Rose Wilson Bragg,' Yvonne C h o ~Lawrence ,~ Dennis, Lloyd N. Ferguson, Susan H o ~ e l lGeorge ,~ Morga,' Craig Ogino, Harriet pugh,' and Manque Winters' California State University-Los Angeles Los Angeles, 90032
I Sweet Organic Chemistry
...to examine some observed structure-taste correlations, and to explore one or two generalizations which might help elucidate the mechanism of taste stimulation. What molecular properties do compounds (I-VI) have in common to make them sweet?3
P" H
Sodium
Saccharin 300X
Sodium C~elamate 33X
1111) S R Oxime ~ 450X
lVll Neohe eridin Dihvdrochalcone 2000??
Even more difficult, why are compounds (VII-X) sweet whereas the structurally similar compounds (XI-XIII) are not?
Sweet
in the study of sweetening agents. A wide variety of structures have long been known to be sweet but several new extremely sweet types have emerged during the last decade (I).As with other physiological responses to chemicals (drug reactions), there are several approaches taken in attempting to account for the tastes of sweet substances. This article is written to examine some observed structure-taste correlations and to explore one or two generalizations which might help elucidate the mechanism of taste stimulation. I t is not an exhaustive tabulation of structures reported to he sweet. Some Sweet Molecular Types Natural Products (2)
The carbohydrates such as sucrose and dextrose, of course, are the universal sweeteners, with sucrose taken as the reference standard for relative intensity of sweetness. An excellent review of the structure-taste relationships among sugars has appeared very recently (3). Several glycosides have been isolated which are extremely sweet ( 4 ) , such as Stevioside (280-300X) from an herb of Paraguay, Osladiu (3000X) from a species of fern (51, and Glycyrrhizin (50-100X) from a licorice root (6). Sweet substances have been extracted from tropical plants, which in some cases served as a source of sweetness for the local inhabitants (7). For example, two similar proteins, named thaumatin I and thaumatin I1 have been isolated from a west African plant. Hofh protein8 h w e molecular weights of ahout 21.000 and are a~nroximatel\~ 1600 times as sweet as sucrose on a weight basis*). Their amino acid contents have been published (9). Another sweet protein, monellin, is obtained from a plant found in Ghana (10). With amolecular weight of 10,700, it is approximately 3000 times sweeter than sucrose on a weight basis. Interestingly, monellin consists of two unlike, noncovalently bound chains, neither of which is sweet alone (11). However, solutions of the two separated chains slowly recover the sweetness when mixed. The tertiary structure of monellin is important to its sweetness (12). A diterpene acid (XIV) has been isolated from pine tree rosin with a relative sweetness of 1600-2000X (13). Only the isomer shown, among the four stereoisomers isolated, is sweet.
Not Sweet -
qb
H,C
10 H~O.H 1XIVI
These and other questions about the sense of taste regarding sweet substances are of intense interest these days. The recent ban of cyclamates and threatened cancer risk of saccharin have produced a surge of international activity
'National Institutes of Health Minority Student Biomedical Research Trainee. 2National Science Foundation Undergraduate Research Trainee.
3Relativesweetness. Sucrose = 1X. Volume 55. Number 5, May 1978 1 281
.
..the sweetest compounds known to date are dipeptides; Laspartyl-aminomalonic diester is reported to have a relative sweetness o f ZZ,OOO-33,ZOOX. Svnthntics
There is an extensive, international search for synthetic sweeteners, many of which are amides, imides, glycosides, oximes, or polymers. Shallenberger has convincingly shown that virtually all sweet compounds have a molecular fragment designated A-H, B, in which A-H and B are proton-donor and -acceptor, respectively, in H-bonding partnership with the receptor site (14).
Suhstantial evidence continues to be reported in support of this hypothesis. However, not all compounds with such a structural fraement are sweet. The A-H, B moiety, then, is " necessary for sweetness but does not guarantee sweetness. Many A-H, B groupings can be found in a carbohydrate fragment with its 12-glycol structure, although all carbohydrates are not sweet, of course. I t is desirable therefore, to locate the A-H, B system in a sweet carbohydrate which is resoonsible for its taste. Birch and his coworkers (15)have identified the active glycol fragment in gluco-pyranosides hy systematically replacing the OH groups ur the hydroxyl protons. Their results indicate that the hydroxyl proton on C4 and the oxygen atom on C3 are important fur a sweet taste. Thus. 3 deoxv IH for OH) elucose is not sweet. whereas I-, 2-, 4-, and 6-deoxyglucose are. A useful exoerimental tool here is ir spectroscopy. If the A-H group is i o be availableasa proton d;,nor for H bonding it cannot be tied U D in intramolecular H bonding and should exhibit a free 0 - H band in the ir. Shallenberger has shown this to he the case (16). The free OH hand of polyols is usually a sharp peak in the 3400-3600 cm-1 region. Thus, sucrose and fructose, which are sweet, have sharp bands near 3500 cm-I whereas raffinose, which is essentially tasteless, has only a composite band below 3400 cm-I.
