Review The adulteration of expensive commodities such as alcoholic beverages and flavouring materials is a costly type of economic fraud, when successful. In many cases, the process of adulteration has become so sophisticated that conventional means of detecting adulteration, such as chromatography or proximate analysis, are no longer of value. In recent years, the traditional methods have been supplemented with techniques based on the analysis of carbon and hydrogen isotopes in low molecular weight molecules such as water, ethanol or flavour molecules. The most specific of these techniques uses nuclear magnetic resonance spectroscopy to determine sitespecific isotope ratios.
Many analytical techniques have been used successfully to establish the authenticity of flavour molecules and beverages. For example, the determination of a4C levels has been used to differentiate synthetic products derived from fossil fuels (which have relatively low ~4C levels) from natural products derived from plants. However, such determinations cannot be used to discriminate semisynthetic molecules derived from natural precursors from their natural counterparts. Furthermore, ~4C levels of products may be manipulated to resemble expected levels. Thus, more sophisticated methods are required to unequivocally establish the origin of flavour molecules and ethanols. This review examines the applications of deuterium nuclear magnetic resonance (2H NMR) spectroscopy and the mass.spectrometric determination of L~C contents in characterizing such molecules. Deuterium NMR Quantitative -'H NMR spectroscopy at the natural abundance level is a powerful tool for investigating the internal distribution of 2H in various organic compounds ~,2. Despite the low sensitivity of -~H NMR spectroscopy, 2H has interesting properties for NMR experiments, due to its spin number of ! and its relatively small quadrupolar coupling constant. Relaxation of the 2H nucleus is dominated by quadrupolar interactions and, consequently, proton decoupling is very efficient and does not involve any nuclear Overhauser effects. The relaxation times of 2H nuclei in small flavour molecules are generally of the order of two to five seconds, but -'H nuclei in associated molecules (e.g. ethanol or acetic acid molecules) are characterized by T~ values of one second or less. A sample of any hydrogen-containing molecule will naturally contain a certain proportion of 2H atoms in the place of ~H. In contrast to the case of ~-~C, which is
Application of deuterium NMR and isetopic analysis to the characterization
of foods and beverages Claude Guillou, G6rald Remaud and G6rard J. Martin usually assumed to be randomly distributed in molecules, the distribution of ZH differs markedly from that which would be predicted statistically. Thus, several different deuterated forms of the molecule will exist, and sitespecific natural isotope fractionation (SNIF) determination can be used to discriminate molecules from different sources. Since, under appropriate technical conditions, the area and the intensity of the NMR signal are proportional to the number of nuclei resonating at a particular frequency, isotope abundance data are directly accessible from peak area measurements or, in strictly defined conditions, from signal height measurements. The measured sitespecific ZHPH isotope ratio for each non-equivalent chemical site (i) can be compared to the isotope ratio of the international reference standard (SMOW - 'standard mean ocean water' - for the 2H/JH isotope ratio)'~: pws MA rnws tws (:H/'H)? = ~ x ~-~-s x ~ x F x r, x (-'H/'H) *s (I)
where A is the sample and WS is the working standard (calibrated against the international standard); P~ is the number of hydrogen positions in the sample at site i; pws is the number of hydrogen positions in the working standard; M ^ and M ws are the molecular weights of the sample and working standard; m A and mws are the masses of the sample and working standard; t A and t ws are the purities (%, w/w) of the sample and working standards; and Ti is the ratio of the signal intensities (S~lSWS). Alternatively, the measured 2H/~H isotope ratio (R) for each chemically non-equivalent site (i) of the sample can be transformed to 't~' units with respect to the -~H/~H ClaudeGuillouandG~raldRemaudare with LaboraloireEurofins,Sitede la G6raudi~re,CP 4001, 44073 NantesC~dex03, France.Girard I. Martin is isotope ratio of the working reference standard (Rear.): at I'Universit6 de Nantes - CNRS, URA 472, 2 rue de la Houssini~re, t~(ZH/IH)i = (e, \ ~ - , e.o,~ l X 1000 (2) 44072 NantesC~dex,France. Trends in Food Science & TechnologyApril 1991
©1991.ElsevierStiencePublishersLid,(UKI 0924-2244191/$1)2.00
85
