Meeting report - Roche - Nature Medicine Translational Neuroscience Symposium 2009: Autism and Other Developmental Brain Disorders
Roche - Nature Medicine Translational Neuroscience Symposium
Strategies for identifying and developing new drugs have traditionally relied on serendipity to discover small molecules with potential biological activity, followed by laboratory evaluation and ultimately clinical use. As spending on health care grows, there is an increasing demand for medicines with an improved risk:benefit ratio. Translational medicine, which has proved its worth in oncology, offers a targeted strategy for identifying new medicines for psychiatric and neurological conditions.
Dr. Thomas Insel, Director of the National Institute of Mental Health, USA, explained the 'reverse translation cycle' by which the clinical diagnosis of a genetic disorder leads to discovery of the causative gene, which is then used to develop an animal model of the clinical disorder. This in turn helps to clarify the molecular pathophysiology of the disease, revealing a putative target for treatment and allowing candidate drugs to be evaluated before proceeding to clinical trials. Examples of where this has worked 'with spectacular success' include the development of Gleevec for acute myelogenous leukaemia and the discovery that losartan may be an effective treatment for Marfan syndrome. He noted another three promising examples: mGluR5 antagonists for Fragile X syndrome, rapamycin for tuberous sclerosis and statins for neurofibromatosis 1.
Neurodevelopmental disorders as synaptic diseases
The symposium reviewed current thinking about tuberous sclerosis, Fragile X syndrome, Rett syndrome and autism. But each of these specific conditions offers more general lessons about the molecular mechanism and function underlying neurodevelopmental disorders. Some—for example, Fragile X syndrome and autism but possibly also schizophrenia and mood disorders—can be thought of as disorders of synaptic function. We have long known that synaptic plasticity is important for learning and memory, and alterations of the formation and function of synaptic connections is now believed to underlie autism spectrum disorders. Synaptic plasticity depends critically on an increase in synaptic numbers and also on the long-term maintenance of that increase. Professor Alcino Silva, Departments of Neurology, Psychiatry and Psychology at UCLA, pointed out that autism, epilepsy and other forms of learning disabilities all share some form of synaptic dysfunction.
Symposium speakers and organisers: back row, left to right: Jeffrey Neul, Jamel Chelly, Petrus de Vries, Geraldine Dawson, Randi Hagerman, Thomas Südhof, Chris Walsh, Stephen Warren and Mark Bear. Front row, left to right: Matthew State, Thomas Insel, Alcino Silva, Juan Carlos Lopez, Luca Santarelli, Thomas Bourgeron and Anthony Bailey. Not pictured: Robert Manilow, Morgan Sheng, Gerald Fischbach and Christer Nordstedt.
Recent research has revealed the complexity of variations in synaptic structure and function. These include variations in receptors for glutamate or other neurotransmitters but postsynaptic scaffold proteins seem to also play an important role in certain types of familial autism. Among these, SHANK proteins promote maturation and enlargement of synapses and dendritic spines and SHANK1 has been implicated in specific cognitive processes that may have a parallel in autism.
Synapse organisation is regulated by the trans-synaptic cell adhesion complex comprising the neurexins and neuroligins, both of which have been implicated in autism. Professor Thomas Südhof, Avram Goldstein Professor at Stanford University School of Medicine, suggested that cognitive disorders such as schizophrenia and autism are neural circuit disorders in which alterations of neurexins and neuroligins is associated with synaptic dysfunction. These molecules present themselves in many different isoforms, resulting in a wide range of hierarchies. In vivo studies have shown that the neurexin-neuroligin complex plays an important, activity-dependent role in synapse function. "This is a crucial system that organises synapses," Professor Südhof said. "There is accumulating evidence that this complex, with its associated proteins, is involved in autism." In animal studies, mutations of neuroligin genes have been associated with changes in behaviour and learning and a large number of neurexin and neuroligin mutations have been associated with autism. Multiple families with familial autism have been shown to carry mutations of neuroligin-4 gene, though only a single family has been shown to have a point mutation in the neuroligin-3 gene; several families have copy number variations in the neurexin 1-alpha gene. "Although these mutations are rare, taken together they are not that rare," Professor Südhof added. "There's a good case to be made for a synaptic phenotype, at least for a limited group of people with autism. Possibly that is paradigmatic that many other genes causing heterogeneity in autism may in some way have an impact on synaptic transmission."
