The Interplay Between Environmental Chemical Exposures and Obesity: Proceedings of a Workshop (2016)

Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

4 Biological Pathways and Environmental Influences Given the evidence from epidemiological and animal studies that exposure to various environmental chemicals in utero or shortly after birth can affect metabolism, body composition, and weight later in life, the obvious question is, How? The workshop’s third session was devoted to exploring the various biological pathways and mechanisms by which environmental influences may have their effects. HIGH-THROUGHPUT SCREENING OF ENVIRONMENTAL CHEMICALS The first speaker in the session was Scott Auerbach, a molecular toxicologist with the National Toxicology Program (NTP) of the National Institute of Environmental Health Sciences (NIEHS). Tox21 and ToxCast Auerbach began by describing Toxicology in the 21st Century (Tox21) and ToxCast, two complementary and collaborative projects aimed at generating a community data resource on compounds of concern to the public health community. Both use high-throughput data screening and are essentially data-mining efforts. They are carried out by various contractors or through a collaboration between NTP and the National Center for Chemical Genomics (NCCG). The result will be a public resource. Auerbach explains that these projects did not begin as an effort to identify chemicals that caused obesity. Instead, they began because no toxicity data are available for many of the chemicals used. Indeed, he said that for the tens of thousands of chemicals in use at significant levels, there is little or no safety or toxicological information that can be used to evaluate the health risks that they pose. 45 

46 ENVIRONMENTAL CHEMICAL EXPOSURES AND OBESITY In response to concerns about this situation, NTP began a collaboration with NCCG in 2005 to screen chemicals. In 2007, the U.S. Environmental Protection Agency (EPA) started screening chemicals through its program called ToxCast. The Tox21 program was formed as a collaboration among EPA, the National Institutes of Health, and the U.S. Food and Drug Administration (FDA). Then, in 2008, Tox21 and ToxCast joined forces to form the “Tox21 community.” The Tox21 community had several goals: to identify patterns of compound-induced biological responses to characterize toxicity and disease pathways, facilitate cross-species extrapolation, and model low- dose extrapolation; to prioritize compounds for more extensive toxicological evaluation; and to develop models that could predict the biological response in humans. In Phase I, which ran from 2005 to 2010, the collaboration screened more than 3,000 compounds using a variety of in vitro assays. In Phase II, which started in 2011, ToxCast screened about 700 compounds in about 700 assays plus about 1,000 compounds in endocrine activity assays. Meanwhile, Tox21 screened the chemicals in a 10,000-compound library that was created as a part of the interagency collaboration. The 10,000-compound library includes drugs, drug-like compounds, and active pharmaceutical ingredients from FDA; ToxCast Phase I and II compounds and compounds from such programs as the Antimicrobial Registration Program and the Endocrine Disruptor Screening Program from EPA; and a wide variety of compounds supplied by NTP. The diversity of chemical structures found in the library is quite large, Auerbach said. These 10,000 compounds were screened three times at 15 concen- trations in a variety of quantitative high-throughput screening assays, such as assays for nuclear receptor activation or inhibition and assays for cellular stress response pathways. A main difference between Tox21 and ToxCast, he noted, is that Tox21 covers a lot more chemicals, but ToxCast has a lot more assays. There is an overlap between the two data sets, with certain chemicals being included in both Tox21 and ToxCast. Mining of the Tox21 Data Auerbach’s main focus in his presentation was on what is being done to mine the data from the ToxCast and Tox21 programs to find chemicals that might pose an obesogenic or diabetogenic risk. 

BIOLOGICAL PATHWAYS AND ENVIRONMENTAL INFLUENCES 47 Auerbach described the analysis that he presented as a hypothesis- generating exercise. The goal is to find chemicals that might activate certain biological pathways that may lead to obesity or diabetes and that can be further assessed in complex assay systems. They are doing this in two ways. The first approach Auerbach described employs the ToxPi ranking system. This approach starts by defining key biological processes associated with obesity or diabetes (e.g., adipocyte differentiation). Expert input and literature mining are then used to match ToxCast and Tox21 assays to the biological processes (e.g., peroxisome proliferator-activated receptor γ [PPARγ] would be associated with a biological process related to adipocyte differentiation). The selected assays are then used to build a ToxPi (i.e., a toxicity pie) for the biological process of interest, where each assay represents a slice of the ToxPi. The more potent the effect of the chemical on the assay is, the larger the slice is. The ToxPis are then ranked by their area, with the ToxPis with the larger areas being ranked higher. When the process was tested on the original 309 chemicals in ToxCast Phase I to identify chemicals involved in such biological processes as adipocyte differentiation and feeding behavior, it did indeed pull out the appropriate chemicals. For example, when ToxPi was applied to feeding behavior—which is particularly interesting, because it has a large number of neurological functions associated with it—the top eight compounds identified by the tool were all known to be associated with obesity in humans. This success offers confidence that ToxPi can effectively screen for chemicals from the much larger 10,000-compound library. The second approach being used to mine the Tox21 data is what Auerbach called the “sentinel chemical correlation.” This approach is purely data driven, he said, and does not involve input from experts. The approach is based on a particular premise: chemicals that exhibit similar biological properties across high-throughput screening assays will likely exhibit similar biological properties in vivo. Thus, for example, one can start with a known obesogen and look for chemicals in the database that have similar results by the collection of assays; such chemicals are likely candidate obesogens themselves. One of the challenges is to determine how to measure the similarity of the findings from the two assays. After all, the results of the assays are not numbers but rather are functions that map such things as the size of a response to various concentrations of a chemical. The details are somewhat technical, but these mappings can be characterized in various 

48 ENVIRONMENTAL CHEMICAL EXPOSURES AND OBESITY ways that make it easier to compare different assays. One can look at certain thresholds, for example. Once this is done, it is possible to run correlation software on the data set, looking for assays with similar responses to the same chemical. From there, one builds relationship networks that reflect how closely correlated various chemicals are. The result is a map of the relational space for the biological activity of the different chemicals, and one can use that map to pick out chemicals with biological activity similar to that of a target chemical. As an example, Auerbach showed the network for a PPARγ activator called rosiglitazone, which triggers adipocyte differentiation. The chemicals in the network that are closest to rosiglitazone are all pharmacologically related to rosiglitazone. This shows that the approach can be used to identify chemicals that are likely candidates for having particular properties. The tools are publicly available and a browser allows users to input a chemical and quickly find other chemicals in the 10,000-compound library with similar properties. Thus, a researcher can, for instance, input a chemical that is a prototype agent causing adipocyte differentiation and find other chemicals in the data set that behave very similarly to it to test those other chemicals. To make the tool even more valuable, the developers created a collection of chemical-to-biological annotations for the various chemicals in the data set. This makes it possible to do “chemical annotation enrichment analysis,” which is similar in spirit to the commonly used gene annotation enrichment analysis. It offers a way of judging the value of the chemicals that one identifies through the sentinel chemical correlation. As Auerbach described it, “I go fishing. I throw out that line and I pull in a bunch of chemicals. How do I know if I have … any truth in that bag? Am I just getting random association?” The chemical annotation enrichment analysis helps evaluate the plausibility that the prioritized chemicals have an effect. Discussion In the brief question-and-answer session following Auerbach’s presentation, Bruce Blumberg of the University of California, Irvine, said that his group had done a great deal of validation testing of the sorts of predictions that Auerbach had described and that the results had not been as good as they had hoped. The problem lies, he suggested, in the disparity between the activity and the different assays. “For example, if 

