Date: June 4th, 2009

Seminar Title: "Genetic Influences on Human Ability to Taste Bitter and Fat: Implications for Obesity Risk?"

Speaker's Name and Affiliation:

• Kathleen L. Keller
• New York Obesity Research Center, St. Luke's Roosevelt Hospital
• Columbia University College of Physicians and Surgeons

Chair: Harry R. Kissileff, Ph.D.

Rapporteur: Kathleen L. Keller, Ph.D.

Attendees and their Affiliation:

Kathleen Keller	Columbia/Obesity Research Center
Harry Kissileff	Obesity Research Center
David L. Williams	Newscorp
Nina Tamayo	CU/ORC
Sehera Himani	CU/ORC
John Chen	CU/ORC
Phedra Penn	CU/ORC
Adam Shavit	GSAS
Lisa Liang	CU/ORC
Sonia Singh	IHN
Carol Maggio	ORC
Tony Sclafani	Brooklyn College
Karen Ackroff	Brooklyn College
Heewon Lee	Teacher's College
Susan Ettinger	ORC
James Greenberg	Brooklyn College
Susan Carnell	ORC/CU
Allan Geliebter	CU/ORC
Sally Ann Lederman	CU/IHN
Chris Ochner	CU/IHN
Saskia Sanderson	Mount Sinai School of Medicine

Summary: (Prepared by the Speaker/Rapporteur)

The long-term goal of this work is to identify reliable eating-related phenotypes and genetic markers of taste that when exposed to a high-fat diet, provoke hyperphagia and obesity. My primary interest in these phenotypes and genetic markers is to use them for a screening tool to identify children who may be "at risk" for obesity. Presently, the working hypothesis is that genes for the perception of basic taste, like CD36 which may play a role in fat perception and TAS2R38 which plays a role in bitter perception, also impact the ability to discriminate fat in foods. There are data that TAS2R38 primarily impacts the ability to perceive the textural components of fat (eg. creaminess, oiliness, etc), while CD36 may impact the ability to "taste" fatty acids. Our data suggest that reduced fat discriminability is a risk factor for excess fat consumption and obesity. This is because a reduced fat discriminability may result in an individual seeking either higher fat food sources or adding more discretionary fats to the diet to achieve the same level of reward that an individual with increased ability to discriminate fat would achieve at a lower fat level. Alternatively, a reduced ability to discriminate fat may also affect one's cognitive perceptions about the amount of fat that you are consuming, and make you think you are consuming less than you actually are. This, in turn, may result in an increased intake of fat in the long-run, followed by obesity.

The first part of the talk addressed new data on the relationship between PROP taster status, TAS2R38 variation, sex, and weight status in a cohort of 4-6 year-old children. Previous data from Keller and Tepper (2004) suggested that in children, PROP taster status interacts with child sex to influence obesity risk. In this study, nontaster-males had higher BMI z-scores than taster-males, while in females, the relationship was the opposite. We set out to repeat these findings in a lower income, more ethnically diverse cohort of 4-6 year-old children. Our findings revealed that again, PROP status interacts with sex to influence child BMI z-score. However, we found a 3-way interaction between TAS2R38 variation, PROP phenotype, and sex. Male nontasters who were also homozygous for the bitter insensitive allele at TAS2R38 had the highest BMI z-scores (over 2 standard deviations above the mean), but this relationship was not true for children who were homozygous for the bitter sensitive allele, or for those that were heterozygous. While there is still controversy on the relationship that PROP status plays in obesity risk in adults, these findings partially clarify previous inconsistencies in children. Despite finding differences in BMI z-score, we found no differences in ad libitum laboratory energy or macronutrient intake as a function of PROP taster status, TAS2R38 genotype, sex, or interactions between all the above.

In the second part of the talk, the relationship between CD36 variation and the ability to discriminate differences in fat content was explored. There is now good evidence from both animals and humans that there is a taste component to dietary fat (specifically fatty acids). There are several putative fatty acid taste receptors or transport proteins, including CD36 (a fatty acid translocase), fatty acid sensitive delayed rectifying channels, and G-coupled protein receptors. CD36 is a candidate because it is expressed in taste papillae and functions (in multiple cell types) to transport fatty acids across cell membranes. The objectives of this study were to 1) phenotype healthy African-American adults into two groups: fat discriminators and fat non-discriminators, 2) assess the relationship between 5 common SNPs at CD36 and the fat discriminability phenotype, and 3) assess the relationship between CD36 variation, the fat discriminability phenotype, and reported fat preferences/intake. Fat discriminability was assessed by using Italian salad dressings that ranged in fat-by-weight content from 5-55% as the test stimuli. Participants were presented with a total of 7 pairs of salad dressings and asked to taste both dressings and mark whether they were the "same" or "different." Fat discriminability scores ranged from 1-7, with higher scores indicating greater fat discriminability. The top 10% of fat discriminability scores were classified as "fat discriminators," and the bottom 10% were classified as "fat non-discriminators." Discriminators and non-discriminators were genotyped at 5 common haplotype tagging polymorphisms at CD36. After sequencing half of the sample, one CD36 SNP (rs1761667) was significantly associated with the fat discriminability phenotype. Individuals who were carriers of two As at this SNP were more likely to be fat nondiscriminators. Carriers of one or both As at this SNP also had higher waist circumferences, suggesting that variation at this SNP may be a marker of obesity risk in African-Americans.