The clinical phenotypes do not always correlate with specific neuroligin mutations and there is still much to do to fully understand the relationship between these synaptic abnormalities and the clinical expression of autism. However, it is clear that we should consider the synapse as a dynamic entity associated with variations in scaffolding and other proteins in different areas of the brain. We have made significant progress since the 1970s, when autism was attributed to dysfunctional parenting, with the result that children were forcibly removed from families, Dr. Insel commented. "We've gone from sin to synapses in the course of my lifetime," he said.
Do simple Mendelian disorders help us understand more complex disorders?
The overwhelming message from the symposium was that Mendelian disorders are anything but simple. These disorders and their underlying mechanisms are tremendously complex: even a single gene disorder may be associated with over a thousand mutations and a single mutation can give rise to variable phenotypes. Despite this complexity it is clear that genetic disorders may respond to relatively simple treatments. The reverse translation cycle has worked well in developing treatments for simple gene disorders—will it do the same for more complex conditions such autism? Dr. Insel pointed out that for many disorders we still have not identified the genes that allow us to run the translation cycle but we do not need complete genetic architecture to make a start. If we could develop a treatment for only two percent of people affected we could perhaps begin to identify small molecules that would help the others.
Dr. Stephen Warren, William Patterson Timmie Professor of Human Genetics, Professor of Biochemistry and Professor of Paediatrics, Emory University School of Medicine, described Fragile X syndrome as a dendritic disorder. This X-linked disorder, which affects about one in 5,000 newborn males, is associated with cognitive impairment and causes autism in about 30 percent of affected children. The genetic abnormality responsible for the syndrome is an excessive repeat length of CGG triplets on the FMR-1 (Fragile Mental Retardation-1) gene. Whereas normal alleles have repeat lengths of 7 - 55, a premutation phenotype occurs with repeats of 55 - 200. As the premutation gene is transmitted down the generations the repeat length increases. Fragile X is associated with repeat lengths of more than 200, when the allele becomes heavily methylated and is silenced. FMR-1 encodes a selective RNA-binding protein, FMRP, which suppresses translation of target RNA. FMRP is associated with ribosomes, especially in dendritic spines where local control of protein synthesis is critical in synaptic plasticity. The absence of FMRP results in over-translation of mRNA in the postsynaptic space following type 1 mGluR stimulation. This causes abnormal AMPA receptor trafficking, which is probably the immediate cause of synaptic defects associated with Fragile X syndrome and perhaps the cognitive impairment itself.
"This idea of a pathway for Fragile X led to a number of thoughts about therapeutic interventions," Dr. Warren explained. Candidate drugs currently include GABA inhibitory agonists and mGluR 5 antagonists and clinical trials are now underway. The strategy of identifying a gene implicated in a phenotype and using it as a toe-hold to explore the basic science means that therapeutic approaches become obvious, Dr. Warren continued. "Using basic science to understand mechanisms with a human disease as a phenotype can lead, sometimes over years, to unexpected insight." Twenty years ago, it would have been difficult to imagine that an orally active small molecule could offer a treatment for a disorder such as Fragile X syndrome, he said.
In 1999, Rett syndrome, an X-linked neurodegenerative disorder, was linked with mutations of a single gene—in this case Methyl-CpG-binding protein2 (MecP2), which controls the expression of biogenic amine synthetic enzymes and amines. Dr. Jeffrey Neul, Cynthia and Anthony Petrello Scholar, Jan and Dan Neurological Research Institute, Texas Children's Hospital and Assistant Medical Director at the Blue Circle Rett Center, Baylor College of Medicine explained that affected girls initially develop normally but undergo a period of regression in which they lose skills and develop autistic features and neurological abnormalities such as seizures and autonomic dysfunction. Evidence from animal models and humans suggests that MecP2 dysfunction in neurone populations associated with different neurotransmitters such as dopamine, serotonin and norepinephrine can explain specific behavioural and neurological abnormalities associated with Rett syndrome. Consistent with this, selective serotonin reuptake inhibitors can be used to alleviate anxiety in girls with Rett syndrome. Additionally, this can improve some of the characteristic features of the disease, such as such as hand wringing and breathing abnormalities, which are worsened in novel environments, although it does not abolish them completely. In contrast, uncontrolled studies have not suggested that levodopa is helpful.
