A Method to Measure Taste Qualities, Taste Intensity, and Temporal

---

Downloaded by UNIV OF MICHIGAN ANN ARBOR on May 19, 2016 | http://pubs.acs.org Publication Date: March 4, 2008 | doi: 10.1021/bk-2008-0979.ch013

Chapter 13

A Method to Measure Taste Qualities, Taste Intensity, and Temporal Profile of Compounds Aimed at Human Consumption by Taste Nerve Recordings in Monkeys Göran Hellekant and Yiwen Wang Department of Physiology and Pharmacology, Medical School, University of Minnesota, Duluth, M N 55812

A method is described to measure taste qualities and intensity of compounds aimed at human consumption by recording from taste nerve fibers of monkeys. Here we demonstrate its usefulness by presenting results of a comparison of sweetness of brazzein derivatives and bitterness of denatonium benzoate analogs as assessed by a human taste panel and recorded from monkey single taste fibers. The correlation between the responses in sweet sensitive fibers in monkeys and the estimates of sweetness by a human taste panel was 0.78 for 25 analogs of the sweet protein brazzein, and 0.9 between the responses of bitter sensitive fibers and human bitterness estimates for 6 analogs of the bitter compound denatonium benzoate.

© 2008 American Chemical Society Weerasinghe and DuBois; Sweetness and Sweeteners ACS Symposium Series; American Chemical Society: Washington, DC, 2008.

185

186

Downloaded by UNIV OF MICHIGAN ANN ARBOR on May 19, 2016 | http://pubs.acs.org Publication Date: March 4, 2008 | doi: 10.1021/bk-2008-0979.ch013

Introduction Evaluation of the taste of compounds aimed for oral consumption is a necessary step in the development and marketing process. Data acquisition from human taste panels can be tedious, especially since toxicological concerns have to be put to rest prior to the taste tests in humans. This is not a concern in taste nerve recordings from animals because the compounds never enter the body of the animal. Instead, the main concern is: how applicable are the animal data to human taste? With regard to the sweet and bitter taste qualities, all mammals tested so far, recognize these taste qualities, as judged by behavioral tests, but the compounds that elicit these qualities differ among species. Although some earlier investigators clearly were aware of the existence and importance of species differences in taste ' , it was not generally well understood, as judged by the many attempts to relate human taste qualities with the taste fibers from cats, rodents and other non-primate species ' . This makes the choice of animal model crucial. l

2

3

4

We became aware of both qualitative and quantitative species differences in taste when our taste nerve recordings from human and monkey showed a large enhancement of the taste nerve response to acids after miraculin, whereas none was recorded in rat . Miraculin is a taste modifier that adds sweetness to acids in humans and monkeys but not rats. This and later studies ' lead us to the conclusion that mammals have different types of sweet receptors . Our further studies suggested that the sense of taste is more similar among primates, than between primates and non-primates, and that combinations of behavioral observations and taste nerve recordings from non-human primates could be used to elucidate human taste mechanisms, including how information from the taste buds is mediated to the brain, that is, how taste is coded. Recordings from taste nerves of several primates over more than 30 years, in particular from three higher primates, have revealed that indeed there are nerve fibers in the taste nerves, whose response spectrum mirrors the human taste qualities. Thus, we and others have found that sweet and bitter taste are linked to activity in two groups of taste fibers, one responding to sweet, the other to bitter tasting compounds, and that nerve impulses in these fibers evoke either a sweet or bitter taste quality * . This coding of taste has been called labeled lines. These findings refute the idea that all taste fibers have to participate in creating a taste quality, the across-fiber pattern theory, which originally was proposed to explain the lack of connection between human taste qualities and single fiber response spectrum in the cat . It was later expanded, using rodent recordings and recently supported again " Since then supportive data for the labeled line theory, obtained with other techniques, have been published. Genetic engineering, combined with behavioral 5

6

7

8

9

16

3

17

4

19

Weerasinghe and DuBois; Sweetness and Sweeteners ACS Symposium Series; American Chemical Society: Washington, DC, 2008.

187 experiments in mice, have shown that bitter and sweet tastes are sensed by separate sets of receptors, and, what is more important from the point of coding, these receptors are located in different, not overlapping taste bud cells ~ . This is an important condition, because i f bitter and sweet receptors were colocalized, it is difficult to visualize how a taste fiber, synapsing with a cell with two different receptors, could cany only one taste quality. In the following we present data from an Old-World primate, the rhesus monkey, M mulatta. It provides data on the intensity of a sweet and a bitter compound, which are well correlated with assessments by a human taste panel.