BIOLOGICAL PATHWAYS AND ENVIRONMENTAL INFLUENCES 49 you take the PPARγ assay, you add a gene assay that ToxCast uses and the original NCCG assays that ToxCast used and the NovaScreen,” he said. “There is absolutely no correlation among those three. If you now plug in the GeneBLAzer assays that NCCG is doing for Tox21, those do not correlate either.” Given that the same receptor can have such a disparate set of results, he asked, what does that imply about the quality of the results that a researcher can get from the approach that Auerbach described? Auerbach acknowledged that this is a significant challenge. “Part of the challenge that we are dealing with is the lower-potency environmental chemicals,” he said. “What I can tell you from reviewing the Tox21 data is that pharmacology works. It works very well. The positive controls run very well. When you start getting up into the higher-micromolar dosing and start looking at activity there, there is some variability.” And that variability can affect the predictions made by ToxPi, he said. Another workshop participant commented that the ToxCast and Tox21 assays are mostly for gene expression. There are no good assays to determine whether a particular chemical has obesogenic-type activity, for example—no good assays for lipid accumulation, no metabolic competence assays, and so on. This lack, the participant suggested, may be part of the reason for the problems in the results, and it makes it more difficult to identify chemicals that have some sort of obesogenic effect. Auerbach agreed and said that the next generation of Tox21 would include more complex emergent cellular phenotypes, such as lipid accumulation. In general, he said, the commenter was correct. “We are typically measuring the point of contact between chemical and biology, which is the receptor,” and the assays use things like transformed cell lines that do not have metabolic activation. “The next generation of assays that we will be working on will have those capabilities.” EFFECTS OF PERSISTENT ORGANIC POLLUTANTS ON ADIPOSE TISSUE The next speaker was Robert Barouki, a biochemist and a molecular biologist at Paris Descartes University who studies the toxicity mechanisms of environmental pollutants, with a particular focus on the dioxin receptor aryl hydrocarbon (AhR). He discussed the effects of persistent organic pollutants (POPs) on the function of adipose tissue. While scientists used to consider adipose tissue little more than a storage organ, Barouki said, in recent years researchers have learned that adipose tissue has a number of very interesting physical, metabolic, 

50 ENVIRONMENTAL CHEMICAL EXPOSURES AND OBESITY endocrine, and maybe even toxicological functions. Thus, the consequences can be serious if the functioning of adipose tissue is altered or interrupted by environmental chemicals. Barouki’s research comes at the relationship between adipose tissue and pollutants from two different angles. First, he seeks to understand how the presence of increased adipose tissue mass affects the uptake and kinetics of pollutants and the toxicity of pollutants. Second, he examines how pollutants affect adipose tissue, including endocrine disruption, metabolic disruption, and inflammation and oxidative stress. In particular, he said, his talk would focus on the effects of POPs on inflammation and inflammatory processes. Protective Function of Adipose Tissue Adipose tissue plays a number of roles in toxicology, Barouki said. One of them is a protective function, as some POPs have an affinity for adipose tissue, which holds them and protects other organs from their toxic effects. This would be helpful mostly for acute exposure, he noted. The evidence for this role of adipose tissue comes from various studies in which aquatic or terrestrial animals were exposed to dioxins or other POPs. Those animals with the largest amount of fat mass had the highest levels of protection from the exposures, which led one research to term the phenomenon “survival of the fattest” (Lassiter and Hallam, 1990). Recent evidence indicates that the same effect may be seen in humans. In particular, epidemiological studies have found that if you examine death rates among people with very high levels of POPs in their blood, people who are obese have a significantly lower risk of death than thin people. Of course, among those with low levels of POPs in their blood, the obese are at a high risk of dying, but having large amounts of fat tissue does seem to have a protective effect against the toxicity of POPs (see Figure 4-1). 

BIOLO OGICAL PATH HWAYS AND ENVIRONMEN EN NTAL INFLUEN NCES 51 FIGUR RE 4-1 Protective effect of fat mass in huumans when seerum POP levels are high. NOTE: PCB = polych hlorinated biph henyl. SOURC CES: Barouki presentation too workshop, MMarch 2, 2015. Data from Hoong et al., 2012. 2 Buut what happ pens in the lo ong run to obbese people whose adipoose tissue has stored to oxic chemicaals? They maay be protected from acuute exposu ures, but whaat happens if the chemicalls stored in thhe fat tissue aare later reeleased? Baro ouki set out to o find out. Onne approach is i to study ob bese individuuals who have undergone or who are a undergoin ng massive weight w loss, suuch as those who have hhad bariatrric surgery, soo Barouki and d his colleaguues recruited 886 patients whho were undergoing u this t surgery along a with 223 lean contrrols. They first measu ured the POP Ps in their subjects s befoore the surgeery. When tthe concen ntration of thhe chemicals in i adipose tisssue was meaasured in term ms of the amount of POPs per gram m of lipid, Baarouki and coolleagues fouund that obese o subjectts had signifficantly loweer concentraations than tthe controols. In short, the t POPs weere distributedd more diffussely in their ffat tissue. However, when w the totall amount of P POPs that an individual w was carryinng was calcullated, the obeese individualls had a signiificantly highher burden n because they y had so mucch more fat tisssue. When W the reseaarchers examiined the correelations between the amouunt of POP Ps present in serum s or adipoose tissue verssus the variouus phenotypes of the inddividuals in thhe study, theyy found that thhe higher thaat the amount of serum POPs was, th he higher that the levels of cholesterol annd triglyceriddes 

52 ENVIRONMENTAL CHEMICAL EXPOSURES AND OBESITY were. Furthermore, liver toxicity was positively correlated with higher levels of POPs in the blood. Gene expression studies showed that the levels of expression of some of the genes that are the classical targets of POPs were increased in the adipose tissue of the obese subjects. This may indicate that the POPs were having some effect on the adipose tissue, Barouki said. Barouki’s team then examined the obese subjects after their bariatric surgery. They found that the amount of POPs in the blood increased steadily over time in the year after the surgery. “These people lose 30 or 40 kilograms of weight during the few weeks and few months [after surgery],” he said. “There is an increase in POPs in serum, which suggests that when you lose weight, you probably release these pollutants in your serum.” Ideally, once the pollutants were released into the blood, they would be eliminated from the body, Barouki said, but this did not happen, at least to a large degree. At most they saw a decrease in the total burden of some pollutants of about 4 percent. What probably happens, he said, is that the POPs released into the blood are taken back up by the adipose tissue because these people still have a lot of fat, even after the weight loss following the surgery. In general, the obese subjects saw improvements in their serum lipid levels and liver parameters after the surgery, but the improvement was significantly smaller in those with higher levels of POPs in their blood. The one unexpected, even paradoxical effect that Barouki and his colleagues observed was related to sensitivity to insulin, Barouki said. Those subjects who had higher levels of POPs in their blood had greater improvement in their insulin sensitivity than those with lower levels of POPs. “It is not so easy to explain,” Barouki said. “We have an idea about the effect on gluconeogenesis [that is, production of glucose in the liver, which could be decreased by certain pollutants] that could explain part of it, but this is still a paradox, I have to say.” Barouki also noted that other researchers have also observed higher levels of POPs in the blood of people who have experienced long-term weight loss. The bottom line, he said, is that adipose tissue plays an important role in protecting against acute exposure to environmental chemicals. However, the fact that fat tissue takes up these chemicals and releases them over time—following weight loss or in other situations—means that this process can lead to toxicity in the long run. 

