Nuclear Magnetic Resonance Methodology for the Analysis of Regular and Non-Alcoholic Lager Beers

1 Nuclear Magnetic Resonance methodology for the analysis of regular and non-alcoholic lager beers 1 Cristina Sánchez-Estébanez,a Sergio Ferrero,b Celedonio M. Alvarez,b Fernando Villafañe,b Isabel 2 Caballero,a and Carlos A. Blancoa,* 3 4 5 a Dpto. Ingeniería Agrícola y Forestal (Área de Tecnología de los Alimentos). E.T.S. Ingenierías 6 Agrarias. Universidad de Valladolid, 34004 Palencia, Spain. 7 b, GIR MIOMET-IU CINQUIMA-Química Inorgánica, Facultad de Ciencias, Campus Miguel 8 Delibes, Universidad de Valladolid, 47011 Valladolid, Spain. 9 10 11 12 *Corresponding author. C.A. Blanco 13 E-mail address: cbla[email protected].es 14 15 2 ABSTRACT 16 The presence of seven main agents responsible for beer aroma and taste (n-propanol, isobutanol, 3- 17 methylbutanol, tyrosol/tyrosine, ethyl acetate, isoamyl acetate, and acetaldehyde) is determined by 18 different NMR techniques (1H PRESAT, zTOCSY, HSQC, and HMBC) in five regular and five low- or 19 free-alcoholic beers. The new methodology includes the identification of the 1H and 13C NMR chemical 20 shifts of the analytes by a standard addition method, and the consequent identification of the compounds 21 studied in regular and non-alcoholic beers. The chemical composition is different depending on whether 22 the beer is regular or non-alcoholic, therefore affecting the organoleptic characteristics of each type of 23 beer. 24 25 Keywords: lager beer, NMR, non-alcoholic beer, free-alcoholic beer, beer compounds 26 27 3 1. INTRODUCTION 28 Beer is an alcoholic drink obtained by fermentation of a starch-rich wort coming from cereal grain 29 such as malted barley, wheat, maize and rice. According to the fermentation process, beers are classified 30 as top or high, and bottom or low fermentation beers. Lagers, the most consumed type of beer, are 31 produced by ‘‘low” fermentation, which is carried out under refrigeration (usually between 6 and 15 ºC). 32 After fermentation, yeast cells deposit at the bottom of the fermenter and are usually removed. In contrast, 33 ale type beers are produced by ‘‘high” fermentation, occurring between 16 and 24 ºC. (Bamforth, 2003). 34 Although the main steps of their processing are common, there is a wide variety of lager beers with 35 pronounced differences among them. However, several differential aspects in the composition and 36 organoleptic characteristics among the different lager styles may be established. These differential 37 features depend both on the raw material used, and on the parameters employed in the subsequent steps of 38 the production processes. Thus, some taste defects in alcohol-free beer come from the alcoholic absence 39 (Blanco et al. 2014; Andrés-Iglesias et al. 2016). 40 Each country has established its required alcohol by volume (ABV) maximum thresholds, which are 41 diverse. In the United States, alcohol-free beer (AFB) means that there is not any alcohol present, while 42 the upper limit of the so-called non-alcoholic beer or ‘‘near-beer’’ is 0.5% ABV. However, in most of the 43 EU countries beers with low alcohol content are divided into free-alcohol beers, which contain less than 44 0.5% ABV, and low-alcohol beers, with less than 1.2% ABV (Brányik et al. 2012). In Spain, beers with 45 low alcohol content have less than 3% ABV, whereas non-alcoholic beers must contain no more than 1% 46 ABV. Commercially, non-alcolholic beers are divided into "0,0%", which is the label for those beers 47 containing less than 0.1% alcohol; and "free", where the alcohol content must not exceed 1% (R. D. 48 53/1995). 49 Nowadays, low-alcoholic beers are generating an increasing technological and economic interest. 50 Some of the dealcoholization processes so far described (vacuum distillation, reverse osmosis, 51 evaporation, fermentation control...) expose the beer to severe conditions. This may cause the loss of the 52 original aroma, since the chemical and/or physical processes when high temperatures are used may 53 transform the original aroma compounds. Hence, the sensorial quality of the final brew may be distinct 54 from the original one, what is not recommended, since the success of low alcohol drinks lies on an aroma 55 profile as close as possible to the original/alcoholic brew (Sohrabvandi et al. 2010). 