Downloaded by UNIV OF MICHIGAN ANN ARBOR on May 19, 2016 | http://pubs.acs.org Publication Date: March 4, 2008 | doi: 10.1021/bk-2008-0979.ch013

20

27

Methods, Animals, Subjects and Stimuli The taste nerve responses were recorded under general anesthesia from the front of the tongue of rhesus monkeys, M. mulatta, through the chorda tympani proper (CT) and the back of the tongue, through the glossopharyngeal nerve (NG). Some 30 taste stimuli, representing the sweet, bitter, salty, sour and umami taste qualities, were applied to the tongue while the responses in single taste fibers were recorded . The responses in the fibers were subjected to hierarchical cluster analyses. The analyses resulted in three to four clusters, which were linked to the sweet, bitter, salty or sour tastes. The fibers in these clusters are labeled S, Q, Ν and Η fibers. The responses in single nerve fibers, belonging to the S cluster, were then used as a tool to assess the sweetness of the sweet protein Brazzein and 25 2 8

2 9

in

structurally closely related derivatives of brazzein . As a measure we used the impulses over 5 sec of stimulation. Similarly we used the responses in Q fibers to assess the bitterness of denatonium benzoate and 6 of its derivatives. The neurophysiological data were then compared with data from a human taste panel using the Labeled Magnitude Scale to estimate the sweetness or bitterness of the same compounds · . The panel consisted o f 14 volunteers, 6 females and 8 males.

Results Figures 1 and 2 present the result of the cluster analyses of the fibers in the C T and N G nerves as dendrograms. The analyses took into account the responses to all stimuli. The cluster analyses distinguished four major clusters, which, based on the fibers' response to salty, sour, sweet and bitter compounds, were labeled, either the N , H , S or Q cluster. The dendrograms show that the taste fibers can be arranged according to their responses to the compounds humans consider salty, sour, bitter and sweet.

Weerasinghe and DuBois; Sweetness and Sweeteners ACS Symposium Series; American Chemical Society: Washington, DC, 2008.

Downloaded by UNIV OF MICHIGAN ANN ARBOR on May 19, 2016 | http://pubs.acs.org Publication Date: March 4, 2008 | doi: 10.1021/bk-2008-0979.ch013

188

NRH9 2U20E Ν RH92U24 L HRH92U23P HRH92U24D Η RH92A23 Η N R » 5D07B NRH92U27I Ν RH92U20 Η Ν RH92Y0SB NRW2U27B NRHJ2U27E Ν RH95D12C NRH95D9H Ν RH95D12E NRHJ2U27K Η RH92A23B NR W2A23D Ν RH95D12D NRH95D12F HRH92U230 HRH92U23D Η RH92U27 Η Η RH92U2 7F HRH92U20F Η RH92A23A HRH95DBA Η RH92U27A Η RH9SD13G S RH92U24H S RH92U24F SRH92U27G SRH95D07C S RH95D19C SRH92U27D S RH92Y05C S RH92U27 Ρ S RH92U27L S RH92U 23G S RH92 U23R S RH92 U24E S RH92A23C S RH92A23 F SRH92A23Q QRH95D20E Q RH92U2 7N Q RH95D12B Q RH92U2 7J 1 .4

Figure 1. Hierarchial cluster analysis of the response profiles for 47 CT nerve fibers. Fiber number and response category on the basis of its response to the 4 basic solutions are listed on the left.

Weerasinghe and DuBois; Sweetness and Sweeteners ACS Symposium Series; American Chemical Society: Washington, DC, 2008.

189

Downloaded by UNIV OF MICHIGAN ANN ARBOR on May 19, 2016 | http://pubs.acs.org Publication Date: March 4, 2008 | doi: 10.1021/bk-2008-0979.ch013

M-du ster

Q-clu8ter

11

S RH94Y23H

S-cluster

S R H94E27C S RH94E07II S RH94E07 F SRH94E21I S RH94F.2 H S R H 9 4 E 27D S RH94Y23C

—Γ­

1.4

Ι.2

—τ­ ι .0

C lu ste r sim ila rity 0.8

—Γ0.6

"Τ­ 0.4

Ο.2

0.0

Figure 2. Hierarchial cluster analysis of the response profiles for 33 NG nerve fibers. Fiber number and response category on the basis of its response to the 4 basic solutions are listed on the left.

Weerasinghe and DuBois; Sweetness and Sweeteners ACS Symposium Series; American Chemical Society: Washington, DC, 2008.

Downloaded by UNIV OF MICHIGAN ANN ARBOR on May 19, 2016 | http://pubs.acs.org Publication Date: March 4, 2008 | doi: 10.1021/bk-2008-0979.ch013