BIOLOGICAL PATHWAYS AND ENVIRONMENTAL INFLUENCES 53 Inflammatory Effects of Pollutants on Adipocytes Barouki then turned to the issue of how POPs affect adipocytes. The adipose tissue of obese individuals is different from that of lean individuals in a variety of ways. There are more cells, and the cells are larger. There are differences in the presence of inflammatory cells, in the presence of fibrosis, in the vascularization of the fat tissue, and so on. Barouki’s focus is on differences in inflammation. To study this, he uses a human cell line descended from an adipose tissue-derived stem cell. The cells can be differentiated, so it is possible to examine both the preadipocyte stage and the fully differentiated adipocyte. The adipocytes in the cell line accumulate fat over time as they age and grow. After the addition of dioxin or polychlorinated biphenyls (PCBs) to some of the cells, it is possible to use large-scale transcriptomic studies to see what pathways are modified in the cells in response to the pollutants. Research has shown that the pathway most affected is the inflammatory pathway, with its cytokines, chemokines, and so forth. Repeating these in vitro studies in mice, the researchers treated a group of mice with 10 micrograms per kilogram of dioxin and assayed them for cytokines. What they found was that there was an increase in the adipose tissue cytokine levels. This was not observed, however, in AhR-knockout mice, in which the gene for the AhR dioxin receptor had been removed. This implies that the response to dioxin was through the AhR pathway. Another important observation, Barouki said, was that when the adipose tissue in these dioxin-challenged mice was examined, it was possible to see macrophages accumulating in the adipose tissue. Thus, exposure to dioxin leads to an increase in the number of macrophages in the adipose tissue, indicative of an inflammatory response. In general, Barouki said, the more inflammation in the adipose tissue that there is, the more metabolic consequences that there are. It is very important to note that the pollutants can provoke this effect in an acute way, Barouki said. “This is not a chronic or obesogenic effect,” he said. “This is acute. It can provoke an increase in the inflammation of the adipose tissue and the invasion by macrophages.” Barouki also mentioned recent work by a colleague, Jerome Ruzzin, who has studied the levels of POPs in metabolically abnormal versus metabolically healthy obese individuals (Gauthier et al., 2014). Many obese individuals are actually metabolically healthy, he noted, while 

54 ENVIRONMENTAL CHEMICAL EXPOSURES AND OBESITY others are metabolically abnormal. Ruzzin found that higher POP levels made it more likely for an individual to be metabolically abnormal. In terms of inflammation, this makes sense. “If the mechanism is inflammation,” Barouki said, “then it will have more metabolic consequences.” Summing up his presentation, Barouki said that it seems likely that adipose tissue has a protective effect under acute conditions. It binds the POPs and keeps them from going to other places in the body. “It is better to have them in the adipose tissue than in your brain,” he said. But over time the adipose tissue can release the POPs into the bloodstream, potentially leading to chronic problems elsewhere in the body. On the flip side—talking about how POPs affect adipose tissue rather than how adipose tissue affects POPs—Barouki said that the data indicate that POPs result in a relatively acute effect on the inflammation of the adipose tissue and probably have some metabolic effects as well. They can also affect the programming of the individual and of obesogens. Discussion In the question-and-answer session following Barouki’s presentation, Sarah Rothenberg of the University of South Carolina asked for more details on when chemicals are released from adipocytes. What are the life stages at which they are released? Are they only released when you lose weight? Are they released during pregnancy? Are they released during old age? There has been little research on that topic, Barouki said. It is not known, for instance, exactly where the POPs are stored inside the adipocytes. He said he suspects that the release of POPs from the adipocytes is simply a matter of physical chemistry, with adipocytes constantly releasing and taking up the chemicals, and that when a person loses weight and the adipocytes shrink, they release some of the pollutants they have been holding onto. He also noted that during breast- feeding, some of the POPs go into the milk, so that the mother loses some of them then, with the child gaining them. TRANSGENERATIONAL EFFECTS OF OBESOGENS: TRIBUTYLTIN The session’s third speaker was Bruce Blumberg, a professor of developmental and cell biology and pharmaceutical sciences at the University of California, Irvine. He discussed how the effects of 

BIOLOGICAL PATHWAYS AND ENVIRONMENTAL INFLUENCES 55 obesogens can become permanent and heritable by reprogramming a set of stem cells in the body. To set the stage, Blumberg spoke in general about why people become obese. He indicated that the prevailing wisdom is simply that people consume more calories than they burn, and while that is a major part of it, he explained that is not all there is to it. In one set of experiments, for example, two sets of mice were fed the same number of calories per day, but one set could eat whenever they wanted, while the other was given their food in three pulses. The first group of mice— which could eat whenever they wanted—got fatter than the ones who ate in three pulses. Indeed, it is well recognized that there are other factors than simply the energy balance, Blumberg said. These include stress, inadequate sleep, prenatal nutrition, “thrifty” genes that have evolved to make the most of scarce calories, and exposure to various chemicals in the environment. Tributyltin It is this last factor that Blumberg and his colleagues have been studying in recent years—in particular, whether disturbances in signaling pathways caused by endocrine-disrupting chemicals play a role in adipogenesis and obesity. The specific obesogen that they have been focusing on is tributyltin, a chemical known primarily for its effect on snails. It causes imposex, which is the imposition of male sexual characteristics on female snails. Blumberg said he first became interested in tributyltin a dozen years ago when he heard that it could genetically reverse female fish into male fish. He set out to determine which hormone receptors might be targets for tributyltin, expecting a steroid receptor to be involved. Instead, he found that tributyltin activates two nuclear receptors, retinoid X receptor (RXR) and PPARγ, which are critical in adipogenesis, or the formation of fat cells. It turns out that if you expose mice to tributyltin while they are still in the womb, they grow up to be fatter than normal by about 15 percent. Strangely enough, the mice are no heavier than the control animals and actually weigh a little less. What happens is that the amount of fat increases at the expense of other types of body tissue, so that the overall weight of the animal does not increase. “What we have been trying to understand for probably almost 10 years now is how does this [tributyltin] exposure cause weight gain,” he 

56 ENVIRONMENTAL CHEMICAL EXPOSURES AND OBESITY said. “Does it change the hormonal control of appetite and satiety? Are these extrahungry mice? Do they eat more? We have not noticed any big changes.” Indeed, other studies have shown that these mice actually consume less food than the controls. So perhaps the adipocytes are doing a better job of storing lipids. Or perhaps there are more fat cells or more pre-fat cells. The latter is the favored model, Blumberg said. The reason that he and his colleagues favor this model is because of the existence of a particular type of cell called a mesenchymal stem cell, which can be induced to become an adipocyte by exposure to rosiglitazone, a PPARγ antagonist. Mesenchymal stem cells, he explained, can follow many different paths. They can make muscle, cartilage, fat, or bone. PPARγ controls the choice between becoming a fat cell or becoming a bone cell. PPARγ expression in a mesenchymal stem cell commits it to the adipocyte lineage, Blumberg said, while PPARγ knockout commits it to the bone lineage. Tributyltin’s Transgenerational Effects The first experiments that Blumberg’s group carried out examined what happened to mesenchymal stem cells when mice were exposed prenatally to tributyltin or to rosiglitazone. They found that exposure to either chemical caused a significant increase—about twofold—in the ability of those stem cells to become fat cells. This was reflected by an increase in such fat markers as fatty acid binding protein 4, leptin, and a half a dozen others, he said. At the same time, the ability of the cells to differentiate into bone was inhibited. This was indicated by the downregulation of a variety of bone markers, such as osteopontin, osteocalcin, and Runx2, by the prenatal treatment. Even when the cells were induced to become bone cells, they still expressed fat markers. “These cells are predisposed to become fat cells after this prenatal exposure to tributyltin,” Blumberg said. Perhaps the most surprising finding, though, was that some of the effects of tributyltin exposure were passed down to succeeding generations. Blumberg’s team exposed pregnant mice to tributyltin throughout their pregnancies by giving it to them in their drinking water. The doses were very low, about 5 to 50 times less than the reported no-observed-adverse- effects level. They raised the mice born to those mothers, bred them to 

BIOLOGICAL PATHWAYS AND ENVIRONMENTAL INFLUENCES 57 create a second generation, and bred those mice to create a third generation. An effect in either the first generation (F1) or the second generation (F2) is called a multigenerational effect, reflecting the fact that these animals themselves were exposed to the tributyltin: the first-generation mice as embryos inside the mothers and the second-generation mice as germ cells inside the first-generation mice. An effect in the third generation (F3) is called transgenerational because there is no such exposure. What Blumberg found was that the tributyltin had transgenerational effects. For example, perirenal and interscapular fat deposits were substantially larger in the F2 and F3 mice than in the original mice or in the F1 mice. Furthermore, all three generations of mice, but particularly the third-generation mice, had substantially higher levels of expression of fat-related genes in the mesenchymal stem cells than control mice. In other words, the genes related to the production of fat cells were much more active in the exposed mice and their descendants than in unexposed mice. Blumberg has since been exploring the mechanism by which the effects on the mesenchymal stem cells are passed from generation to generation. He said that he suspects that it is an epigenetic effect, and he has been studying the methylation of the DNA in the stem cells. In some cases, he said, genes in the stem cells are more highly methylated in exposed mice and their descendants—and, thus, the expression of the genes is decreased—while other genes are less methylated in the exposed mice and their descendants. This indicates, Blumberg said, that there are indeed epigenetic effects resulting from the original exposure of pregnant mice to tributyltin. His group is now looking very deeply into what is going on, he said, and is carrying out a full genomic analysis of the mesenchymal stem cells in the sperm from four generations of mice. “The goal,” he said, “is to link changes in transcription with changes in DNA and histone methylation to identify epimutations responsible for this transgenerational inheritance. As you probably know, the existence of DNA methylation changes that persist from generation to generation is quite controversial. There are many people who simply do not believe that. We would like to know if it is true or if we are seeing perhaps the effects of a long noncoding RNA or histone methylation that is being transmitted. We really do not know the answer.” 