56 4 Recently, the consumption of low- and alcohol-free beer is being significantly increased. This may be 57 explained considering health reasons, safety rules at the workplaces or driving, or even strict social 58 regulations. Moreover, alcohol consumption is completely forbidden by law in some countries 59 (Sohrabvandi et al. 2010). 60 As fresh flavour is one of the most appreciated sensory characteristics of beer (Bravo et al. 2008), 61 flavour stability is one of the main quality criterias for beer, and as well a concern for the brewing 62 industry (Caballero et al. 2012; Moreira et al. 2013). 63 Beer aroma profile is made by many volatile organic compounds at very low concentration (ppm 64 level), which are responsible for its unique flavor (Catarino et al. 2007). Levels of different chemical 65 compounds, such as alcohols, esters, aldehydes, ketones, organic acids and phenols, can be found on beer 66 composition, giving a specific flavor that contributes to the overall organoleptic properties of the final 67 beer (Karlsson and Trägårdh 1997). Among them, esters and alcohols are the main groups of aroma 68 compounds. 69 The sensorial evaluation of the beer organoleptic characteristics, such as color, taste, appearance, 70 flavor, and aroma is the usual method to evaluate the beer quality control. Some analytical measurements, 71 such as through photometry (for color and bitterness), enzymatic analyses (for organic acids), and gas 72 chromatography (for higher alcohols) should contribute also to this evaluation (Bamforth 2003). 73 At present, traditional analytical reference methods tend to be replaced by others faster and more 74 economical. Screening methods seem to be the most advantageous for this purpose, since they guarantee a 75 very high sample throughput. Conventional methods are often focused on the analysis of few specific 76 components, but in contrast, Nuclear Magnetic Resonance (NMR) spectroscopy enables to register most 77 of the constituents of the foodstuff in a single experiment (Lachenmeier et al. 2005). Thus, the advantages 78 of NMR is the rapid information which provides, when compared to other common analytical tools, such 79 as high pressure liquid chromatography, gas chromatography or mass spectrometry (Marcone et al. 2013). 80 The combination of mass spectrometry analysis with multivariate statistical analysis as a suitable 81 method to find out differential metabolites between regular and non-alcohol beers has been previously 82 reported by us (Andrés-Iglesias, Blanco, Blanco and Montero, 2014). We have also described a 83 simulation program which predicts the flavor compounds present in beer once dealcoholized via vacuum 84 distillation (Andrés-Iglesias et al. 2015). Here we are extending these studies by using NMR spectroscopy 85 in order to recognize the differences between regular and non-alcohol beers. 86 5 Nowadays, NMR is a leading technique (Mattaruchi et al. 2010), which is being applied to a wide 87 range of liquid and solid matrices. The main advantages of NMR are: easy sample preparation (sample 88 degassing and occasional pH adjustment were not necessary and were not used, even though both 89 methods have been previously reported), and rapid analysis, thus envisaging potential industrial 90 applications. This technique also enables to carry out a rapid and non-invasive characterisation of foods 91 and beverages, and therefore provides information about the compounds therein present (Belton et al. 92 1996; Monakhova et al. 2012). However, compound quantification remains less than straightforward 93 application in complex mixtures such as foodstuffs, even though NMR is a quantitative technique. This 94 may be solved by the traditional method of NMR signals integration vs. the signal area of a reference 95 compound, an approach previously described for vinegars (Caligiani et al. 2007), wines (Lopez-Rituerto 96 et al. 2009), juices (Berregi et al. 2007) and beer (Almeida et al., 2006; Petersen et al., 2013). However, 97 using internal references for quantification in complex mixtures has potential difficulties, such as signal 98 overlapping or formation of chemical interactions between the reference and the sample components. 