Archive for the ‘Epigenetics’ Category

The Lake Effect by Nancy Nichols

By the time the PCB problem was isolated in January 1976, the Illinois Environmental Protection Agency believed that Outboard Marine was delivering approximately nine to ten tons of PCBs to the harbor each day. The PCB content of the sludge at the bottom of the harbor ranged from 240,000 to 500,000 parts per million depending on when and where the sample was taken. That means that either one in two or one in four grains of sand or silt at the bottom of the harbor was not actually sand or silt, but was a PCB instead. page 43

Waukegan would take its turn on the national stage two years later, in 1984,when a U.S. Environmental Protection official, Rita Lavelle, was accused of secretly meeting with lakefront polluters in an effort to strike a cleanup deal that heavily favored industry… In the aftermath of the scandal, the full extent of Waukegan’s chemical contamination was revealed… Eventually, three separate Superfund sites, named after the 1980 federal legislation that allocated funds to clean them up, were designated in Waukegan. Two of the sites are adjacent to the lake… In addition, more than a dozen other sites form what federal and state regulators call an expanded study area, which stretches along the lakefront from one end of town to the other. These smaller sites contain the waste products from a tannery, a steel company, a paint factory, a pharmaceutical company, and a scrap yard. Together these sites contain not just PCBs, but an alphabet soup of pollutants. “Just about every chemical we know to be dangerous to human health is in one of those sites,” Says Margaret Quinn, a professor at the University of Massachusetts, Lowell, who specializes in human exposure assessment. In addition to PCBs, these chemicals include benzene and other volatile organic compounds, arsenic. lead, asbestos, polycyclic aromatic hydrocarbons (PAHs), dioxins, vinyl chloride, and ammonia. Various chemicals among these have been associated with reproductive diseases, learning and attention deficits in children, birth defects, immune system deficiencies, and some forms of cancer.

Was there a relationship between my sister’s cancer and the toxins of our childhood? My sister certainly thought so. And many other people have suspected, often correctly, that elements in their environment have had an effect on their health. Yet because of the long time it takes for a cancer to develop and because of relative mobility of our lives today, it can be challenging to establish a casual link between a disease and its origin.

pages 5 -6

“Ovaries are approximately three centimeters long by one and one-half centimeters wide by one centimeter thick,” writes Ethel Sloan in, “The Biology of Women.”… Whichever edition you consult will tell you that the ovary is about the size of an almond and that it produces the female hormone estrogen. During the monthly menstrual cycle, each ovary forces an egg through a wall of tissue and afterward repairs that rupture in a process called ovulation. “The ovary is no beauty,” writes Natalie Angier in “Woman: An Intimate Geography, “It is scarred and pitted, for each cycle of ovulation leaves behind a blemish where an egg follicle has been emptied of its contents. The older the woman, the more scarred her ovaries will be. It is this continual bursting and repairing–part and parcel of the ovarian life cycle–that makes the ovary vulnerable to cancer.

Scientists have long theorized that as cells multiply each month to repair the breach in the ovarian wall, more opportunities are created for mistakes in the DNA copying process, which in turn increases the chances of a malignant mutation. More ovulations, in other words, mean more chances for mistakes.

Risk factors for the disease therefore include never giving your ovaries a break by being pregnant or having a child. The other risk factor is having a close relative with the disease. That would be my sister, of course, and that would bring our story back home….

Doctors at this hospital and elsewhere have long speculated that there were significant environmental factors associated with ovarian cancer. The vagina provides a runway to the ovaries not simply for sperm but for many other substances as well. Significantly, women who have their tubes tied experience a lower rate of ovarian cancer than those who do not. Some have theorized that this may be because the pathways to the ovaries has been blocked, keeping outside agents at bay.

For example, some researchers have found a link between talcum powder and ovarian cancer–though several other studies have produced conflicting results. Some early forms of talcum may have contained asbestos and thus given researchers their positive findings. Indeed, at least one retrospective study found a much higher disease rate among women who used talc prior to 1960 than those who used is after–giving at least some credence to the idea that the use of asbestos-laden talc increases a woman’s risk of ovarian cancer.

My sister speculated that asbestos had contributed to her illness. A group of naturally occurring fibrous materials that are fire-resistant, asbestos has been thought to cause adverse health effects since the first century. Yet, as writer Paul Brodeur tells us in his book on asbestos, Outrageous Misconduct, its role in causing the disease asbestosis, a noncancerous condition in which the lungs scar so badly that they won’t expand and contract properly, was not well established in medical literature until the 1970s.

In the years before my sister died, when I was an editor for the Harvard Business Review, I worked on a piece written by Bill Sells, the man who had run the Johns-Manville plants in Waukegan in the early 1970s–a time when deaths from asbestosis and other asbestos-related diseases were beginning to occur in the workforce at an alarming rate. After noting that his job included the unenviable task of visiting his sick and dying employees at the local hospital, he offered this description of his first visit to the factory: “The plant lay at the back of a sprawling complex built in the 1920s. Its view of Lake Michigan was obscured by a landfill several stories high. A road wound through this mountain of asbestos-laden scrap, and as I drove through it for the first time I stopped to watch a bulldozer crush a 36-inch sewer pipe. A cloud of dust swirled around my car.” Inside the plant, he said, he found “asbestos-laden dust coating almost every visible surface.”

An EPA official charged with overseeing the cleanup of the Johns-Manville plant, Brad Bradley, has a similar recollection. Standing at the edge of the 350-acre Superfund site that overlooks Lake Michigan, Bradley recalled his first visit there in 1982. He remembers asking an asbestos expert where he thought they would find the fibers. “I think they are everywhere,” said the expert. Indeed, virtually anywhere on the site that Bradley scuffed the ground with his boot, he found the telltale fibers.

People are more likely to connect the fiber with asbestosis than with ovarian cancer. However, a thirty-year study of nearly two thousand women who worked with asbestos while manufacturing gas masks during World War II showed these women to be seven times more likely to die from ovarian cancer than a control group. My sister’s medical history seems to tell a different story, though, and the link between asbestos and ovarian cancer in general does not appear to be a strong one. The ovarian cancer specialist I saw at the clinic was quick to point out that my sister’s record indicated that her cancer was preceded by endometriosis.

The phrase “painful periods” does not begin to describe the torture that my mother and sister endured during menstruation. White and sweating, doubled over with pain, they retreated to the bed or the couch until the pain and the bleeding passed. When I recounted my mother’s experience, the ovarian cancer specialist suggests that my mother also likely suffered from endometriosis.

Endometriosis is a once rare disease that is now common. When the disease was first named and discovered in 1921 by a New York physician, there were only twenty reports of the illness in the medical literature. Today, the National Institutes of Health estimates that roughly 5.5 million women suffer from the disease in the United States, and as many as 89 million women may have it worldwide. An exact number is hard to come by, since the disease can only properly be diagnosed during surgery. Still, about one-third of women of childbearing age suffer some symptoms–including pelvic pain and infertility–and in the United States at least, the average age of onset has been declining…

Endometriosis is a complex condition, and no one is certain what causes it. Some scientists believe it is an immune system disorder. Others believe that women with endometriosis lack the ability to shed cells that have migrated and are growing where they should not be. Other scientists have focused on a genetic component of the disease since it can run in families. A woman with a sister or mother with endometriosis, for example, is three to seven times more likely to get the disease.

The mechanisms of endometriosis are not that different from those that create cancer: they involve cell proliferation, the migration of cells, and a change in their cellular nature. Endometriosis grows unchecked and invades surrounding tissues, and the body’s immune system fails to rid itself of the misplaced lesions. In the same way, the body fails to rid itself of cancerous lesions.

It is often but not always the case that the kind of cancer my sister suffered from, ovarian clear-cell adenocarcinoma, is preceded by endometriosis, and many believe that there is a relationship between the two diseases. Some scientists believe that endometriosis–in certain cases–is a kind of precancerous condition, and others believe that the two diseases spring forth in unison. Other experts theorize that the endometrial cells themselves drive the proliferation of cancer once it has started by producing their own estrogen. Each lesion is capable of increasing the local production of estrogen, so that once the disease takes hold it is capable of feeding itself.

In my sister’s case, cancerous growths arose within her endometrial lesions. Whatever the exact mechanism of disease development, women with the type of ovarian cancer that my sister suffered from have higher rates of endometriosis that the general female population. In one study, about 70 percent of the women with clear-cell ovarian cancer also had endometriosis.

Scientists have long suspected that chemicals of the type found in Waukegan–dioxins, PCBs, and polycyclic aromatic hydrocarbons (PAHs)–play a role in human endometriosis.

pages 75 – 81

Carson died in 1964, but her work and her life serve as a warning to everyone who struggles with cancer. “As we pour millions into research and invest all our hopes in vast programs to find cures for established cases of cancer,” she wrote, “we are neglecting the golden opportunity to prevent, even while we seek to cure.”

Carson’s favorite quote, from Abraham Lincoln, can be found snuggled into her almost daily letters to Freeman, where she explains what keeps her going through her treatments and on to finish her groundbreaking book. It reads: “To sin by silence when they should protest, makes cowards of men.”

page 122

Read Full Post »

How a Big Agribusiness Firm Infiltrated the EPA and Made a Mockery of Science By Kamil Ahsan for AlterNet

Expensive coverups have kept a dangerous chemical in America’s water supply.

June 5, 2014

Earlier this year, in an exposé in The New Yorker, Rachel Aviv detailed the story of Syngenta, an agribusiness firm that was sued by the community water systems of six states in a class-action lawsuit over the firm’s herbicide atrazine.

Atrazine is the second most commonly used herbicide in the US and is used on more than 50% of all corn crops. It is one of Syngenta’s most profitable chemicals with sales at over $300 million a year. Banned in the EU, atrazine remains on the market in the US despite scores of scientific publications demonstrating its role in abnormal sexual development. Almost insoluble in water, atrazine contaminates drinking water supplies at 30 times the concentration demonstrated to cause severe sexual abnormalities in animal models.

Recently unsealed court documents from the lawsuit have disclosed how Syngenta launched a multimillion-dollar campaign to disrepute and suppress scientific research, and influence the US Environmental Protection Agency to prevent a ban on atrazine.

Tyrone Hayes, a professor of Integrative Biology at UC Berkeley has demonstrated in his research that atrazine leads to health problems, reproductive issues and birth defects. Hayes is a vocal proponent of legislative action to ban the dissemination of atrazine in water supplies. The court documents showed that Syngenta specifically attacked Hayes’ work with its smear campaign.

In addition to smear campaigns, Syngenta hired a private detective agency to look into the personal backgrounds of scientists on an advisory panel at the EPA, the judge presiding over the lawsuit, and Hayes. The documents also reveal a host of third-party organizations and independent “experts” who were on Syngenta’s payroll and supplied with Syngenta’s data in order to make public statements or write op-ed pieces in support of atrazine. Often, these experts were supplied directly with material that company employees edited or wrote.

Syngenta’s Coverup

It all started in 1997 when Hayes was employed by Syngenta to study atrazine, which was under review by the EPA. Hayes’ experimental research on the developmental growth of frogs began to reveal that even at levels of atrazine as low as 0.1 parts per billion (ppb), the chemical was capable of causing males to develop as hermaphrodites. Some males developed female organs and were even capable of mating with normal males and producing eggs. As reported in top peer-reviewed journals such as PNAS and Nature, at exposure to 0.1 ppb atrazine the frogs showed extremely reduced levels of testosterone and feminized voice boxes.

As Hayes amassed data, Syngenta downplayed his results, citing problems with statistics or asking him to repeat studies, often nitpicking or questioning his credibility or scientific skills.

In 2000, Hayes resigned from the panel. He continued to speak at conferences, publicizing his ongoing research in the lab. Meanwhile, Syngenta employees began to show up at conferences to publicly besmirch his data. Sporadically, the campaign turned into threats of violence. In a Democracy Now interview with Amy Goodman, Hayes said:

“Tim Pastoor, for example, before I would give a talk, would literally threaten, whisper in my ear that he could have me lynched, or he said he would send some of his ‘good ol’ boys to show me what it’s like to be gay,’ or at one point he threatened my wife and my daughter with sexual violence.”

Shockingly, even though Syngenta settled the lawsuit for $105 million in late 2012 after eight years of litigation, it still maintains that amount of atrazine present in the water is much lower than would be required to cause damage. In an article in Forbes published a week after the New Yorker story, Jon Entine criticized Hayes and claimed that “after numerous follow up studies by the EPA and a score of scientists… evidence of endocrine related problems Hayes claimed to have identified… are nowhere to be found.”

