Environmental exposures and autism: decoding brain transcriptional patterns

Recent evidence indicates that individuals with autism exhibit characteristic gene expression changes in the brain. Specifically, they often have increased expression of genes related to immune and microglial function and decreased expression of genes related to synaptic transmission. Recognizing such patterns, researchers at the University of North Carolina-Chapel Hill asked an intriguing question: can these distinctive changes be used to identify chemicals that might contribute to risk of Autism Spectrum Disorder (ASD)?

For this work (recently published in Nature Communications), the researchers screened 294 chemicals from the US EPA ToxCast Phase I library in mouse cortical neuron-enriched cultures. While such 2D cultures do not represent the complexity of a fully functioning brain, it should be noted that gene expression profiles of their cultures revealed that their system was highly reflective of a whole embryonic brain during mid to late gestation (a critical period of development – the disruption of which has been linked to ASD).

After treatment, they monitored gene expression and created six clusters of chemicals based on the patterns of changes observed. While all of the chemical clusters represent potentially consequential biological changes, the researchers focused in particular on cluster 2, which up-regulated expression of immune and cytoskeletal-related genes and down-regulated expression of ion channel and synaptic genes. These patterns mirror changes observed in post-mortem ASD brains. (Interestingly, the gene expression changes also correlated with patterns observed in Alzheimer’s disease and Huntington’s disease, which suggests that neurodevelopmental and neurodegenerative diseases may share common pathways and pathology.) In addition to the observed gene expression changes, the cluster 2 chemicals also led to oxidative stress and microtubule disruption – effects that are implicated in neurodevelopmental and neurodegenerative disorders.

The results of this work are as unsettling as they are groundbreaking. The chemicals in cluster 2 are EPA-approved pesticides and fungicides: famoxadone, fenamidone, fenpyroximate, fluoxastrobin, pyraclostrobin, pyridaben, rotenone, and trifloxystrobin. (Rotenone might sound familiar to you; it has already been linked to Parkinson’s disease.) Several of these chemicals can be found at high concentrations on conventionally grown food crops, such as leafy green vegetables – yet another reason to buy organic. And, given that usage of some of these chemicals seems to be increasing (see Figure 7 of their paper), exposure across the population is a concern.

Correlation does not mean causation, however, and therefore this study does not prove that these chemicals trigger autism. Furthermore, as noted above, this study was conducted in cell culture using mouse neurons and therefore is not fully representative of what would happen in an actual human brain. But, we can use these findings as an important warning and should now prioritize these chemicals for further evaluation in animal studies (and also, perhaps, develop relevant epidemiological studies to monitor population-level effects given existing exposures).

While we lack general hazard information on most of the thousands of chemicals in commerce, the absence of information about potential developmental neurotoxicity is a particular problem. This study demonstrates that evaluation of gene expression changes could provide a screening-level assessment that might help to fill some of these concerning gaps. In addition, the authors suggest that their approach could be used to help identify therapeutics that could counter these disease-related gene expression changes – perhaps the first step towards treatment for this increasingly common condition.

Assessing new methods for detecting obesogens

The Environmental Protection Agency (EPA) has invested tremendously in its new toxicity-testing program, ToxCast, which aims to use in vitro high-throughput screening (HTS) assays to evaluate the effects of thousands of chemicals and prioritize them for further in vivo testing. Yet, many questions remain regarding the reliability and relevance of these assays. For example, are they providing accurate predictions about the effects of interest? Are the assays consistent over time and between laboratories? And, ultimately, do we have enough confidence in the results to use them as the basis for decision-making?

While EPA has begun to evaluate some of their assays, a recently published article in Environmental Health Perspectives reports specifically on the performance of ToxCast assays and related tools in detecting chemicals that promote adipogenesis. Such “obesogenic” chemicals interact with pathways involving the peroxisome proliferator activated receptor (PPARγ), among others, to alter normal lipid metabolism and contribute to abnormal weight gain. (Note: the term “obesogen” was coined by Bruce Blumberg, the senior author of this paper).

For the first part of this work, the researchers evaluated ToxCast results for one specific pathway in adipogenesis. Of the top 21 chemicals that were reported to bind to PPARγ in ToxCast Phase I, only 5 were actually found to activate PPARγ in their own laboratory.

Next, they examined the predictive power of multiple ToxCast assays representing various pathways related to adipogenesis. The researchers chose eight biologically relevant targets (including PPARγ) and generated a ToxPi (Toxicological Priority Index) graphic based on assay results for the chemicals (see figure from the paper, below). Each color represents a specific target evaluated by one or more assays, and larger slices correspond to higher relative activity in those assays. In this way, they could combine the results of multiple assays for each chemical and easily compare adipogenic potential.

Screen Shot 2016-01-31 at 9.26.29 AM
Adipogenesis ToxPis. Supplemental Figure from: Janesick et al, 2016: On the Utility of ToxCast and ToxPi as Methods for Identifying New Obesogens. Environmental Health Perspectives.