-
Dipeptides
Following an accidental discovery of the sweet taste of aspartylpbenylalanine methyl ester, a variety of molecular changes were made to explore the relationship of structure to taste (17). Indeed, the sweetest compounds known to date are among the dipeptides, with the L-aspartyl-aminomalonic diester (IV) reported to have a relative sweetness of 22,00&33,2M)X (18). The member to get closest to the market in the US. was aspartame [XV, RI = CHzCsHs, Rz = COzCHa, 150-200x1 (19), although FDA approval was subsequently withdrawn. Some of the structure-taste relationships observed for the dipeptides (XV)
aminomalonic acid but the other amino acid may be one of several without losing sweetness. The L-L isomer is generally the sweetest although some L-D isomers are sweet; (3) the peptide link must be unchanged. Thus replacement of the NH by 0 or N-alkylation, or reversing the CO-NH sequence destrovs the aweet taste:. (4) . . R,. and RI must be hvdrophobic . . groups. Thus RI may be an ester group but sweetness is lost ro when R Iis a carboxyl, amide, or hydrazide group (wing their hydrophilic characwr~;( 5 ) the size and shnpe of the HI-CH-R?e r o u ~ i n eis imponant. Sweetness increases with &eater difkrence the sizes of R1 and Rz (18,20). Different approaches have been used to discern the molecular geometry of the R1-CH-R2fragment that is necessary for a sweet taste. For example, Brussel, Peer, and Heijden used molecular models and found a correlation of shape and sweetness for R2-CH-C02MEanalogs of (XV). The dipeptide is sweet when Rz has a length of 4.8-8.8& a volume greater than 30A3, and has a maximum bulkiness at 2-5A from the -CH- carbon (21). Temussi and coworkers determined the minimum energy conformation of the apolar group at the asymmetric carbon in aspartame from nmr measurements and about the CO-NH group by energy calculations (22). The preferred configuration has a rather coplanar polar portion, amide, and benzyl groups, whereas the ester group is perpendicular to the rest of the molecule. Accordingly, the molecule is fairly flat with a short orthogonal projection. They believe that the receptor site is a deep, narrow cleft with two interacting parts, one for locking the sweet molecule and the other for triggering the nerve impulse. However, it is hard to imagine the fenchyl group of (IV), the sweetest compound yet reported, fitting into this cleft in place of the benzyl group. Ariyoshi has rationalized the tastes of the dipeptides through . aspartyldipeptides Fischer projection formulas ( 2 0 ~ )Sweet have structure (XVIA), where R7 is larger than RI. ~
~~
d~
~
~
COH ,
I
H-
F" 'i 7"
dNH,
NH H-
I-R,
7
R.
F""
F" H-P-NHA F" I R'-P-H NH
R,
(XVI A l
IXVI B l
Sweet
Nonsweet
OH
I
B./N A
(XVIII
Perillartine R.S. = 2000X
can he summarimd as follows (20):(1) The :I-carboxyl und the amino erouns must he unsubstituted (e.e. CO,ME for COOH or (CH3)zN for H2N destroys the sweettaste). No more than two carbons may separate the carboxyl from the peptide bond. Apparently, to elicit a sweet taste, the amino and carboxyl groups need to be intramolecularly H-bonded; (2) the N-terminal amino acid is virtually limited to L-aspartic acid or
- ~ ~ ~ . ~ ~
282 1 Journal of Chemical Education
Although several oximes have been reported UI be sweet, Perillartine is the only one to have a limited commercial use, being used to sweeten tobacco in Japan (23). Acton and Stone (24), however, took it as a starting point and made systematic changes with the objective of maintaining sweetness but increasing water solubility and eliminating bitterness. After
making scores of structural changes, it was observed: (1)that the syn-a,&enaldimine, free hydroxyl, molecular fragment (XVII) is essential for sweetness. This, then, contains the sweet A-H, B sapophore (a group that elicits a sweet taste). The rest of the molecule, termed the carrier, affects the taste qualities; (2) polar groups in the carrier, such as OH, increase Hz0 solubility but markedly lower sweetness; (3) mildly polar groups, such as ethers, provide a good intermediate balance between reduced sweetness and increased water soluhilitv. although sweetness is markedly affected by their positions: Ring oxygen atoms destroy the sweet taste; (4) a cyclic carrier produces greater sweetness than a flexible open chain; however the sweet taste was lost with a hicvclic riun, -. and burnine: or hitter tastes accompany an aromatic ring. As a result of trial and error structural modifications following these empirical guidelines, Acton and Stone (24) find SRI Oxime V (111) to he a potentially very useful sweetening agent, which is undergoing toxicological studies. Solfarnates
The most well-known of the sulfamates are the cyclamates. Following the accidental discovery of the sweet taste of cyclamate by Sveda in 1937, many sulfamates have been synthesized for evaluation of their tastes. Most of the structureactivity relationships have been summarized by Benson and Spillane (25), which can he outlined as follows:
saccharins and the oxathiazinone dioxides. Functionally, the latter differs from the saccharins
"
0
IXXIV)
IXXV)
Saccharin salt
Orathiazinane dioxides
through a separation of the SO2 group from an electron pi system by an oxygen atom. Based on the tastes of approximately 80 saccharin derivatives, Hamor has noted three generalities (27): (1)Suhstitution of the proton on the imide nitrogen, other than to form the salt, destroys the sweetness; (2) electron withdrawing groups on the benzene ring usually produce bitterness; (3) sweetness is sometimes retained when there is an electrondonating group attached to the benzene ring (NH?, CH3). Changes in the imide function may or may not destroy the sweet taste. The tastes which accompany some structure modifications are shown below (28).
Saccharin Sweet
0 Nonsweet
H
Sweet
(XIX)
(XWI) n =I-I1
(1) normal and medium-size rings attached to the sulfamate function -NH-SO; M+ yield sweet salts (XVIII, n = 3-7) hut small and large ring derivatives are not sweet (XVIII, n = 1,2,10); (2) insertion of a methylene group between the ring and sulfamate function may or may not destroy the sweet taste. Thus. N-cvclohexvlmethvlsulfamate is not sweet hut N-cyclope~itylm;thylsul~amat~(~~~) is sweeter than cyclamate WJ"2: ( 3 ) N-alkvlation of the cvclohexvlsulfamates lXXl . . causes loss of the sweet taste hu
