Anal. Chem. 1998, 65, 3119-3120
3119
Quantitative Structure-Retention and Structure-Odor Intensity Relationships for a Diverse Group of Odor-Active Compounds Leanne M. Egolf and Peter C. Jurs’ Department of Chemistry, 152 Davey Laboratory, The Pennsylvania State University, University Park, Pennsylvania 16802
Numerical representations of structure-based features are used to estimate both the retention indexes and sweetnesses of a diverse set of industrially important fragrance compounds. Retention indexes measured on nonpolar as well as polar stationary phases are modeled with accuracies of 3.6% and 5.6% at the mean of the respective retention ranges. Similar success was achieved when the developed equations were applied to predict the retention indexes of external data set compounds. Finally, the implications of using strictly 2-D structural information versus incorporatinggeometrical informationare explored and discussed. The intensity of sweetness attributed to each compound is quantitatively predicted using identical multiple linear regression techniques. Difficulties encountered in this portion of the study warranted a critique of the procedures used to gain access to the odor data. As a consequence, the limited control exerted over several experimental variables is questioned. INTRODUCTION Odor-active compounds are found in a wide variety of consumer products ranging from foods and perfumes to health care products and medicines. In combinationor alone, flavor and/or fragrance compounds are used to induce consumers to associate favorable impressions with a given product. In some cases, products have one predominant componentwhich provides the characteristic odor. In most cases, however, products containing odors include a complex mixture of fragrant compounds.’ The goal of the fragrance researcher is to be able to separate and identify the individual mixture components and to characterize the quantitative as well as qualitative aspects of each isolated compound. The primary objectives of this project will be to build two types of relationships. The first will allow us to differentiate the fragrance compounds on the basis of their chromatographic retentions while the second will allow us to estimate the intensity of a specific odor characteristic associated with each compound. In a chromatographic system, the molecular structure of a compound determines its relative affinity for the mobile and stationary phases and, therefore, its retention characteristics. Much research has been directed at using topological, geometrical, electronic, and physicochemical parameters to generate predictive equations which will increase our understanding of the relationship between structure and retention.- Previous retention studies from our group have ~~
(1) Remington’sPharmceutical Sciences, 16thed.; 0801, A,,Ed.; Mack Publishing Co.: Easton, PA,.1980, pp 1230-1239. (2) Kaliamn, R. Quantitative Structure-Chromatographic Retention Relationships; John Wiley & Sons: New York, 1987. 0003-2700/93/03655119$04.00/0
mainly been targeted at homogeneous data sets.”2 Consequently, the equations are limited in practical utility. Industrial applications would profit from a more universal model. Since many consumer products include numerous fragrance components, each of which may contain multiple functionalities, separation and subsequent identification would be greatly accelerated with a single predictive equation. Furthermore, developing a more general relationship could provide unique structure-retention insight which would aid in experimental contexts as well. The research presented here will extend analytical capabilities. Using numerical representations of molecular structure (i.e., descriptors) as variables in linear regression equations, the chromatographic retention indexes of a highly diverse set of industrially important odor-active compounds will be predicted. Our second area of study, the measurement of odor intensity, is based on human perception and is subjective in nature. Thus, the relationship between molecular structure and odor characteristics has remained much more elusive. In the perfume industry for example,discriminationis provided through qualitative prose rather than quantitative scoring.18 For years investigators have been attempting to convert from analyses dealing with purely psychological impressions to those which incorporate concrete physiological data. In the physiological model, odor is defined as the result of the interaction between the odorant molecule and the olfactory receptor.14 Both the quality and intensity of the odor are known to be a function of the molecular structure of the stimulant.14-19 Consequently,these odor indexes should also be amenableto our structure-basedcomputationaltechniques. This hypothesiswill be explored as we also attempt to predict the sweetness of each odorant in the data set.
METHODOLOGY All computations were performed on a Sun 4/110workstation using the ADAPT software system.m (3) Kaliszan, R. Anal. Chem. 1992,64,629A-631A. ( 4 ) Hansch, C.; Fujita, T. J.Am. Chem. SOC.1964,86,1616-1619. (6)Sander, L. C.; Wise, S. A. Crit. Rev. Anal. Chem. 1987,18,299-415. (6) Bassler, B. J.; Kaliszan, R.; Hartwick, R. A. J. Chromatogr. 1989, 461, 139-147. (7) Walczak, B.; Chretien, J. R.; Dreux, M.; Morin-Allory, L.; Lafoase, M.; Szymoniak, K.; Membrey, F. J. Chromtogr. 1986,353,123-137. (8) Cserhati, T.; Valko, K. J. Biochem. Biophys. 1990,20,81-95. (9)Anker, L. S.; Jurs, P. C.; Edwards, P. A. Anal. Chem. 1990, 62, 2676-2684. (IO)Hasan, M. N.; Jurs, P. C. Anal. Chem. 1990,62, 2318-2323. (11) Stanton, D. T.; Jurs, P. C. AnaZ. Chem. 1989,61, 1328-1332. (12) Rohrbaugh, R. H.; Jurs, P. C. Anal. Chem. 1987,59,1048-1054. (13) Carsch, G.Soap, Cosmet., Chem. Spec. 1991, 67, 40-58. (14) Beets, M. G. J. Structure-Activity Relationships in Human Chemoreception; Applied Science Publishers: London,England, 1978; Chapters 1 and 3. (15) Klopman, G.;Ptchelintsev, D. J. Agric. Food Chem. 1992, 40, 2244-2251. (16) Mihara, 5.;Masuda, H.J. Agric. Food Chem. 1988,36,1242-1247. (17) Vial, C.; Thommen, W.; NU,F. Helv. Chim. Acta 1989,72,139& 1399. (18) Weyerstahl, P.; Krohn, K. Tetrahedron 1990,46,3503-3514. (19) Tateba, H.; Mihara, S.Agric. Biol. Chem. 1990,54, 2271-2276.
0 1993 American Chemical Society
3120
ANALYTICAL CHEMISTRY, VOL. 65, NO. 21. NOVEMBER 1, 1993
-0
Compound 36 Compound 61
Compound 28
Compound 2 Compound 56
v
Compound 52
Jo/ -S
0
Compound73 Compound 50 Compound 13 Figure 1. Representative structures showing the diversity of the data
set. The Data Set. (a) Compounds. A data set of 73 compounds was compiled from compounds which were listed in both our retention index and odor character sources. Representative structures from the final set are shown in Figure 1. While most of the odorant compounds possess a single, sterically accessible polar group, some molecules contain two or more such functionalities. These include alcohol, amino, ester, ether, imine, sulfide, thiophene, and carbonyl moieties. (b)Retention Indexes. The retention indexeswere generated by Jennings and Shibamoto.21 The data are reported using the Kovats (logarithmic) scale, where the retention index of a compoundis determined relative to the retentions of the n-alkane series on a specific chromatographic system. The retention indexes were obtained on two smooth-bore wall-coated opentubular glass capillary columns. The nonpolar column, 0.28mm X 50 m, was coated with methyl silicone OV-101,admixed with 1 % Carbowax 20M as an antitailing additive, and was temperature programmed from 80 to 200 OC at 2 OC/min. The polar column, 0.20 mm X 80 m, was coated with poly(ethy1ene glycol) Carbowax 20M and was programmed from 70 to 170 OC at 2 OC/min. A variance of