58 ENVIRONMENTAL CHEMICAL EXPOSURES AND OBESITY Recently, Blumberg’s postdoctoral student, Raquel Chamorro- Garcia, discovered a surprising effect in fourth-generation (F4) mice in the tributyltin experiment. The mice were raised on a normal chow diet, which is about 13 percent fat by calories, for the first 19 weeks of life. They were then switched to a breeder chow, which has 21 percent fat by calories, for 6 weeks, and then they were switched back to the normal chow. The body weight of these mice did nothing surprising, but the effects of the slightly higher-fat diet on the fat mass in the male F4 mice were striking. Immediately after the diet was switched, the male descendants of the tributyltin-exposed mice got fatter, and they continued to get fatter than the controls for the entire 6 weeks, and then they stayed fatter after the switch back to the normal diet. A much higher percentage of the body weight of fourth-generation male mice was fat tissue and a much lower percentage was lean tissue. The effect was much less striking in females. There was a small effect, but it was just barely statistically significant. Mechanisms Blumberg and his team also looked for genes whose expression was affected by tributyltin, which should offer an indication of which genes are responsible for the effects that they were seeing. Many of them were involved in lipid storage and lipid transport, which was to be expected. What is important to keep in mind, Blumberg said, is that there are actually two processes at work. One is the commitment of the mesen- chymal cells to the adipogenic pathway, that is, to becoming fat cells. The second is the actual differentiation of the stem cells into fat cells. In terms of the genes involved, those are two separate pathways. What Blumberg’s experiments have shown is that multiple generations of mice exposed to tributyltin all have undifferentiated mesenchymal stem cells with adipogenic gene expression profiles, meaning that they are poised to become fat cells. Other work has shown that rosiglitazone does very little to cause such differentiation: the highest dose of rosiglitazone can just barely push these cells down that pathway. On the other hand, tributyltin is very effective in inducing these cells toward the adipogenic pathway. Indeed, 50 nanomolar tributyltin is vastly more potent than 500 nanomolar rosiglitazone. In terms of the actions of individual genes, PPARγ does not push mesenchymal stem cells to commit to becoming fat cells. It can differen- 

BIOLOGICAL PATHWAYS AND ENVIRONMENTAL INFLUENCES 59 tiate cells that are already committed, but it cannot drive the process itself. Other genes, however, are induced by tributyltin and are involved in inducing these cells to commit to the adipogenic lineage. Are Environmental Chemicals Making Us Fat? Given everything that is now known about the effects of tributyltin, could it and other organotins be contributing to today’s obesity epidemic? It is known, Blumberg said, that adult exposure to tributyltin rapidly induces adipogenic genes. It is also known that drugs that activate PPARγ increase obesity. Furthermore, prenatal exposure to tributyltin permanently alters the adult phenotype, and it recruits mesenchymal stem cells to the adipocyte lineage and diverts them away from the bone lineage. However, Blumberg continued, a key question is whether humans are exposed to sufficient levels of tributyltin to lead to these effects, and that is controversial. There are few data on the subject, and the data that exist are conflicting. It is clear, however, that tributyltin does exist in the environment. He mentioned that the New York Department of Public Health has found it in house dust, so it is likely that humans are exposed to it. Polyvinyl chloride (PVC) plastic also has a certain amount of tributyltin in it. “I would like to argue that human exposure to organotins might reach levels sufficient to activate these high-affinity receptors, RXR and PPARγ,” he said. “Although I cannot say that organotins are making us fat, I think it is probably a fair statement to say that there are such things as obesogens that modify our response to calories such that we are more likely to get fat than we would be if we were not exposed.” There is a great deal that is not known about environmental exposures and obesity, Blumberg said. It is not known how many obesogens there are, although he thinks that it could be many hundreds. “We do not know what the body burdens and the population are in most of these chemicals,” he said. “We do not know what all the molecular targets are.…We do not know a lot about critical windows of exposure.…We do not know how prenatal exposure alters the adult phenotype.” It is pretty clear, Blumberg said, that diet and exercise by themselves are not sufficient to explain the obesity epidemic, particularly given the example of the obesity epidemic in 6-month-old infants studied by Rob Lustig at the University of California, San Francisco. So what else is going on? 

60 ENVIRONMENTAL CHEMICAL EXPOSURES AND OBESITY The research clearly indicates that obesogens exist and that certain chemicals make people fat or increase susceptibility to weight gain, and research has identified a number of candidates. Prenatal exposure to tributyltin and another chemical that Blumberg has tested, triflumizole, reprograms exposed animals to be fat. The existence of these obesogens implies, Blumberg said, that scientists and clinicians need to shift the paradigm from trying to treat obese adults to preventing children from becoming obese in the first place. This can be done by optimizing their nutrition, making sure that they get enough exercise and sleep, and all the standard things, but reducing their exposure to obesogens will be an important component, he said. A growing number of chemicals have been shown to have trans- generational effects. Blumberg’s lab has shown it for tributyltin, and other researchers have shown it for such chemicals as vinclozolin, certain plastics, jet fuel, and dichlorodiphenyltrichloroethane. The existence of transgenerational effects raises the stakes for the argument concerning which chemicals should be regulated and which standard should be used to determine which chemicals should be taken off the market, Blumberg said. “That is not a decision for me to make, but that is a discussion that we need to have.” Discussion In the question-and-answer session following Blumberg’s presentation, Robert Barouki asked Blumberg about the body composition of the lab animals that gained body fat without gaining weight. What was lost? Blumberg answered that it was lean mass, not water, and it was probably mostly muscle because it would be unlikely that the animals lost bone mass so quickly. Another participant asked that if mice were treated with different obesogens, would some common genes be upregulated or downregulated for every chemical while the expression of other genes would be affected differently for different individual obesogens and, if so, if that would be a way to get insights into the mechanisms. Blumberg answered that in treating with different chemicals one would expect to see some common and some distinct effects and added that the experiment would more likely be carried out in mesenchymal stem cells than animals. Al McGartland of EPA commented that until the mid-2000s, when this use was banned, tributyltin was used on the bottom of boats, and he asked if this might be one source of exposure, perhaps from people 

