Overlapping Differential Expression

Underground Fat Loss Manual

Best Weight Loss Programs That Work

Get Instant Access

Because of the limited amount of fly material in this pilot study, we did not have sufficient RNA from each sample to perform a dye-swap experiment as is normally done in two-channel array experiments. Thus, it is impossible for us to distinguish between effects due to diet vs effects due to labeling with different dyes. Nevertheless, we do not think that there are significant effects due to differential labeling with different dyes because, in contrast to direct labeling of probes with Cy3 and Cy5-labeled nucleotides, indirect labeling of probes is more efficient and, in our experience, does not introduce as much bias.

With the above caveats, we note a number of potentially revealing gene expression differences. Using an FDR cutoff of 0.1 for all three data sets and an additional p value cutoff of 0.003 for the beef data set (to reduce the number of candidate genes), we compiled a list of array elements and associated genes that appeared to be differentially expressed relative to the controls. We used the corresponding Flybase 'gn' codes (when available) to retrieve Gene Ontology (GO) functional annotations and identify candidate genes that may warrant future study. For this, we used Drosphila GO annotations downloaded from www.geneontology.org in March, 2005; the file we obtained (gene_associa-tion.fb) is labeled "version 1.9." FlyBase ids were unavailable for a few (around 30 per experiment) array elements; we discarded these in the subsequent analysis because it is unclear what these array elements represent.

We examined the overlap between sets of differentially expressed genes between and among the different dietary data sets. The Venn diagram in Fig. 6 tallies the number of genes found to be differentially expressed under each diet

Fig. 6. Venn diagram of genes with altered expression in the three diets compared with the control diet (FDR = 0.1). The overlap between the microarray results is shown at the intersection points. Note that this presentation does not indicate when a gene is significantly over- or under-expressed, only that the expression is significantly altered versus the control microarrays.

Fig. 6. Venn diagram of genes with altered expression in the three diets compared with the control diet (FDR = 0.1). The overlap between the microarray results is shown at the intersection points. Note that this presentation does not indicate when a gene is significantly over- or under-expressed, only that the expression is significantly altered versus the control microarrays.

and combinations of diets. The intersection of all three diets contains 18 genes, including one false-positive consisting of several LacZ array element controls that were grouped under one Flybase "gn" id and which ought not to give a differential expression result. Drosophila do not have a lacZ gene, but their gut bacterial florae presumably do (based on positive X-gal staining), so it is possible that the bacteria grow to different levels when their hosts are in the different diets. We mention this clear-cut false-positive in order to emphasize that experiments such as these that involve many thousands of statistical tests almost certainly produce a number of false-positives that must be screened in subsequent experiments.

Until further work is done, we hesitate to draw conclusions regarding how the different diets affect global gene expression patterns; however, these results do suggest that DNA microarrays have the potential to implicate specific pathways as being particularly responsive to diet. Complete microarray results are available in supplementary files at http://www.transvar.org/fly_chow/analysis.

4. Discussion and Future Prospects

All of the microarray results discussed in this chapter are available at www.transvar.org/fly_chow/analysis. Software packages discussed in this chapter are available in the web site for the UAB Section on Statistical Genetics (www.soph.uab.edu/SSG), the Jackson Laboratory (http://www.jax.org/staff/ churchill/labsite/), and by contacting X.C. and A.L.

The results of these studies are, surprisingly, that triglyceride and total protein levels are significantly decreased by the beef diet in all adults, and total protein levels are significantly increased in male flies fed the soy diet. Furthermore, we found that all three experimental diets significantly decrease longevity, increase the length of time to develop from egg to adult, and alter global gene expression patterns compared with the control high-sucrose diet. Preliminary microarray analyses suggest that a total of 60 genes have significantly (FDR = 0.1) altered gene expression diets by all three experimental diets compared with the control diet. The life-shortening effect of palmatic acid confirms previous studies by Driver's laboratory (26).

The triglyceride-lowering effects of beef are consistent with recent studies in humans comparing high-fiber (HC), high-fat (HF - Atkins), and high-protein (HP—Zone) diets (85). According to the authors, "Body weight, waist circumference, triglycerides and insulin levels decreased with all three diets but, apart from insulin, the reductions were significantly greater in the HF and HP groups than in the HC group" (85). Soy isoflavones have also been reported to decrease triglyceride levels in mammalian models (47), but we did not observe this effect in Drosophila fed the high-soy diet.

