The development of the brain involves many thousands of different gene products acting in hundreds of distinct molecular and cellular processes, all tightly coordinated in space and time – from patterning and proliferation to cell migration, axon guidance, synapse formation and many others. Large numbers of proteins are involved in the biochemical pathways and networks underlying each cell biological process. Each of these systems has evolved not just to do a particular job, but to do it robustly – to make sure this process happens even in the face of diverse challenges.

Robustness is an emergent and highly adaptive property of complex systems that can be selected for in response to particular pressures. These include extrinsic factors, such as variability in temperature, supply of nutrients, etc., but also intrinsic factors. A major source of intrinsic variation is noise in gene expression – random fluctuations in the levels of all proteins in all cells. These fluctuations arise due to the probabilistic nature of gene transcription – whether a messenger RNA is actively being made from a gene at any particular moment. The system must be able to deal with these fluctuations and it can be argued that the noise in the system actually acts as a buffer. If the system only worked within a narrow operating range for each component then it would be very vulnerable to failure of any single part.

Natural selection will therefore favour system architectures that are more robust to environmental and intrinsic variation. In the process, such systems also indirectly become robust to the other major source of variation – mutations.

Many individual components can be deleted entirely with no discernible effect on the system (which is why looking exhaustively for a phenotype in mouse mutants can be so frustrating – many gene knockouts are irritatingly normal). You could say that if the knockout of a gene does not affect a particular process, that that means the gene product is not actually involved in that process, but that is not always the case. One can often show that a protein is involved biochemically and even that the system is sensitive to changes in the level of that protein – increased expression can often cause a phenotype even when loss-of-function manipulations do not.

Direct evidence for robustness of neurodevelopmental systems comes from examples of genetic background effects on phenotypes caused by specific mutations. While many components of the system can be deleted without effect, others do cause a clear phenotype when mutated. However, such phenotypes are often modified by the genetic background. This is commonly seen in mouse experiments, for example, where the effect of a mutation may vary widely when it is crossed into various inbred strains. The implication is that there are some genetic differences between strains that by themselves have no effect on the phenotype, but that are clearly involved in the system or process, as they strongly modify the effect of another mutation.

How is this relevant to understanding so-called complex disorders? There are two schools of thought on the genetic architecture of these conditions. One considers the symptoms of, say, autism or schizophrenia or epilepsy as the consequence of mutation in any one of a very large number of distinct genes. This is the scenario for intellectual disability, for example, and also for many other conditions like inherited blindness or deafness. There are hundreds of distinct mutations that can result in these symptoms. The mutations in these cases are almost always ones that have a dramatic effect on the level or function of the encoded protein.

The other model is that complex disorders arise, in many cases, due to the combined effects of a very large number of common polymorphisms – these are bases in the genome where the sequence is variable in the population (e.g., there might be an “A” in some people but a “G” in others). The human genome contains millions of such sites and many consider the specific combination of variants that each person inherits at these sites to be the most important determinant of their phenotype. (I disagree, especially when it comes to disease). The idea for disorders such as schizophrenia is that at many of these sites (perhaps thousands of them), one of the variants may predispose slightly to the illness. Each one has an almost negligible effect alone, but if you are unlucky enough to inherit a lot of them, then the system might be pushed over the level of burden that it can tolerate, into a pathogenic state.

These are the two most extreme positions – there are also many models that incorporate effects of both rare mutations and common polymorphisms. Models incorporating common variants as modifiers of the effects of rare mutations make a lot of biological sense. What I want to consider here is the model that the disease is caused in some individuals purely by the combined effects of hundreds or thousands of common variants (without what I call a “proper mutation”).

Ironically, robustness has been invoked by both proponents and opponents of this idea. I have argued that neurodevelopmental systems should be robust to the combined effects of many variants that have only very tiny effects on protein expression or function (which is the case for most common variants). This is precisely because the system has evolved to buffer fluctuations in many components all the time. In addition to being an intrinsic, passive property of the architecture of developmental networks, robustness is also actively promoted through homeostatic feedback loops, which can maintain optimal performance in the face of variations, by regulating the levels of other components to compensate. The effects of such variants should therefore NOT be cumulative – they should be absorbed by the system. (In fact, you could argue that a certain level of noise in the system is a “design feature” because it enables this buffering).