BIOLOGICAL PATHWAYS AND ENVIRONMENTAL INFLUENCES 61 eating shrimp or other seafood that had been exposed. It is important to determine the source of exposure, Blumberg said, but he suspects that seafood is a relatively small contributor to human exposure compared with PVC plastic. Jerry Heindel of NIEHS commented that the key element in the obesogen hypothesis is that exposure to a chemical must alter a person’s set point or sensitivity for gaining weight. Up to now, he said, no one had demonstrated a potential obesogen that changed the set point in this way, but Blumberg’s experiments with the higher-fat diet pushing tributyltin- exposed mice, even in the fourth generation, to gain weight when controls did not was just such a demonstration, Heindel said. “I think that is really a proof of principle for the obesogen hypothesis,” he said. “It is really a key piece of data.” Scott Auerbach pointed out that one of the known properties of organometallics such as tributyltin is that they tend to be toxic to mitochondria, the energy sources of the cell. One way that they exert their obesogenic effect, he speculated, could be by damaging the body’s ability to burn calories. Blumberg responded that this was a good suggestion and that he will soon be able to test such ideas by putting the F4 mice in metabolic cages and testing their metabolism. EFFECTS OF PERINATAL EXPOSURE TO BISPHENOL A ON OBESITY AND METABOLIC DISEASE LATER IN LIFE The next speaker was Beverly Rubin, an associate professor of integrative physiology and pathobiology at the Tufts University School of Medicine and the Sackler School of Graduate Biomedical Sciences. She discussed the mechanisms by which perinatal exposure to the chemical bisphenol A (BPA) can lead to obesity and metabolic disease later in life. Rubin began with some basic information about BPA. It is ubiquitous in the environment, and 93 percent of the population has detectable levels of BPA in their urine. The chemical is found in cash register receipts, in the resin lining tin cans, in dental materials, in polycarbonate plastic often used in food and beverage storage containers, and in many other products that humans come into contact with daily. BPA is also an endocrine disrupter that has been documented to have a large number of effects on humans, including reproductive effects, development of neoplasias, and disruption of brain development. The ones of most relevance to the workshop are increased adiposity, changes in glucose homeostasis, and an increased risk of diabetes and liver pathology. 

62 ENVIRONMENTAL CHEMICAL EXPOSURES AND OBESITY Some years back, Rubin said, there was some controversy regarding BPA’s effects on body weight, with some research showing that it led to weight gain, some showing that it led to weight loss, and some showing that it led to no change. At present, there are many more papers reporting weight gain in animals exposed to BPA, but there are still some conflicting reports. Human studies have reported an association between exposure to BPA and increased body weight, increased waist circumference, and increased body mass index (BMI). Studies have also found an association between urine BPA levels and obesity-associated metabolic changes, including insulin resistance and diabetes. So in carrying out their research, Rubin and her colleagues had a series of questions that they wished to answer: Does developmental exposure to BPA alter body weight and body composition and contribute to other associated components of metabolic disease? What BPA doses cause weight gain and increased fat mass? What is the critical window for BPA exposure? Are there different effects in males versus females? And does early exposure interact with the Western diet in adulthood? Part 1: Effects of Perinatal and Peripubertal Doses of BPA In the first part of the study, Rubin’s group examined the dose– response using four different doses of BPA and a control in mice. They also looked at two different exposure windows: a perinatal exposure window that went from day 8 of gestation through lactational day 16 and a perinatal plus peripubertal exposure window that took the perinatal exposure window and added an additional exposure after the animals were weaned, with the animals being given drinking water with BPA from postnatal day 20 to postnatal day 35. They looked at males and females separately, and they measured internal dose levels to understand what kinds of exposures that the animals were actually getting. Three of the BPA doses were below the tolerable daily intake level of 50 micrograms per kilogram of body weight, and one was a bit higher but still well below the lowest observed adverse effect level of 50 milligrams per kilogram of body weight. For male mice exposed perinatally to BPA, the researchers measured the percentage of body fat and that of lean tissue at 50, 90, and 130 days after birth. The control mice, which had not been exposed to BPA, had the lowest percentage of body fat and the highest percentage of lean tissue. With increasing doses of BPA—0.25, 2.5, and 25 micrograms per 

BIOLOGICAL PATHWAYS AND ENVIRONMENTAL INFLUENCES 63 kilogram of body weight—the mice had increasing percentages of body fat and decreasing percentages of lean tissue. But beyond that, the effect turned around, so that the mice given a dose of 25 micrograms per kilogram of body weight had the highest percentage of body fat and the lowest percentage of lean tissue, while mice given 250 micrograms per kilogram of body weight actually had less body fat and more lean tissue that all of the other mice except the controls. Something very similar was seen in females exposed perinatally and peripubertally. The females exposed to 2.5 micrograms per kilogram of body weight had the largest amount of fat mass, while those given doses of 250 micrograms per kilogram of body weight had noticeably less fat mass, and, indeed, there was little difference between the control animals and the mice given 250 micrograms per kilogram of body weight. Looking at just the animals given 250 micrograms per kilogram of body weight, the difference was not noticeable, Rubin commented. Early into the data collection for this part of the experiment, she said, it seemed that the animals exposed perinatally (the P group) and those exposed both perinatally and peripubertally (the P+P group) were developing in a very similar way. In both groups, the midrange exposures were leading to increased body fat. However, she said, as the animals got older, it became apparent that that second exposure peripubertally had exacerbated the adverse effects in the females but not in the males. For example, at 40 weeks the P+P females were significantly more insulin resistant than those exposed just perinatally. Furthermore, data from serum assays suggested that the P+P females had more pronounced metabolic issues than the P females. In particular, the serum triglyceride levels were significantly higher in the P+P females than in the P females. These data suggest that the peripubertal period may be another critical window for the effects of BPA exposure, at least in females. Although the effects during the peripubertal period have been poorly studied to date, Rubin said that her studies suggest that it is an important time during which BPA has an influence in females. Furthermore, she said, in a recent study by Kim and colleagues, a methylation analysis in 10- to 13-year-old girls suggested that BPA exposure during preadolescent development may affect the specific epigenomic modification of genes in pathways relevant to human health (Kim et al., 2013). Summing up the results from the first part of her study, Rubin said that the key points were that BPA leads to changes in body weight and body composition in lab animals, that these changes depend on the sex of 

64 ENVIRONMENTAL CHEMICAL EXPOSURES AND OBESITY the animal, and that although the P and the P+P groups at first appeared to be similarly affected, as the animals aged, both the body composition and the metabolic parameters of the P+P females were more affected. Furthermore, she added, internal BPA doses that were measured at the Centers for Disease Control and Prevention in collaboration with Antonio Calafat suggested that the doses of BPA used in the studies were environmentally relevant, that is, that they were within the range that has been observed in humans. Part 2: Effects of BPA Combined with a High-Fat Diet In the second part of the study, Rubin and her colleagues looked at the effects of early BPA exposure in mice combined with a high-fat diet (in which 45 percent of calories were from fat) in adulthood. They did not use the lowest or the highest dose from Part 1 and instead gave the animals either 2.5 or 25 micrograms per kilogram of body weight (and included controls that did not receive BPA). “The exposure window we chose was the perinatal window,” she said. “It was most effective for the males, probably not the best choice for the females, but we had to make a decision.” They took littermates matched for body weight and body composition at 8 weeks of age. One male and one female from each litter were on the high-fat diet, and one male and one female from the same litter were placed on the normal chow diet for the remainder of the study. For the male mice fed normal chow, the animals that were exposed to either dose of BPA were significantly heavier than the control animals. The same was true for fat mass: the exposed animals had more fat mass than the controls. All the males on the high-fat diet weighed much more and had more fat than the ones on the usual chow. The BPA-exposed animals were heavier, but the difference was not as large as that for the chow diet. “We were really disappointed when we saw [these] data,” Rubin said, “because we thought that the high-fat diet would really make them fat. But then again, when you take a look at how fat these animals already are, I am wondering if we have maxed out their capacity because they are up there. The mean weight here is 73 grams for a CD1 male. That is pretty big.” In the females, on the other hand, there did seem to be some synergy between the BPA exposure and the high-fat diet, and the females given the dose of 25 micrograms per kilogram of body weight and fed the high- 