This is a patently false assertion. A mere scientific literature search shows dozens of peer-reviewed articles showing atrazine-induced defects in animal models. A number of papers on salmon and fish find similar results to those in frog: fish exposed to atrazine showed major reproductive abnormalities in both males and females, low sperm counts and low testosterone levels in males. Similar defects have been observed in reptiles. Research in rats has demonstrated decreased fertility, effects on sperm count, increased prostate disease in males and poor mammary development. A collaborative effort of an international team of scientists confirmed these studies by demonstrating feminization of male gonads across vertebrate species.

All signs point toward the same being true for humans. Said Hayes:

“A number of epidemiological studies in humans have associated atrazine with impaired reproduction and a decline in sperm count and fertility. Another study looking at increased prostrate disease in workers who are exposed to atrazine in the production plant in St. Gabriel, Louisiana. A number of studies now show birth defects in humans exposed to atrazine: gastroschisis where the intestines are on the outside of the baby when it’s born, choanal atresia, an effect where the oral cavity and the nasal cavity close up. Most recently, there’s been work showing atrazine associating with three different types of genital abnormalities in males.”

Corruption Within the EPA

Interestingly, the scientific advisory panel to the EPA recognizes this wealth of scientific data. In a memo from the 2012 review the advisory panel repeatedly calls attention to the biased methodology employed by the EPA. In fact, the advisory panel disagreed with almost every conclusion the EPA made.

Hayes explained: “The panel was only making recommendations, they don’t make decisions and so the EPA doesn’t need to listen to them. This really undermines the role of the scientific advisory panel.”

Syngenta was closely involved with the EPA’s decision. The EPA mainly considered just one study that found inconclusive effects of atrazine. This was the sole premise for the EPA’s decision. It was based on the research of a group led by Kloas Werner. Said Hayes:

“Kloas Werner was originally on the EPA scientific advisory panel that I presented my data to. He at that time was hired by Syngenta and subsequent to being on the panel he conducted a study in collaboration with the EPA and Syngenta and reported back to the panel that he was on. The panel’s conclusion was that more work needed to be done, and then he presented back to that panel. Essentially, his previous decision helped him get the money for his study. Furthermore, they selected a strain of frogs that don’t respond even to estrogen, which was acknowledged by the advisory panel which reviewed their work.”

But Syngenta wasn’t satisfied with bad science and corruption within the EPA. As Syngenta was hiring Werner, a scientific advisory panel member who could sway the EPA review process, it also held scores of closed-door meetings with panel members. As the documents reveal, Syngenta also hired a communications consultancy, the White House Writers’ Group, to set up meetings with members of Congress and Washington bigwigs to discuss upcoming EPA reviews.

The information about Syngenta’s misdeeds has had little to no effect. The fiction that Hayes is a scientific hack continues to pervade the work of pro-Syngenta writers like Entine. These columnists, who write from corporation-apologist perspectives, bolster the fiction by glossing over critiques of the EPA and pretending like its conclusions represent uncontroversial scientific consensus.

Time and time again, these “third-party allies” of Syngenta hyperbolically talk about the “scientific method,” and suggest that science is science, regardless of the angle of the investigator (none have much to say about Werner’s estrogen-insensitive frogs). For them, it seems, there is no conceivable way Syngenta employed techniques that would furnish them with results to protect its multimillion-dollar profits.

In other words, for them, “conflict of interest” means nothing. Scientific publishing is uncompromising about this: journals require the disclosure of conflicts of interest in publications. Obviously, political and financial incentives are sufficient criteria to change scientific results because they deeply influence the way experimenters do science.

Unsurprisingly, the Kloas paper failed to declare any conflict of interest.

“How can you declare no conflict of interest when clearly the manufacturer benefits from the conclusions drawn by that paper as well as benefits from the decisions made by the EPA advisory panel?” Hayes said. “Especially when the member was both on the panel and was paid by Syngenta.”

Corporation v. Science

Syngenta frequently alleges that Hayes never made his data on atrazine publicly available, a damning indictment that makes it seem like his data could have been fabricated. Hayes said this is not the case.

“The work that I did for Syngenta, Syngenta owns all that raw data,” he said. “This includes the generated raw data, the transcribed typed data, and really everything. The EPA actually visited my lab. Members of the EPA actually were in my laboratory, they observed all of our processes and data collection. Mary Frankenberry, a statistician, actually analyzed the data herself.”

Syngenta and its supporters also rely heavily on the vitriol that Hayes hardly seems like a disinterested, objective scientist. Rich criticism from a company that hires people to obtain the scientific results it wants.

Hayes has spoken widely, set up a website AtrazineLovers.com and rapped about Syngenta’s powerful lobbying to keep atrazine on the market. There is, however, a fundamental distinction between a company lobbying to get its favored scientific result, and a scientist who vocally defends his scientific results. Hayes’ response isn’t surprising or unusual. Scientists often claim ownership over their results and will doggedly defend them at conferences.

The actions of big corporations like Syngenta, especially when dealing with highly profitable products, reveal a broader truth about the nature of corporate power. There is a dangerous trend in which corporate fiat is used to call scientific research into question and sway governmental policy. This trend puts millions of lives at risk as hazardous products avoid regulation and remain on the market.

One wonders why the burden isn’t on Syngenta for proving without a doubt that atrazine has no effects before plying the entire population with a highly dangerous chemical. Even if it wasn’t a near-certainty that atrazine causes birth defects, why wouldn’t we require regulatory bodies such as the EPA to err on the side of caution?

Today, atrazine remains legal and in the water supplies of millions of Americans, despite evidence from scores of labs outside Tyrone Hayes’ showing it to be hazardous.

“In the 15 plus years that I’ve had experience with the EPA, I don’t really have a lot of faith that we’re going to get an objective review that’s really going to focus on environmental health and public health with regards to atrazine, or any other chemical for that matter,” Hayes said.

Who can blame him?


The Expose


Read Full Post »

Scientists discover DNA body clock

Newly discovered mechanism could help researchers understand ageing process and lead to ways of slowing it down.

by Ian Sample, science correspondent

A US scientist has discovered an internal body clock based on DNA that measures the biological age of our tissues and organs.

The clock shows that while many healthy tissues age at the same rate as the body as a whole, some of them age much faster or slower. The age of diseased organs varied hugely, with some many tens of years “older” than healthy tissue in the same person, according to the clock.

Researchers say that unravelling the mechanisms behind the clock will help them understand the ageing process and hopefully lead to drugs and other interventions that slow it down.

Therapies that counteract natural ageing are attracting huge interest from scientists because they target the single most important risk factor for scores of incurable diseases that strike in old age.

“Ultimately, it would be very exciting to develop therapy interventions to reset the clock and hopefully keep us young,” said Steve Horvath, professor of genetics and biostatistics at the University of California in Los Angeles.

Horvath looked at the DNA of nearly 8,000 samples of 51 different healthy and cancerous cells and tissues. Specifically, he looked at how methylation, a natural process that chemically modifies DNA, varied with age.

Horvath found that the methylation of 353 DNA markers varied consistently with age and could be used as a biological clock. The clock ticked fastest in the years up to around age 20, then slowed down to a steadier rate. Whether the DNA changes cause ageing or are caused by ageing is an unknown that scientists are now keen to work out.

“Does this relate to something that keeps track of age, or is a consequence of age? I really don’t know,” Horvath told the Guardian. “The development of grey hair is a marker of ageing, but nobody would say it causes ageing,” he said.

The clock has already revealed some intriguing results. Tests on healthy heart tissue showed that its biological age – how worn out it appears to be – was around nine years younger than expected. Female breast tissue aged faster than the rest of the body, on average appearing two years older.

Diseased tissues also aged at different rates, with cancers speeding up the clock by an average of 36 years. Some brain cancer tissues taken from children had a biological age of more than 80 years.

“Female breast tissue, even healthy tissue, seems to be older than other tissues of the human body. That’s interesting in the light that breast cancer is the most common cancer in women. Also, age is one of the primary risk factors of cancer, so these types of results could explain why cancer of the breast is so common,” Horvath said.

Healthy tissue surrounding a breast tumour was on average 12 years older than the rest of the woman’s body, the scientist’s tests revealed.

Writing in the journal Genome Biology, Horvath showed that the biological clock was reset to zero when cells plucked from an adult were reprogrammed back to a stem-cell-like state. The process for converting adult cells into stem cells, which can grow into any tissue in the body, won the Nobel prize in 2012 for Sir John Gurdon at Cambridge University and Shinya Yamanaka at Kyoto University.

“It provides a proof of concept that one can reset the clock,” said Horvath. The scientist now wants to run tests to see how neurodegenerative and infectious diseases affect, or are affected by, the biological clock.

“These data could prove valuable in furthering our knowledge of the biological changes that are linked to the ageing process,” said Veryan Codd, who works on the effects of biological ageing in cardiovascular disease at Leicester University. “It will be important to determine whether the accelerated ageing, as described here, is associated with other age-related diseases and if it is a causal factor in, or a consequence of, disease development.

“As more data becomes available, it will also be interesting to see whether a similar approach could identify tissue-specific ageing signatures, which could also prove important in disease mechanisms,” she added.

Read Full Post »

Methylomic analysis of monozygotic twins discordant for autism spectrum disorder and related behavioural traits
Open – Molecular Psychiatry advance online publication 23 April 2013; doi: 10.1038/mp.2013.41

C C Y Wong1, E L Meaburn1,2, A Ronald1,2, T S Price1,3, A R Jeffries1, L C Schalkwyk1, R Plomin1 and J Mill1,4


Autism spectrum disorder (ASD) defines a group of common, complex neurodevelopmental disorders. Although the aetiology of ASD has a strong genetic component, there is considerable monozygotic (MZ) twin discordance indicating a role for non-genetic factors. Because MZ twins share an identical DNA sequence, disease-discordant MZ twin pairs provide an ideal model for examining the contribution of environmentally driven epigenetic factors in disease. We performed a genome-wide analysis of DNA methylation in a sample of 50 MZ twin pairs (100 individuals) sampled from a representative population cohort that included twins discordant and concordant for ASD, ASD-associated traits and no autistic phenotype. Within-twin and between-group analyses identified numerous differentially methylated regions associated with ASD. In addition, we report significant correlations between DNA methylation and quantitatively measured autistic trait scores across our sample cohort. This study represents the first systematic epigenomic analyses of MZ twins discordant for ASD and implicates a role for altered DNA methylation in autism.


Autism spectrum disorder (ASD) defines a collection of complex childhood neurodevelopmental disorders affecting ~1% of the population and conferring severe lifelong disability.1 ASD is characterized by a triad of impairments: (1) deficits in social interactions and understanding, (2) non-social impairments, such as repetitive behaviour and interests, and (3) impairments in language and communication development. Quantitative genetic studies indicate that ASD has a strong heritable component,2 which is supported by the recent identification of several susceptibility loci and an emerging literature implicating the relevance of de novo and inherited copy number variants (CNVs) in the disorder.3 Despite intense research effort during the past decade, however, no definitive biological or clinical markers for ASD have been identified. This can be partly explained by the highly heterogeneous nature of ASD, both clinically and aetiologically. The clinical manifestation of ASD displays considerable individual variability in the severity of impairments and quantitative genetic studies also report genetic heterogeneity between the three trait domains of ASD.4, 5, 6

Despite the high heritability estimates for ASD, there is notable discordance within monozygotic (MZ) twin pairs for diagnosed ASD, and often considerable symptom severity differences within ASD-concordant MZ twins,2 strongly implicating a role for non-genetic epigenetic factors in aetiology. Epigenetic mechanisms mediate reversible changes in gene expression independent of DNA sequence variation, principally through alterations in DNA methylation and chromatin structure.7 Epigenetic changes in the brain have been associated with a range of neurological and cognitive processes, including neurogenesis,8 brain development9 and drug addiction.10 Emerging evidence implicates epigenetic modifications in several neuropsychiatric disorders, including ASD.11, 12 In particular, epigenetic dysregulation underlies the symptoms of Rett syndrome and Fragile X syndrome, two disorders with considerable phenotypic overlap with ASD.11 Although few empirical studies have systematically examined the role of altered epigenetic processes in ASD, recent analyses provide evidence for altered DNA methylation and histone modifications in disease pathology.13, 14, 15

The use of disease-discordant MZ twins represents a powerful strategy in epigenetic epidemiology because identical twins are matched for genotype, age, sex, maternal environment, population cohort effects and exposure to many shared environmental factors.16, 17 Recent studies have uncovered considerable epigenetic variation between MZ twins,18, 19, 20 and DNA methylation differences have been associated with MZ twin discordance for several complex phenotypic traits, including psychosis21 and Type 1 diabetes.22 In ASD, Nguyen and co-workers23 recently examined lymphoblastoid cell lines derived from peripheral blood lymphocytes collected from three ASD-discordant MZ twin pairs, reporting several ASD-associated differentially methylated loci.23 Two loci (RORA and BCL2) reported as hypermethylated in ASD were found to be downregulated in RNA from post-mortem autism brains. These findings support a role for DNA methylation in ASD and highlight the successful use of peripherally derived DNA from discordant MZ twins to identify disease-associated epigenetic changes. Given the highly heterogeneous nature of ASD, however, more comprehensive genome-wide analyses across larger numbers of samples are warranted to investigate the extent to which ASD-associated epigenetic variation is individual- and symptom-specific.