99 Both may lead to changes of the integration with subsequent erroneous quantification. 100 Therefore, extensive compositional information in just a few minutes (Duarte et al. 2002) and 101 automation or low injection technology (Lachenmeier,et al. 2005 ) are the main advantages of NMR 102 technology. The analysis method has been already adapted for routine beer analysis, and high-resolution 103 NMR and hyphenated NMR (LC-NMR and LC-NMR/MS) have enabled to establish a significant 104 database of compounds found in beers, with particular emphasis on carbohydrates (Duarte et al. 2003) 105 and aromatic compounds (Gil et al. 2003). This technique has also found broad applications in the wine 106 industry (Giménez-Miralles et al. 1999; Ogrinc et al. 2001), and has now emerged as an important tool for 107 wine quality control. In the case of beer, ethanol and water D/H ratios have been measured by deuterium 108 NMR, and have been correlated to beer quality parameters, such as the beer geographical origin 109 (Rossmann 2001), environmental factors (Franconi et al. 1989), or characteristics of the raw materials and 110 of the brewing process (Franconi et al. 1989). In brewing science, NMR has mostly been applied so far to 111 solve specific problems, such as the identification and quantitation of malt and hop constituents, like 112 polyphenols (Friedrich and Galensa 2002) or isohumulones (Nord et al. 2003). 113 NMR spectroscopy may give a direct and fast overview of the chemical composition of beer (Duarte 114 et al. 2002), which can be obtained without any pre-treatment, aside from degassing. Assigning signals in 115 mono- and bidimensional NMR spectra from beer samples has facilitated the identification of ca. 30 116 6 compounds, including organic acids, amino acids, and alcohols, or even higher molecular weight 117 compounds, such as lipids, and large aromatic compounds, as polyphenols. However, as indicated above, 118 a full assignment of the beer spectra is not possible, mainly due to strong signal overlap, even though 119 techniques like diffusion-ordered spectroscopy (DOSY) have been used for this purpose (Gil et al. 2004) 120 We have found only one previous report where non-alcoholic beers, along with ales and lager beers, 121 are included. The study describes the combination of NMR and FTIR data to provide information about 122 different factors affecting beer production (Duarte et al. 2004). However, no previous studies have been 123 performed on the specific use of NMR to differentiate between regular and noalcohol beers. In this 124 work, this new approach focuses on the compounds, mainly alcohols and esters, which are responsible for 125 the characteristic flavour of regular beer. Thus, high-resolution NMR spectroscopy is here used to 126 identify the presence of selected compounds in different commercial alcoholic and low- or non-alcoholic 127 beers. 128 129 2. Materials and methods 130 2.1. Beer samples 131 Ten different commercial national lager beers, five regular beers, labeled as A, two 0.0% beers (1B, 132 4B) and three alcohol-free (2B, 3B, and 5B) were used for this study (see Table 2). All analyses were 133 carried out from newly opened bottles. 134 2.2. Sample preparation and spectra measurement. 135 2.2.1. Sample preparation 136 Sample preparation (both for the standard addition method and for test samples) was carried out as 137 follows: a 500 μl aliquot of commercial beer was transferred to a 5 mm NMR tube, and then 50 μl of D2O 138 was added for internal lock. The samples thus prepared were then used without any additional treatment. 139 One sample per beer was analyzed Degasification was not performed, as we have verified that the NMR 140 spectra of degassed samples and of non-degasified samples were exactly the same. The absolute value 141 lock deuterium was used as reference for the chemical shifts. 