BIOLOGICAL PATHWAYS AND ENVIRONMENTAL INFLUENCES 65 fat diet were significantly fatter than unexposed females on a high-fat diet. When her team looked at the efficiency with which the animals gained weight, they found that, for a normal chow diet, the males exposed to BPA gained more weight by eating the same amount of food. Although there was no synergy between BPA exposure and a high- fat diet in males in terms of weight gain, the researchers did see such an effect when they looked for evidence of altered glucose homeostasis. In particular, the high-fat diet appeared to exacerbate the effects of BPA on glucose homeostasis. Furthermore, perinatal exposure to BPA in the male mice resulted in increased fasting insulin levels in the chow-fed animals when measured at 13 to 15 weeks of age and at 22 weeks of age. The glucose levels that they measured were similar across the various groups, but the amount of insulin that it took to maintain these levels was a lot higher in the BPA-exposed mice than in the controls. In the animals fed a high-fat diet, the glucose levels were, again, similar across the three groups at 13 to 15 weeks of age, but the exposed animals required higher levels of insulin to maintain their glucose levels. And by 22 weeks of age, the glucose levels in the exposed animals fed a high-fat diet were no longer being maintained and they were developing severe hyperglycemia. The hyperglycemia was strikingly bad, Rubin said, and many of the animals became diabetic. The BPA-exposed animals also had increased inflammation in their adipose tissue, which was exacerbated by the high-fat diet. The livers of the exposed animals had higher lipid levels. There was also increased expression of lipogenic and adipogenic genes in the livers of the BPA- exposed animals as well as very significant increases in the levels of expression of cholesterol-synthesizing genes. All these effects of BPA on body weight, body composition, and elements of metabolic disease differed by BPA dose, sex, exposure window, and diet, Rubin said. Future Work Since she and her colleagues began this work, Rubin said, there has been a growing body of data from studies with mice and rats that corroborate their findings that early BPA exposure may act as an obesogen, may alter body composition and glucose homeostasis, and may affect liver function. But, she said, many important questions remain. 

66 ENVIRONMENTAL CHEMICAL EXPOSURES AND OBESITY Probably the most important question is what happens with humans. Epidemiological studies have found associations between levels of BPA and various parameters of metabolic disease in adults. In children and teens there is a correlation between BPA levels and body weight. A correlation between maternal BPA levels and children’s BMI at 4 years of age has been reported. And a recent report has described an association between early BPA exposure and increased leptin levels in 9- year-old boys. Still, few clear conclusions can be drawn from the currently available data. It is probably too early to understand all of the potential consequences of prenatal or perinatal exposure to BPA in humans, Rubin said. Those studies are being done now, and it will take a while to see effects, particularly if the changes are occurring in adulthood, because that will require waiting a long time between exposure and the appearance of effects. Another important question, she said, is what are the mechanisms through which BPA has these various effects in animals? BPA is known to have estrogenic actions, and other estrogen-related chemicals, such as diethylstilbestrol, are known to increase body weight when given perinatally. BPA can directly affect adipose tissue and the pancreas and liver. It has effects on various endocrine components, including the thyroid and adrenal glands, and these are very important in weight regulation. The circuits that control food intake and metabolism are developing during the period in which the laboratory animals were exposed in the studies. “We really want to take a look at this,” Rubin said, “not so much for food intake, but for metabolism.” BPA has also been observed to cause epigenetic changes. It is an open question whether BPA affects the microbiome, but the animals in the studies were exposed during periods that are very important for microbiome development. Preliminary data from a pilot study suggest that good bacteria are decreased in 5-month-old female animals, and the bacteria seen in intestinal pathologies are increased, but Rubin and colleagues would like to explore that finding further. Finally, she said, BPA seems to alter metabolic pathways. Rubin concluded by saying that she and her colleagues are working to determine exactly what kinds of changes are occurring in metabolic pathways that influence body weight. 

BIOLOGICAL PATHWAYS AND ENVIRONMENTAL INFLUENCES 67 Discussion During the short question-and-answer period that followed Rubin’s presentation, Melissa Perry of the Milken Institute School of Public Health at George Washington University asked for details about the effects of the different doses. Rubin responded that she had initially expected the 250-microgram-per-kilogram dose to give better effects than the 25-microgram-per-kilogram dose, but that was not the case at all. “It is very clear that the low doses of BPA are the effective ones here,” she said. “Once you get across a certain border, it no longer has the effect, which may explain why there is controversy in the literature. I think it really depends on strain, dose, time of exposure, and lots of different factors that can all enter into this picture.” She added that her team had tried to maintain stable levels of BPA in their animals and eventually used osmotic minipumps to accomplish that. Liza Makowski from the University of North Carolina at Chapel Hill asked about the effects of BPA on macrophages and whether that seemed to be a direct effect of the BPA on the macrophages themselves or if it was an indirect effect through adipocytes. Rubin said that based on the observations from her studies, she believes that the BPA makes the adipose tissue dysfunctional. The adipocytes in the exposed animals become much larger than those in control animals because they pack in a lot of lipids, and eventually they just give out. “I believe what we are looking at is the macrophages coming in to clean up the mess,” she said. EFFECTS OF ENVIRONMENTAL CHEMICALS ON ENERGY METABOLISM AND INSULIN SECRETION The final speaker in this session was Barbara Corkey, the Zoltan Kohn Professor of Medicine and Biochemistry at the Boston University School of Medicine. Corkey began by saying that her research is rooted in a specific problem: that there is neither a cure for obesity or diabetes nor an understanding of the molecular basis for these diseases. The majority of people in the field focus on insulin resistance as the primary pathology in the development of diabetes, she said, but she and her colleagues started to question that because it was not leading to answers. Perhaps, she said, that focus is wrong. Instead, she suggested, hyperinsulinemia rather than insulin resistance might be the problem—in other words, a defect at the level of the beta cell. And if that is the case, then it is a natural question to ask whether there are 

68 ENVIRONMENTAL CHEMICAL EXPOSURES AND OBESITY changes in our environment that affect basal insulin secretion, although, as Corkey pointed out, almost no studies have been done to examine what regulates basal insulin secretion. Is Hyperinsulinemia the Problem? In exploring whether hyperinsulinemia might underlie diabetes, Corkey began by describing data from studies of patients who undergo bariatric surgery. One of the things that the data showed is that subjects with severe obesity and type 2 diabetes had basal fasting insulin secretion levels that were nine times greater than those of lean subjects (Pories et al., 1992). To see what would happen if insulin levels were artificially raised, researchers implanted insulin pumps in rats for a 10-day period and afterward gave the rats a glucose tolerance test. It showed that the higher insulin levels were associated with impaired glucose tolerance—just the opposite of what one might expect according to the usual understanding (Juan et al., 1999). In a test with human subjects who were put on a weight loss regimen, half were also given diazoxide, which is an inhibitor of insulin secretion. The weight loss during the diet was much greater when insulin secretion was inhibited (Alemzadeh et al., 1998). Again, Corkey said, this is not what one would expect. In reality, Corkey said, whenever either insulin resistance or hypersecretion is present, so is the other, but if one wishes to find a solution to diabetes, it is important to know which comes first. So, she said, she and her colleagues began to look at a model in which beta cell hypersecretion was the problem. This led to hyperinsulinemia, which in turn led to obesity, diabetes, and insulin resistance. “That is an okay hypothesis,” she said, “because what we know is that obesity, diabetes, and elevated fat all cause hypersecretion and insulin resistance. As far as we know, they all occur together. Any one of these could be primary.” It is actually not a surprise that insulin infusion should cause insulin resistance, she said. It has been known for a long time that insulin downregulates its receptor. Inhibition of the secretion of insulin improves insulin resistance and increases weight loss, and under all of these different manipulations of insulin, which is supposed to be controlling glucose, normal glucose levels are maintained. 