Materials and methods

Samples for methylomic analysis

Participants were recruited from the Twins’ Early Development Study (TEDS), a United Kingdom-based study of twins contacted from birth records.24 For this study, a total of 50 MZ twin pairs were identified within TEDS using the Childhood Autism Symptom Test (CAST), which assesses dimensional ASD traits, at age 8 years. The CAST25 is a 31-item screening measurement for ASD, designed for parents and teachers to complete in non-clinical settings to assess behaviours characteristic of the autistic spectrum. Items within the CAST are scored additively and a score of 15 (that is, answering ‘yes’ on 15 items) is the cutoff for identifying children ‘at risk’ for ASD. On the basis of the Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition criteria for autism, CAST items can be divided into three subscales: impairments in social symptoms (12 items); impairments in non-social symptoms (that is, restricted repetitive behaviours and interests (RRBIs); 7 items); and communication impairments (12 items).6 The CAST has been widely used in population-based studies of singletons25 as well as in twin studies.26 Within TEDS, the CAST has been shown to have good reliability and validity.27 Supplementary Figure 1 shows the distribution of total CAST and its three subscale scores within samples selected using parent ratings. Supplementary Table 1 provides a summary of the samples included in the analyses. Whole-blood samples were collected from subjects at age 15 years by a trained phlebotomist for DNA extraction and blood cell-count analysis. Blood cell counts were assessed for all collected samples and found to be within the normal range.

Genome-wide analysis of DNA methylation

For each individual, genomic DNA (500 ng) extracted from whole blood was treated with sodium bisulphite using the EZ 96-DNA methylation kit (Zymo Research, Irvine, CA, USA) following the manufacturer’s standard protocol. The bisulphite conversion reaction was performed in duplicate for each sample to minimize potential bias caused by variable conversion efficiency, and pooled bisulphite-treated DNA was used for subsequent array analysis. Genome-wide DNA methylation was assessed using the Illumina Infinium HumanMethylation27 BeadChip (Illumina, San Diego, CA, USA), which interrogates the DNA methylation profile of 27 578 CpG sites located in 14 495 protein-coding gene promoters and 110 microRNA gene promoters, at single-nucleotide resolution.28 Illumina GenomeStudio software (Illumina, San Diego, CA, USA) was used to extract signal intensities for each probe and perform initial quality control checks, with all data sets (except two individuals) being considered to be of high quality and included in subsequent analyses. To ensure stringent data quality, probes with a detection P-value >0.05 in any of the samples were removed across all individuals (N=1161 probes) in addition to a set of probes (N=2923) that were reported as nonspecific and potentially unreliable in a recent survey of all probes on the microarray.29

Methylation microarray data processing

All computations and statistical analyses were performed within the R statistical analysis environment (http://www.r-project.org), and all analysis scripts are available on request from the authors. A customized pipeline was used for the analysis of Illumina 27K methylation data as described in a previous study of psychosis-discordant MZ twins.21 Briefly, signal intensities for each probe were normalized using quantile normalization to reduce unwanted interarray variation. The relative methylation level of each interrogated CpG site was calculated as the ratio of the normalized signal from the methylated probe to the sum of the normalized signals of the methylated and unmethylated probes. This gave an average DNA methylation value, described as average ‘β-value’ for each CpG site, ranging from 0 (unmethylated) to 1 (fully methylated). A density plot of β-values for every sample revealed that, as expected given the known distribution of probes on the array, the data followed a bimodal distribution (Supplementary Figure 2). An empirical variance stabilizing transformation was used to adjust for the bimodal distribution of the data.21 Raw microarray data are available for download from http://epigenetics.iop.kcl.ac.uk/ASDTwins/.

Identification of ASD-associated DMRs

Two major analysis strategies were used to identify DMRs associated with ASD and related traits. First, DNA methylation differences within pairs of MZ twins were examined in MZ twin pairs discordant for ASD and ASD-related traits. Second, case–control comparisons of DNA methylation were performed between groups of individuals scoring high and low for ASD traits. With the aim of identifying real, biologically relevant within-twin and between-group DNA methylation differences, we used an analytic approach that incorporates both the significance (that is, t-test statistic) and magnitude (that is, absolute delta-β (Δβ)) of any observed differences to produce a ranked list of DMRs.21 A summary of the analysis strategy is presented in Supplementary Figure 3. This combined approach, where data are interpreted based on the combination of fold change and statistical significance, is routinely used in genome-wide gene expression studies and has been shown to produce gene lists of higher reproducibility and biological relevance.30 We recently used a similar approach successfully to identify disease-associated epigenetic changes in a psychosis-discordant MZ twin study.21 Given the known phenotypic and aetiologic heterogeneity, we also screened for large Δβ-values within each discordant MZ twin pair to examine the possibility that disease-associated epigenetic changes are potentially private and not consistent across all families. Finally, we examined whether quantitative CAST scores are correlated with DNA methylation at specific loci. The association between each of the quantitatively rated CAST subscale variables and DNA methylation at each CpG site was assessed using Pearson’s product–moment correlation.

Global DNA methylation analysis

Global levels of DNA methylation were quantified using the LUminometric Methylation Assay (LUMA).31 This method relies on DNA cleavage by methylation-sensitive and -insensitive restriction enzymes, followed by the quantification of the resulting restriction fragments using pyrosequencing.31 Positive controls, including both artificially methylated and artificially unmethylated samples, were included in all experimental steps to ensure unambiguous restriction enzyme digestions and to calibrate the experimental data, with each sample being processed in duplicate.

Fine mapping of DNA methylation using bisulphite pyrosequencing

Although the Illumina 27K array has been well validated for detecting differences in DNA methylation, we further tested specific regions nominated from the genome-wide microarray analysis using bisulphite pyrosequencing. Independent verification analyses were performed on two CpG sites (cg16474696, MGC3207; cg20507276, OR2L13) that demonstrated a large significant ASD-associated difference from the case versus control analysis. In each case, the assay spanned multiple CpG sites, including the specific CpG interrogated on the Illumina 27K array. Briefly, 500 ng DNA from each individual was independently treated with sodium bisulphite in duplicate using the EZ 96-DNA methylation kit as described above. Bisulphite-polymerase chain reaction amplification was performed in duplicate. Quantitative DNA methylation analysis was conducted using the PyroMark Q24 pyrosequencer (Qiagen, Valencia, CA, USA). The correlation between DNA methylation estimates obtained from Illumina 27K array and bisulphite pyrosequencing was assessed using Pearson’s moment–correlation coefficient. In addition, Sanger sequencing using BigDye v.3.1 terminator mix (Applied Biosystems, Foster City, CA, USA) was performed on the regions targeted by the MGC3207 pyrosequencing assay to ensure that the Illumina probe sequences and the primer binding sites for the pyrosequencing assay were free of any DNA sequence variation. The primers and assay conditions are given in Supplementary Table 2.

CNV analysis using genotyping arrays

Genomic DNA (200 ng) extracted from whole blood was genotyped using the Illumina HumanOmniExpress BeadChip (Illumina) targeting >730 000 single-nucleotide polymorphisms and Illumina GenomeStudio software was used to call genotypes based on predefined genotype cluster boundaries to denote cluster positions (HumanOmniExpress-12v1_C.egt). CNVs were identified from the genotyping data using two independent algorithms, PennCNV32 and QuantiSNP,33 with default parameters, and GC content signal preprocessing was applied. Stringent quality control steps were used to ensure that only high-confidence CNVs, that is, those >1 kb in size, covered by >5 probes and detected by both programs, were included for further analysis.


ASD is not associated with systemic differences in global DNA methylation
As expected, within-twin patterns of DNA methylation were highly correlated across all MZ twin pairs (average within-twin r across all probes=0.99), indicating that ASD and related traits are not associated with systemic changes in epigenetic programming. Supplementary Figure 4a shows the correlation between genome-wide DNA methylation across all probes on the array and one example ASD-discordant MZ twin pair; data for the other ASD-discordant MZ pairs are available for download from http://epigenetics.iop.kcl.ac.uk/ASDTwins/. These data were corroborated by global DNA methylation analysis using LUMA, which identified no significant difference between affected ASD twins and their co-twins (affected ASD twins mean=65.1%, unaffected co-twins mean=65.9%; P=0.817) (Supplementary Figure 4b).

Site-specific DNA differences are widespread in MZ-discordant ASD twins

In contrast to global levels of DNA methylation, DNA methylation at individual CpG sites demonstrated considerable variability within ASD-discordant MZ twin pairs. Figure 1a shows the distribution of average absolute differences in DNA methylation (Δβ) within all MZ twins discordant for ASD and ‘control’ MZ twin pairs concordant for low autistic trait score (unaffected). The overall distribution of average within-pair DNA methylation differences showed a highly significant skew to the right in ASD-discordant twins (P<2.2e−16, Kolmogorov–Smirnov test), with a higher number of CpG sites demonstrating a larger average difference in DNA methylation. Using an analysis method designed to identify the largest and most significant differences in DNA methylation at individual CpG sites, we identified multiple CpG sites across the genome exhibiting significant ASD-associated differential DNA methylation (Table 1). Of note, variability at these sites appears to be specific to ASD-discordant twin pairs; for the 50 top-ranked ASD-associated DMRs, we observe significantly higher average within-pair differences for MZ twin pairs discordant for ASD (P<0.01; see Figure 1b). The top differentially methylated site (cg13735974) across all ASD-discordant MZ twin pairs located in the NFYC promoter was consistently hypermethylated in affected individuals compared with their unaffected co-twins (mean Δβ=0.08, range=0.04–0.10, P<0.0004). For the top 10 DMRs, Figure 2 indicates highly consistent differences across all six ASD-discordant twin pairs.

Large DNA methylation differences are observed at specific loci within individual ASD-discordant MZ twin pairs

Because ASD is a highly heterogeneous disorder,3 it is probable that many disease-associated DMRs are family-specific. We therefore screened for the largest family-specific DNA methylation differences within each discordant ASD twin pair, identifying numerous loci (average=37.4 per twin pair) showing large DNA methylation differences (Δβ0.15) within each discordant twin pair (Supplementary Figure 5 and Supplementary Table 3). Although the majority of DMRs of large magnitude are family-specific, several are common across two or more discordant twin pairs in the same direction: cg12164282, located in PXDN promoter, showed ASD hypomethylation in twin pair 2 (Δβ=−0.19) and twin pair 4 (Δβ=−0.28); cg04545708, located in exon 1 of C11orf1, showed ASD hypermethylation in both twin pair 3 (Δβ=0.23) and twin pair 6 (Δβ=0.35); cg20426860, located in exon 1 of TMEM161A, showed ASD hypermethylation in twin pair 4 (Δβ=0.21) and twin pair 6 (Δβ=0.27); and cg27009703, located in HOXA9 promoter, showed ASD hypermethylation in twin pair 1 (Δβ=0.19) and twin pair 4 (Δβ=0.21).

DNA methylation differences are observed in MZ twins discordant for ASD-related traits

We detected significant DNA methylation differences between MZ twin pairs discordant for the three ASD-associated traits: that is, social autistic traits (N=9 MZ pairs), autistic RRBIs (N=9 MZ pairs) and communication autistic traits (N=8 MZ pairs). The top-ranked DMRs for each trait are shown in Supplementary Figure 6 and Supplementary Table 4. Interestingly, these included several genes previously implicated in the aetiology of ASD, including GABRB3, AFF2, NLGN2, JMJD1C, SNRPN, SNURF, UBE3A and KCNJ10.
As ASD is composed of a triad of all three impairments, we also examined if any CpG sites are differentially methylated across all discordant twin pairs (N=32 pairs, 64 individuals), regardless of their focal impairment. The top DMRs across all discordant twin pairs are shown in Supplementary Figure 7 and Supplementary Table 5. The top-ranked DMR located in the promoter region of PIK3C3 (cg19837131) was significantly hypomethylated in affected individuals compared with their unaffected co-twins (mean Δβ=−0.04, P<0.00004). Interestingly, while the overall average difference at this locus is small, the range of within-twin methylation difference is much greater (Δβ ranges from −0.12 to 0.6) and that the direction of effect is strikingly consistent across the majority of individual twin pairs, with 25 out of 32 discordant pairs (78%) demonstrating trait-related hypomethylation (Supplementary Figure 7).