142 NMR spectra were recorded on an Agilent DD2 500 instrument equipped with cryoprobe, operating at 143 499.81 MHz for 1H and at 125.69 MHz for 13C. 144 7 All experiments were performed at 25ºC using the 1H PRESAT pulse sequence by selective low- 145 power irradiation in order to suppress both the water (4.71 ppm), and the ethanol (3.46 and 0.99 ppm) 146 signal resonances. Obviously, for free-alcohol beers, only the residual water resonance was irradiated. 147 Homonuclear 2D experiment zTOCSY (TOtal Correlation SpectroscopY) with a zeroquantum 148 filter was used for artifact suppression (Trippleton and Keeler, 2003). Heteronuclear 2D experiments 149 HSQC (Heteronuclear Single Quantum Correlation) and HMBC (Heteronuclear Multiple Bond 150 Correlation) with gradient coherence selection CRISIS (Hu and Krishnamurthy, 2008) having BIP 151 (Broadband Inversion Pulse) pulses in both 13C and 1H channels were used. These homo- and y 152 heteronuclear 2D experiments allowed to determine unequivocally the 1H and 13C chemical shifts. 153 In order to completely identify the signals, a standard addition method was designed previously 154 to record the sample beer tests. The acquisition parameters for these two experiments are different, since 155 sample tests required a higher number of transients to be sensitive, whereas fewer transients are required 156 for the standard addition method, since the target compounds are more concentrated. The acquisition 157 parameters are summarized in the following paragraphs. 158 2.2.2. Standard addition method 159 A 50 μl aliquot of a 5∙10-2 M ethanolic solution of the compound to be determined was added to a 160 sample of 1A. This method was used for all the compounds collected in Table 2, except for 161 tyrosol/tyrosine and acetaldehyde, which could be assigned directly from the spectra. 162 Each 1D PRESAT 1H spectrum was recorded with 16 transients, a spectral width of 8012 Hz, 16384 163 acquired points, and an acquisition time of 2.044 s. Selective irradiation at water and ethanol frequencies 164 during the recycle delay was carried out during 2 s. 165 zTOCSY experiments were acquired in the phase sensitive mode with PRESAT solvent suppression. 166 A total of 4 transients for each of the 300 t1 increments were collected, using a spectral width of 8012 Hz 167 for both dimensions, with a mixing time of the DIPIS2 spin lock of 100 ms. The data were apodized with 168 gaussian functions window in both dimensions. 169 Gradient CRISIS2 HSQC (gc2hsqc) were recorded with inverse detection and carbon decoupling 170 during acquisition in the phase sensitive mode with PRESAT solvent suppression. A nominal value of 171 146 Hz was used for one-bond coupling constants JCH. A total of 8 transients for each of the 256 t1 172 increments were collected, using spectral widths of 8012 Hz in the F2 dimension, and 25133 Hz in the F1 173 dimension. The data were apodized with gaussian functions windows in both dimensions. 174 8 2.2.3. Sample tests 175 Each 1D PRESAT 1H spectrum was recorded with 128 transients, a spectral width of 8012 Hz, 16384 176 acquired points, and an acquisition time of 2.044 s. Selective irradiation at water and ethanol frequencies 177 during the recycle delay was carried out during 2s. 178 zTOCSY experiments were acquired in the phase sensitive mode with PRESAT solvent suppression 179 when necessary. A total of 12 transients for each of the 200 t1 increments were collected, using a spectral 180 width of 5707 Hz for both dimensions, with a mixing time of the DIPIS2 spin lock of 100ms, and an 181 acquisition time of 300 ms. The data were apodized with gaussian window in both dimensions. 182 Gradient CRISIS2 HSQC (gc2hsqc) were recorded with inverse detection and carbon decoupling 183 during acquisition in the phase sensitive mode with PRESAT solvent suppression when necessary. A 184 nominal value of 146 Hz was used for one-bond coupling constants JCH. A total of 32 transients for each 185 of the 200 t1 increments were collected, using spectral widths of 5707 Hz in the F2 dimension, and of 186 25133 Hz in the F1 dimension. The data were apodized with gaussian functions windows in both 187 dimensions. 