BIOLOGICAL PATHWAYS AND ENVIRONMENTAL INFLUENCES 69 “That got us to think even further,” she said, “that maybe insulin resistance is beneficial.” It may be an adaptive response aimed at maintaining normal levels of glucose in the presence of high levels of insulin, and so improving insulin sensitivity, without doing something about the hypersecretion, might lead to the problem of hypoglycemia. Possible Causes of Hyperinsulinemia If hyperinsulinemia is the problem, Corkey asked, what might cause insulin secretion in the absence of stimulatory fuel? One possibility was suggested by an experiment in which she and her colleagues incubated rat islets in the presence of fatty acid and compared their response to glucose to the response of cells incubated without the fatty acids present. What they found was that the islets incubated with fatty acid ramped up their insulin response much faster so that their half-maximal response occurred at much lower levels of glucose. There occurred hypersecretion where normally none would occur, Corkey said. Work done by one of her students has shown a similar effect when incubating cells are exposed to high levels of glucose: the insulin response ramps up faster than normal. And in cells incubated with both high levels of fat and high levels of glucose, the ramp-up is even faster. Could something in the modern human diet be having a similar effect on human islet cells? Certainly, the modern diet has high fat and high sugar contents, Corkey noted, but there are many other aspects to be considered. “There are processed foods. There are thousands of new agents. The interesting thing, which I am sure everyone in this room is clearly aware of, is that almost none of these have been evaluated as [a] potential cause of metabolic disease.” To discover what environmental agents might be causing beta cell hypersecretion of insulin, which would then lead to hyperinsulinemia, obesity, diabetes, and insulin resistance, Corkey and her colleagues began to do high-throughput screening of a variety of agents. “It did not last very long because the hit rate was so high,” she said. To offer some examples, she described the first three chemicals that her screening identified to lead to hypersecretion. The first was mono- acylglycerides, which are commonly added to food products as emulsifiers and preservatives. Many dairy products contain them, for instance. They are very effective in elevating insulin secretion without any stimulatory glucose, she said. 

70 ENVIRONMENTAL CHEMICAL EXPOSURES AND OBESITY The second group they looked at was artificial sweeteners. Tests showed that sucralose, aspartame, and saccharin all increase basal insulin secretion in the absence of stimulatory glucose. Then they looked at iron. There are many reports in the literature that people with elevated levels of iron are more prone to diabetes, Corkey said, and the modern diet has more iron than historic diets. “Our red meats have more iron than they ever did before because our food animals have been modified so that they have more lean, which means they have more iron,” she said. Again, her team found that iron induced insulin stimulation and more iron stimulated more insulin. It also had an effect in the presence of glucose, causing the production of much more insulin than normal. So what are these various agents doing to stimulate insulin secretion? Corkey and her colleagues examined various known molecular steps involved in glucose-induced insulin secretion and did not find any that were triggered by these various agents. What they did find, however, was that both monoglycerides and glucose lead to a change in redox state in rat islets. The fact that glucose causes this change in redox state in islet cells was well known, but they discovered that monoglycerides had almost exactly the same effect. The apparent mechanism is that the monoglycerides are causing the generation of reactive oxygen species (ROS) in the islet cells. Experiments showed that both iron and saccharin similarly led to the production of ROS. Another experiment showed that the addition of ROS scavengers, which removed the ROS, prevented monoglycerides from triggering insulin secretion. Thus, Corkey said, it seems that the agents that they were studying trigger insulin secretion through changes in the redox state and the production of ROS in the islets. This insulin production can take place in the absence of glucose or any other of the usual fuels that signal the islets to produce insulin. Effects on Other Systems in the Body There is no reason to believe that the agents that increase the level of ROS in the islets do not have effects elsewhere in the body, Corkey noted. After all, they are transported around the body through the circulatory system and thus interact with all of the body’s organs. Therefore, she and her colleagues asked whether such agents might have effects elsewhere in the body. 

BIOLOGICAL PATHWAYS AND ENVIRONMENTAL INFLUENCES 71 In experiments with hepatocytes, she and her colleagues varied the redox state outside the cells and found that as it went to a more oxidized condition, there was an increase in ROS production inside the cells. “This implies—it does not prove—that changes in the circulating redox state can be communicated to the inside of a cell,” Corkey said. Furthermore, there is significant evidence that changes in redox state and ROS can alter function. In has been shown in hepatocytes, for instance, that decreases in the redox state increase ROS and inhibit hepatic glucose production, adipocyte lipolysis, and triglyceride synthesis. They have also shown that changes in the redox state alter function in fat cells and, as described above, beta cells. The bottom line, she said, “is that there is a circulating system that is a reflection of the metabolic state that informs all the cells in the body of what this situation is.” Although she has not yet done the experiments, she said that she expects that when she tests toxic or obesogenic agents in the same way, they will change the parameters in the bloodstream, which will then have an effect on all the various cells that are sensitive to such changes. Summing up, Corkey said that the current hypothesis that eating too much and exercising too little causes obesity has not worked very well in the sense that it has not led to successful solutions to the current obesity epidemic. “I think we forgot the most important variable,” she said, “which is involuntary control of energy metabolism.” Extreme examples of such involuntary control of metabolism include hibernating mammals, which decrease their energy expenditure fourfold but not volitionally, and migrating birds, which have a sevenfold increase in their energy expenditure. In the well-known Vermont prisoner study, lean subjects worked to increase their body weight (Salans et al., 1971). Although it was not the point of the study, an interesting finding was that the prisoners required between 6,000 and 8,000 calories per day to gain 20 percent excess weight, Corkey mentioned. The prisoners required so many more calories because they were wasting a lot of energy. On the other hand, dieters are known to decrease their energy expenditure quite dramatically so that they do not require as many calories to maintain their weight. It is possible to show the same sort of energy variation in cells. In one experiment, Corkey said, she and her colleagues were able to measure the “leak” in beta cells—the oxygen that was being consumed but that was not used to make adenosine triphosphate (ATP), the body’s main source of energy. By adding a combination of oleate and palmitate 

72 ENVIRONMENTAL CHEMICAL EXPOSURES AND OBESITY to the cells, they were able to double the leak. Similarly, the provision of excess fuel to a variety of types of cells has been shown to cause the cells to become less efficient and to waste energy. In short, it is possible to vary cells’ energy efficiency by various means. “Our current hypothesis,” Corkey said, “is that ROS can control both energy efficiency and respiration” and that it is the dysregulation of energy efficiency rather than overeating that causes obesity. “We should at least consider it.” DISCUSSION After the presentations, there was a wide-ranging discussion involving all of the panelists, the Roundtable members, and the members of the audience, both those who were physically at the workshop and those who were attending via the webcast. Barbara Corkey began the discussion by asking her fellow panelists if they had heard any data that contradicted their own data or if there seemed to be relative consistency. Beverly Rubin responded that she was struck by how all of the presenters had similarities in their data. Robert Barouki said that one problem is that the body has many targets for pollutants, including the estrogen receptors, adipocytes, the pancreas, and others. What is not clear, he said, is whether there really are so many different targets or whether there are just a few targets that are influencing various things in the body so that, depending on the experimental setting, some researchers see one thing and others see another. “In Tox21,” he said, “you have so many pathways that you are testing. And sometimes it does not fit with the biological experiment that you do afterwards. The question is, Are we really looking at all the relevant pathways? How many are still lacking?” Corkey replied that it would be useful to get input from a systems biologist. One major problem, she said, is that single signals can affect many different systems. She said that she expects that obesogens generate some general signaling molecules that affect many different types of cells. “These are not simple one target diseases,” she said. “There have been many targets identified. None of them has worked.” Xiaoyan Pang from the University of Illinois suggested that there is a good reason why a single environmental agent circulating through the body would affect different tissues or cell types differently. The various epigenomes among individuals may explain different tissue responses, Pang said. 