Others have argued precisely the opposite – that robustness permits cryptic genetic variation to accumulate in populations. Cryptic genetic variation has no effect in the context in which it arises (allowing it to escape selection) but, in another context – say in a different environment, or a different genetic background – can have a large effect. This is exactly what robustness allows to happen – indeed, the fact that cryptic genetic variation exists provides some of the best evidence that we have that the systems are robust as it shows directly that mutations in some components are tolerated in most contexts. But is there any evidence that such cryptic variation comprises hundreds or thousands of common variants?

To be fair, proving that is the case would be very difficult. You could argue from animal breeding experiments that the continuing response to selection of many traits means that there must be a vast pool of genetic variation that can affect them, which can be cumulatively enriched by selective breeding, almost ad infinitum. However, new mutations are known to make at least some contribution to this continued response to selection. In addition, in most cases where the genetics of such continuously distributed traits have been unpicked (by identifying the specific factors contributing to strain differences for example) they come down to perhaps tens of loci showing very strong and complex epistatic interactions (1, 2, 3). Thus, just because variation in a trait is multigenic, does not mean it is affected by mutations of small individual effect – an effectively continuous distribution can emerge due to very complex epistatic interactions between a fairly small number of mutations which have surprisingly large effects in isolation.

(I would be keen to hear of any examples showing real polygenicity on the level of hundreds or thousands of variants).

In the case of genetic modifiers of specific mutations – say, where a mutation causes a very different phenotype in different mouse strains – most of the effects that have been identified have been mapped to one or a small number of mutations which have no effect by themselves, but which strongly modify the phenotype caused by another mutation.

These and other findings suggest that (i) cryptic genetic variation relevant to disease is certainly likely to exist and to have important effects on phenotype, but that (ii) such genetic background effects can most likely be ascribed to one, several, or perhaps tens of mutations, as opposed to hundreds or thousands of common polymorphisms.

This is already too long, but it begs the question: if neurodevelopmental systems are so robust, then why do we ever get neurodevelopmental disease? The paradox of systems that are generally robust is that they may be quite vulnerable to large variation in a specific subset of components. Why specific types of genes are in this set, while others can be completely deleted without effect, is the big question. More on that in a subsequent post…

It takes a lot of genes to wire the human brain. Billions of cells, of a myriad different types have to be specified, directed to migrate to the right position, organised in clusters or layers, and finally connected to their appropriate targets. When the genes that specify these neurodevelopmental processes are mutated, the result can be severe impairment in function, which can manifest as neurological or psychiatric disease.

How those kinds of neurodevelopmental defects actually lead to the emergence of particular pathological states – like psychosis or seizures or social withdrawal – is a mystery, however. Many researchers are trying to tackle this problem using mouse models – animals carrying mutations known to cause autism or schizophrenia in humans, for example. A recent study from my own lab (open access in PLoS One) adds to this effort by examining the consequences of mutation of an important neurodevelopmental gene and providing evidence that the mice end up in a state resembling psychosis. In this case, we start with a discovery in mice as an entry point to the underlying neurodevelopmental processes.

In just the past few years, over a hundred different mutations have been discovered that are believed to cause disorders like autism or schizophrenia. In many cases, particular mutations can actually predispose to many different disorders, having been linked in different patients to ADHD, epilepsy, mental retardation or intellectual disability, Tourette’s syndrome, depression, bipolar disorder and others. These clinical categories may thus represent more or less distinct endpoints that can arise from common neurodevelopmental origins.

For a condition like schizophrenia, the genetic overlap with other conditions does not invalidate the clinical category. There is still something distinctive about the symptoms of this disorder that needs to be explained. I have argued that schizophrenia can clearly be caused by single mutations in any of a very large number of different genes, many with roles in neurodevelopment. If that model is correct, then the big question is: how do these presumably diverse neurodevelopmental insults ultimately converge on that specific phenotype? It is, after all, a highly unusual condition. The positive symptoms of psychosis – hallucinations and delusions, for example – especially require an explanation. If we view the brain from an engineering perspective, then we can say that the system is not just not working well – it is failing in a particular and peculiar manner.

To try to address how this kind of state can arise we have been investigating a particular mouse – one with a mutation in a gene called Semaphorin-6A. This gene encodes a protein that spans the membranes of nerve cells, acting in some contexts as a signal to other cells and in other contexts as a receptor of information. It has been implicated in controlling cell migration, the guidance of growing axons, the specification of synaptic connectivity and other processes. It is deployed in many parts of the developing brain and required for proper development in the cerebral cortex, hippocampus, thalamus, cerebellum, retina, spinal cord, and probably other areas we don’t yet know about.