We interpret that time required from egg laying to eclosion as being inversely proportional to the "quality" of food, at least as it applies to proper Drosophila larval development. If this interpretation is correct, then it suggests that the control (high sucrose) food also has the highest "quality." This is not surprising because fruit flies typically lay eggs in rotting fruit, hence their name. Rotting fruits are high in fructose and fermentation products, which like sucrose, are quickly broken down and easily digested. Likewise, we interpret the longevity results as indicating that sucrose has the highest "quality" in terms of longevity. It might be that the other diets require more energy for digestion and energy expenditure might decrease life span. We do not think that this is the case because mouse models heterozygous for the insulin receptor have an increase in energy expenditure, but also an increase in mean and maximal life span (reviewed in ref. 86). Therefore, "quality" likely means micronutrients that are essential for optimal Drosophila development.

The sucrose-yeast diet is the highest "quality" possibly because, as mentioned previously, this diet most closely resembles the natural diet of Drosophila, to which they are adapted. One could test this idea by performing longevity and microarray analyses on insects that specifically eat soy beans, for instance, or insects that primarily eat meat, such as carrion beetles. The prediction would be that the shortest larval stages and the longest life span would be observed when these insects are reared on diets that most closely resemble their natural diets.

Another possibility is that components in soy, such as the isoflavones genistein and daidzein, interfere with ecdysone steroid-hormone signaling in Drosophila. Sharpe's laboratory, which has studied the effects of environmental estrogens on mammalian hormonal signaling for several decades, published an influential paper in 2002 that showed that soy formula decreases testosterone levels in male marmoset monkeys (87). Sharpe also suggested in a review on the "estrogen hypothesis" that this could lead to testicular abnormalities and cancer (88). The marmoset result was a primary reason for the recent recommendation to discourage the use of soy-based infant formula for babies under 6 mo of age in the United Kingdom (89).

Drosophila do not have estrogens or testosterones, but the steroid hormone ecdysone is the primary molting hormone that is required for inducing the larval molts and pupation. If soy isoflavones, for instance, interfere with ecdysone signaling, this could help explain why the times from egg laying to eclosion are prolonged when larvae are on the soy diet. However, this would not explain why similar prolongation effects are seen in the beef and palmitic acid diets. Also, a decrease in ecdysone signaling in adult flies increases longevity (90-92), whereas we see a decrease in longevity when flies are fed soy, which is opposite to the predicted result.

The preliminary microarray results are informative in determining how gene expression changes when Drosophila are fed the various diets. Many of the lipid catabolism genes have altered regulation in palmitic acid, beef, and soy diets. This is not surprising because all three of these foods are high in lipids. However, most of the genes that have altered gene expression profiles in the various diets are not as easily interpreted. For example, one intriguing finding is that several olfactory pathway genes have altered expression patterns in the various diets. It would be interesting to conduct behavioral experiments to determine if mutations in these genes affect preference for the various diets.

Because of space limitations, we did not exhaust the types of microarray analyses that we could have performed. For example, the Badger laboratory recently performed microarray experiments to analyze gene expression changes in particular tissues caused by rats fed a high-soy diet (93). A meta-analysis comparing gene expression changes in rats and Drosophila fed high-soy diets would be interesting. Also, the Partridge laboratory has published a comprehensive study on microarray analyses of aging and calorically restricted Drosophila (23). In this thorough and well conducted paper, they used several analysis techniques that we did not present in this chapter. For example, in addition to survivorship curves, these investigators also showed curves representing the mortality rates vs age (23). Also, they performed a time course of gene expression changes in control or calorically restricted conditions (23). Additionally, the analyses of the data in terms of gene ontogeny (GO) codes and the number of up- or downregulated genes with a particular GO code was more extensively determined in their paper than in this chapter (23).

