CALHM1 Deletion in Mice Affects Glossopharyngeal Taste Responses, Food Intake, Body Weight, and Life Span

Animals

Calhm1 +/− founder mice were generated at GenOway. Calhm1 exon 1 deletion was performed by homologous recombination in 129Sv cells. The resultant embryonic stem cells were injected into blastocysts derived from C57BL/6J mice to obtain chimeric mice that possess germ line transmission of the targeted Calhm1 locus. WT (+/+) and Calhm1 KO (−/−) mice on the mixed 129Sv×C57BL/6J genetic background of F₁ obtained from littermate mice were used in this study. Loss of Calhm1 expression in Calhm1 −/− taste buds was verified by real-time polymerase chain reaction and immunohistochemistry (cf. Taruno et al. 2013). The WT and Calhm1 null mice (KO) males and females were housed individually in the same room and given the same feed (Purina 5001). These studies were conducted in accordance to institutional and national guidelines for the care and use of animals and were approved by the University of Minnesota IACUC.

Stimuli and stimulation

Our choice of compounds and concentrations was influenced by our and other studies in mice and we strived to include a variety of compounds to represent all taste qualities. Hence, we used NaCl and NH₄Cl for salty taste; ascorbic and citric acid as sour tastants; quinine hydrochloride, denatonium benzoate (DB), and sparteine as bitter compounds; SC45647, acesulfame-K, saccharin, 2 concentrations of sucrose, fructose, glucose as sweet tastants; and monosodium glutamate (MSG), monocalcium glutamate, and disodium-5′-inosinate (IMP) as umami tasting compounds (see Table 1).

The same TBP method and equipment were used as in our previous studies in mice (e.g., Finger et al. 2005). Hence, one bottle contained the tastant, the other water. The side of the cage for which the bottles were presented was switched every 24 h. Intake was measured every 24 h for at least two readings.

During taste nerve recordings, the stimuli were delivered to the tongue with an open flow system, controlled by a computer under conditions of constant flow and 33°C (cf. Hellekant and Roberts 1995; Danilova et al. 2002). A 10-s stimulus time was used and a 50-s rinse between stimuli. Generally, the order of Table 1 was followed to minimize cross-adaptation (Hellekant 1968).

Nerve recordings

The mice were anesthetized; a tracheotube was inserted and ligated caudally to the larynx. Isoflurane or halothane anesthesia was maintained at surgical levels to prevent muscle artifacts. Body temperature and temperature of the surgical table, blood oxygen and CO₂, anesthesia concentration, and heart rate (electrocardiogram) were continuously monitored.

The lingual branch of the NG nerve was exposed after mobilizing and freeing M. digastricus’ caudal belly and dissection along the hypoglossal nerve until the NG nerve was seen dorsally as it dips under hyoglossal muscle. The central end was then cut and the nerve mobilized.

Nerve impulses were recorded between a silver wire electrode and an indifferent electrode touching the walls of the wound, fed into a custom-made amplifier, monitored over a loudspeaker and an oscilloscope. The nerve impulses were processed by a smoothed absolute value circuit (Hellekant and Roberts 1995), which generates a direct current potential, the product of nerve impulse frequency and amplitude, here called the summated signal. This signal and a binary code for the tastant used were fed to an oscilloscope and computer. Stimulus code, individual nerve impulses, and summated signal were preserved on a recorder (Gould ES 1000). The response to a stimulus was calculated by subtracting the level of nerve activity preceding stimulation from the peak response during stimulation.

In each mouse, the responses to all compounds were expressed relative to the response to 0.1M NH₄Cl. The average in each animal and group was calculated, variance determined, and significance between the NG responses of KO and WT mice determined. Student’s t-test of log-transformed electrophysiological data was performed with significance level set at P < 0.05.

Measurements of food intake and body weight

Food intake and body weight in adult KO and WT mice were measured once a week for several weeks. We repeated these measurements in 3 different groups of WT and KO mice and found the differences between KO and WT mice consistent.