Despite widespread cellular disorganisation and miswiring in their brains, Sema6A mutant mice seem overtly pretty normal. They are quite healthy and fertile and a casual inspection would not pick them out as different from their littermates. However, more detailed investigation revealed electrophysiological and behavioural differences that piqued our interest.

Because these animals have a subtly malformed hippocampus, which looks superficially like the kind of neuropathology observed in many cases of temporal lobe epilepsy, we wanted to test if they had seizures. To do this we attached electrodes to their scalp and recorded their electroencephalogram (or EEG). This technique measures patterned electrical activity in the underlying parts of the brain and showed quite clearly that these animals do not have seizures. But it did show something else – a generally elevated amount of activity in these animals all the time.

What was particularly interesting about this is that the pattern of change (a specific increase in alpha frequency oscillations) was very similar to that reported in animals that are sensitised to amphetamine – a well-used model of psychosis in rodents. High doses of amphetamine can acutely induce psychosis in humans and a suite of behavioural responses in rodents. In addition, a regimen of repeated low doses of amphetamine over an extended time period can induce sensitisation to the effects of this drug in rodents, characterised by behavioural differences, like hyperlocomotion, as well as the EEG differences mentioned above. Amphetamine is believed to cause these effects by inducing increases in dopaminergic signaling, either chronically, or to acute stimuli.

This was of particular interest to us, as that kind of hyperdopaminergic state is thought to be a final common pathway underlying psychosis in humans. Alterations in dopamine signaling are observed in schizophrenia patients (using PET imaging) and also in all relevant animal models so far studied.

“Scientists discover gene for autism” (or ovarian cancer, or depression, cocaine addiction, obesity, happiness, height, schizophrenia… and whatever you’re having yourself). These are typical newspaper headlines (all from the last year) and all use the popular shorthand of “a gene for” something. In my view, this phrase is both lazy and deeply misleading and has caused widespread confusion about what genes are and do and about their influences on human traits and disease.

The problem with this phrase stems from the ambiguity in what we mean by a “gene” and what we mean by “for”. These can mean different things at different levels and unfortunately these meanings are easily conflated. First, a gene can be defined in several different ways. From a molecular perspective, it is a segment of DNA that codes for a protein, along with the instructions for when and where and in what amounts this protein should be made. (Some genes encode RNA molecules, rather than proteins, but the general point is the same). The function of the gene on a cellular level is thus to store the information that allows this protein to be made and its production to be regulated. So, you have a gene for haemoglobin and a gene for insulin and a gene for rhodopsin, etc., etc. (around 25,000 such genes in the human genome). The question of what the gene is for then becomes a biochemical question – what does the encoded protein do?

But that is not the only way or probably even the main way that people think about what genes do – it is certainly not how geneticists think about it. The function of a gene is commonly defined (indeed often discovered) by looking at what happens when it is mutated – when the sequence of DNA bases that make up the gene is altered in some way which affects the production or activity of the encoded protein. The visible manifestation of the effect of such a mutation (the phenotype) is usually defined at the organismal level – altered anatomy or physiology or behaviour, or often the presence of disease. From this perspective, the gene is defined as a separable unit of heredity – something that can be passed on from generation to generation that affects a particular trait. This is much closer to the popular concept of a gene, such as a gene for blue eyes or a gene for breast cancer. What this really means is a mutation for blue eyes or a mutation for breast cancer.

The challenge is in relating the function of a gene at a cellular level to the effects of variation in that gene, which are most commonly observed at the organismal level. The function at a cellular level can be defined pretty directly (make protein X) but the effect at the organismal level is much more indirect and context-dependent, involving interaction with many other genes that also contribute to the phenotype in question, often in highly complex and dynamic systems.

If you are talking about a simple trait like blue eyes, then the function of the gene at a molecular level can actually be related to the mutant phenotype fairly easily – the gene encodes an enzyme that makes a brown pigment. When that enzyme is not made or does not work properly, the pigment is not made and the eyes are blue. Easy-peasy.

But what if the phenotype is in some complex physiological trait, or even worse, a psychological or behavioural trait? These traits are often defined at a very superficial level, far removed from the possible molecular origins of individual differences. The neural systems underlying such traits may be incredibly complex – they may break down due to very indirect consequences of mutations in any of a large number of genes.

