Basic Characteristics of Glutamates and Umami Sensing in the Oral Cavity and Gut The Representation of Umami Taste in the Taste Cortex1,2 Edmund T. Rolls Department of Experimental Psychology, University of Oxford, Oxford OX1 3UD, England ABSTRACT To investigate the neural encoding of glutamate (umami) taste in the primate, recordings were made from taste-responsive neurons in the cortical taste areas in macaques. Most of the neurons were in the orbitofrontal cortex (secondary) taste area. First, it was shown that there is a representation of the taste of glutamate that is separate from the representation of the other prototypical tastants, sweet (glucose), salt (NaCl), bitter (quinine) and sour (HCl). Second, it was shown that single neurons that had their best responses to sodium glutamate also had good responses to glutamic acid. Third, it was shown that the responses of these neurons to the nucleotide umami tastant inosine 59-monophosphate were more correlated with their responses to monosodium glutamate than to any prototypical tastant. Fourth, concentration-response curves showed that concentrations of monosodium glutamate as low as 0.001 mol/L were just above threshold for some of these neurons. Fifth, some neurons in the orbitofrontal region which responded to monosodium glutamate and other food tastes, decreased their responses after feeding with monosodium glutamate to behavioral satiety, revealing a mechanism of satiety. In some cases, this reduction was sensory-speci®c. Sixth, it was shown in psychophysical experiments in humans that the ¯avor of umami is strongest with a combination of corresponding taste and olfactory stimuli (e.g., monosodium glutamate and garlic odor). The hypothesis is proposed that part of the way in which glutamate works as a ¯avor enhancer is by acting in combination with corresponding food odors. The appropriate associations between the odor and the glutamate taste may be learned at least in part by olfactory to taste association learning in the primate orbitofrontal cortex. J. Nutr. 130: 960S±965S, 2000. KEY WORDS: c taste cortex c orbitofrontal cortex c insular cortex c glutamate c umami c primates c nucleotide c olfaction To understand how appetite and food intake are controlled secondary taste cortex in which neurons are activated by the by the human brain, and how disorders in appetite and feeding taste of food (Baylis et al. 1994, Rolls et al. 1990, Rolls 1989, develop, the underlying neural mechanisms are being analyzed 1995 and 1997). These orbitofrontal taste neurons can be in primates (Rolls 1994 and 1999). A reason for performing tuned quite ®nely to gustatory stimuli (Rolls et al. 1990). these experiments with primates is that the primate taste Moreover, their activity is related to food reward, in that those system may be organized anatomically in a manner different that respond to the taste of food do so only if the monkey is from that of nonprimates (Beckstead et al. 1980, Norgren and hungry (Rolls et al. 1989). These neurons show effects of Leonard 1973, Norgren 1984, Rolls 1989 and 1995). For sensory-speci®c satiety, an important mechanism in the con- example, unlike rodents, macaques have no subcortical set of trol of feeding (Rolls 1989, 1994, 1995 and 1997, Rolls et al. pathways from the brainstem; instead, there exists an obliga- 1989, Rolls and Rolls 1997). This region is implicated in the tory relay from the nucleus of the solitary tract via the taste control of feeding because it is the ®rst part of the taste system thalamus to the taste cortex. of primates in which neuronal responses to the taste of food In the orbitofrontal cortex of primates, there is a region of occur when hunger exists, but not after satiation (Rolls et al. 1989, Rolls 1989, 1995 and 1997). The orbitofrontal cortex also contains neurons with multi- 1 Presented at the International Symposium on Glutamate, October 12±14, modal representations, for example, neurons that respond to 1998 at the Clinical Center for Rare Diseases Aldo e Cele Dacco , Mario Negri olfactory and taste stimuli, or to visual and taste stimuli (Rolls Institute for Pharmacological Research, Bergamo, Italy. The symposium was sponsored jointly by the Baylor College of Medicine, the Center for Nutrition at the 1989, 1995 and 1997, Rolls and Baylis 1994). A neuronal University of Pittsburgh School of Medicine, the Monell Chemical Senses Center, representation of ¯avor appears to be formed in the orbito- the International Union of Food Science and Technology, and the Center for frontal cortex. Approximately 40% of the olfactory neurons in Human Nutrition; ®nancial support was provided by the International Glutamate Technical Committee. The proceedings of the symposium are published as a the orbitofrontal cortex have activity that depends on the supplement to The Journal of Nutrition. Editors for the symposium publication association of the olfactory input with a taste reward, in that were John D. Fernstrom, the University of Pittsburgh School of Medicine, and some categorize odors depending on whether they are associ- Silvio Garattini, the Mario Negri Institute for Pharmacological Research. 