190 Figures 3 and 4 show the responses of the individual taste fibers of the C T and N G fibers. The stimuli have been arranged along the X-axis in order of salty, umami, sour, bitter and sweet stimuli. The identity of the fibers is presented along the Y-axis to allow their identification in the dendrogram. The heights of the bars depict the response during the first 5 sec of stimulation of the compounds listed along the X-axis. Figure 3, from the C T nerve, shows one group of fibers responding only to NaCl. These fibers did not respond to KC1. This indicates that the taste of NaCl and of KC1 are very different to the rhesus monkey, a difference, which also is very evident to humans. A second group in Figure 3, part of the Ν cluster, responded to monosodium glutamate (MSG), but not to KC1. Citric and aspartic acid stimulated the largest range of C T fibers, which also included some S fibers. We noticed the same feature in our first single fiber study in rhesus monkey . This is not surprising, since these acids also have a sweet component, in contrast to HC1, which lacks sweetness, and consequently with one exception, did not stimulate any S fibers. A small number of C T fibers responded to the bitter compounds, quinine, caffeine and denatonium. Instead Q fibers are found in and dominate the N G recordings in Figure 4, where there really is no well defined S cluster. This indicates that bitter is the dominant taste quality on the back of the tongue. The fact that the largest cluster of S fibers were found in the C T nerve supports the notion that sweet is perceived from the tip of the tongue. The final observation, which we think is important from the point of view of taste quality, is that the S fibers responded to virtually every sweet compound applied. This corroborates our conclusion that these fibers are responsible for the sweet taste quality. 33

The S fibers were then used to measure the intensity of the brazzein compounds. Figure 5 compares the results obtained from humans and monkeys, when sweetness scores and nerve responses are expressed relative to those of brazzein. As can be seen, the results are highly correlated (r=0.78, pO.001). Figure 5 shows that the same four mutants, Asp29Ala, Asp29Lys, Asp29Asn, and Glu41Lys, were scored significantly sweeter than W T brazzein by both the human panel and the monkey data. Similarly, both methods suggested that the sweetness of 8 other mutants was significantly decreased. Ten brazzein mutants were scored similar to water by both methods. The response to three brazzein mutants and both types of monellins did not differ from that of the WT. It is notable that the results of the nerve recordings stressed the differences in sweetness between the compounds more than the psychophysical method. However, most interesting in the present context is that the results of the two methods were highly correlated. We used the same approach to estimate bitterness of 8 derivatives of denatonium benzoate by recording from fibers, which cluster analyses had

Weerasinghe and DuBois; Sweetness and Sweeteners ACS Symposium Series; American Chemical Society: Washington, DC, 2008.

Downloaded by UNIV OF MICHIGAN ANN ARBOR on May 19, 2016 | http://pubs.acs.org Publication Date: March 4, 2008 | doi: 10.1021/bk-2008-0979.ch013

191

Figure 3. Overview of the response profiles of 51 CT single fibers. The stimuli were arranged along the X- axis in order of salt, sour, bitter and sweet The fibers were arranged along the Y- axis in groups: NaCl, acid, Quinine hydrochloride and sucrose best fibers. MSG denotes monosodium glutamate; GMP, guanosine 5'-monophosphate. (See color insert in this chapter.)

Weerasinghe and DuBois; Sweetness and Sweeteners ACS Symposium Series; American Chemical Society: Washington, DC, 2008.

Downloaded by UNIV OF MICHIGAN ANN ARBOR on May 19, 2016 | http://pubs.acs.org Publication Date: March 4, 2008 | doi: 10.1021/bk-2008-0979.ch013

192

Figure 4. Overview of the response profiles of 33 NG single fibers. The stimuli and fibers were arranged as for the CT. (See color insert in this chapter.)

Weerasinghe and DuBois; Sweetness and Sweeteners ACS Symposium Series; American Chemical Society: Washington, DC, 2008.

Downloaded by UNIV OF MICHIGAN ANN ARBOR on May 19, 2016 | http://pubs.acs.org Publication Date: March 4, 2008 | doi: 10.1021/bk-2008-0979.ch013

193

Figure 5. Comparison between human psychophysical and monkey electrophysiological results for brazzein, brazzein mutants, monellin, single chain monellin and water. (See color insert in this chapter.)

identified as bitter responding fibers, Q fibers. In a similar manner as for the sweet compounds above, we obtained data on the taste intensity of these compounds from a human panel using the Label Magnitude Scale. A s was done for brazzein, the average response to denatonium benzoate was used as standard and assigned the value 100. The bitterness scores of the psychophysical experiments and number of impulses in electrophysiological experiments were then expressed in percent of the standard. Figure 6 shows the results. It is evident that the human assessment of intensity of bitterness and the responses in the bitter sensitive fibers of the monkey model are highly positively correlated. The correlation coefficient is 0.9 . To summarize, here we demonstrate a close positive correlation between estimates of sweet and bitter by humans and the nerve response in monkey S and Q taste fibers. This suggests that recordings from S and Q fibers can be used to assess the taste to humans of sweet and bitter compounds. We suggest that the method can replace human taste panels. The results serve also as a further verification of the relevance of the labeled line theory in sweet and bitter taste. 3 4

Weerasinghe and DuBois; Sweetness and Sweeteners ACS Symposium Series; American Chemical Society: Washington, DC, 2008.

194

.a S S

ι Downloaded by UNIV OF MICHIGAN ANN ARBOR on May 19, 2016 | http://pubs.acs.org Publication Date: March 4, 2008 | doi: 10.1021/bk-2008-0979.ch013

•S »

S

Co ft)

^ ο

*"··»» ··»»

SD .S

fc

•S

1 ι