BIOLOGICAL PATHWAYS AND ENVIRONMENTAL INFLUENCES 73 “I don’t do epigenetics,” Corkey responded, “but it certainly makes a lot of sense to me that that is yet another factor that has to play in. It is sort of in the same category as the microbiome.” In all of these things, she said, it is important to determine whether the differences that are observed are a cause or a consequence, but researchers generally do not spend a lot of time on that. A webcast audience member asked Scott Auerbach if there were ever any pathways not picked up by the chemical enrichment analysis that he had expected would be picked up, such as pathways identified in the expert analysis. Auerbach said that there were. “One of the things about the chemical enrichment analysis or the sentinel chemical analysis is that it goes after one specific mechanism typically because you pick a prototype chemical,” he said. “I picked rosiglitazone, which is a well- known PPARγ activator, … whereas with the ToxPi approach, we take into account all the assays that the experts have recommended. These assays are diverse. You get a different set of chemicals because you are looking at a diversity of assays as opposed to using just correlation and biological activity with one chemical with one very specific or a pattern of activity.” Nik Dhurandhar of Texas Tech University asked the presenters if they had any suggestions for how to reduce the contribution of these various chemicals to obesity. Bruce Blumberg suggested prevention: “Avoid them as much as you are able.” Barouki said that because it appears that inflammation may play a role, anti-inflammatory agents might be useful, but ultimately, nothing equals prevention in terms of public health issues. Corkey said that the biggest challenge is how to communicate these findings and provoke the appropriate changes in policy. Henry Anderson of the Wisconsin Division of Public Health asked about the nature of obesity as a disease. Is it, for instance, really a collection of different diseases? Barouki said that clinicians clearly see it as different diseases. Some obese individuals, for example, do not have metabolic issues, while others, even those with a lower BMI, do have metabolic problems. Frank Biro of the University of Cincinnati mentioned a paper published in the New England Journal of Medicine that examined adiponectin levels in a group of young boys in India. Those who had higher levels of adiponectin were far less likely to develop insulin resistance and type 2 diabetes even if they attained higher BMIs. 

74 ENVIRONMENTAL CHEMICAL EXPOSURES AND OBESITY A workshop participant noted that most of the discussion about the factors leading to obesity had ignored behavioral issues. Are there environmental chemicals that have a direct effect on the brain? Auerbach commented that when he searched through the side effects of various pharmaceutical drugs to look for drugs known to cause weight gain and obesity, he found that 90 percent of such drugs were neuroactive substances. So, yes, some environmental chemicals are likely to have direct effects on the brain, but it is harder to study such effects, he said, because it is difficult to characterize behavior, he said. Beverly Rubin said that, having been trained as a neuroendocrinologist, she thinks about the potential effects of chemicals (particularly endocrine- disrupting chemicals) on the developing brain. She is convinced, she said, that BPA is having effects because the developing brain is very sensitive to it. “We have seen reports of changes in the hippocampus, changes in cortical development, and changes in neurotransmitter development,” she said, “and, actually, in the metabolic fingerprinting we have been doing on our animals, we do see at postnatal day 2 differences in neurotransmitters in the brain.” In short, she said that the brain is likely to be one of the major targets of environmental chemicals. The problem will be to determine exactly what those chemicals are doing to the brain. “These are very complex pathways that still aren’t completely worked out,” she said. “If they were, the drug companies would be doing really well at getting us a drug that works for weight loss.” Corkey asked the other presenters if anyone has looked to see if any of the obesogens act on the parts of the brain that play a role in appetite or in satiety. Jerry Heindel of NIEHS responded that BPA exposure during development increases food intake and that this has been correlated with increases in the number of appetite neurons and decreases in the number of satiety neurons. He also commented that one place where people are not looking is the hedonic pathway, the part of the brain that is involved with food cravings. “A lot of people think overweight is emotional eating,” he said. “It is like a food addiction. There [are] a lot of data on addiction and changes in dopamine receptors in the brain and all of that. I think that is a whole field that people doing obesogen research need to get up to speed on and to look for effects in those areas.” Rubin said that there are some data suggesting that early exposure to BPA does change dopamine levels in the brain. This would also fit with the reports of hyperactivity, she said, because that could also be caused by changes in the dopaminergic system. 

BIOLOGICAL PATHWAYS AND ENVIRONMENTAL INFLUENCES 75 Dhurandhar asked the speakers if any of their research had found any of the chemicals to cross the blood–brain barrier. Rubin replied that BPA, particularly during development, can easily cross the blood–brain barrier. Blumberg noted that at least some of the organotins are neurotoxins and thus must cross the blood–brain barrier. Corkey said that some of the monoglyceride-type compounds resemble 2-arachidonoylglycerol or the endocannabinoids and that those chemicals do enter the brain. Auerbach also said that small (that is, low-molecular-weight) molecules can easily pass the blood–brain barrier. The same is true of greasy molecules and amino acids. “Generally speaking,” he explained, “those are the rules of the blood–brain barrier.” Al McGartland of EPA asked what it would take to get a dose– response curve for humans, rather than lab animals, that would apply to a subpopulation in the United States. This would be important information for setting policy. Corkey responded that researchers have access to a large variety of human tissue and they could examine in detail the precise time course, dose–response, and targets within any given tissue. She believes that there is an increasing emphasis on the use of human or humanized tissues. Blumberg said that he was part of a working group that was studying the cost of endocrine disrupters. The work was focused on the European Union because it had legislation on that topic coming up. “One of the surprising and sobering things for me,” he said, “was despite the huge volume of data that we have, how little of it we could use for the exercise of trying to figure out what is the attributable fraction of disease burden due to chemicals. We need good longitudinal studies. We need a lot more biomonitoring than we have. Those two things would go a long way toward being able to link … laboratory animal studies with human outcomes.” An audience member offered some context to McGartland’s question. EPA, the member noted, must try to quantify the health benefits that a regulation is expected to produce. This means getting numbers for such things as the number of cases of type 2 diabetes prevented or the number of strokes avoided. At this point in time, that sort of quantification is very challenging. 

76 ENVIRONMENTAL CHEMICAL EXPOSURES AND OBESITY REFERENCES Alemzadeh, R., G. Langley, L. Upchurch, P. Smith, and A. E. Slonim. 1998. Beneficial effect of diazoxide in obese hyperinsulinemic adults. The Journal of Clinical Endocrinology & Metabolism 83(6):1911–1915. Gauthier, M.-S., J. R. Perusse, M.-E. Lavoie, R. Sladek, S. R. Murthy Madiraju, N. B. Ruderman, B. Coulombe, M. Prentki, and R. Rabasa-Lhoret. 2014. Increased subcutaneous adipose tissue expression of genes involved in glycerolipid-fatty acid cycling in obese insulin-resistant versus -sensitive individuals. The Journal of Clinical Endocrinology and Metabolism 99(12):E2518–E2528. Hong, N.-S., K.-S. Kim, I.-K. Lee, P. M. Lind, L. Lind, D. R. Jacobs, and D.-H. Lee. 2012. The association between obesity and mortality in the elderly differs by serum concentrations of persistent organic pollutants: A possible explanation for the obesity paradox. International Journal of Obesity 36:1170–1175. Juan, C.-C., V. S. Fang, C.-F. Kwok, J.-C. Perng, Y.-C. Chou, and L.-T. Ho. 1999. Exogenous hyperinsulinemia causes insulin resistance, hyperendothelinemia, and subsequent hypertension in rats. Metabolism 48(4):465–471. Kim, J. H., L. S. Rozek, A. S. Soliman, M. A. Sartor, A. Hablas, I. A. Seifeldin, J. A. Colacino, C. Weinhouse, M. S. Nahar, and D. C. Dolinoy. 2013. Bisphenol A-associated epigenomic changes in prepubescent girls: A cross- sectional study in Gharbiah, Egypt. Environmental Health 12:33. Lassiter, R. R., and T. G. Hallam. 1990. Survival of the fattest: Implication for acute effects of lipophilic chemicals on aquatic populations. Environmental Toxicology and Chemistry 9(5):585–595. Pories, W. J., K. G. MacDonald, Jr., E. J. Morgan, M. K. Sinha, G. L. Dohm, M. S. Swanson, H. A. Barakat, P. G. Khazanie, N. Leggett-Frazier, S. D. Long, K. F. O’Brien, and J. F. Caro. 1992. Surgical treatment of obesity and its effect on diabetes: 10-year follow-up. American Journal of Clinical Nutrition 55(2 Suppl):582S–585S. Salans, L. B., E. S. Horton, and E. A. H. Sims. 1971. Experimental obesity in man: Cellular character of the adipose tissue. The Journal of Clinical Investigation 50(5):1005–1011.