In addition to associations between CD36 variation and the fat discriminability phenotype, we also found that this phenotype was associated with reported fat intake. For example, fat non-discriminators reported greater intake of high-fat foods, high-fat meats, and high-fat sweets. These data suggest that both CD36 and TAS2R38 variation may be associated with dietary fat intake and obesity, but follow-up experiments should be conducted to determine the exact mechanisms by which these genes function.

Discussion:

Q. How much evidence is there to suggest that people who are less discriminating of fat seek additional fat in the diet to achieve a higher level of reward?
A. There is not a great deal of evidence for this, but it is a working hypothesis at this point, and one that we are trying to get evidence for at this point. Comment: It would be easy to design an experiment using sweet taste because you could use nutritive and non-nutritive sweeteners, but it may not be the best model for this experiment. I suppose you could use nutritive and non-nutritive oils.

Q. Why would a change in the number of taste receptors also be associated with a change in the number of taste cells?
A. The number of taste papillae isn't necessarily associated with the change in the number of taste cells. We're not sure what causes this change. It is associated with the ability to taste PROP/PTC, but not under the control of the same gene.

Q. Does a non-taster always have fewer taste papillae, and can't you classify a non-taster by using number of taste papillae (instead of sensory measures of taste perception)?
A. That is not always the case. You will see non-tasters that have lots of papillae.

Q. When you say that this bimodal distribution of basic taste is rare, what do you mean?
A. The ability to taste most compounds is distributed in a Gaussian fashion. The stronger it is, the more people are able to taste it. PROP/PTC are not like that. You have separate threshold curves for tasters and non-tasters.

Q. When you say "hot" are you talking about temperature or spicy hot? Has anyone looked at temperature sensitivity?
A. I'm referring to spicy hot. I'm not sure, but I think Harry Lawless has looked at temperature sensitivity.

Q. If you have an AVI/AVI parent, can we assume that they could only have a child with a moderate PTC sensitivity at most?
A. This would be true, but genotype does not always predict phenotype. You have other factors that affect overall taste perception (eg. other genes, taste pathology, incidence of earaches, etc.)

Q. With respect to studies that show a relationship between the inability to taste PTC/PROP and obesity, what about Bartoshuk's study?
A. I didn't include that study because the effects were very small and I'm not sure they published that in a full paper. I think it might be in a review.

Q. Could you give us a little background on what you think the sex difference in the relationship between PROP status and obesity could be attributed to? What evidence is there?
A. This is from empirical evidence in our lab in children. Male-nontasters seem to have higher BMIs than female nontasters. While I don't have a complete explanation, I think it might be because in females, social and cognitive factors show stronger associations with body weight than genetic factors.

Q. In your breakdown of participants, what does the p value refer to under gender?
A. That is the frequency of girls and boys in each of the taster groups. There are no significant differences between groups.

Q. Do any of the meal items have a bitter taste, and if so, did tasters eat less of those?
A. I wasn't really interested in bitter taste components, but the string beans were the only item that had any bitter component. We didn't find differences in any of the individual foods.

Q. Did you calculate how large your groups would have to be to see differences in the meal intake?
A. We did for the follow-up study, but this first study was not really focused on the meal aspects. We also did not realize that we would get a 3-way interaction between PROP status, PTC genotype, and sex. We might have to recalculate our sample size.

Q. If you give children a choice of foods to choose from, they may be distracted and eat one food one day, but ignore it the next. Do you think you would get any better results if you used a single item?
A. In the studies we are currently doing, we are manipulating fat content, and we use meals with fewer items. But we still have children that don't eat sometimes because they are distracted. It's difficult to predict what will happen and it may not be the best measure of intake.

Q. Is the lack of correlation between genotype and phenotype due to effects from other genes (aside from TAS2R38)?
A. PROP does not bind with 100% affinity to the PTC receptor. And, taste perception is impacted by other factors as well, such as oral pathology, incidence of earaches, etc.

Q. Can you explain why the AG and GG are almost equal in proportion?
A. We would expect them to be equal in proportion because this is not a dominant or recessive gene. The allele frequencies are similar between A and G.