Between-group analyses identified additional ASD-associated DMRs

Our study design also permitted us to examine group-level DNA methylation differences between ASD cases and controls. Unlike the within-pair discordant MZ twin design, between-group DNA methylation differences can be attributable to both genetic and environmental factors. Given the known gender difference in DNA methylation across the X chromosome, these analyses were restricted to probes on the autosomes (N=22 678) to minimize gender-induced biases.

Numerous DNA methylation differences were observed between ASD cases and controls. Supplementary Table 6a and Supplementary Figure 8 highlight the CpG sites showing the largest absolute DNA methylation differences (mean Δβ0.15) between ASD cases and unrelated control samples. The top case–control ASD-associated DMR was located upstream of MGC3207 (cg16474696), which was significantly hypomethylated in ASD cases compared with control samples (mean Δβ=−0.24, P<0.0002). In addition to MGC3207, large significant ASD-associated differences were observed in several other loci, including CpG sites near OR2L13 (cg20507276; mean Δβ=0.18) and C14orf152 (cg20022541; mean Δβ=−0.16, data not shown). Verification experiments were conducted on MGC3207 and OR2L13 using bisulphite-pyrosequencing confirming a high correlation (r=0.91 and 0.86, for MGC3207 (total N=33) and OR2L13 (total N=35), respectively) in DNA methylation levels, detected using the Infinium microarray and pyrosequencing platforms. Although our list of DMRs was stringently filtered to exclude probes containing known polymorphic SNPs,29 several of the top-ranked case–control DMRs, including cg16474696 and cg20507276, demonstrated patterns of DNA methylation consistent with DNA sequence effects, suggesting that they may be mediated by cis effects on DNA methylation34 or potentially reflect technical artefacts caused by uncatalogued sequence variation in probe binding sequences. To exclude the latter for MGC3207, we sequenced genomic DNA across the DMR in a range of samples showing differential methylation and identified no obvious polymorphic DNA sequence variation in the immediate vicinity of the probe.

Epigenetic differences identified between sporadic and familial ASD cases

ASD is an aetiologically heterogeneous syndrome and can occur both as a sporadic and a familial disorder. Recent CNV analyses report considerably higher frequencies of de novo variation in simplex compared with multiplex ASD families,35 suggesting that they represent genetically distinct classes. To test whether these are epigenetically distinct, we compared DNA methylation between individuals with sporadic ASD (where ASD is reported in only one member of the MZ twin pair; N=6) and individuals with familial ASD (as observed in concordant ASD MZ twin pairs; N=10). The genes most proximal to the 50 top-ranked differentially methylated CpG sites from this analysis are listed in Supplementary Table 6b. The top differentially methylated CpG site (cg07665060) is located upstream of C19orf33, which was significantly hypomethylated in individuals affected by sporadic ASD compared with those affected by familial ASD (mean Δβ=−0.12, P<0.0008) (Supplementary Figure 9). Interestingly, significant DNA methylation differences were also observed near several genes that have been previously implicated in ASD, including MBD4, AUTS2 and MAP2.

There is some overlap in DMRs across analytical groups

Table 2 provides a full list of top-ranked DMRs demonstrating overlap between analytical groups and highlighting their potential relevance to different autism-associated phenotypes. Interestingly, the top-ranked locus from the ASD-discordant twin analysis, located near NFYC, was also differentially methylated in the case–control analysis (mean Δβ=0.04, P<0.003). Furthermore, we identified significant DNA methylation differences in the MBD4 promoter in both ASD-discordant twin analysis and sporadic versus familial ASD analysis, suggesting that MBD4 methylation may have functional relevance to sporadic ASD. For each of the 50 top-ranked probes in each analysis category, Supplementary Table 7 lists their corresponding rank across the other analysis groups; although there is some overlap across groups (and each ranked list is positively, although modestly, correlated; Supplementary Table 8), few CpG sites are consistently altered across multiple analytical groups.

Quantitative autistic trait scores are correlated with DNA methylation at multiple CpG sites

Supplementary Figure 1 shows the distribution of total CAST and its three trait subscale scores across our samples. Initial analyses highlighted a strong correlation between DNA methylation and CAST score at multiple CpG sites (Supplementary Table 9). Further analysis showed that many of these correlations are influenced by extreme DNA methylation levels and phenotypic scores exhibited by one male ASD-concordant MZ twin pair (Figure 3a and Supplementary Figure 10). These twins are extreme outliers for CAST score (both scored 29 out of a maximum score of 31) and DNA methylation at multiple CpG sites (Supplementary Figure 11), and both have a history of pervasive developmental problems, with severe behavioural phenotypes and early-appearing IQ deficits, with special deficits in language. Given the existing link between highly penetrant CNVs and severe ASD, we tested whether the extreme patterns of DNA methylation in these two twins were associated with the presence of genomic alterations. Interestingly, high-density SNP microarray analysis revealed significant structural genomic alterations at multiple loci, with CNVs detected in regions previously implicated in ASD (Supplementary Table 10).

DNA methylation at multiple CpG sites remained significantly correlated with CAST scores even after this extreme twin pair was excluded from analyses (Supplementary Table 11 and Figure 3b), suggesting that they do not necessarily represent epigenetic/phenotypic ‘outliers’ but have DNA methylation levels (and phenotypic scores) at the extreme end of a true quantitative spectrum. For example, there is a strong correlation between DNA methylation at cg07753644 in P2RY11 and total CAST score in both analyses (with extreme twin pair: r=0.44; P=0.000009; without extreme twin pair: r=0.35; P=0.0006). Furthermore, DNA methylation at cg16279786 in the known ASD susceptibility locus, NRXN1, is significantly correlated with social autistic trait score in both analyses with (r=−0.41; P=0.00003) and without (r=−0.28; P=0.007) the extreme twin pair.


This study represents the first comprehensive analysis of DNA methylation differences in MZ twins discordant for ASD and autism-related traits using a genome-wide approach. We report ASD-associated DNA methylation differences at numerous CpG sites, with some DMRs consistent across all discordant twin pairs for each diagnostic category and others specific to one or two twin pairs, or one or two autism-related traits. Although sporadic cases of ASD appear to be epigenetically distinct to familial cases of ASD, some DMRs are common across both discordant MZ twin and case–control analyses. We also observed that DNA methylation at multiple CpG sites was significantly correlated with quantitatively rated autistic trait scores, with our analyses identifying one MZ twin pair, concordant for a very severe autistic phenotype, that appear to represent epigenetic outliers at multiple CpG sites across the genome. Interestingly, both individuals harbour numerous CNVs in genomic regions previously implicated in autism. Given the important role of epigenetic mechanisms in regulating gene expression, it is plausible that, like CNVs, methylomic variation could mediate disease susceptibility via altered gene dosage. Our hypothesis-free experimental design allowed us to identify disease-associated DNA methylation differences at loci not previously implicated in ASD, although we also found evidence for epigenetic changes at several genes previously implicated in autism.

Our findings have several implications for our understanding about the aetiology of ASD. First, they document the presence of numerous DNA methylation differences in MZ twins discordant for ASD and ASD-related traits, as well as between autistic individuals and control samples. This concurs with findings from a previous ASD-discordant twin study23 and further supports the association of variable DNA methylation with phenotypic differences between genetically identical individuals.21, 22 Second, the observed DNA methylation differences in MZ twins discordant for ASD and ASD-related traits, who are otherwise matched for genotype, shared environment, age, sex and other potential confounders, highlight the role of non-shared environmental and stochastic factors in the aetiology of autism. These findings concur with mounting data suggesting that environmentally mediated effects on the epigenome may be relatively common and important for disease.36 Third, our data suggest that although DNA methylation at some CpG sites is consistently altered across the entire set of discordant twins, differences at other CpG sites are specific to certain symptom groups, with considerable overall epigenetic heterogeneity between the three domains of autistic traits. These findings are in line with recent genetic research demonstrating significant genetic heterogeneity between the three core symptoms of ASD.4, 6 Fourth, the analysis of individual ASD-discordant twin pairs suggests that there is also considerable familial heterogeneity, with rare epigenetic alterations of large magnitude being potentially associated with ASD. These findings are not entirely surprising given the known heterogeneous nature of ASD revealed by molecular genetic studies,3 with an important role for highly penetrant rare genomic alterations, especially de novo mutations. Fifth, the identification of significant correlations between DNA methylation and autism symptom scores across our sample cohort suggests that there is a quantitative relationship between the severity of the autistic phenotype and epigenetic variation at certain loci. This reinforces the view of autism as the quantitative extreme of a phenotypic spectrum and highlights the potential use of epigenetic biomarkers as a predictor for severity of symptoms, although the accuracy, sensitivity and specificity of such predictors would require extended investigation. Finally, in addition to implicating a number of novel genes in the aetiology of ASD, we identified ASD-associated differential DNA methylation in the vicinity of multiple loci previously implicated in the pathogenesis of autism in genetic studies, including AFF2, AUTS2, GABRB3, NLGN3, NRXN1, SLC6A4 and UBE3A (see Supplementary Table 12 for a comprehensive list).

This study has several strengths. First, our unique sample consisted of MZ twin pairs discordant for autism and ASD-related-traits, in addition to age-matched concordant MZ twin pairs (for both ASD and low CAST score). It allowed us to perform a comprehensive analysis of the role of DNA methylation in ASD and ASD-related traits controlling for genotype, age, sex and other potential confounders. Second, by undertaking a genome-wide approach using a robust and reliable array platform, we were able to uncover phenotype-relevant differentially methylated loci in genomic regions that are both novel and have been previously associated with ASD. Third, our analysis of 32 discordant MZ twin pairs is relatively large compared with other discordant twin studies performed for other complex disease phenotypes; in this regard, for example, the only other ASD-discordant twin study assessed only three MZ twin pairs.23 Finally, we were able to complement our discordant-twin analyses by assessing group-level differences between ASD cases and controls, and also examining the relationship between DNA methylation and quantitatively rated trait scores across our entire sample cohort.

This study also has a number of limitations that should be considered when interpreting the results. First, although this is the largest and most comprehensive study of epigenetic variation in ASD performed to date, the sample size for each subgroup is small, in part because truly discordant MZ twin pairs are relatively rare. Although none of the reported differentially methylated loci reached a Bonferroni-corrected P-value cutoff (P=2.13E–05 for discordant-twin analysis and P=2.20E–05 for between-group analysis), this statistical approach is likely to be too conservative, especially given the non-independence of CpG sites37 and the small numbers of samples tested in each group. In this study, a combined analytic approach, taking into account the significance and the extent of methylation change, was used to identify differentially methylated loci that have potentially real, biological relevance to ASD. This analytic approach is widely used in genome-wide gene expression studies and is reported to produce gene lists of higher reproducibility and biological relevance compared with the convention method that relies solely on statistical significance.30 This notion is supported by the identification of differentially methylated loci near numerous genes previously implicated in ASD. Nonetheless, given the relatively small subgroup sample size, replication in larger samples is needed. Second, genome-wide DNA methylation profiling was performed on DNA extracted from whole blood, controlled for cell count, rather than the brain. Unfortunately, there is no archived collection of post-mortem brain samples from ASD-discordant MZ twins. Although there are known tissue-specific differences in DNA methylation profiles, recent studies suggest that disease-associated epimutations may be detectable across tissues,38 and our recent work suggests that some between-individual epigenetic variation is conserved across brain and blood.39 Furthermore, ASD-associated epimutations have been demonstrated to be detectable both in the brain and in peripheral tissues (that is, blood).13, 23 Moreover, our identification of DMRs in the vicinity of genes previously implicated in autism supports the notion that disease-relevant gene network and pathways can be identified from peripheral samples. Nonetheless, it would be informative for future studies to assess whether disease-associated epimutations reported from this study are also present in brain samples from ASD patients. Third, informations pertaining to the amniotic and chorionic status of our twin samples are unavailable, preventing us from further dissecting the epigenetic similarity/dissimilarity between twins sharing their placenta and/or amniotic sac. Fourth, the genome-wide platform used for this study (the Illumina 27K array), although robust and highly reliable,28 has a somewhat limited density of probe coverage, assaying only one or two CpG sites per gene. Future studies should take advantage of recent advances in genomic profiling technology and perform a more in-depth examination of methylomic differences associated with ASD. Finally, it is difficult to draw conclusions about causality for any of the ASD-associated DMRs identified in this study, in part, because we do not have corresponding RNA expression data, or DNA samples from the twins taken before they became discordant for ASD. It is thus plausible that many of the identified changes have occurred downstream of ASD, for example, resulting from exposure to medications commonly used to treat autistic symptoms. In fact, there is mounting evidence that many drugs used to treat neuropsychiatric disorders induce epigenetic changes.40 Such medication-induced changes could still be interesting; an understanding of the pathways via which these drugs work may provide information about the neurobiological processes involved in disease. The ideal study design, however, would assess DNA methylation changes in the brain longitudinally during individuals’ transition into ASD, although such a study does not appear feasible at present.

