. Marcone, M. F., Wang, S., Albabish, W., Nie, S., Somnarain, D., & Hill, A. (2013). Diverse food-based applications of nuclear magnetic resonance (NMR) technology. Food Research International, 51(2), 729–747. Matera, J. A., Cruz, A. G., Raices, R. S. L., Silva, M. C., Nogueira, L. C., Quitério, S. L., Conte Júnior, C. A. (2014). Discrimination of Brazilian artisanal and inspected pork sausages: Application of unsupervised, linear and non-linear supervised chemometric methods. Food Research International, 64, 380–386. Monakhova, Y. B., Schäfer, H., Humpfer, E., Spraul, M., Kuballa, T., & Lachenmeier, D. W. (2011). Application of automated eightfold suppression of water and ethanol signals in 1H NMR to provide sensitivity for analyzing alcoholic beverages. Magnetic Resonance in Chemistry, 49(11), 734–739. Oliveri, P., & Simonetti, R. (2016). Chemometrics for Food Authenticity Applications. In Advances in Food Authenticity Testing (pp. 701–728). Elsevier Ltd. Petersen, B. O., Meier, S., & Duus, J. O. (2012). NMR assignment of structural motifs in intact β-limit dextrin and its α-amylase degradation products in situ. Carbohydrate Research, 359, 76–80. Poreda, A., Czarnik, A., Zdaniewicz, M., Jakubowski, M., & Antkiewicz, P. (2014). Corn grist adjunct application and influence on the brewing process and beer quality. Journal of the Institute of Brewing, 120(1), 77–81. Schievano, E., Tonoli, M., & Rastrelli, F. (2017). NMR Quantification of Carbohydrates in Complex Mixtures. A Challenge on Honey. Analytical Chemistry, 89(24), 13405–13414.
12
Silva, G. A. da, Augusto, F., & Poppi, R. J. (2008). Exploratory analysis of the volatile profile of beers by HSSPME-GC. Food Chemistry, 111(4), 1057–1063. Souza, S. S., Cruz, A. G., Walter, E. H. M., Faria, J. A. F., Celeghini, R. M. S., Ferreira, M. M. C., Sant’Ana, A. de S. (2011). Monitoring the authenticity of Brazilian UHT milk: A chemometric approach. Food Chemistry, 124(2), 692–695. Spreng, S., & Hofmann, T. (2018). Activity-Guided Identification of in Vitro Antioxidants in Beer. Journal of Agricultural and Food Chemistry, 66(3), 720–731. Vogels, J. T. W. E., Terwel, L., Tas, A. C., van den Berg, F., Dukel, F., & van der Greef, J. (1996). Detection of Adulteration in Orange Juices by a New Screening Method Using Proton NMR Spectroscopy in Combination with Pattern Recognition Techniques. Journal of Agricultural and Food Chemistry, 44(1), 175–180. Wold, S., Esbensen, K., & Geladi, P. (1987). Principal component analysis. Chemometrics and Intelligent Laboratory Systems, 2(1–3), 37–52.
13
FIGURE CAPTIONS Figure 1: Overlay of 1H NMR spectra. (a) Full spectral range (0-10 ppm), ethanol signals, residual water and sodium 3-(trimethylsilyl)propionate (TSP) are indicated, (b) anomeric proton signal α(1-6). Figure 2: PCA of sugar region (3.2-5.5 ppm). (a) Scores scatter plot of PC1 = 62.2% and PC2 = 10.1%, (b) Scores scatter plot of PC1 = 62.2%, PC2 = 10.1% and PC3 = 9.2%, blue circles; Standard American Lager, red circles; Premium American Lager. Figure 3: The loadings bi-plot in the sugar region. Loadings plot (PC1 black line and PC2 blue line). Figure 4: PLS-DA predictions. Calibration, training samples (a) and Prediction, test samples (b). Standard American Lager (blue diamond) and Premium American Lager (red diamond)
14
15
16
17
18
Table 1: Physicochemical attributes of Brazilian beers of different styles analyzed in this study (n=40).
Sample no. Beer style 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40
Premium American Lager Premium American Lager Premium American Lager Premium American Lager Premium American Lager Standard American Lager Standard American Lager Standard American Lager Standard American Lager Standard American Lager Standard American Lager Standard American Lager Standard American Lager Standard American Lager Standard American Lager Premium American Lager Premium American Lager Premium American Lager Premium American Lager Premium American Lager Premium American Lager Premium American Lager Premium American Lager Premium American Lager Premium American Lager Standard American Lager Standard American Lager Standard American Lager Standard American Lager Standard American Lager Standard American Lager Standard American Lager Standard American Lager Standard American Lager Standard American Lager Premium American Lager Premium American Lager Premium American Lager Premium American Lager Premium American Lager
Carbohydrate sourcea
Ethanola (%, v/v)
Pure barley malt Pure barley malt Pure barley malt Pure barley malt Pure barley malt Barley malt and adjunct* Barley malt and adjunct* Barley malt and adjunct* Barley malt and adjunct* Barley malt and adjunct* Barley malt and adjunct* Barley malt and adjunct* Barley malt and adjunct* Barley malt and adjunct* Barley malt and adjunct* Pure barley malt Pure barley malt Pure barley malt Pure barley malt Pure barley malt Pure barley malt Pure barley malt Pure barley malt Pure barley malt Pure barley malt Barley malt and adjunct* Barley malt and adjunct* Barley malt and adjunct* Barley malt and adjunct* Barley malt and adjunct* Barley malt and adjunct* Barley malt and adjunct* Barley malt and adjunct* Barley malt and adjunct* Barley malt and adjunct* Pure barley malt Pure barley malt Pure barley malt Pure barley malt Pure barley malt
4.4 4.5 5.0 5.5 5.1 5.0 4.7 4.9 4.8 4.5 4.5 4.6 5.0 4.5 5.0 6.0 4.8 4.8 5.5 5.0 5.5 4.8 4.8 5.0 4.5 4.7 4.7 4.5 4.6 4.7 4.5 4.5 5.0 4.7 4.6 6.2 4.2 5.0 4.7 5.0
a
As reported on the label Mean coefficient of variation = 1.2% *There is no information of the type of adjunct employed in brewing b
19
pHb 4.40 4.51 4.56 4.48 4.40 4.06 4.10 4.07 4.09 3.96 4.19 4.16 4.03 3.98 4.03 4.59 4.35 4.45 4.21 4.45 4.69 4.80 4.43 4.39 4.59 4.42 4.20 4.24 4.15 4.30 3.91 4.27 4.27 4.01 4.27 4.61 4.91 4.66 4.28 4.38
Table 2: Results of PLS-DA and SIMCA Method Class Mathematical (Lager) transformation Premium American NNormalize and PLS-DA Standard American Smooth Overall
LV/PC* LLV 3 3
Correct classification rate (%) Training Prediction 100% (14/14) 100% (6/6) 100% (14/14) 100% (6/6) 100% (28/28) 100% (12/12)
Premium American NNormalize and 4 92.9% (13/14) 100% (6/6) Standard American Smooth 4 100% (14/14) 83.3% (5/6) Overall 96.4% (27/28) 91.6% (11/12) *LV: number of latent variable in the case of PLS-DA and PC: number of principal component in the case of SIMCA SIMCA
20
HIGHLIGHTS
Differentiation of Brazilian lager beers based on carbohydrate source.
Partial replacement of barley malt with adjuncts alters the NMR spectral profile.
NMR spectroscopy allied chemometrics was used to discriminate styles of beer.
Discriminant (PLS-DA) analysis was useful in separating beers of different styles.
21