2 Supported by Medical Research Council Programme Grant 8513790 to ated with glucose or saline in an olfactory discrimination task E.T.R. and by a grant from the International Glutamate Technical Committee. (Critchley and Rolls 1996a, Rolls et al. 1996b). Moreover, 0022-3166/00 $3.00 © 2000 American Society for Nutritional Sciences. 960S PRIMATE ORBITOFRONTAL CORTEX 961S these olfactory responses may be modi®ed during the reversal taste), just as other cells were found with best responses to of this olfactory discrimination task (Rolls et al. 1996a). The glucose (sweet), sodium chloride (salty), HCl (sour) and qui- responses of some of these orbitofrontal cortex olfactory and nine HCl (bitter). Examples of single neurons tuned to gluta- visual neurons are also modulated by hunger and contribute to mate taste are shown in Figure 1. sensory-speci®c satiety (Critchley and Rolls 1996b, see also Across the population of neurons recorded by Rolls and Rolls and Rolls 1997). Baylis (1994), the responsiveness to glutamate was poorly The orbitofrontal cortex is thus a region that is involved in correlated with the responsiveness to NaCl; thus the represen- taste, olfactory and ¯avor information processing in nonhu- tation of glutamate was clearly different from that of NaCl. man primates. Using functional magnetic resonance imaging, Further, the representation of glutamate was shown to be we have also demonstrated the existence of corresponding approximately as different from each of the other four tastants taste and olfactory regions in the human orbitofrontal cortex as they are from each other, as shown by multidimensional (Francis et al. 1999, Rolls et al. 1997). scaling and cluster analysis. Moreover, it was found that glu- An important food taste that appears to be different from tamate is approximately as well represented in terms of mean that produced by sweet, salt, bitter or sour is the taste of evoked neural activity and the number of cells with best protein. At least part of this taste is captured by the Japanese responses to it as the other four stimuli, i.e., glucose, NaCl, word umami, which is a taste common to a diversity of food HCl and quinine. Baylis and Rolls (1991) concluded that in sources including ®sh, meats, mushrooms, cheese and some primate taste cortical areas, glutamate, which produces umami vegetables including tomatoes. Within these food sources, it is taste in humans, is approximately as well represented as are the the synergistic combination of glutamates and 59-nucleotides tastes produced by glucose (sweet), NaCl (salty), HCl (sour) that creates the umami taste (Ikeda 1909, Yamaguchi 1967, and quinine HCl (bitter). Yamaguchi and Kimizuka 1979). Monosodium L-glutamate 9 (MSG), GMP and inosine 5 -monophosphate (IMP) are ex- Glutamic acid amples of umami stimuli. Umami does not act by enhancing the tastes of sweetness, These studies indicated that a separate mechanism from saltiness, bitterness or sourness in foods, but instead may be a that for other tastes operates for the neurophysiological pro- ¯avor in its own right, at least in humans. For example, cessing and for the perception of umami. However, the role Yamaguchi (1967) found that the presence of MSG or IMP did played by the sodium cation in the MSG molecule remains an not lower the thresholds for the prototypical tastes (produced interesting issue; the degree to which this contributes to by sucrose, NaCl, quinine sulfate and tartaric acid), suggesting umami taste quality has not been clari®ed completely. We that umami did not improve the detection sensitivity for the therefore performed a neurophysiological investigation in four basic taste qualities. Also, the detection thresholds for which glutamic acid was used; its effects on a population of MSG were not lowered in the presence of the prototypical neurons in the orbitofrontal cortex were compared with those taste stimuli. This suggests that the receptor sites for umami of the prototypical tastants as well as monosodium glutamate substances are different from those for other prototypical stim- (Rolls et al. 1996c). It was possible to complete the testing for uli (Yamaguchi and Kimizuka 1979). (A synergistic effect was 70 taste-responsive cells. It was found that some of the cells found when IMP was added to MSG in that the detection had large responses to 0.05 mol/L glutamic acid. The cells that threshold for MSG was dramatically lowered.) Yamaguchi and responded to glutamic acid also typically responded to MSG Kimizuka (1979) tested the ªsingularityº of umami by present- and did not necessarily have large responses to 0.01 mol/L ing human subjects with 21 taste stimuli including single and HCl. (The pH of the glutamic acid was 2.1.) mixture solutions of MSG and sucrose, NaCl, tartaric acid and To test how similarly the whole population of 70 cells quinine sulfate.