For example, mutations in the genes encoding two related proteins, neuroligin-3 and neuroligin-4 have been found in patients with autism and there is good evidence that these mutations are responsible for the condition in those patients. Does this make them “genes for autism”? That phrase really makes no sense – the function of these genes is certainly not to cause autism, nor is it to prevent autism. The real link between these genes and autism is extremely indirect. The neuroligin proteins are involved in the formation of synaptic connections between neurons in the developing brain. If they are mutated, then the connections that form between specific types of neurons are altered. This changes the function of local circuits in the brain, affecting their information-processing parameters and changing how different regions of the brain communicate. Ultimately, this impacts on neural systems controlling things like social behaviour, communication and behavioural flexibility, leading to the symptoms that define autism at the behavioural level.

So, mutations in these genes can cause autism, but these are not genes for autism. They are not even usefully or accurately thought of as genes for social behaviour or for cognitive flexibility – they are required, along with the products of thousands of other genes, for those faculties to develop.

But perhaps there are other genetic variants in the population that affect the various traits underlying these faculties – not in such a severe way as to result in a clinical disorder, but enough to cause the observed variation across the general population. It is certainly true that traits like extraversion are moderately heritable – i.e., a fair proportion of the differences between people in this trait are attributable to genetic differences. When someone asks “are there genes for extraversion?”, the answer is yes if they mean “are differences in extraversion partly due to genetic differences?”. If they mean the function of some genetic variant is to make people more or less extroverted, then they have suddenly (often unknowingly) gone from talking about the activity of a gene or the effect of mutation of that gene to considering the utility of a specific variant.

This suggests a deeper meaning – not just that the gene has a function, but that it has a purpose – in biological terms, this means that a particular version of the gene was selected for on the basis of its effect on some trait. This can be applied to the specific sequence of a gene in humans (as distinct from other animals) or to variants within humans (which may be specific to sub-populations or polymorphic within populations).

While geneticists may know what they mean by the shorthand of “genes for” various traits, it is too easily taken in different, unintended ways. In particular, if there are genes “for” something, then many people infer that the something in question is also “for” something. For example, if there are “genes for homosexuality”, the inference is that homosexuality must somehow have been selected for, either currently or under some ancestral conditions. Even sophisticated thinkers like Richard Dawkins fall foul of this confusion – the apparent need to explain why a condition like homosexual orientation persists. Similar arguments are often advanced for depression or schizophrenia or autism – that maybe in ancestral environments, these conditions conferred some kind of selective advantage. That is one supposed explanation for why “genes for schizophrenia or autism” persist in the population.

Natural selection is a powerful force but that does not mean every genetic variation we see in humans was selected for, nor does it mean every condition affecting human psychology confers some selective advantage. In fact, mutations like those in the neuroligin genes are rapidly selected against in the population, due to the much lower average number of offspring of people carrying them. The problem is that new ones keep arising – in those genes and in thousands of other required to build the brain. By analogy, it is not beneficial for my car to break down – this fact does not require some teleological explanation. Breaking down occasionally in various ways is not a design feature – it is just that highly complex systems bring an associated higher risk due to possible failure of so many components.

So, just because the conditions persist at some level does not mean that the individual variants causing them do. Most of the mutations causing disease are probably very recent and will be rapidly selected against – they are not “for” anything.

Jamain S, Quach H, Betancur C, Råstam M, Colineaux C, Gillberg IC, Soderstrom H, Giros B, Leboyer M, Gillberg C, Bourgeron T, & Paris Autism Research International Sibpair Study (2003). Mutations of the X-linked genes encoding neuroligins NLGN3 and NLGN4 are associated with autism. Nature genetics, 34 (1), 27-9 PMID: 12669065

That this discovery was made using classical “cytogenetic” techniques (physically looking at the chromosomes down a microscope) and in a single family is somehow pleasing in an age where massive molecular population-based studies are in vogue. (A win for “small” science).

The discovery of the DISC1 translocation clearly showed that disruption of a single gene could lead to psychiatric disorders like schizophrenia. This was a challenge to the idea that these disorders were “polygenic” – caused by the inheritance in each individual of a large number of genetic variants. As more and more mutations in other genes are being found to cause these disorders, the DISC1 situation can no longer be dismissed as an exception – it is the norm.