188 Gradient CRISIS2 HMBC (gc2hmbc) were recorded with inverse detection and no carbon decoupling 189 was applied during acquisition with PRESAT solvent suppression when necessary. A nominal value of 8 190 Hz was used for the multi-bond coupling constant JnCH. A total of 52 transients for each of the 256 t1 191 increments were collected, using spectral widths of 5707 Hz in the F2 dimension and of 30165 Hz in the 192 F1 dimension. The data were apodized with sinebell window in the F2 dimension and gaussian function 193 window in the F1 dimension. 194 Forward linear prediction was employed to improve digital resolution in the F1 dimensions of all 2D 195 experiments. The resulting spectra were processed and manipulated using VnmrJ3.2 Agilent Software. 196 197 3. Results and Discussion 198 3.1. Experimental design 199 Hundreds of compounds may be detected in the 1H NMR spectra of beers and other natural products. 200 Although NMR-based urine metabolic profiling has been recently used with in order to identify and 201 quantify a wide range of compounds (Emwas et al. 2015), the identification of individual components in 202 beer samples is very difficult due to significant signal overlapping. 203 9 Figure 1 shows PRESAT 1H NMR spectra of a regular beer (above) and an alcohol-free (below). In 204 the regular beer spectrum the aliphatic region of the spectra (0-3 ppm) shows signals arising from 205 alcohols (e.g. propanol, isobutanol, isopentanol), organic acids (e.g. citric, malic, pyruvic, acetic, 206 succinic), amino acids (e.g. alanine, γ-aminobutyric, proline), and fatty acids. The contribution of 207 fermentable sugars (e.g. glucose, maltose), and dextrins (glucose oligomers with different degrees of 208 polymerization and branching) is observed in the midfield region (3-6 ppm). The aromatic region (6-10 209 ppm) shows the presence of aromatic amino acids (tyrosine, phenylalanine, tryptophan), nucleosides 210 (cytidine, uridine, adenosine/inosine), aromatic alcohols (2-phenylethanol, tyrosol/tyrosine, tryptophol), 211 and polyphenolic compounds. The latter give rise to underlying broad humps between 6.7 and 8.7 ppm 212 (Lachenmeier et al. 2005). 213 Comparing the whole PRESAT 1H NMR spectra of regular beer and alcohol-free beer, the different 214 pattern displayed by each type of beer can be observed. Alcohol-free beers (Figure 1, below) display 215 more signals at the 3-4 ppm region, which indicates a higher concentration of carbohydrates (sugars), and 216 fewer signals at 1-2 ppm, where the aliphatic protons of alcohols resonate. As indicated above, the 217 assignment of the signals of all the compounds present is not possible, except for a handful of 218 compounds. 219 Some acids, acetic, succinic and pyruvic, could be readily identified from literature reports (Almeida 220 et al, 2006; Duarte, Godejohann, Braumann, Spraul and Gil, 2003; Rodrigues and Gil, 2011; Rodrigues et 221 al, 2011). These acids are generated from metabolic by-products or intermediates excreted by yeast cells. 222 The assignment made could not be completed, since the two doublets observed at ca. 6.7 and 7.0 ppm 223 may be assigned to the para-substituted phenyl group protons of either tyrosol or tyrosine (Almeida et al 224 2006; Rodrigues et al. 2011). In this case, two-dimensional (2D) NMR spectra do not help to overcome 225 this uncertainty. Figure 2 collects the PRESAT 1H NMR spectra of samples 1A and 1B divided in the 226 three representative regions of the spectra, showing the signals of some the above mentioned compounds. 227 Given the complexity of the spectra, we decided to focus our work on the esters and higher alcohols, 228 which are the main agents responsible for their aroma and taste in lager beers. The concentration of these 229 compounds is very different depending on whether the beer is regular or non-alcoholic, thus affecting 230 their organoleptic characteristics (Montanari et al. 2009). Table 1 collects these compounds, as well as the 231 main features of their aromas (Kobayashi et al. 2010; Olmedo et al. 2014; Tian 2009). 232 16 Moreira N, Meireles S, Brandao T, de Pinho PG (2013) Optimization of the HS-SPME-GC-IT/MS 410 method using a central composite design for volatile carbonyl compounds determination in beers. Talanta, 411 117, 523-531. 412 Nord LI, Sorensen SB, Duss JO (2003) Characterization of reduced iso-α-acids derived from hops 413 (Humulus lupulus) by NMR. Magnetic Resonance in Chemistry, 41, 660–670. 