Since the 2H NMR spectroscopy is conducted at the photosynthetic pathways used, of plant physiology, and natural abundance level, a high signal-to-noise ratio is a of the geographical and climatological environment of prerequisite in the SNIF-NMR protocol. The low sensi- the plant. However, the main source of differences in tivity of 2H NMR is offset by signal accumulation and the carbon isotope contents of sugars is the type of Fourier transformation of the free induction decay (FID) metabolism of the plants from which they originate5. signal. On the other hand, the chemical shift discrimiIt is well known that the sugars synthesized by C4 nation of 2H for a given magnetic field is approximately plants (e.g. maize and cane) have higher '3C contents six times smaller than that of tH, and the ability to dis- than those found in C3 plants (e.g. beet, grape and barley). tinguish deuterated forms depends on the degree of The '3(2 contents of alcohols formed from the fermenoverlap of the signals. With a magnetic field strength of tation of sugars are related to the t3C contents of the 7.04T (corresponding to a 300-MHz tH signal), ethanol, initial sugars and, therefore, may be used to discriminate and acetic, lactic and propionic acids can be analysed between fermentation alcohols from C3 and C4 plants. effectively. In contrast, a field strength of l l.6T The 2H contents of sugars are also an appreciable source (500MHz for ~H) is necessary to study chemical species of information, and it has been shown that the 2H conthat have larger numbers of deuterated forms, such as tents of the non-exchangeable sites of sugars are related terpenes, or those with slow molecular dynamics, such to plant type. In particular, beet sugar is characterized by as sugars or phenoxy derivatives. a lower 2H content than cane Sugar or grape sugar. If the signal peaks (e.g. from the -CHD and -CH2D Forms of ethanol that are monodeuterated at the methyl, groups of ethanol, and from the trimethyisilane (TMS) methylene or hydroxyl sites are distinguished on the reference) have similar line widths and are well 2H NMR spectrum of ethanol, but only the methyl and resolved, it is possible to consider the signal heights as methylene deuterated forms are of practical interest for intensity indicators. In most other situations, different the characterization of ethanol since, in an ethanolic signals do not possess the same intrinsic line width, and solution, the hydroxyl groups undergo fast chemical some lines may even overlap. In such cases, the height exchange with water. is meaningless and peak areas have to be used. Curve The 2HPH ratios at the methyl [(2HPH)c2] and fitting is often necessary to ensure repeatability of methylene [(2H/IH)cl] sites of ethanol are closely related (2H/IH)i values. to those of the starting products, water and sugar. It has been shown that the isotope ratio (2HPH)c2 is about five Determination of '3(: content times more sensitive to the 2H content of the nonThe ~C contents are measured using an isotope-ratio exchangeable sites of glucose than to the 2H content of mass spectrometer. The samples are heated in sealed tin the starting water. In contrast, the isotope ratio containers in a combustion chamber that contains a pure (2HPH)c, is highly dependent on the 2H content of the oxygen atmosphere at 10500C. At this temperature, the water of the furmentation medium, and is practically tin container melts. The tin vapours that are produced insensitive to the 2H content of the starting glucose. provoke a violent exothermic reaction, and the tempera- Therefore, these two parameters are very useful for ture reaches 1800-2000"C. Under such conditions, determining the origin of ethanol resulting from the ferorganic samples are completely oxidized, and the prod- mentation of sugars. For example, due to the high rate ucts of the combustion (CO2 and H20) are eluted using of transpiration in the vine, the 2H content of the water helium through chromium and cobalt oxide filters that in grapes is higher than that of water in the environcomplete oxidation and retain the sulphur- and halogen- ment, and leads to relatively high (2HPH)ct values for containing compounds. The CO2 is introduced into the grape alcohols (~130ppm). In contrast, alcohols from mass spectrometer; the '3C content is determined and beet or cereals are generally characterized by lower the '3CP2C ratio is compared to that of a working refer- (2H/'H)cl values (~120ppm), since surface or ground ence standard previously calibrated against the inter- water is used in the fermentation medium. Comparison national '3C reference standard ('PDB' - carbonate from of (2HPH)c2 ratios not only enables the discrimination of the Pee Dee belemnite formation in South Carolina, alcohols from C3 and C4 plants, but also makes it possUSA). Results are usually expressed in the 8 notation4: ible to distinguish between alcohols from different kinds of C3 plants. Alcohol from sugar beet is characterized ( "-',o, ) × 1000 by low (2HpH)c2 values, while alcohols from C3 cereals RRef. have intermediate (2HPH)c2 values between those of where R is the ~3Cp2Cratio of the sample and RRa. is the grape alcohols and beet alcohols. '3C/'2C ratio of the working reference standard.