TBP tests

We used TBP tests to elucidate differences in ability to taste between KO and WT mice. These data complement also taste nerve data by providing insight to the hedonic value of a taste created by a compound. Figure 1 presents a summary of the results of the two-bottle tests using these compounds. The mean preference ratio was calculated for each mouse and then averaged for the WT and KO mice groups. The left staple in each pair displays averaged data from the KO mice, whereas the right staple shows the average of the WT group. Asterisks indicate statistically significant differences between the responses of the 2 groups (*P < 0.05). Error bars show standard error (SE). A value between 40 and 60 indicates no preference or rejection of the compound.

Nerve responses to sweet, sour, bitter, umami, and salty compounds

Figure 2 presents a continuous sequence of summated responses from the NG nerve to NaCl, SC47647, IMP, acesulfame-K, citric acid, QHCl, sucrose, saccharin, and MSG. The upper trace was obtained from a WT mouse and the lower trace was obtained from a KO mouse. Approximately 20 s of recordings during rinses were not included in the figure. The recordings demonstrate complete loss of NG response to all sweeteners in the KO mouse, but no significant difference in responses of the WT and KO mouse to stimulation of salts and acids. The umami compounds, IMP and MSG, elicited a smaller response and the bitter compound, QHCl, almost no response in the KO mouse.

Figure 3 presents averages of 9 consecutive recordings of NG in KO mice (dark columns) and 9 NG recordings in WT mice (light columns). All responses were normalized to the response of NH₄Cl. The columns depict peak responses to each compound with the baseline activity deducted. The columns show a significant difference between WT and KO mice for sweet and bitter stimuli, indicated with asterisks, but none for salts and acids. Two out of the 3 umami compounds gave a diminished response in KO mice. The exception is that MSG gave a significant response in the NG recordings but not in our previous recordings of the CT (Taruno et al. 2013). This will be discussed later.

Comparison between non-normalized data from KO and WT mice does not change these conclusions, features already evident in the 2 recordings of previous Figure 2.

Body weight, food intake, and survival times

We discovered that CALHM1 KO mice had leaner bodies than WT mice. We therefore began weighing the mice. We also reasoned that differences in food intake caused the difference in body weight between WT and KO mice. In Figure 4, we present measurements of food intake over 12 weeks of 8 adult WT (black columns) and 15 adult KO mice (white columns). The columns show that WT mice consumed more food.

A difference in food intake between KO and WT mice should over time increase the difference in body weight between the 2 groups, so we measured and then calculated the mean body weights of 11 adult WT and 12 adult KO mice <1-year old and 14 WT and 7 KO >1-year old. Figure 5 shows that the KO mice weighed less than the WT mice and the weight difference increased with time. In the mice >1-year old, the difference between KO and WT mice was statistically significant.

Given the association between life span and excessive body weight, we compared life span of 9 WT and 6 KO mice, which coexisted in the same room and type of cages. The KO mice lived on the average 981 days, SE 41, whereas the average life span of the WT mice was 571 days, SE 91.

In summary, the study focuses on effects of CALHM1 on the posterior tongue. We evaluated NG nerve responses in CALHM1 KO and WT mice to an array of taste solutions. The NG in KO mice showed no response to sweet, a slight response to umami, a much smaller response to bitter, but no significant difference in responses to salty and sour compounds. These data are similar to our results of CALHM1 KO in CT nerve recordings except in regard to the umami response (Taruno et al. 2013).

CALHM1 and CNS

Given the expression of CALHM1 in the brain of humans, and particularly in hippocampus, with the latter’s central role in short term memory, it can be surmised that the behavioral data of the KO mice were caused by memory deficits (Khachaturian 1994; Dreses-Werringloer et al. 2008; Lambert et al. 2008; Yu et al. 2009). This is however an unlikely explanation because our taste nerve recordings here and in (Taruno et al. 2013) clearly show that the primary defects of CALHM1 deficiency are at the periphery, at least for sweet, bitter, and umami taste sensing. Consequently, regardless of the CNS effects of deletion of CALHM1, no taste nerve impulses were conveyed to the CNS of KO mice, which therefore had no ability to react to sweet.