In summary, this is the first large-scale study to examine the role of genome-wide DNA methylation in ASD and ASD-related traits. Our findings show that: (1) there are numerous DNA methylation differences between MZ twins discordant for ASD and ASD-related traits, as well as between autistic individuals and control samples; (2) many of these DMRs are located in the vicinity of both novel genes and loci that have been previously implicated in ASD; (3) the nature of ASD-associated epimutations is complex with high heterogeneity between individuals; (4) there is high epigenetic heterogeneity between the triad of impairments that define ASD; and (5) there is a quantitative relationship between the severity of the autistic phenotype and DNA methylation at specific CpG sites across the genome. Overall, our findings from this study provide further support for the potential role of DNA methylation in ASD and ASD-related traits.

To review the figures you can access the full study here http://www.nature.com/mp/journal/vaop/ncurrent/full/mp201341a.html
Methylomic analysis of monozygotic twins discordant for autism spectrum disorder and related behavioural traits

Read Full Post »

Some women actually have men on the brain By Melissa Healy, Los Angeles Times
For the Booster Shots Blog
September 27, 2012

For decades after a woman has carried a male child in her womb or shared her mother’s womb with a brother, she carries a faint but unmistakable echo of that intimate bond: male fetal DNA that lodges itself in the far recesses of her brain.

That astonishing finding, published Wednesday in the journal Public Library of Science One (PLoS One), suggests that the act of having a child is no mere one-way transmission of genetic material and all that goes with it: There is an exchange of DNA that passes into the part of us that makes us who we are. That, in turn, may alter a woman’s health prospects in ways her own DNA never intended.

In the study, researchers from the Fred Hutchinson Cancer Research Center and the University of Washington examined, post-mortem, the brains of 59 women. In 63% of the brains, they found fetal DNA that could only have come from a male. While scattered throughout the brain, the genetic traces of this other individual were clustered heavily in the brain’s hippocampus — a region crucial to the consolidation of memories — and in the parietal and temporal lobes of the brain’s prefrontal cortex, areas that play roles in sensation, perception, sensory integration and language comprehension.

When a person takes on the DNA of another, as happens, for instance, in bone marrow transfusions, she is called a “chimera” — in mythology, a beast that is the fusion of two or more creatures. The discovery that a person can carry the fetal DNA of another person has given rise to a variant: This is dubbed microchimerism.

This line of research, says rheumatologist J. Lee Nelson, coauthor of the study, “suggests we need a new paradigm of the biological self” and how it is formed. We think of ourselves as the product of two biological parents and a one-time roll of the genetic dice. That, says Nelson, appears to be wrong: In the womb, we may also pick up the DNA of older siblings left over from their stay, or of a fetal twin who never made it to daylight. In the course of our lives, we may take on the DNA of the sons we bear, or even of the sons we conceived and miscarried. And that DNA can stay with us long after our big brothers have moved on and our sons have grown up and moved away.

The sources of our DNA “are much more diverse than we know,” said Nelson in an interview. And these exchanges of DNA may play an evolutionary role far greater than we have ever imagined, she added. Walt Whitman once wrote, “I contain multitudes,” and Nelson says she and her colleagues now glean new meanings from the observation.

The new study shows that this evolutionary X-factor is also at work in the brain.

It hasn’t been many years since scientists first learned that a baby’s DNA could cross the placental barrier from baby to mother and lodge itself in her blood and organs. The current study finds that it can also penetrate the vaunted “blood-brain barrier,” which is thought to protect the brain from toxins and foreign invaders.

Once there, Nelson said, the DNA of another person may alter a woman’s propensity to certain brain diseases — conferring protection in some cases and vulnerability in others. It may carry the switches that turn brain cancers on — or off. It may harden the brain against trauma or psychiatric disease — or make it less resilient. Future research will need to determine how, say, carrying a male fetus may influence a mother’s likelihood of developing Alzheimer’s disease or auto-immune diseases such as multiple sclerosis.


Male Microchimerism in the Human Female Brain

William F. N. Chan1¤*, Cécile Gurnot1, Thomas J. Montine2, Joshua A. Sonnen2, Katherine A. Guthrie1, J. Lee Nelson1,3
1 Clinical Research Division, Fred Hutchinson Cancer Research Center, Seattle, Washington, United States of America, 2 Department of Pathology, University of Washington, Seattle, Washington, United States of America, 3 Division of Rheumatology, University of Washington, Seattle, Washington, United States of America


In humans, naturally acquired microchimerism has been observed in many tissues and organs. Fetal microchimerism, however, has not been investigated in the human brain. Microchimerism of fetal as well as maternal origin has recently been reported in the mouse brain. In this study, we quantified male DNA in the human female brain as a marker for microchimerism of fetal origin (i.e. acquisition of male DNA by a woman while bearing a male fetus). Targeting the Y-chromosome-specific DYS14 gene, we performed real-time quantitative PCR in autopsied brain from women without clinical or pathologic evidence of neurologic disease (n = 26), or women who had Alzheimer’s disease (n = 33). We report that 63% of the females (37 of 59) tested harbored male microchimerism in the brain. Male microchimerism was present in multiple brain regions. Results also suggested lower prevalence (p = 0.03) and concentration (p = 0.06) of male microchimerism in the brains of women with Alzheimer’s disease than the brains of women without neurologic disease. In conclusion, male microchimerism is frequent and widely distributed in the human female brain.


During pregnancy, genetic material and cells are bi-directionally exchanged between the fetus and mother [1], following which there can be persistence of the foreign cells and/or DNA in the recipient [2], [3]. This naturally acquired microchimerism (Mc) may impart beneficial or adverse effects on human health. Fetal Mc, which describes the persistence of cells and/or DNA of fetal origin in the mother acquired during pregnancy, has been associated with several different autoimmune diseases as well as implicated in tissue repair and immunosurveillance [4]–[6].

Although there is a broad anatomical distribution of Mc in humans that varies in prevalence and quantity [7]–[13], whether the human brain harbors fetal Mc and with what frequency is not known. Fetal Mc has recently been described in the mouse brain [14], [15]. In limited studies, maternal Mc was described in the human fetal brain [9].

In this study, we performed real-time quantitative PCR (qPCR) to detect and quantify male DNA in multiple brain regions of women, targeting the Y-chromosome-specific DYS14 gene sequence as a marker for Mc of fetal origin. Deceased female subjects had no clinical or pathologic evidence of neurologic disease. We also tested brain specimens from women with Alzheimer’s disease (AD) for Mc. This is because AD has been reported as more prevalent in parous vs. nulliparous women [16], [17], increasing with higher number of pregnancies that also correlated with a younger age of AD onset [17], [18].


Subjects and Specimens

This study was approved by the institutional review board of the Fred Hutchinson Cancer Research Center (Number 5369; Protocol 1707). Subjects of the study were women without neurologic disease or with AD, totaling 59 deceased individuals. Twenty-six women had no neurologic disease. Thirty-three women had AD (Table 1). Brain autopsy specimens from these women came from one of two institutions: the Department of Pathology at the University of Washington in Seattle, Washington, or the Harvard Brain Tissue Resource Center established at McLean Hospital in Belmont, Massachusetts. Specimens from the University of Washington were obtained from adult women who had no clinical history of neurologic disease within two years of death and whose brain histology showed no evidence of disease, and from women who were diagnosed with probable AD during life [19] and met the National Institute on Aging-Reagan Institute consensus criteria for a neuropathologic diagnosis of AD [20]. Similarly, specimens from the Harvard Brain Tissue Resource Center were obtained from adult women without neurologic disease or who met clinical and pathologic criteria for AD. Age at death ranged from 32 to 101 (Table 1). Age at disease onset was known for subjects with AD from the University of Washington (median: 77 years; range: 64–93 years). Following autopsy, brain specimens were either formalin fixed or frozen in liquid nitrogen. Depending on availability, samples from two to twelve brain regions were obtained from each subject. Brain regions investigated included frontal lobe, parietal lobe, temporal lobe, occipital lobe, cingulate gyrus, hippocampus, amygdala, caudate, putamen, globus pallidus, thalamus, medulla, pons, cerebellum, and spinal cord. Subjects with AD contributed more specimens per person than subjects without neurologic disease, but this was not statistically significant (means: 3.6 vs. 2.5, respectively; p = 0.05). Combining subjects from both institutions, subjects with AD were significantly older at death (p<0.001); the post-mortem intervals (PMIs) were not significantly different (p = 0.06; Table 1). The most likely source of male Mc in female brain is a woman’s acquisition of male DNA from pregnancy with a male fetus. Limited pregnancy history was available on the subjects; pregnancy history on most subjects was unknown. Nine women were known to have at least one son, eight with AD and one without neurologic disease. Two women were known to have no history of having sons, one with AD and one without neurologic disease.

DNA Extraction

Genomic DNA was extracted from brain tissues using the QIAamp® DNA Mini Kit (QIAGEN, Valencia, CA) according to the manufacturer’s tissue protocol.

Real-time qPCR

Male DNA was quantified in female brain tissues by amplifying the Y chromosome-specific sequence DYS14 (GenBank Accession X06325) [21] using the TaqMan® assay and the ABI Prism® 7000 Sequence Detection System (Applied Biosystems, Foster City, CA). Primer and probe sequences for quantifying DYS14 [22], as well as preparation of standard curves, composition of the qPCR mixture and thermal profile [23] have all been described previously. Square of the correlation coefficient for all standard curves was always greater than 0.99. Every experiment included no template controls to test for male DNA contamination during plate setup and all controls were negative. A minimum of six wells was tested for each specimen. Mean Ct was 36, with a range between 30 and 39 for all specimens except those of B6388, which was between 26 and 29. A representative amplification plot is provided in (Figure S1). Only wells in which amplification occurred at Ct0.5 gEq/105. Thus, estimates of male Mc might be lower than the true values. On the other hand, since detection of male DNA did not account for Mc potentially contributed by female fetuses, this could result in underestimation of the overall Mc in the brain. HLA-specific qPCR, as previously reported [25], [26], is another approach to Mc detection that is not sex-dependent. It requires participation of family members which was not possible for the current studies. As a supplementary study, we tested autopsied brain from a female systemic sclerosis patient by HLA-specific qPCR for whom we had familial HLA genotyping, targeting the child’s paternally transmitted HLA as previously described [26], [27]. These data are provided in (Tables S1 and S2). All qPCR data were analyzed using the 7000 System Sequence Detection Software.


Subject and Mc measurement characteristics were compared across groups by Chi-squared test for categorical data and t-test for continuous data. Mc prevalence and concentrations were analyzed according to disease status. A logistic regression model was used to estimate the association between Mc prevalence and disease status, with adjustment for total gEq tested, age at death, and PMI. The estimates were also adjusted for possible correlation between repeated measures from the same subject via generalized estimating equations. Association was reported as an odds ratio (OR) along with p value to indicate significance. As an example, OR of 0.30 could be interpreted to say that the odds of having AD for a subject who tested positive for Mc was 70% lower than the odds for a subject who tested negative. We also analyzed Mc concentrations as the outcome in Poisson log-linear regression models, assuming that the number of gEq detected as Mc was directly proportional to the number of total gEq tested. By definition, Mc occurs rarely, thus the data distribution is skewed to the right. We utilized negative-binomial models to account for the high degree of over-dispersion in the data; interpretation of the resulting estimates is identical to those of a Poisson model. Adjustments for potential confounders and for possible correlation between repeated measures were made as described above. The rate ratio (RR), along with p value to indicate significance, was used to compare the observed rates of Mc detection in the two groups. As an example, RR of 0.30 could be interpreted to say that the rate of Mc detection in subjects with AD was 70% lower than the rate of Mc detection in subjects without neurological disease. Secondary analysis was conducted to determine whether disease status was associated with Mc prevalence or concentration in a subset of samples from brain regions thought to be most affected by AD. Furthermore, we investigated whether Mc prevalence or concentration correlated with the Braak stage, which describes the extent of neurofibrillary tangles in subjects with AD [28], or with HLA-DRB1*1501, a human leukocyte antigen allele that has been reported in association with AD [29]. Two-sided p-values from regression models were derived from the Wald test. Analyses were performed on SAS software version 9 (SAS Institute, Inc., Cary, NC).