Determining the origin of ethanols Differentiation between synthetic and fermentation ethanols in beverages and spirits may be achieved by measuring '4(2 levels, but the discrimination of alcohols from different plant sources is much more complex. From a biochemical point of view, the isotopic composition of the sugars synthesized by a plant is representative of the 86
Detecting the chaptalization of wine and characterizmg mixtures of alcohols Chaptalization is the addition of beet or cane sugar to the grape must, before or during fermentation~ in order to increase the alcoholic grade of the wine6. Although the addition of sugar from a (24 plant can be detected by mass spectrometry measurements of the overall '3C/'2C ratio, no efficient method for detecting the enrichment Trendsin Food Science& TechnologyApril 1991
or watering-down of wines was available before the On the other hand, labelling a product as being of development of the SNIF-NMR procedure7.s. natural origin enhances its cmnmercial value. Indeed, a Since the isotope 'fingerprint' of a wine reflects the method proving the authenticity of natural propionic origins of the precursor sugar(s) and the must water, the acid, for example, would probably show the preservdetection of exogenous sugars and water is based on ative in a new light. weighted equations in which the (2HPH)c2 or (2H/'H)cm On the basis of work done on ethanols, a new proratios of ethanol are expressed in terms of the contri- cedure has been developed for the authentication of acetic, propionic and lactic acids. We have shown that butions of alcohols from several origins: the fermentation of sugars leads to a depletion in the 2H (2H/IH)c~ = x(2H/*H)V2+ (1 -x)(2H/IH)& (4) content of the final ethanol with respect to the 2H conwhere M, V and B refer to the chaptalized wine, pure tent of the starting glucose, and that the 2HPH ratio of grape wine and beet-sugar ethanols, respectively; x is the -CH3 group of acetic acids correlates well with that the mole fraction of grape ethanol, and (1 -- x) is the of the corresponding ethanols from which they are mole fraction of beet-sugar ethanol. Equation 4 can also derived by chemical or bacterial oxidation9. Similarly, be expressed in terms of the degree of chaptalization the 2H contents of the -CH3 groups of propionic acid (C) and the total alcoholic grade (t) of the chaptalized and lactic acid reflect the origin of their glucose precursors; the 2H contents of the CH2 group (CHOH for lactic wine: acid) reflect the composition of the medium in which t C x= ; (I - x) = C (5) the fermentation took place. It should be noted that S t t and R lactic acids obtained from the same medium but Combining Eqns 4 and 5 gives: using different microorganisms have different 2H/tH t [ (2H/JH)V2 - (2H/IH)c~ 1 c = t ~ (2H/'n)g2J (6) ratios at the chiral centre, CHOH. Indeed, naturally occurring lactic acids can adopt either the S or R configurSimilar equations can be derived for nC/'-~C isotope ation; Lactobacillus leichmannii gives the S isomer ratios, and for chaptalization using cane sugar. Matrix exclusively, while Rhizopus oryzae gives the R isomer. methods may be used to characterize complex mixtures Moreover, the -CH3 group is only slightly affected by of ethanols; (n - 1) isotope ratios may be used to resolve chemical transformations occurring on the other parts of a mixture containing ethanols from n different origins. the molecule. Therefore, it is possible to understand the In approximate terms, a change of ~0.8ppm in the relationship between a starting material and the final (-~H/~H)c2 value of an unknown wine with respect to that product of a process when lactic acids are involved. The of a natural (unehaptalized) wine corresponds to ~1 ° of relationship between the -CH3 group of ethanol and the -CH3 group of acetic acid is of great importance. exogenous alcohol. The large amount of information on ethanols from all over the world that is available can be used for Acetic acids, propionic acids and derivatives Organic acids such as acetic, propionic and lactic determining the origins of acetic acids in acids are commonly used in the food industry as vinegars t°. Applying both the SNIF-NMR method and preservatives and seasonings. All may he obtained by ~3C/~2C determination to a wide variety of vinegars the fermentation of precursors containing sugars. provides an unambiguous method of authentication. Knowledge of the origin of such compounds is of Natural and synthetic acetic acids can be clearly distinparticular interest to the manufacturer, vendor and con- guished (Fig. 1). Furthermore, the botanical origin of sumer. For instance, vinegar is defined as a product each acid can be determined. A careful study of the manufactured from naturally occurring ethanol. A repeatability of the analytical procedure (extraction of diluted aqueous solution of commercial acetic acid does acetic acid from vinegars, 2H NMR measurements and not fit this definition. Methods for the characterization t3Cr2C ratio determinations) using known mixtures of of ~ynthetic acetic acid in vinegars have been developed natural and synthetic acetic acids suggests that the and, until now, most have been based on the calculation method can detect as little as 5% synthetic acid in a of 13C/12C ratios, on ~4C scintillation counting, and on natural vinegar. At the moment, a sensitivity level of the determination of uaee or minor compounds in 20% may be expected if no information on the origin of vinegars. However, the d~termination of minor com- the precursors is available. This detection limit should pounds is of little value when the vinegar has been decrease as more data on vinegars becomes available. fermented from distilled ethanol, which contains only low amounts of minor compounds. Authentication Aromasand flavourings The SNIF-NMR method can also be applied to molmethods based on the determination of stable and ecules larger than those that have been described above. radioactive carbon isotopes have been popular. However, adulteration may not be detected if the (natural) The main limitations of its use are sensitivity and signal product of interest has a higher isotope content than that resolution. Molecules such as benzene derivatives (e.g. of the synthetic one; the addition of compounds en- benzaidehyde, anethole, vanillin and thymol), terpenes riched with respect to a particular isotope (stable or (e.g. pinenes, linalool, citral and menthol), aliphatic radioactive) is a common procedure in the adulteration esters, steroids, alkaloids and purine bases (e.g. caffeine) are easily analysed n. At present, 2H NMR is the only of ethanols 4, vanillins and benzaldehydes7'a. Trendsin FoodScience& ]i~chnologyApril 1991
87
-10"
•O
--20"
0
°~,
=° •
-30"
•
O
site-specific isotope ratio (2HrH)i depends on the origin of the vanillin molecule: it constitutes the 'fingerprint' of the molecule. Measuring the site-specific 2HPH ratios affords an excellent means of discriminating between several vanillin sources. Selective deuterium enrichment and depletion is a rather difficult and expensive task that is not justified by the price of natural vanillin.