How well do these ToxPis – created based on weighted results from ToxCast assays – predict actual PPARγ activation and overall adipogenic activity? The researchers found that only 2 out of 11 highest scoring ToxPi chemicals could activate PPARγ in their laboratory assays, and only 7 out of 17 top and medium scoring ToxPi chemicals were active in cell culture adipogenesis assays. In addition, 2 of the 7 chemicals that appeared negative in ToxPi actually promoted adipogenesis in culture.

EPA had previously recognized the potential for false positive and false negatives in the testing program and had begun to implement correction methods, such as z-score adjustments, in more recent ToxCast phases. Unfortunately, problems remained even after researchers considered these supposed improvements. While many false positives and false negatives were removed, the true positives were also eliminated.

These results are concerning, to say the least. Why are the ToxCast assays performing so poorly in predicting PPARγ activity and overall obesogenic potential? The researchers suggest several possible reasons, including 1) the fact that there are relatively few specific obesogenic assays that have been developed (especially compared to estrogen and androgen receptor assays), and 2) the inherent difficulties in using simple receptor binding tests to reflect the complexities of the endocrine system.

These issues must be resolved if we are to move forward with the goal of using these assays for prioritization and risk assessment. Last year, EPA announced (see here and here) that they would allow the use of a combination of ToxCast estrogen receptor assays to replace several existing tests in the Endocrine Disruptor Screening Program (EDSP). Clearly, however, other areas of the ToxCast program need additional refinement and validation before they can be used confidently for regulatory purposes.

While it is discouraging to read about these weaknesses in ToxCast, such external assessments are essential and will motivate important improvements. With more input from and collaboration with the scientific community, we can be hopeful that EPA’s ToxCast program will be able to fulfill its goal of efficiently evaluating thousands of chemicals and serving as the basis for decision making to protect public health.

A sensitive test for skin sensitization

As the European Union moves away from animal testing for cosmetics, validation of alternative methods to assess the safety and hazards of compounds in such products is vital. Individual in vitro tests can provide key information on specific parts of the mechanism of disease, yet they may not be able to represent the multiple, sequential steps (what toxicologists refer to as the “adverse outcome pathway”) that result in the ultimate disease endpoint. An additional useful tool for chemical hazard identification is in silico modeling, which uses data on the structure or properties of compounds to predict their interactions with biological systems. There are many challenges with this approach, though, and previous in silico models have demonstrated limited accuracy.

However, researchers at George Washington University have developed a new modeling platform specifically for skin sensitization, CADRE-SS, that seeks to use different information in the prediction process. By incorporating data on molecular properties, rather than only on structure, to model the behavior of compounds in a biological environment, they have made significant improvements in the predictive capacity of such in silico tools.

To develop CADRE-SS, the researchers examined each part of the skin sensitization adverse outcome pathway — skin penetration, enzymatic activation, and protein binding — and then determined the specific physical-chemical properties or energy states that would lead to progression along the pathway. Linking these key chemical parameters for each part of the pathway allowed them to develop the final CADRE-SS model representing the whole skin sensitization process.

Initial tests demonstrated that this new model is highly accurate. Furthermore, not only is the model able to predict chemicals likely to cause skin sensitization, but it is also able to categorize chemicals based on their degree of sensitization potential (extreme, moderate, or weak) according on international classification systems.

If we ever hope to obtain health and safety information on the growing number of chemicals in commerce, then we must utilize methods other than traditional rodent testing (which is costly and time-intensive). Key data will likely come from a combination of in vitro, in silico, and high-throughput alternative animal assays. Thus, by improving the methods by which chemical activity in biological systems can be predicted, these researchers have moved us one step closer towards closing the existing data gap. In the future, these tools could also be used early in the chemical design process to screen out potentially problematic chemicals at the outset and direct companies towards the development of safer products.

Cancer Risk Assessment: Are We Missing the Forest for the Trees?

In recent years, national and international environmental public health organizations (including the US Environmental Protection Agency and the World Health Organization) have begun to use the adverse outcome pathway (AOP) and/or mode of action (MOA) as unifying frameworks for chemical testing and risk assessment. While the details of these frameworks vary, their underlying ideas are similar: researchers link specific molecular changes caused by environmental chemicals with adverse outcomes at the organism level (ie: disease), and then risk assessment is conducted based on the premise that preventing the early molecular disruption will prevent the development of the end-stage adverse event.

While there are practical advantages and real logic to this mechanism-based approach, a new review article published in Carcinogenesis suggests that this strategy may be overly simplistic and could potentially hinder our ability to adequately identify chemicals that contribute to the development of cancer.

This international team of cancer biologists and environmental health scientists organized their discussion around the “Hallmarks of Cancer,” a list of acquired characteristics that commonly occur in cancer (for example: continued growth, resistance to cell death, and tissue invasion). For each key characteristic, they identified typical target sites for disruption as well as environmental chemicals that have been shown to act on those targets. The researchers focused their discussion solely on chemicals that were not already categorized as human carcinogens by the International Agency for Research on Cancer (IARC), and they took careful note of effects observed at low doses. In addition, they specifically mapped connections between different pathways to highlight cases in which alterations leading to a given cancer hallmark could also lead to another.