MecP2 mutations are not specific: they are not detected in all patients with Rett syndrome and they have been identified in other neurodevelopmental disorders. Rett syndrome therefore remains a clinical diagnosis. Our understanding of its molecular pathophysiology lags behind that of, for example, Fragile X syndrome and neurofibromatosis, Dr. Neul said. The molecular mechanism of MeCP2 function is still being determined, but it does appear that MeCP2 controls the expression of a large number of genes, including Brain Derived Neurotrophic Factor (BDNF). This insight has lead to preclinical testing of both BDNF and Insulin-like growth factor-1 (IGF-1) in animal models. "Evidence from the mouse shows that IGF-1 does not change the disease course," Dr. Neul commented. "It gets shifted to a later time but they still develop the disease and it's still as severe because the animals still die."
Neurodevelopmental disorders can be reversible
"Probably the most exciting thing to happen in the field of Rett syndrome since identifying the causative gene," is how Dr. Neul described recent work showing that the early effects of MecP2 dysfunction could be reversed and lifespan restored in the mouse. Noting other encouraging data from animal models of Fragile X, neurofibromatosis and tuberous sclerosis, Dr. Insel described reversibility as the most profound idea to emerge from the symposium and, even though the clinical potential is still uncertain the early data seem promising. "It's difficult to identify anything in the last ten years that's quite as transformative as this concept," he said. However, he cautioned against raising unrealistic expectations of medicines and emphasised the fundamental role that training will play. Plasticity appears to be greater than we previously suspected and cognitive training, learning and memory are "synaptic plasticity in action." We should not hope for magic pills that could cure neurodevelopmental disorders but instead aim to develop new treatments that will help individuals benefit more from cognitive retraining and other behavioural interventions.
The conference room at Buonas
Modelling neurodevelopmental disorders
Animal models are essential tools for exploring the impact of gene modification on the manifestations of neurodevelopmental disorders. Some are very close to the human disorder—the null mouse model for Rett syndrome is one example with close but not exact correspondence of behavioural, physical and neurological changes. But even less precise models have proved highly informative about pathophysiological mechanisms and much about resilience can be inferred from the absence of a behavioural phenotype. Furthermore, clinical models are far from perfect due to the wide phenotype variability in humans.
Neurofibromatosis is due to mutations on the neurofibromatosis type 1 (NF1) gene which encodes a p21Ras GTPase activating protein (GAP) and enhances Ras/MAPK signalling in inhibitory neurones in the prefrontal cortex. The resulting increase in GABA release probably causes failure of persistent neuronal activity. Functional imaging studies in patients have demonstrated hypoactivation of the prefrontal cortex which probably results in deficits in working memory and attention. Clinically, this affects cognition causing attention deficits and impairing spatial skills, reading and vocabulary; it is also associated with increased rates of autism and epilepsy.
Modulation of the NF1 gene in the mouse has created a valuable model for understanding these attentional and executive disabilities. Learning and memory deficits in this model are due to impairment of long-term potentiation, explained Professor Alcino Silva. This, he continued, has its origins in excessive ras/MAPK signalling leading to hyperphosphorylation of synapsin 1, leading to enhanced GABA release. Studies in patients with NF1 have shown corresponding changes, including hypoactivity in key brain regions that is consistent with increased GABA-mediated inhibition. In these patients, deficits in working memory are similar to those observed in patients with schizophrenia (and there is evidence of increased prefrontal expression of NF1 in patients with schizophrenia).
These data demonstrate parallels between the mouse model and patients with NF1 in behaviour and regional involvement. The next step, Professor Silva noted, is to determine whether there are also molecular parallels to build a compelling case that the mechanism in mice and patients is the same and from there to develop new treatments. In the mouse model, both picrotoxin and statins reverse enhanced p21Ras activity, rescue the deficit in long term potentiation and reverse the deficit in spatial learning and attention. Pilot studies conducted in Rotterdam have now provided some evidence that statins may reverse cognitive deficits in children with NF1, though the size of this effect depended on the magnitude of the deficit and has proved difficult to demonstrate consistently. A larger trial is now underway to clarify this finding.