The studies presented in this chapter also set the stage for studying the possible long-term "metabolic imprinting effects" of various diets. "Metabolic imprinting" is the hypothesis that long-term global alterations in chromatin organization, and therefore in the expression of genes, are induced by dietary components (94,95). During metabolic imprinting, it has been hypothesized that chromatin alterations persist long after the inducing components are removed from the diet, thereby causing a stably altered metabolic state. Some evidence suggests that metabolic imprinting can persist for a significant portion of an organism's life span, and possibly even in subsequent generations, by heritable-epigenetic alteration of genes. It has been proposed that some influences on obesity may occur in utero by the mother's diet or even by the grandmother's diet (96). Further evidence that trans-generational effects of obesity might be regulated by epigenetic (e.g., DNA methy-lation) events is the recent finding that cloned mice tend to be obese yet do not pass on this obesity to their offspring (97).

We note that the interpretation of the diet experiments presented in this chapter is arguably limited by our choice of using only one particular concentration of each nutrient (sucrose, palmitic acid, soy, or beef). Life span data from flies maintained on a range of concentrations of each nutrient might be required before a meaningful inference can be developed. For example, one can examine published data from flies experiencing dietary restriction (98,99). When high-nutrient medium composed of 15% w/v yeast-sucrose medium is diluted, life span increases while reproduction decreases (the dietary restriction effect) until the food level reaches roughly 5-7% w/v concentration (98,99). Further dilution results in a decrease in life span, presumably as a result of malnutrition. Thus, there is a parabolic (concave down) relationship between sucrose-yeast concentration and life span (98,99). Therefore, one could logically conclude that two diets could be chosen such that the lower concentration is either short lived, long-lived, or has identical lifespan in comparison to higher concentration.

However, an argument supporting our approach of using different isocaloric diets at only a high concentration is that it has recently been reported that dilution of food is, in many cases, almost fully compensated for by an increase in food uptake (100). Carvalho and colleagues demonstrated that dietary restriction elicits robust compensatory changes in food consumption (100). Therefore, feeding behavior and nutritional composition act concertedly to determine fly life span. We argue that feeding isocaloric diets of different compositions, as we describe in this chapter, might be a better controlled experiment than dietary restriction by dilution favored by the majority of the longevity scientific community. Although both dietary restriction and dietary composition studies assume equal volumes of food intake, evidently an incorrect assumption in the dietary restriction studies (100), it is possible that this assumption is valid with the isocaloric diets described in this chapter. We advocate that whatever approach is used, proper studies are needed to control for the volume of food intake (100).

Despite the obvious benefits illustrated previously of using Drosophila as a model to combine nutrigenomic and longevity studies, one might still have valid arguments against conducting these studies in Drosophila. One could raise the point, for instance, that we use in our studies a variety of "nonnatural" food sources (palmitic acid, soy, beef), and we find that they slow down development and decrease adult lifespan. Because Drosophila eat yeast, this is unsurprising because they presumably suffer some degree of malnutrition on these unnatural food sources. However, this argument misses the main point of these studies, which is ultimately a comparative metabolic study between humans and Drosophila. Drosophila in the wild eat primarily yeast on fermenting fruit, and tofu and beef are not "natural" foods for this species, but what is a "natural" food for humans? Elaine Morgan, in her controversial book The Aquatic Ape: A Theory of Human Evolution, argues that "hairless" human ancestors, like dolphins and whales, were originally aquatic animals who ate primarily fish (101). Evidence cited for this controversial hypothesis is that humans must have enough eicosanoic acid (a 20 carbon mono-unsaturated fatty acid) and other omega-3 fatty acids, primarily found in fish, for proper brain and body development (102-105). Regardless of which of the many hypotheses for human evolution are correct, tofu and beef are almost certainly "nonnatural" diets for both humans and Drosophila. The unnaturalness of many human foods, and especially extreme diets favored by many in the United States, further illustrates the need to conduct comparative nutrigenomics studies with both "natural" and "nonnatural" foods.

Nevertheless, one should be careful in applying what has been learned about Drosophila nutrigenomics to humans. Dietary requirements are certainly quite different between flies and man, and even between two insect or mammalian species. Nevertheless, we believe that our findings contribute to the field of nutrigenomics, and will help identify evolutionarily conserved mechanisms connecting diet to metabolism.

Was this article helpful?

0 0
Get The Body Of Your Dreams

Get The Body Of Your Dreams

Everybody wants to lose weight. This is one fact that is supported by the countless weight loss programs on the market along with the numerous weight loss products, ranging from snack bars, powdered juices, shakes and even slimming soaps and lotions.

Get My Free Ebook


Post a comment