CALHM1 and sweet taste coding

Our previous study (Taruno et al. 2013) and the present one show that blocking transmitter release of Type II TRCs in KO mice abolishes the taste nerve response to compounds that, lacking for better description, are sweet to WT mice. The absence of impulses of taste fibers that respond to compounds described as sweet by humans brings up 2 points. One is, can these compounds be labeled as sweet to a mouse? Mice are phylogenetically distant to humans, as are their food choices. However, a sweet taste quality seems to exist in mice because Ninomiya and Imoto (1995) using the sweet taste blocker gurmarin demonstrated that sweet compounds stimulate a dedicated group of taste fibers. Further, calcium imaging of geniculate ganglion neurons demonstrates a relationship between tastants and subsets of tastes fibers, of which one subset responded to sweet (Barretto et al. 2015). It is likely that these were S fibers synapsing with Type II TRCs. There is no definite answer to the question if taste qualities exist in mice, but the sweeteners used here trigger intake in both mouse and human. Together, these data justify the conclusion of a taste quality also in mice, which, for lack of better description and in analogy with human perception, can be called sweet.

The second point is the long-lived question of how taste information is coded in nerves. Defining the relationship between TRCs and their sensory fibers is critical for understanding how compounds on the tongue are perceived in the CNS. The importance of elucidating taste coding may be prioritized if impulses in sweet taste fibers are found to determine the borderline between necessary consumption and excessive consumption. This knowledge can lead to a better understanding of factors that lead to obesity.

A brief recapitulation of the 2 theories of taste fiber coding is provided in the following. This has been an open-ended question long before the discovery of receptors for sweet/umami and bitter taste qualities. As mentioned, earlier investigators found, using laboratory species that taste fibers generally responded to stimuli representing more than one human taste quality. Thus, there was no group of fibers that responded only to a single taste quality, for example, sweet. This gave rise to the across-fiber pattern theory in which taste qualities are thought to be the result of responses across a large number of fibers (Pfaffmann 1941, 1955; Erickson 1963).

The first single fiber recordings of rhesus monkeys showed a better relationship between taste fibers and human taste qualities (Gordon et al. 1959) than earlier studies. Similar conclusions could be drawn from recordings of Japanese macaques (Sato et al. 1975; Sato et al. 1977). But the limitations of these early studies were that a compound might elicit more than one taste quality. For example, saccharin elicits, in addition to sweet taste, a bitter taste. Other sweeteners, such as acesulfame-K, stevioside, sucralose, or even carbohydrates suffered of similar limitations. The answer to this problem was the identification of a compound that elicits sweet taste without side tastes. Miraculin fulfills this requirement because it adds a sweet taste quality to sour taste (Brouwer et al. 1968; Kurihara and Beidler 1968). This effect was first identified in humans, but because we cannot record single taste fibers in humans, we searched for other species. We found that miraculin has similar effects in simian primates (Diamant et al. 1972; Hellekant et al. 1981). We therefore used rhesus monkeys to elucidate taste coding by combining results of behavioral tests with single taste fiber recordings before and after miraculin (Brouwer et al. 1983); after miraculin exposure, the same cluster of fibers responding to sweet compounds also responded to acids. Thus, the existence of taste fibers, which specifically conveyed the sweet taste quality, was demonstrated.

Later investigations involving single taste fiber studies especially in primates, using a larger number of stimuli representing human taste qualities alone (Hellekant et al. 1974, 1976; Danilova et al. 1998b, 1998c), or in combination with gymnemic acid (GA; Hellekant et al. 1996, 1997b, 1998), miraculin (Danilova et al. 1998a; Hellekant et al. 1998; Danilova and Hellekant 2006), or lactisole (Wang et al. 2009), demonstrated the existence of taste fiber clusters with high specificity to sweet, umami, sour, bitter, and salty compounds. Together, these studies indicate that a one-to-one relationship exists between taste fibers responding to sweet and T1R receptors or bitter and T2R receptors in the Type II TRCs. These data demonstrated that taste coding in animals is similar to human; in all species the labeled-line theory is supported, although the compounds inside a taste quality may differ from one species to another (Kare 1960, 1961, 1966; Danilova et al. 1999; Hellekant and Danilova 1999; Hellekant et al. 2010).