Mc Prevalence and Concentration According to Brain Regions

he median number of specimens tested per subject was three, with a range of one to 12. Table 2 summarizes the specimen-level prevalence of male Mc according to brain region for all subjects. Per brain region, between two and 35 specimens were tested for male DNA. Although there were few specimens available, we did not detect male DNA in the frontal lobe and the putamen, and found the highest prevalence in the medulla. Considering all subjects together, Mc concentrations ranged from 0–512.5 gEq/105, with a median value of 0.2 and a 90th percentile of 3.7 gEq/105 (Figure 1). One subject from the Harvard Brain Tissue Resource Center who was without neurologic disease (coded as B6388; age of death 69 years) had three specimens with the highest concentration values in our dataset (296.1, 481.8, and 512.5 gEq/105 in the temporal lobe, cingulate gyrus, and pons, respectively). Using fluorescence in situ hybridization, we indeed found rare male cells in the brain of B6388 (Figure S2). The remaining concentration values in the dataset were 29.4 gEq/105 or less. Regarding the relationship between pregnancy history and Mc prevalence, five of nine subjects who were known to have at least one son harbored male Mc in at least one of their brain regions (Table S3). All positive individuals had AD; among the negatives were three with AD and one without neurologic disease. One of two women without history of having sons was also positive for male Mc in her brain and without neurologic disease; the negative individual had AD.

Prevalence and Concentration of Male Mc in Human Brain: Women without Neurologic Disease or with AD

Of 183 specimens, 64 (35%) tested positive for Mc (Table 2). Eighteen of 26 subjects without neurologic disease (69%) had at least one positive value, with 30 positive results in 65 specimens (46%). Nineteen of 33 subjects with AD (58%) had at least one positive value, with 34 positive results in 118 specimens (29%). The estimated OR from a univariate model was 0.47 (95% confidence interval (CI) 0.21–1.08, p = 0.08). After adjustment for total gEq tested, age at death, and PMI, AD was significantly associated with lower Mc prevalence: OR 0.40 (95% CI 0.17–0.93, p = 0.03). Thus, the odds of having AD for a subject who tested positive for Mc was 60% lower than the odds for a subject who tested negative. When Mc concentrations were analyzed according to whether subjects had no neurologic disease or had AD, the estimated RR from an adjusted model was 0.05 (95% CI 0.01–0.39, p = 0.004). However, exclusion of brain specimens from subject B6388, who was without neurologic disease and whose level of male Mc was 10-fold greater than the next highest concentration from a different subject, changed the estimate dramatically: RR 0.41 (95% CI 0.16–1.05, p = 0.06). Thus, the rate of Mc detection in subjects with AD was 59% lower than the rate of Mc detection in subjects without neurological disease, but was not statistically significant. Age at death was also not statistically significantly associated with Mc prevalence, either in univariate or adjusted models (adjustments for disease status and total gEq tested; p = 0.79). However, any relationship between age at death and male Mc from prior pregnancies with a male fetus could not be evaluated because pregnancy history and the time interval from pregnancies to death were generally unknown from our subjects.

Prevalence and Concentration of Male Mc in Brain Regions Affected by AD

We conducted a secondary analysis considering specimens only from the five brain regions thought to be most affected by AD: amygdala, hippocampus, frontal lobe, parietal lobe, and temporal lobe [30], [31]. Considering only these regions, 12 of 24 subjects without neurologic disease (50%) had at least one positive value, with 12 positive results in 24 specimens (50%). Thirteen of 33 subjects with AD (39%) had at least one positive value, with 15 positive results in 44 specimens (34%). The adjusted OR describing the association of Mc prevalence and disease status was 0.48 (95% CI 0.14–1.62, p = 0.23). Therefore, the odds of having AD for a subject who tested positive for Mc in brain regions most affected by this disease was 52% lower than the odds for a subject who tested negative, but was not statistically significant. However, none of the subjects without neurologic disease contributed specimens of the amygdala or the frontal lobe. Comparing Mc concentrations across groups, excluding one specimen from subject B6388, the adjusted RR was 0.27 (95% CI 0.13–0.56, p<0.001). Thus, the proportion of positive specimens was not significantly different between groups, but Mc concentrations in this subset of brain specimens from subjects with AD tended to have lower values than those found in subjects without neurologic disease. In other analyses, there was no significant association between Mc prevalence or concentration and the Braak stage (Table 1; p = 0.99 and 0.93, respectively), and no significant association between Mc prevalence and HLA-DRB1*1501 (8 of 11 DR15-bearing subjects positive for Mc (73%) vs. 16 of 31 subjects without DR15 who also had Mc (52%); p = 0.13).


n this study, we provide the first description of male Mc in female human brain and specific brain regions. Collectively with data showing the presence of male DNA in the cerebrospinal fluid [32], our results indicate that fetal DNA and likely cells can cross the human blood-brain barrier (BBB) and reside in the brain. Changes in BBB permeability occur during pregnancy [33] and may therefore provide a unique opportunity for the establishment of Mc in the brain. Also unique to our study are the findings that male Mc in the human female brain is relatively frequent (positive in 63% of subjects) and distributed in multiple brain regions, and is potentially persistent across the human lifespan (the oldest female in whom male DNA was detected in the brain was 94 years).

That Mc can penetrate the human BBB and reside in the brain was first indicated by murine studies that showed the presence of both foreign cells and DNA in mouse brains [14], [15]. However, prevalence of brain Mc in mice has not been well defined, as the frequency reported varies depending on the study [14], [15], [34]–[36], and in one investigation, Mc was not observed [37]. Similar to mouse data, our study of humans found that brain Mc was not present in all individuals tested. Even in those who showed positivity overall, not all of their brain regions had Mc. Mc concentration also showed considerable variability. Overall, our data complement and extend on other reports describing Mc in the general human population, in peripheral blood and at the level of the tissue/organ studied within and between subjects [9]–[13]. It is currently not possible to meaningfully compare Mc prevalence or concentration in human brain to other tissues because other tissues were not available from our subjects. Moreover, prior studies that evaluated Mc in other organs used diverse methods, some of which were not quantitative.

The most likely source of male Mc in female brain is acquisition of fetal Mc from pregnancy with a male fetus. In women without sons, male DNA can also be acquired from an abortion or a miscarriage [22], [23], [38]–[40]. The pregnancy history was unknown for all but a few subjects in the current studies, thus male Mc in female brain could not be evaluated according to specific prior pregnancy history. In addition to prior pregnancies, male Mc could be acquired by a female from a recognized or vanished male twin [41]–[43], an older male sibling, or through non-irradiated blood transfusion [44].

Because AD is more prevalent in women than men and an increased risk has been reported in parous vs. nulliparous women and correlated with younger age of onset [16]–[18], we also investigated male Mc in women with AD. AD is a neurodegenerative disease characterized by elevated levels of amyloid plaques, cerebrovascular amyloidosis, and neurofibrillary tangle [30]. Our results suggesting women with AD have a lower prevalence of male Mc in the brain and lower concentrations in regions most affected by AD were unexpected. However, the number of subjects tested was modest and, as discussed previously, pregnancy history was largely unknown. The explanation for decreased Mc in AD, should this observation be replicated in a larger study, is not obvious. In other diseases, both beneficial and detrimental effects of Mc of fetal origin have been described depending on several factors including the specific type and source of Mc [6]. A significant limitation of the current study was the inability to distinguish the type and source of male Mc, and further studies that distinguish genetically normal from abnormal Mc would be of potential interest.

At present, the biological significance of harboring Mc in the human brain requires further investigation. Mc appears to persist in the blood, bone, and bone marrow for decades [2], [45] and is present among different hematopoietic lineages [46]. Moreover, Mc appears to integrate and generate specific cell types in tissues [10], [11], [47]–[49]. In murine studies, fetal Mc in the maternal brain has been observed to resemble perivascular macrophages, neurons, astrocytes, and oligodendrocytes both morphologically and phenotypically and occupy the respective niches [15], [36]. Thus, it is possible that Mc in the brain is able to differentiate into various mature phenotypes or undergoes fusion with pre-existing cells and acquires a new phenotype, as suggested by murine and human studies in which bone marrow-derived cells circulated to the brain and generated neuronal cells by differentiation, or fused with pre-existing neurons [50]–[53]. Lastly, a few studies have reported an association between parity and decreased risk of brain cancer, raising the possibility that Mc could contribute to immunosurveillance against tumorigenic cells as has been suggested for some other types of malignancy [6], [54]–[56].

In conclusion, male Mc is frequent and widely distributed in the human female brain. Although the relationship between brain Mc and health versus disease requires further study, our findings suggest that Mc of fetal origin could impact maternal health and potentially be of evolutionary significance.

For more on this important study http://www.plosone.org/article/info%3Adoi%2F10.1371%2Fjournal.pone.0045592

Read Full Post »

Unnatural selection: How humans are driving evolution by Michael Le Page
New Scientist – 27 April 2011

Humans are not only causing a mass extinction – we are also the biggest force in the evolution of the species that will survive

THE Zoque people of Mexico hold a ceremony every year during which they grind up a poisonous plant and pour the mixture into a river running through a cave (pictured below). Any of the river’s molly fish that float to the surface are seen as a gift from the gods. The gods seem to be on the side of the fish, though – the fish in the poisoned parts of the river are becoming resistant to the plant’s active ingredient, rotenone.

If fish can evolve in response to a small religious ceremony, just imagine the effects of all the other changes we are making to the planet. We are turning grassland and forests into fields and cities, while polluting the air and water. We are hunting species to the brink of extinction and beyond, as well as introducing new pests and diseases to just about every part of the world. And that’s not to mention drastically altering the climate of the entire planet.

It is no secret that many – perhaps even most – species living today are likely to be wiped out over the next century or two as a result of this multiple onslaught. What is now becoming clear is that few of the species that survive will live on unchanged.

Far from being a slow process, evolution can occur extremely rapidly when the environment changes (New Scientist, 2 April, p 32). So, as we alter the planet ever faster and more drastically, we are becoming the main force driving evolution. “The intensity of the ecological effect of man is pretty obvious,” says Stephen Palumbi of Stanford University in California. “There is an amazing amount of evolutionary change as a result.”

Some of the fastest rates of evolution ever measured in the wild are in plants and animals harvested by humans. The few populations for which we have data are, on average, evolving three times as fast as populations subject only to natural pressures, for example.

Over the following pages, we look at the many ways in which plants and animals are already evolving in response to human pressures. Some of these changes, such as animals evolving to survive in highly polluted areas, can be seen as a positive thing. Others are bad from our point of view, such as animals we hunt losing the traits we value most in them, or pests becoming immune to poisons. What is clear is that whether the issue is growing enough food, conserving wild animals or keeping our beds bug-free, human-driven evolution is a factor we can no longer afford to ignore.

Unnatural selection: Hunting down elephants’ tusks

Most predators target the young or the weak. We are different, targeting the biggest and best, or those with characteristics we desire, such as large antlers. Combine this with our ability to kill in great number and the result is extremely rapid evolution of our prey.

The first clear evidence was published in 1942, and since then many examples have emerged of how hunting can transform the hunted. The targeting of large animals has resulted in a fall in the average size of caribou in some areas, for instance, while trophy hunting has led to bighorn sheep in Canada and mouflon in France evolving smaller horns.

Perhaps the most dramatic example is the shrinking of tusks in elephants, or even their complete loss. In eastern Zambia, the proportion of tuskless female elephants shot up from 10 per cent in 1969 to nearly 40 per cent in 1989 as a result of poaching (African Journal of Ecology, vol 33, p 230). Less dramatic rises in tusklessness have been reported in many other parts of Africa, with some bull elephants losing tusks too.

Humans have had an even bigger impact in Asia. Only male Asian elephants have tusks, and the proportion of tuskless bulls has soared in many areas. In Sri Lanka, where there has been a lot of poaching, under 5 per cent of males now have tusks, says Raman Sukumar of the Indian Institute of Science in Bangalore, who studies Asian elephants. Simulations by Ralph Tiedemann of the University of Potsdam in Germany and colleagues suggest that female elephants’ preference for tuskers has partly counteracted the effect of hunting. However, even if all poaching stopped, it would take a very long time for the percentage of tuskers to rise again.