O OR
0
00
Anethole The identification of anetholes was hitherto based on 0 00 0 the chromatographic determination of minor products, 0 but, with improvements in purification methods and the 0 -40' B sophisticated distillation of anethole oils, the method has O become unreliable. The anethole molecule contains six non-equivalent sites, and the 2H NMR spectrum recorded at a field strength of I I.4T shows six monodeuterated o --50 forms. As in the case of vanillin, the analysis of specific 90 1;0 ;Ao (2H/'H)= isotope ratios provides interesting information ~(2H/1H)c2(%o) regarding the origin of the product. Clear subsets of different plant origins can be identified for star anise, Fig. 1 fennel and green anise. Since the values of (2HPH)i Plotof 8(nC/12C)versus5(2HI=H)at the methyl site,for natural(O) and ratios arc sensitive to fractionation due to distillation, synthetic(~) aceticacids. which is a problem commonly encountered for liquid samples, intramolecu!ar isotope ratios (Ritj - the 2H/~H method that can unambiguously determine the origin ratio at site i divided by that at site j) are also taken into (botanical, semisynthetic or synthetic) of an aromatic account. The Ri~j parameters are measured from the molecule. The technique has been successfully used NMR signal heights of peaks i and j, These parameters to determine the origins of vanillin 12, anethole t3 and are, to some extent, independent of the physical treatbenzaldehyde. ments to which the sample has been subjected. Vanillin VanUlin can be obtained from three main sources: vanilla beans (natural), gualacol (synthetic) or lignin (semisyntbetic). Other uncommon sources +are eugenol and isoeugenol. Vanillins from all five sources can De distinguisbed by 2H NMR, but this discussion will be limited to vanillins from vanilla bean, gualacol and lignin. Mass spectrometry measurement of ~-~C levels has long been the only way of identifying the natural origin of a vanilla extract. The detection of minor products (e.g. parahydroxybenzaldehyde and vanillic acid) can be used only on the extract, not on the final product (or pure vanillin). A vanillin is considered to be natural if its 8(~3C/L=C) deviation is greater than -21.5%~ The total a3C contents of vanillins from guaiacol or lignin are less than that of natural vanillin (from Vanilla planifolia). Consequently, it is obvious that a synthetic vaniilin that has been enriched in ~3C can appear to be natural on the basis of the ¢~(~3C/~2C) measurements alone. Measurement of the overall 2H/~H ratio has been used to supplement information supplie.4 by the 13CP2C ratio. The overall 2HPH ratios for vanillins from various sources are ranked as follows: lignin < natural < guaiacol. However, again, an appropriate mixture of vanillins from lignin and guaiacol would lead to an erroneous conclusion regarding the origin of the vanillin sample. At a high magnetic field strength (11.4 T; 76.77 MHz), vanillin exhibits five non-equivalent hydrogen sites (two sites are equivalent); therefore, vanillin contains five different monodeuterated forms. The value of each 88
Benzaldehyde The natural source of benzaldehyde is bitter almond. In addition, a semisynthetic product is obtained from cassia oil, in which the natural cinnamic aldehyde is chemically transformed into ~nzaldehyde. Biotech+ nology-based synthetic methods are being developed to manufacture benzaldehyde that would be considered as a natural product. Literature data have shown that 14C scintillation counting and determining the 8(13C/'2C) deviation can help to discriminate between benzaldehydes of synthetic and natural origin. However, it has not been possible to obtain a clear-cut differentiation between benzaldehydes from botanical and synthetic sources. The SNIF-NMR method can be applied to the problem of the authentication of benzaldehyde because the (2HPH)~ and Rj]j values of each fragment of the molecule are representative of the corresponding biosynthetic pathway. Thus, a benzaldehyde originating from cinnamic aldehyde can be distinguished from natural benzaldehyde by examining the 2H content of the aromatic ring; the 2H contents of both the aromatic ring and the aldehyde function of synthetic benzaldehyde are different from those of natural benzaldehyde. Again, it is better to supplement the use of (2H/IH)i ratios with Rilj values, as benzaldehyde is often purified by distillation. Conclusions The methods described represent a new battery of tests that can be used for certifying the origin of expensive Trendsin FoodScience& TechnologyApril 1991