Q. Which allele do they need to be in order to be tasters?
A. Nontasters are typically the AAs, so the APs and PPs are typically tasters.

Q. Is that allele coding for some protein that is necessary in order for bitter taste to be perceived?
A. This is part of a haplotype. The protein folds a little differently when they have one haplotype than another. The nontaster (AVI haplotype) folds in such a way that the receptor does not bind to PTC or PROP that well.

Q. There's a difference between males and females. Does that mean that females are less likely to be non-tasters?
A. While females are in general more sensitive to many sensations, we actually did not find that in our data.

Q. If someone is a nontaster, they actually have more options in terms of what foods to each. They are less picky. So it could make sense that they would weigh more.
A. That is one thought. They are less picky and they do not have as many food restrictions.

Q. You mentioned the idea that you can taste fatty acids, but are their free fatty acids in foods?
A. Yes, there are small quantities of free fatty acids in foods, and probably enough to bind to CD36 (if it is a taste receptor, ie, Tony's comments at the end of the talk). We use Triglycerides in our experiments, and there is no evidence that they bind to CD36.

Q. Is the quantity of free fatty acid in foods related to the amount of triglyceride?
A. This is not know, but the work I'm going to be talking about is not addressing this mechanism. We attempted to simplify this behavior by looking at a measure of fat discriminabilility.

Q. Have there been any genome wide association studies to show that any of these CD36 SNPs emerge as being associated with obesity?
A. I'm not sure if there have been any genome-wide association studies, but as you will see, we find some nice relationships with waist circumference.

Q. For your salad dressings, do you do anything to mask the visual and textural cues?
A. We use red light to mask the visual cues and a thickener in the lower fat salad dressings to mask the texture differences.

Q. For the fat discriminability test, were subjects asked to state which is greater or less, or did they just respond with "same" and "different?"
A. Just same or different. We had them rate the size of the difference, but I'm not sure that everyone understood that measure.

Q. Did you use the same order for each subject?
A. Yes.

Q. Were these ingested or spit out?
A. We tried to get people to ingest them, but some of them spit them out. It was kind of gross though.

Q. How much did they ingest? How many calories?
A. I'm not sure how many calories? They were given 2-oz souffl¨¦ cups, but they only took a small spoonfulˇ­.enough to taste.

Q. Are these adults?
A. Yes, 137 African-American men, 167 AA women.

Q. I'm not clear on how you calculated the discriminability score?
A. This was the number of correct responses they got on the fat discriminability test (1-7) with higher scores meaning better fat discriminability.

Q. What happens if you run all the data in a regression and not just the extremes of fat discriminability?
A. The data are still significant (the effect of fat discriminability is still significant at least on waist circumference) but the effects are not as robust.

Q. What about Jennifer Nasser's idea that she had about being able to detect fatty acids?
A. My data are not able to address that.

Q. You did not demonstrate that they were actually "tasting" fat.
A. Yes, that is true, we only found differences in fat discriminability.

Q. Did you get them to rate texture in the samples to make sure that they could not tell the difference?
A. No. We should probably do that.

Q. Is this all the data (in the graph on genotype and waist circumference)?
A. No, only half the sample that we have genotyped.

Q. What are the effects of fat discrimination on BMI?
A. They are in the same direction, but not significant.

Q. If you gave subjects fatty acid to taste, would the taste be much stronger than triglyceride?
A. Yes, I think so, but this has not been done. Fatty acids are also bitter and aversive, so it's a different taste quality.

Q. You said there was a problem with using fatty acid in this test. What is the problem?
A. You have to deal with the fact that fatty acids oxidize very quickly. We don't have the analytical equipment in our lab to do these tests. And, more importantly, I question the realworld validity of this procedure. I am interested in something that I can use as a screening test to take into schools or community settings and assess potential risk factors for fat overconsumption. I can't do that with fatty acids.

Q. Are there small quantities of fatty acids in foods?
A. Yes, and there are actually lipases on foods that can also free fatty acids from triglycerides.

Q. Mutations in both TAS2R38 and CD36 could lead to reduced fat discrimination according to your hypothesis.
A. Yes.

Q. Is there a linkage between CD36 and PROP genes?
A. We are going to be doing that sequencing in the same population, but not necessarily linkage analysis. At this point, the genes haven't been shown to be linked.

Q. Are they on the same chromosome?
A. I'm forgetting now, but I don't think so.

Q. Have people found these different CD36 variations in mice strains? It might be a good model.
A. I'm not aware if they have, but it would be a good model.

Q. Is there any evidence that tasters (PROP) have increased cancer risk?
A. There was a paper showing that tasters had slightly higher risk of colon polyps.

Q. You implied that CD36 receptors are found in the gut. Can you elaborate?
A. Tony Sclafani: According to recent data, it's not clear that these are receptors. They might transport fatty acids from the receptor.