Professor Silva described another example of the rescue of neurodevelopmental change by a simple molecule. In the mouse model of tuberous sclerosis, rapamycin reverses learning and memory deficits and corrects underlying biochemical change in adult animals. "This finding is really exciting," Professor Silva said, adding: "It means that even in adults with disorders such as autism or schizophrenia we may be able to ameliorate their phenotypes considerably."
However, the mouse model did not display deficits in social interaction and other behaviours characteristic of autism, prompting the question why do only 40 percent of patients with tuberous sclerosis have autism? There is some evidence that maternal viral infection during the second trimester of pregnancy may be a risk factor for schizophrenia and perhaps autism. Certain strains of mice develop behaviour typical of autism when injected with influenza virus or poly IC (a double-stranded RNA) at prenatal day 12.5. This action is mediated by interleukin-6 (IL-6), a cytokine which is upstream of mTOR (mammalian target of rapamycin) signalling which, in turn, is abnormally enhanced in the mouse model of tuberous sclerosis. IL-6 receptors are expressed in the brain and are involved in plasticity, learning and memory. This finding is supported by experimental data showing that administration of poly IC to pregnant mice induces, in offspring with the mutation for tuberous sclerosis but not in wild-type offspring, social interaction deficits and reduced socialisation consistent with animal models of autism. Poly IC was also associated with a preferential increase in mortality among female pups, raising the possibility that, in humans, the excess prevalence of autism among males may be due to increased female mortality.
"These studies show that it is possible to undergo the voyage from molecule to physiology, to behaviour and to studies in patients and clinical trials," Professor Silva said. "There's a tremendous opportunity to really bridge this huge gulf between molecular and cellular neuroscience and cognitive neuroscience."
Rare variations are common in autism
Autism is a highly heterogeneous developmental brain disorder, possibly but not exclusively due to synaptic dysfunction. Cortical structures are disorganised vertically and horizontally but it is presently unclear whether this is random or has a pattern, or whether abnormal tracts that have been described throughout the brain occur only in more severely handicapped individuals. Genetic studies show that autism is associated with several Mendelian disorders but also with rare, highly penetrant mutations and common alleles with low penetration. However, specific gene mutations are identified in only 20 percent of affected children.
This genetic heterogeneity is matched by the extreme variability in the clinical phenotype of autism and the changes that occur with development, explained Professor Anthony Bailey, Cheryl and Reece Scott Chair of Psychiatry, University of Oxford. Nevertheless, its clinical presentation is similar worldwide. Autism is expressed differently in males and females and it may be associated with developmental regression (which varies even within families); epilepsy affects 25 - 33 percent of cases. Seventy-five percent of affected people have mental retardation but others display high intelligence. Anxiety and depression are common among people with autism and their relatives. "This is an additional burden for individuals to bear which has a huge impact on their daily functioning," Dr. Bailey pointed out.
People with autism achieve adequately in life but use completely different neurological pathways from the norm, and these pathways vary between individuals. "People with autism are not activating the usual parts of the brain," Professor Bailey said. "Sometimes they're activating more parts of the brain, sometimes less. Sometimes the localisation is subtly different, sometimes it moves in a more significant way." We do not understand why some people with autism have mental retardation but such individuals are less likely to be able to utilise alternative neurological pathways and are therefore more likely to be severely affected.
"Development is a very long process and optimising outcomes in autism entails intervention throughout the lifespan," Professor Bailey said. "It's never too late to intervene from a behavioural or pharmacological perspective."
There is wide genetic heterogeneity in neurodevelopmental disorders but the large overlap between autism and mental retardation suggests that they may follow similar genetic rules, said Dr. Christopher Walsh, Bullard Professor of Neurology, Harvard Medical School. Mental retardation is considered a largely Mendelian disorder associated with hundreds of individually rare mutations acting alone with high penetrance. To date, dozens of X-linked genes have been implicated in causing mental retardation but the importance of autosomal genes is increasingly acknowledged. For example, nonsyndromic mental retardation was formerly believed to be an X-linked disorder but now it is clear that the 80 or so X-linked genes identified represent only about 50 percent of families. Most unidentified genes are likely to be autosomal recessive: a handful has been identified to date but the total may exceed 1,000.
Participants on Buonas' terrace during a break.