commodities. In addition, such techniques are effective at detecting the more sophisticated methods of food ingredient and beverage adulteration, such as the manipulation of total isotope ratios. In the future, such methods may find applications in assuring the quality of kosher commodities and in detecting the products of biotechnology-based processes. References
1 Marlin, G.I., Marlin, M.L., Mabon, E and Michon, M.J.(1982)]. Chem. Soc.Ser.Chem.Commun.616-617 2 Marlin, M.L. and Martin, G.J.(1990)in NMR:BasicPrinciplesand Progress(Giinther, H., ecl.),pp. 1-61, Springer-Verlag 3 Gonfiantini,R. (1978}Nature271,534-536 4 Craig,H. (1957} Geochim.Cosmochim.Acta 12, 133-149
5
Marlin, G.I., Marlin, M.L., Mabon, F. and Michon, M.J. (1982) Anal Chem. 54, 2380-2382 6 Marlin, G.I. (1990) Endeavour14, 137-143 7 Marlin, G.I. and Marlin, M.L. (1983) 1. Chim. Phys. 80, 293-297 8 EuropeanCommunity (1989) Regulation on the Organization of the Market of Wines and Vine Products (no. 822/'87), modified 19 June 1989 (no. 20432-2048/'89) 9 Martin, G.J., Zhang, B.L., Naulet, N. and Marlin, M.L. (1986) ]. Am. Chem. Soc. 108, 5116-5122 10 Vallet, C., Arendt, M. and Marlin, G.J. (1988) Biotech. Tech.2, 83-88 11 Hanneguelle, S., Remaud,G. and Marlin, GJ. (1990) Parfums Cosm~t. Ar~mes 94, 95-109 12 Maubert, C., Guerin, C., Mabon, E and Martin, G.J. (1988) Analysis 16(7), 434-439 13 Marlin, G.J., Marlin, M.L., Mabon, F. and Bricout, I. (1982} }. Am. Chem. Soc. 104, 2658-2659
Review
mstenosls: biology and implications for the food industry L O
Transmission, via foods, of Listeria monocytogenes, the causative agent of listeriosis, has become of major concern in recent years. Although the pathogen has been associated with human disease since 1929 (Ref. 1), it was not until recently that foodborne transmission of listeriosis to humans was conclusively demonstrated 2. In its most severe form, listeriosis can be a life-threatening infection; in some cases, it can cause severe complications in pregnancy, leading to spontaneous abortion. The severity of the symptoms and the high mortality rates (-30%) associated with recent outbreaks of listeriosis have led to widespread public concern, and have increased efforts to prevent further cases. Detailed information about the growth characteristics, ecology, virulence mechanisms and clinical aspects of L. monocytogenes is now available. Such information enables more effective control of the transmission of L. monocytogenes via foods.
The genus Listeria consists of small, non-sporeforming Gram-positive rods that have a tendency to form diphtheroid palisade arrangements. They are catalase positive, hydroiyse aesculin, and are motile by means of peritrichous (uniformly distributed) flagellae, showing a distinct tumbling motility at room temperature3. The genus contains seven species, of which L. monocytogenes is the most important human pathogen; L. ivanovii, L. seeligeri and L. welshimeri are rare causes of human disease 2. The ~maining three species, L. innocua, L. murrayi and L. grayi, are considered avirulent'. L. monocytogenes can be distinguished from Cormac G.M. ,Gahan and John K. Collins are at the Deparlment of
Microbiology and the National Food Biotechnology Centre, University CollegeCork, Irish Republic. Trends in Food Science & Technology April i991
i
•
•
Cormac G.M. Gahan and John K. Collins other species of the genus by specific biochemical tests and by a haemolysis test known as the CAMP (Christie-Atkins-Munch-Petersen) test. Strains of L. monocytogenes can be classified into at least 16 different serotypes 4. However, a large proportion of environmental and clinical isolates of L. monocytogenes belong to a small number of serotypes (notably, serotypes 4b, l/2a and 1/2b) 2"4. Phage typing has been of some use in subdividing strains of the pathogen, but many isolates are not amenable to typing by this --1.'--riLl " n...t.~ method4. Consequently, a more apv,~ao,e typing ~y~,~., is required to show definitively when a clinical isolate and a food isolate represent the same swain. Isoenzyme typing is being developed for this purpose4. L. monocytogenes will grow at temperatures of 1-45°C; the temperature range for optimal growth is 30-37°C (Ref. !). Although growth is possible at refrigeration temperatures, the multiplication rate is slow. In chocolate milk, L. monocytogenes has a doubling time of 1.2-1.7 days at 4°C; the doubling time is 0.65-0.69 hours at 35°C (Ref. 5). The pH range for growth of L. monocytogenes is reported to be between 4.5 (Ref. 6) and 9.6 (Ref. 5), but the organism favours @1991,EBevierSciencePublishersUd,(UKI 0924-2244/gl/$02.00
1:]9