It’s not just animals that are being shaped by human preferences: the harvesting of wild plants can have a similar effect to hunting and fishing. In Tibet, for example, the height of the snow lotus at flowering has nearly halved over the past century as a result of the flowers being picked for use in traditional medicine (Proceedings of the National Academy of Sciences, vol 102, p 10218).

To even the balance, some biologists are now promoting the idea of counteracting the evolutionary pressures of hunting through “compensatory culling” – killing animals with undesirable traits. This has actually long been done in some places. In Germany and Poland, for instance, there is a tradition of shooting yearling deer with poor antlers to prevent a decrease in the antler size of mature stags.

Private reserves in countries such as Zimbabwe have a similar policy. They typically charge hunters a smaller “trophy fee” for shooting tuskless elephants – $3000 versus at least $12,500 for a tusker, for example. This is partly because tuskless animals are less valuable, but it is also a deliberate attempt to eliminate the trait.

Unnatural selection: The race against climate change

In Finland, the tawny owl used to be mainly grey. But since the 1960s, the proportion of a brown subtype has risen fast. “The frequency averaged around 12 per cent in the early 60s and has increased steadily to over 40 per cent nationwide,” says Patrik Karell of the University of Helsinki, whose findings were published earlier this year (Nature Communications, DOI: 10.1038/ncomms1213).

His team found that grey owls (pictured above right) have an advantage over brown ones only when there is lots of snow. As winters have become milder, the brown subtype has thrived. It is possible that this is because brown owls are better camouflaged when there is less snow, but it could also be because brown coloration is linked to another characteristic, such as higher energy needs.

There are countless examples of how global warming is affecting life, from plants flowering earlier in spring, to species spreading to areas that were once too cold for them to survive, to birds becoming smaller. The vast majority of these changes are not genetic but due to plasticity: organisms’ built-in ability to change their bodies and behaviour in response to whatever the environment throws at them. At least a few species, however, like the owls of Finland, are already changing genetically – evolving – in response to climate change.

In North America, for instance, pitcher plant mosquitoes lay their eggs in pitcher plants and the larvae enter a state of dormancy in the winter months before resuming development in spring. Dormancy is genetically programmed, triggered not by falling temperature but by the shortening days. As the growing season has lengthened, mutant mosquitoes that keep feeding and growing for longer have thrived. Northern populations now go dormant more than a week later than in 1972, when studies began.

The earlier breeding of red squirrels in North America is also thought to be partly a result of genetic changes. Some families emerge earlier in spring, and they are doing better as the climate shifts.

Plants are changing too. When seed collected from field mustard plants (Brassica rapa) in California in 1997 and 2004 were grown in identical conditions, the 2004 strains flowered 9 days earlier on average (Proceedings of the National Academy of Sciences, vol 104, p 1278). The change was a result of drought – the plants have evolved to reproduce before they run out of water.

Rapid evolution is thus already helping some species adapt to a warming world, but it is no “Get out of jail free” card. For instance, so far pied flycatchers in the UK seem unable to shift to laying eggs earlier in spring. And according to one model that specifically takes rapid evolution into account, global warming will kill off 20 per cent of all lizard species by 2080. The problem for lizards is that as the climate warms, they have to spend more time in the shade and less time feeding.

Organisms with long generation times and slow reproductive rates are the least able to evolve, says Stephen Palumbi at Stanford University. “And they are the ones that are already threatened. It’s a double whammy.”

Even species whose evolution has kept pace with the slight warming so far will not necessarily keep up as the global temperature soars by another 4 °C or more. Rapid evolution generally depends on the existing variation within a population, rather than on new mutations. “It is limited to the kind of changes that can happen quickly,” Palumbi says.

In fact, there is a catch-22 to very rapid evolution – the faster organisms evolve, the less able they are to evolve further. This is because fast change occurs when only a small proportion of each generation manages to reproduce, resulting in a dramatic loss of genetic diversity – the fuel for further evolution. In many cases, the size of populations will also plummet, rendering them vulnerable to extinction. “You could evolve really fast but just not make it,” says Michael Kinnison of the University of Maine in Orono.

Unnatural selection: Living with pollution

Between 1947 and 1976, two factories released half a billion kilograms of chemicals called polychlorinated biphenyls (PCBs) into the Hudson river, in the north-east US. The effects on wildlife weren’t studied at the time, but today some species seem to be thriving despite levels of PCBs, many of which are toxic, remaining high.

At least one species, the Atlantic tomcod – an ordinary-looking fish about 10 centimetres long – has evolved resistance. “We could blast them with PCBs and dioxins with no effect,” says Isaac Wirgin of New York University School of Medicine.

Many of the ill effects of PCBs and dioxins are caused by them binding to a protein called the hydrocarbon receptor (Science, vol 331, p 1322). The Hudson tomcod all have a mutation in the receptor that stops PCBs binding to it, Wirgin and colleagues reported earlier this year. The mutation is present in other tomcod populations too, Wirgin says, but at low levels.

The most famous example of evolution in action was a response to pollution: as the industrial revolution got under way, cream-coloured peppered moths in northern Britain turned black to stay hidden on trees stained by soot. As the tomcod shows, though, most evolutionary changes in response to pollution are invisible.

The spoil heaps of many old mines, for instance, are covered in plants that appear normal, but are in fact growing in soil containing high levels of metals such as copper, zinc, lead and arsenic that would be toxic to most specimens of these and other species. The evolution of tolerance has occurred extremely rapidly in some cases, sometimes within just a few years of the soil being contaminated.

With very widespread pollutants, it is much harder to show that organisms are evolving in response, because all populations change at once. The comparison has been done with a common weed called plantain (Plantago major), though. Ground-level ozone, produced when sunlight strikes car exhaust fumes, greatly impairs the growth of plants. When researchers grew plantain seeds collected in 1985 and 1991 from a site in northern England where ozone pollution reached very high levels in 1989 and 1990, they found that the plants from the 1985 batch grew nearly a third more slowly when exposed to ozone, whereas the growth of those from 1991 fell by only a tenth (New Phytologist, vol 131, p 337).

Since even the remotest parts of the planet are now polluted in one way or another, it is likely that many plants and animal populations have evolved some degree of tolerance, even though few cases have been documented. “Nobody looks for resistance,” says Wirgin. “My guess is that if you look you will find a lot of it.” His own discovery was entirely accidental: the team had set out to study liver cancers, and they only noticed the tomcod’s resistance when blasting the fish with PCBs failed to produce any tumours.

However, there are obviously limits to what evolution can achieve. This is especially true for small populations that reproduce slowly and have few offspring, such as the Yangtze river dolphin. Pollution is thought to have contributed to its extinction.

What’s more, pollution resistance in one species can have unexpected consequences for others. The tomcod’s tolerance allows it to accumulate extraordinarily high levels of PCBs in its body, for instance, which are a threat to animals higher up the food chain – such as humans with a taste for these reportedly delicious fish.

Unnatural selection: Spreading sickness

Perch in Lake Windermere in the UK used to live to a ripe old age. While the average age of fish caught and released by researchers was around 5 years, a few individuals were as old as 20. Then in 1976, an unidentified disease wiped out 99 per cent of adult fish and continued to preferentially kill older fish for years afterwards. Since then, no fish older than 7 have been caught.

According to Jan Ohlberger of the University of Oslo, Norway, the perch (Perca fluviatilis) evolved very quickly in response. They now become sexually mature at an earlier age, which increases their chances of breeding before they get killed by the disease (Proceedings of the Royal Society B, vol 278, p 35).

While the disease is thought to have spread naturally in the lake, Ohlberger points out that many devastating disease outbreaks in plants and animals are a result of human activity. To mention just a few: Dutch elm disease was caused by fungi introduced from Asia; lions were hard hit by canine distemper spread by village dogs, and corals are far more susceptible to diseases when water temperatures are abnormally high, which is happening often as a result of climate change.

Anything that kills a significant proportion of a population has the potential to bring about very fast evolution. In frogs there is now some evidence of this: last year several research groups reported that some populations appear to be becoming resistant to a fungus that has decimated many amphibian species. It is also clear that human populations have sometimes evolved rapidly in response to diseases such as kuru, which attacks the nervous system.

So it seems plausible that by spreading diseases or creating the conditions in which they thrive, humans are indirectly forcing many organisms to evolve. “I think this is a common phenomenon and has not yet been described because it is simply hard to prove,” says Ohlberger. He points out that the long-running capture-and-release programme at Lake Windermere, which began in 1943 and just happened to coincide with the disease outbreak in perch, is pretty unique. In most cases we know too little about what populations were like before disease outbreaks to be able to tell if and how they have evolved in response.

Unnatural selection: The arms race against pests

Had any strange itchy bites or rashes recently? You might have fallen victim to bedbugs. The little bloodsuckers are back in a big way, thanks in part to the fact that, like head lice and human fleas, they have evolved resistance to many common pesticides.

Whatever their drawbacks, there is no doubt that pesticides have made a huge difference to our lives. They have helped eliminate diseases like malaria from some areas and made possible the switch to intensive farming. As soon as we started using them, though, resistance began to evolve.

“Insects that succumb readily to kerosene in the Atlantic states defy it absolutely in Colorado [and] washes that easily destroy the San José scale [insect] in California are ridiculously ineffective in the Atlantic states,” wrote entomologist John Smith in 1897 – the first known report of insecticide resistance.

The use of synthetic pesticides like DDT took off in the 1940s. Resistant houseflies were discovered in 1946. By 1948, resistance had been reported in 12 insect species. In 1966, James Crow of the University of Wisconsin-Madison reported that the count had exceeded 165 species. “No more convincing examples of Darwinian evolution could be found than those provided by the development of resistance in one species after another,” he noted at the time.

It’s not just bugs. Rats and mice around the world have become resistant to the poison warfarin, and in Europe some have even become resistant to warfarin’s replacement, superwarfarin (Journal of Toxicological Studies, vol 33, p 283). In Australia, meanwhile, rabbits are becoming resistant to the poison used to control their numbers, called Compound 1080.

Because of its economic importance, pesticide resistance has been studied far more closely than other kinds of ongoing evolution. In many cases we know which mutations are involved, how they make organisms resistant and sometimes even how the mutations spread through populations.

Resistance often arises due to mutations that alter the shape of proteins and thus prevent insecticides binding to their targets. For instance, DDT and pyrethroid compounds kill insects by opening sodium ion channels in nerve cells, but in the malaria-carrying mosquito Anopheles gambiae, variants of the channels that cannot be opened this way have evolved on at least four separate occasions (PLoS One, vol 2, p e1243).

The other main mechanism of resistance involves enzymes that inactivate pesticides before they can kill. Some resistant strains of A. gambiae, for instance, produce large quantities of an enzyme called CYP6Z1 that can inactivate DDT.

Pesticide resistance is becoming such a serious problem that strategies for preventing it evolving in the first place are taken increasingly seriously. One approach is to alternate the type of pesticide applied, to try to avoid applying sustained selective pressure in one direction.

At present, though, the pests seem to be evolving faster than we can develop new pesticides. In one region of Burkina Faso, A. gambiae has become resistant to all four classes of insecticides used for malaria control.

Unnatural selection: Introducing invaders

In 1935, the South American cane toad was introduced to Australia to control pests feeding on sugar cane. The cane fields were not to the toad’s liking, but the rest of the countryside was. The toad has spread rapidly at the expense of many native species.

The highly poisonous animals are having a big effect on predators. Some, such as the Australian black snake, are developing resistance to cane toad toxins. Others, such as the red-bellied black snake and green tree snake, are changing in a more surprising way – their mouths are getting smaller. The reason is simple: snakes with big mouths can eat large toads that contain enough toxin to kill them.

The toads themselves are also changing. Some are now colonising regions that were too hot for the founder population, suggesting that they are evolving tolerance to more extreme conditions. What’s more, the toads leading the invasion are becoming better colonisers: they have bigger front legs and stronger back legs than toads living in the areas already colonised. Radio tagging has confirmed that these “super-invader” toads can travel faster, as you might expect. They are probably evolving because the first toads to reach new areas benefit from more food and less competition, and thus have more offspring. The changes are likely to be transient, though – once the toads stop spreading, the “super-invader” traits will gradually be lost.

Ships and planes have turned the natural trickle of species spreading to new islands or continents into a raging torrent, and the new arrivals sometimes have a dramatic effect. In areas of the US that have been invaded by fire ants, for instance, native fence lizards have evolved longer legs. They need them: given the opportunity, a dozen fire ants can kill a lizard in minutes.