414 Olmedo Nadal JL. Aldazábal Bardaji LM, Landaluce Fuentes R, García Martí A (2014) Cerveza y 415 Malta. Asociación Española de Técnicos de cerveza y malta, nº201 – Año LI (1º trimestre), Alcalá de 416 Henares, (Madrid, España). 417 Ogrinc N, Kosir I, Kocjancic M, Kidric J (2001) Determination of Authenticity, Regional Origin and 418 Vintage of Slovenian Wines Using a Combination of IRMS and SNIF-NMR Analyses. Journal of 419 Agricultural and Food Chemistry, 49, 1432-1440. 420 Petersen BO, Nilsson M, Bøjstrup M, Hindsgaul O, Meier S (2013) 1H NMR Spectroscopy for 421 Profiling Complex Carbohydrate Mixtures in Non-Fractionated Beer. Food Chemistry, 150, 65-72. 422 Prestch E, Bülhmann P, Affolter C, Herrera A, Martínez R (2001) Determinación estructural de 423 compuestos orgánicos. Masson, Zurich, Tokyo & Madrid. 424 Rodrigues JA, Barros AS, Carvalho B. Brandão T, Gil AM (2011) Probing beer aging chemistry by 425 nuclear magnetic resonance and multivariate analysis. Analytica Chimica Acta, 702, 178– 187. 426 Rodrigues JE, Gil AM (2011) NMR methods for beer characterization and quality control. Magnetic 427 Resonance in Chemistry, 49, S37–S45. 428 Rossmann A (2001) Determination of stable isotope ratios in food analysis. Food Reviews 429 International, 17, 347–381. 430 Smogrovicova D (2004) Beer volatile compounds in dependence on yeast strain used for fermentation. 431 Electronic Journal of Biotechnology, 14(2), 432 Sohrabvandi S. Mousavi SM, Razavi SH, Mortazavian AM, Rezaei K (2010) Alcohol-free beer: 433 Methods of production, sensorial defects and healthful effects. Foods Reviews International. 26 (4), 335- 434 352. 435 Tian J (2010) Determination of several flavours in beer with headspace sampling–gas 436 chromatography. Food Chemistry, 123 (4), 1318-1321. 437 Trippleton MJ, Keeler J (2003) Elimination of Zero-Quantum Interference in Two-Dimensional NMR 438 Spectra. Angewandte Chemie International Edition, 42, 3938-3941. 439 440 17 Tables 441 Table 1. Main compounds responsible for the aromas with their description, and 1H and 13C NMR 442 signals of each molecules (in bold above, and plain below, respectively), recorded on samples of 1A. 443 444 Type Compound Aroma Structure and chemical shifts Alcohols n-Propanol Alcohol, ripe fruit Isobutanol Alcohol, wine, nail polish 3-methylbutanol Oil, alcohol, wine, banana Tyrosol Bitter Esters Ethyl acetate Sweet, fruit Isoamyl acetate Sweet, fruit, banana Aldehydes Acetaldehyde Unripe fruit, winery, mold 445 18 Table 2. Identified compounds in the beers studied (+) 446 447 448 Beer ABV npropanol Isobutanol 3-methyl- butanol Tyrosol (or Tirosine) Ethyl acetate Isoamyl acetate Acetaldehyde 1A 5.5 + + + + + + 2A 5.5 + + + + + + 3A 5.0 + + + + + + 4A 5.4 + + + + + + 5A 4.6 + + + + + + 1B 0.0 + 2B 0.8 + + + + 3B 0.9 + + + + 4B 0.0 + + + + 5B <1 + + + 449 19 Figures captions 450 Fig. 1 PRESAT 1H NMR spectra of 1A (above) and 1B (below) displayed at the same intensity. 451 Fig. 2 PRESAT 1H NMR spectra of samples 1A (above) and 1B (below) divided in the three 452 representative regions. 453 Fig. 3 PRESAT 1H NMR spectra of 1A and assignation of the n-propanol signals: (a) 0.6-1.5 ppm; 454 and (b) 2.9-3.65 ppm. In each part, the regular sample is shown below, whereas above is the spectra after 455 adding an aliquot of n-propanol. The spectra are displayed at the same intensity, and the assignment is 456 shown. 457 Fig. 4 TOCSY spectra of 1A after adding an aliquot of n-propanol showing the signals assigned to 458 protons labeled as a, b, and c. 459 Fig. 5 HSQC spectra of 1A after adding an aliquot of n-propanol showing the crosspeaking signals 460 assigned to proton and carbon atoms labeled as a, b, and c. 461 Fig. 6 HMBC spectra of samples 1A (left) and 1B (right) showing the presence (up: 13C signal of Ca at 462 63.55 ppm correlates with 1H signal of Hc at 0.71 ppm; bottom: 13C signal of Cc at 9.54 ppm correlates 463 with 1H signal of Ha and Hb at 3.38 and 1.36 ppm) and the absence of n-propanol, respectively. 464 465 20 466 467 Figure 1. 468 469 21 470 471 472 473 22 474 475 Figure 2 476 477 478 23 a) 479 480 b) 481 482 483 484 Figure 3. 485 486 24 487 488 489 Figure 4 490 491 25 492 493 494 Figure 5 495 496 497 498