CALHM1 and bitter taste

The NG recordings showed that KO mice have an impaired but not completely absent sense of bitter taste. This seems to contradict TBP data in Figure 1, which showed essentially the same rejection in both KO and WT mice. However, it should be kept in mind that a TBP test is a method to determine the hedonic value of a compound and is less a means to determine how strong the taste is. The subject can choose between paying attention to, or neglecting, taste differences of the compounds. As a result, preference curves for bitter compounds usually suddenly drop from no discrimination to complete rejection (cf. Taruno et al. 2013). This complicates determination of thresholds for aversive compounds more than for attractive compounds. Here, this explains the near identical rejection of QHCl by both WT and KO mice in TBP tests of Figure 1, although the NG responses to QHCl in Figures 2 and 3 differed significantly between KO and WT.

CALHM1 and umami taste

The response to umami was greatly suppressed in CT recordings (Taruno et al. 2013), whereas less suppressed in NG recordings presented here. Figure 2 showed a residual response to umami in the NG recordings of KO mice. This result concurs with data from a number of studies indicating that TRCs with mGluR1 and mGluR4 receptors respond to umami compounds and exist on the posterior tongue of mice (Ninomiya et al. 2000; Shindo et al. 2008; Yasuo et al. 2008; Chaudhari et al. 2009; Kinnamon and Vandenbeuch 2009). These TRCs are not depending on the presence of CALHM1 for their function and gave therefore responses to umami in the NG recordings of KO mice.

CALHM1, food intake, and body weight

The connection between sweet taste and liking exists from birth and the powerful influence of S fibers on intake has been demonstrated in many species, including mice (e.g., Danilova et al. 2006; Taruno et al. 2013), hamsters (Frank et al. 1988; Faurion and Vayssettes-Courchay 1990; Rehnberg et al. 1990; Frank et al. 2005), monkeys (Brouwer et al. 1983; Hellekant and Danilova 1996; Danilova et al. 1998a; Danilova and Hellekant 2004, 2006), and humans (Steiner et al. 2001). The effects of miraculin and GA on S fiber response and intake have been summarized already. Similarly, bitter taste, which is the effect of increased impulses in Q fibers, causes rejection in all species, including human infants (Steiner et al. 2001). Hence, intake is affected by the ratio between S and Q fiber activity.

One example is presented in a study of the common marmoset, where the ratios of S and Q fiber responses in both CT and NG to 32 tastants were recorded and calculated. The results showed that the ratio of S/Q fiber activity was linearly related to consumption of the compounds (Danilova and Hellekant 2006).

Here, the loss of S fiber impulses lowered food palatability and consequently intake. This effect dominated over the effect of diminished Q activity, which should have stimulated food intake, because any aversive bitter taste of the food would be diminished. But loss of pleasant taste, due to the absence of S fiber input to appetite stimulating areas of CNS, appears to be the principal factor responsible for diminished intake in KO mice. In WT mice, with functioning Type II TRCs with T1R1 receptors, the food stimulated S fibers, which gave rise to hedonically positive sensations, triggering further intake in WT mice, even after the energy needs had been satisfied.

A recent study (Damak et al. 2012) in TRPM5 KO mice reports similar effects on food intake as we describe for CALHM1 here. TRPM5 is a calcium-activated cation channel necessary in taste transduction that colocalizes with a high frequency with CALHM1 in Type II TRCs. The authors reported that TRPM5-KO mice weigh less and are leaner than WT mice (Damak et al. 2012). Together, these data indicate a role of Type II TRCs and S fiber activity in food intake and body weight.

CALHM1 and life span

A number of studies in rodents (McCay et al. 1935) and primates indicate that limitation of food intake decreases morbidity and increases life span (Colman et al. 1998; Colman et al. 2009). Here, the WT mice were significantly heavier than the KO and ate more. We suggest that this led to shorter life span of WT than in KO mice. However, final conclusions must wait for the results in larger cohorts of mice.

In summary, our results add corroborative behavioral and NG data to our previous study of CALHM1 deletion on taste. In addition, it demonstrates the close relationship between sweet taste and food intake, body weight, and ultimately mortality. The connection between sweet taste and liking exists from birth, but linking sweet taste and excessive consumption has not generally yielded conclusive results. Our data indicate that the loss of S fiber impulses lowered food palatability and consequently intake. Our results suggest an important link between sweet taste, Type II TRCs with T1R receptors, S fiber activity, and excessive consumption that merits further study.