Their lengthy review provides an important overview of the procarcinogenic effects of numerous common chemicals, but perhaps the most significant conclusion of this work is to emphasize the pitfalls in the status quo for risk assessment. By focusing on categorizing single chemicals as ‘carcinogens,’ we neglect to acknowledge that combinations of chemicals that individually do not meet criteria to be categorized as ‘carcinogenic’ may act in synergistic ways to promote the development of cancer. Even recent efforts to evaluate the effects of chemical mixtures may be inadequate, as they mostly focus on chemicals with common cellular pathways or targets. What about the numerous compounds, as identified in this review, that act on disparate pathways and organs to contribute to a similar disease process in the body?

To address these problems, the authors propose several key principles for an improved framework for cumulative risk assessment, including consideration of the synergistic activity of:

  • chemicals that act via different pathways
  • chemicals that act on different target tissues
  • non-carcinogens that act at low doses to contribute to pro-carcinogenic processes
  • chemicals that are not structurally similar

Carcinogenesis, like many disease processes, is complicated, and identifying the numerous pathways and organs involved is – and will continue to be – an enormous scientific challenge. Slow progress can be made, nevertheless, with a shift towards testing real-world combinations of chemicals and by using the ‘Hallmarks of Cancer’ to guide relevant and appropriate research. New technologies, such as high throughput screening, computational modeling and systems biology-based analysis, can aid in this process. However, the authors stress that traditional in vivo testing still holds an important place in cancer-related research – at least until there is appropriate validation of these emerging tools.

This publication highlights that our current chemical testing and risk assessment system is overly narrow and negates the complexity with which chemicals can interact in the body. We must broaden our approach to acknowledge that distinct chemicals can act in distinct ways at distinct sites – even at low doses – to contribute synergistically to a specific disease process. Reframing our perspective is daunting, and it will emphasize our limited knowledge about the mixtures of chemicals that we are exposed to everyday. But, if we can look up to see the forest, we may begin to make our way towards safer territory.

Mother’s Day Ethics

A few weeks ago, I had the privilege of hearing Dr. Steven Gilbert, founder of the Institute of Neurotoxicology and Neurological Disorders and Toxipedia, give a talk entitled “The Ethics of Epigenetics.” Epigenetics, an increasingly important concept in environmental health, introduces the possibility that environmental exposures can have trans-generational impacts. That is, certain compounds may be able to alter the way in which genes are expressed, so the effects of these exposures can manifest in the children and grandchildren of the exposed population.

As a graduate student in toxicology, the idea of epigenetics was not new to me. However, Dr. Gilbert incorporated a critical ethical dimension to his discussion that was inspiring and moving. He suggested the obvious yet often not explicit enough implication of epigenetics: that it demands an altered framework for policy and action, since what we are being exposed to (most of the time, without our consent) may impact not only ourselves but also future generations.

Mother’s Day may be an especially appropriate time to think about the consequences of epigenetics. As I celebrate my mother, I can’t help but think about how her mother’s exposures may have impacted her life and her health. I wonder the same about myself: how will my life course and health be affected by my mother’s and my grandmother’s exposures to chemicals? And, similarly, what about the effects of my own exposure to unregulated chemicals on my future children? (I should note, though, that epigenetic changes can also be passed down on the paternal side.)

How can we end this cycle? What will it take to ensure that Mother’s Day can be a celebration of family health instead of a reminder of family exposures and disease?

The only real solution is reform of the nation’s severely outdated and ineffective chemical safety law, the Toxic Substances Control Act (TSCA). We need a TSCA reform that is robust and can gain bipartisan support – now. Further delay is unethical, putting ourselves and our children at risk.

Previously Published EDF Health Blog posts

Before starting graduate school, I worked in the Health Program at Environmental Defense Fund (EDF). During that time, I published numerous posts on the EDF Health and EDF Voices blog sites. Links to these posts can be found below (in reverse chronological order):

EDF Health Blog

A gift for mothers (and daughters, and all of us): New tools for breast cancer monitoring and prevention (May 2014)

Unnerving developments in the state of the evidence on developmental neurotoxicity (Feb 2014)

Maybe not surprising, but still upsetting: New report highlights role of election-year politics in OIRA delays (Dec 2013)

NGOs ask Senators to investigate chronic delays in OMB’s review of TSCA regulatory actions (Sept 2013)

EDF comments at EPA workshop on applying systematic review methodology to IRIS assessments (Aug 2013)

My mother is not Angelina Jolie (May 2013)

April brings showers…and a flurry of new studies on the risks of perfluorinated chemicals (April 2013)

“Toxic Clout” shines a much-needed light on the chemical industry’s undue influence over toxic chemical decisions (March 2013)

21st Century on the horizon for endocrine disruptor screening? (Jan 2013)

Variety is the spice of … accurate chemical testing (Jan 2013)

EDF Voices Blog

Why Latinos are disproportionately affected by asthma, and what we can do (April 2014)

Toxic chemicals: The unwanted gifts that keep on giving (Dec 2013)

Either change the system or risk another “Silent Spring” (Dec 2013)