Rather than simply study the results of invasions, Michael Kinnison of the University of Maine in Orono and colleagues have been actively experimenting. In one experiment, his team moved juvenile chinook salmon from one river in New Zealand to another. The salmon were introduced to the country around a century ago, and Kinnison wanted to assess the extent to which they had adapted to conditions in individual rivers. He found drastic differences in survival, even though the fish appear identical (Canadian Journal of Fisheries and Aquatic Sciences, vol 60, p 1). “When a population was locally adapted, it performed twice as well,” he says.

Kinnison suspects that lots of small changes can add up to make a huge difference to a population’s success. “Contemporary evolution may be relatively modest on a trait-by-trait basis, but its overall contribution to the performance of populations may be immense,” he says.

Such findings help explain why there is often a lag between the introduction of new species and their rapid spread. A newly arrived species is likely to find itself in an environment that is not quite ideal, and its population may be very small, meaning there is little genetic diversity. In these circumstances, a species will spread only slowly, if at all.

As the population begins to adapt to local conditions, though – perhaps via invisible changes such as mutations in immune genes – it is likely to start to grow and spread. Because more mutations occur in larger populations, it will then evolve faster, enabling it to spread quicker and further. If this turns out to be common, it is bad news. It suggests that many introduced species that seem to be behaving themselves could yet start spreading explosively and cause serious problems.

To view the original article click on the link below.

Unnatural selection: How humans are driving evolution by Michael Le Page

Read Full Post »

Your Inner Fish by Neil Shubin – 2008

You might all be wondering why I’ve added this book to my collection of Informative Books regarding the synthetic chemical revolution. This is in many ways where epigenetics began. This book documents the very beginning of our understanding of epigenetics. This is where geneticists began learning about gene expression and the triggers that silence them or make them active. It is also where they first discovered the method by which birth defects occur. This is a groundbreaking book with information not yet in our textbooks. The discovery of the Sonic hedgehog gene led to the understanding of how birth defects occur in limb development. You will understand why this book is so critically important to understanding the potential impacts of the synthetic revolution in evolutionary terms regarding humans. Humans share many genes with other species. Our development is extraordinarily similar. You will understand the implications of this field of study after reading the following excerpt.


We begin with an apparent puzzle. Our body is made up of hundreds of different kinds of cells. This cellular diversity gives our tissues and organs their distinct shapes and functions. The cells that make our bones, nerves, guts, and so on look and behave entirely differently. Despite these differences, there is a deep similarity among every cell inside our bodies: all of them contain exactly the same DNA. If DNA contains the information to build our bodies, tissues, and organs, how is it that cells as different as those found in muscle, nerve, and bone contain the same DNA?

The answer lies in understanding what pieces of DNA (the genes) are actually turned on in every cell. A skin cell is different from a neuron because different genes are active in each cell. When a gene is turned on, it makes a protein that can affect what the cell looks like and how it behaves. Therefore, to understand what makes a cell in the eye different from a cell in the bones of the hand, we need to know about the genetic switches that control the activity of genes in each cell and tissue.

Here’s an important fact: these genetic switches help to assemble us. At conception, we start as a single cell that contains all the DNA needed to build our body. The plan for that entire body unfolds via instructions contained in this single microscopic cell. To go from this generalized egg cell to a complete human, with trillions of specialized cells organized in just the right way, whole batteries of genes need to be turned on and off at just the right stages of development. Like a concerto composed of individual notes played by many instruments, our bodies are a composition of individual genes turning on and off inside each cell during our development…

When we compare the ensemble of genes active in the development of a fish fin to those active in the development of a human hand, we can catalogue the genetic differences between fins and limbs. This kind of comparison gives us some likely culprits–the genetic switches that may have changed during the origin of limbs. We can then study what these genes are doing in the embryo and how they might have changed. We can even do experiments in which we manipulate the genes to see how bodies actually change in response to different conditions or stimuli…..We will begin by looking at the structure of our limbs, and zoom all the way down to the tissues, cells, and genes that make it.


Our limbs exist in three dimensions: they have a top and a bottom, a pinky side and a thumb side, a base and a tip. The bones at the tips, in our fingers, are different from the bones at our shoulder. Likewise, our hands are different from one side to the other. Our pinkies are shaped differently from one side to the other. The Holy Grail of our developmental research is to understand what genes differentiate the various bones of our limb, and what controls development in these three dimensions. What DNA actually makes a pinky different from a thumb? What makes our fingers distinct from our arm bones? If we understand the genes that control such patterns, we will be privy to the recipe that builds us.

All the genetic switches that make our fingers, arm bones, and toes do their thing during the third to eighth week after conception. Limbs begin their development as tiny buds that extend from our embryonic bodies. The buds grow over two weeks, until the tip forms a paddle. Inside this paddle are millions of cells which will ultimately give rise to the skeleton, nerves, and muscles that we’ll have for the rest of our lives.

To study how this pattern emerges, we need to look at embryos and sometimes interfere with their development to assess what happens when things go wrong. Moreover, we need to look at mutants and at their internal structures and genes, often by making whole mutant populations through careful breeding. Obviously, we cannot study humans in these ways. The challenge for the pioneers in this field was to find the animals that could be useful windows into our own development. The first experimental embryologists interested in limbs in the 1930s and 1940s faced several problems. They needed an organism in which the limbs were accessible for observation and experiment. The embryo had to be relatively large, so that they could perform surgical procedures on it. Importantly, the embryo had to grow in a protected place, in a container that sheltered it from jostling and other environmental disturbances. Also, and critically, the embryos had to be abundant and available year-round. the obvious solution to this scientific need is at your local grocery store: chicken eggs.

In the 1950s and 1960s a number of biologists, including Edgar Zwilling and John Saunders, did extraordinary creative experiments on chicken eggs to understand how the pattern of the skeleton forms. This was an era of slice and dice. Embryos were cut up and various tissues moved about to see what effect this had on development. The approach involved very careful microsurgery, manipulating patches of tissue no more than a millimeter think. In that way, by moving tissues about in the developing limb, Saunders and Zwilling uncovered some of the key mechanisms that build limbs as different as bird wings, whale flippers, and human hands.

They discovered that two little patches of tissue essentially control the development of the pattern of bones inside limbs. A strip of tissue at the extreme end of the limb bud is essential for all limb development. Remove it, and development stops. Remove it early, and we are left with only an upper arm, or a piece of an arm. Remove it slightly later, and we end up with an upper arm and a forearm. Remove it even later, and the arm is almost complete, except that the digits are short and deformed.

Another experiment, initially done by Mary Gasseling in John Saunder’s laboratory, led to a powerful new line of research. Take a little patch of tissue from what will become the pinky side of a limb bud, early in development, and transplant it on the opposite side, just under where the first finger will form. Let the chick develop and form a wing. The result surprised nearly everybody. The wing developed normally except that it also had a full duplicate set of digits. Even more remarkable was the pattern of the digits: the new fingers were mirror images of the normal set. Obviously, something inside that patch of tissue, some molecule or gene, was able to direct the development of the pattern of the fingers. This result spawned a blizzard of new experiments, and we learned that this effect can be mimicked by a variety of other means. For example, take a chicken embryo and dab a little vitamin A (retinoic acid) on its limb bud, or simply inject vitamin A into the egg. and let the embryo develop. If you supply the vitamin A at the right concentration and at the right stage, you’ll get the same mirror-image duplication that Gasseling, Saunders, and Zwilling got from the grafting experiments. This patch of tissue was named the zone of polarizing activity (ZPA). Essentially, the ZPA is a patch of tissue that causes the pinky side to be different from the thumb side. Obviously chicks do not have a pinky and a thumb. The terminology we use is to the number of digits, with our pinky corresponding to digit five of other animals and our thumb corresponding to digit one.

The ZPA drew interest because it appeared, in some way, to control the formation of fingers and toes. But how? Some people believed that the cells in the ZPA made a molecule that then spread across the limb to instruct cells to make different fingers. The key proposal was that it was the concentration of this named molecule that was the important factor. In areas close to the ZPA, where there is a high concentration of this molecule, cells would respond by making a pinky. In the opposite side of the developing hand, farther from the ZPA so that the molecule was more diffused, the cells would respond by making a thumb. Cells in the middle would each respond according to the concentration of this molecule to make the second, third, and fourth fingers.

This concentration-dependent idea could be tested. In 1979, Denis Summerbell placed an extremely small piece of foil between the ZPA patch and the rest of the limb. The idea was to use this barrier to prevent any kind of molecule from diffusing from the ZPA to the other side. Summerbell studied what happened to the cells on each side of the barrier. Cells on the opposite side often did not form digits; if they did, the digits were badly malformed. The conclusion was obvious. Something was emanating from the ZPA that controlled how the digits formed and what they looked like. To identify that something, researchers needed to look at DNA.


That project was left to a new generation of scientists. Not until the 1990s, when new molecular techniques became available, was the genetic control for the ZPA’s operation unraveled.

A major breakthrough happened in 1993, when Cliff Tabin’s laboratory at Harvard started hunting for the genes that control the ZPA. Their prey was the molecular mechanisms that gave the ZPA its ability to make our pinky different from our thumb. By this time his group started to work in the early 1990s, a number of experiments like the ones I’ve described had led us to believe that some sort of molecule caused the whole thing. This was a grand theory, but nobody knew what this molecule was. People would propose one molecule after another, only to find that none was up to the job. Finally, the Tabin lab came up with a novel notion, and one very relevant to the theme of this book. Look to flies for the answer.

Genetic experiments in the 1980s had revealed the wonderful pattern of gene activity that sculpts the body of a fly from a single-celled egg. The body of a fruit fly is organized from front to back, with the head at the front and the wings at the back. Whole batteries of genes are turned on and off during fly development, and this pattern of gene activity serves to demarcate the different regions of the fly.

Tabin didn’t know at the time, but two other laboratories–those of Andy MacMahon and Phil Ingham–had already come up with the same general idea independently. What emerged was a remarkably successful collaboration among three different lab groups. One of the fly genes caught the attention of Tabin, McMahon, and Ingham. They noted that this gene made one end of a body segment look different from the other. Fly geneticists named it hedgehog. Doesn’t the function of hedgehog in the fly body–to make one region different from another–sound like what ZPA does in making the pinky different from the thumb? That parallel was not lost on the three labs. So off they went, looking for a hedgehog gene in creatures like chickens, mice, and fish.

Because the lab groups knew the structure of the fly’s hedgehog gene, that had a search image to help them single out the gene in chickens. Each gene sequence; using a number of molecular tools, the researchers could scan a chicken’s DNA for the hedgehog sequence. After a lot of trial and error, they found a chicken hedgehog gene.

Just as paleontologists get to name a new species, geneticists get to name new genes. The fly geneticists who discovered hedgehog had named it that because the flies with a mutation in the gene had bristles that reminded them of a little hedgehog. Tabin, McMahon, and Ingham named the chicken version of the gene Sonic hedgehog, after the Sega Genesis video game.

Now came the fun question: What does Sonic hedgehog actually do in the limb? The Tabin group attached a dye to a molecule that would stick to the gene, enabling them to visualize where the gene is active in the limb. To their great surprise, they found that only cells in a tiny patch of the limb had gene activity: the ZPA.

So the next steps were obvious. The patterns of activity in the Sonic hedgehog gene could mimic those of the ZPA tissue itself. Recall when you treat the limb with retinoic acid, a form of vitamin A, you get a ZPA active on the opposite side. Guess what happens when you treat a limb with retinoic acid, then map where Sonic hedgehog is active? Sonic hedgehog is active on both sides–pinky and thumb–just as the ZPA does when it is treated with retinoic acid.

Knowing the structure of the chicken Sonic hedgehog gave other researchers the tools to look for it in everything else that has fingers, from frogs to humans. Every limbed animal has the Sonic hedgehog gene. And in every single animal we have studied, Sonic hedgehog is active in ZPA tissue. If Sonic hedgehog hadn’t turned on properly during the eight week of your own development, then you either would have extra fingers or your pinky and thumb would look alike. Occasionally, when things go wrong with Sonic hedgehog, the hand ends up looking like a broad paddle with as many as twelve fingers that all look alike.

We now know that Sonic hedgehog is one of dozens of genes that act to sculpt our limbs from shoulder to fingertip by turning on and off at the right time. Remarkably, work in chickens, frogs, and mice was telling us the same thing. The DNA recipe to build upper arms, forearms, wrists, and digits is virtually identical in every creature that has limbs.

How far back can we trace Sonic hedgehog and other bits of DNA that build limbs? Is this stuff active in building the skeleton of fish fins? Or are hands genetically completely different from fish fins? We saw an inner fish in the anatomy of our arms and hands. What about the DNA that builds it? ….

The title of the book indicates the answers to those questions

Read Full Post »

Older Posts »