Recessive mutations causing rare neurodevelopmental abnormalities have been identified in the United States, Europe, the Middle East and Asia; they are typically rare but 'founder' mutations may be common in a single community. They are unpredictable but with hindsight they can be seen to reflect ancient patterns of migration and ethnicity. These mutations can be readily mapped if appropriate families are identified; they are usually point changes in the coding region, though larger deletions account for 5 - 10 percent.
"We applied this same rationale to identify genes on the autosomes that may be associated with autism and assessed their role in various populations around the world," Dr. Walsh said. The Homozygosity Mapping Collaborative for Autism has so far ascertained about 200 families throughout the Middle East of which three-quarters, mostly with a single child, are consanguineous. Individual families show linkage to different chromosomes, confirming the expectation that autism would be associated with many genes causing a similar phenotype. Spontaneous large deletions on the chromosome are common causes of autism or mental retardation in many children but in consanguinous families are relatively less frequent than autosomal recessive mutations. The candidate autism genes identified to date differ in their locations and target. They are highly expressed in the hippocampus and their expression is regulated by neuronal activity. They probably have a role in hippocampal synapse plasticity, providing a mechanism for their effects on learning. Some of the deletions identified remove genes entirely whereas others remove noncoding near genes including predicted sites for transcription factors.
"Some of these autism-related mutations may not be null alleles but may be more like conditional or partial loss of function mutations," Dr. Walsh concluded. "That might explain—perhaps—why some children with autism have milder phenotypes, or strengths in certain areas, or later onset of symptoms."
Some cases of autism spectrum disorders are associated with a genetic mutation causing abnormal synaptogenesis. Professor Thomas Bourgeron, Head of Human Genetics and Cognitive Functions, Institut Pasteur, explained that the genes encode for the cell adhesion molecules neuroligin 3 and 4 and neurexin 1 or the scaffolding proteins SHANK2 and 3, all of which have a role in the formation and maintenance of the synapse. Animal models have demonstrated that mutations of neuroligin 3 or 4 are associated with reduced social interaction and ultrasonic vocalisations. Dr. Bourgeron also described mutations of cell adhesion molecules contactin 6, C18 and P9 associated with Asperger's syndrome and high-functioning autism; these molecules are important for axon outgrowth and guidance.
Autism was first associated with low melatonin secretion in 1995 but the cause was unknown and the observation received little attention at the time. Dr. Bourgeron's team identified a deletion in the ASMT gene in some children with autism; ASMT encodes for the last enzyme in melatonin synthesis and is located on pseudo-autosomal region 1. "We found a huge decrease in melatonin in our children compared with controls," Dr. Bourgeron said. "Sixty percent of children had low melatonin activity." Several ASMT mutations have now been identified that decrease gene activity.
Melatonin has been shown to increase and decrease GABA function in different regions of the brain and to modulate the formation of neurites and memory. Low melatonin production occurs in the absence of autism or sleep disorders. However, the sleep-wake cycle may be abnormal in people with autism and several clinical trials have shown that this can be corrected, and sleep improved, by administration of melatonin. The MENDS project (use of MElatonin in children with Neurodevelopmental Disorders and impaired Sleep), involving 350 children, of whom 170 have autism, is now investigating the effects of melatonin on sleep and cognitive function.
These findings suggest that some cases of autism may be associated with altered homeostasis or imbalance of excitatory and inhibitory synaptic currents in some regions of the brain, which may affect or amplify altered circadian rhythms. However, it is important to remember that the interactions of genes and environment are different for each person and that individuals will need treatments appropriate to their particular needs.
Translational medicine is a new science
Dr. Insel noted that the identity of genes associated with some disorders has been known for many years but for others, such as autism, these are still early days. Hopefully, he said, future progress will be rapid. Stem cells will help us understand the biology of common and rare genetic variants and allow us to investigate multiple variants and developmental processes. This will improve screening for small molecules with therapeutic potential, reducing reliance on off-the-shelf compounds and offering the prospect of innovative treatments with greater specificity.
In the case of autism there is a need to develop biomarkers for early detection that will enable us to make pre-emptive interventions. "Personalised treatments are really the holy grail," Dr. Insel said. "What patients want to know is what's good for me? What's going to help my child?" The clinical heterogeneity of autism is a real opportunity to achieve this, he concluded.