It also was the first example of a principle that has since been observed for many other genes – namely that the effects of the mutation can manifest quite variably – not as one specific disorder, but as different ones in different people. Indeed, DISC1 has since been implicated in autism as well as adult-onset disorders. It is now clear from this and other evidence that these apparently distinct conditions are best thought of as variable outcomes that arise, in many cases at least, from disturbances of neurodevelopment.

Since the initial discovery, major research efforts of a growing number of labs have been focused on the next obvious questions: what does DISC1 do? And what happens when it is mutated? What happens in the brain that can explain why psychiatric symptoms result?

We now know that DISC1 has many different functions. It is a cytoplasmic protein – localised inside the cell – that interacts with a very large number of other proteins and takes part in diverse cellular functions, including cell migration, outgrowth of nerve fibres, the formation of dendritic spines (sites of synaptic contact between neurons), neuronal proliferation and regulation of biochemical pathways involved in synaptic plasticity. Many of the proteins that DISC1 interacts with have also been implicated in psychiatric disease.

This new study adds another possible function, and a dramatic and unexpected one at that. This function was discovered from an independent angle, by researchers studying how the two hemispheres of the brain get connected – or more specifically, why they sometimes fail to be connected. The cerebral hemispheres are normally connected by millions of axons which cross the midline of the brain in a structure called the corpus callosum (or “tough body” – (don’t ask)). Very infrequently, people are born without this structure – the callosal axons fail to cross the midline and the two hemispheres are left without this major route of communication (though there are other routes, such as the anterior commissure).

The frequency of agenesis of the corpus callosum has been estimated at between 1 in 1,000 and 1 in 6,000 live births – thankfully very rare. It is associated with a highly variable spectrum of other symptoms, including developmental delay, autistic symptoms, cognitive disabilities extending into the range of mental retardation, seizures and other neurological signs.

Elliott Sherr and colleagues were studying patients with this condition, which is very obvious on magnetic resonance imaging scans (see Figure). They initially found a mother and two children with callosal agenesis who all carried a deletion on chromosome 1, at position 1q42 – exactly where DISC1 is located. They subsequently identified another patient with a similar deletion, which allowed them to narrow down the region and identify DISC1 as a plausible candidate (among some other genes in the deleted region). Because the functions of proteins can be affected not just by large deletions or translocations but also by less obvious mutations that change a single base of DNA, they also sequenced the DISC1 gene in a cohort of callosal agenesis patients and found a number carrying novel mutations that are very likely to disrupt the function of the gene.

While not rock-solid evidence that it is DISC1 that is responsible, these data certainly point to it as the strongest candidate to explain the callosal defect. This hypothesis is strongly supported by findings from DISC1 mutant mice (carrying a mutation that mimics the effect of the human translocation), which also show defects in formation of the corpus callosum. In addition, the protein is strongly expressed in the axons that make up this structure at the time of its development.

The most obvious test of whether disruption of DISC1 really causes callosal agenesis is to look in the people carrying the initial translocation. Remarkably, it is not known whether the original patients in the Scottish pedigree who carry the DISC1 translocation show this same obvious brain structural phenotype. They have, very surprisingly, never been scanned.

This new paper raises the obvious hypothesis that the failure to connect the two hemispheres results in the psychiatric or cognitive symptoms, which variously include reduced intellectual ability, autism and schizophrenia. This seems like too simplistic an interpretation, however. All we have now is a correlation. First, the implication of DISC1 in the acallosal phenotype is not yet definitive – this must be nailed down and replicated. But even if it is shown that disruption of DISC1 causes both callosal agenesis and schizophrenia (or other psychiatric disorders or symptoms), this does not prove a causal link. DISC1 has many other functions and is expressed in many different brain areas (ubiquitously in fact). Any, or indeed, all of these functions may in fact be the cause of psychopathology.

One prediction, if it were true that the lack of connections between the two hemispheres is causal, is that we would expect the majority of patients with callosal agenesis to have these kinds of psychiatric symptoms. In fact, the rates are indeed very high – in different studies it has been estimated that up to 40% of callosal agenesis patients have an autism diagnosis, while about 8% have the symptoms of schizophrenia or bipolar disorder. (Of course, these patients may have other, less obvious brain defects as well, so even this is not definitive).

Conversely, we might naively expect a high rate of callosal agenesis in patients with autism or schizophrenia. However, we know these disorders are extremely heterogeneous and so it is much more likely that this phenotype might be apparent in only a specific (possibly very small) subset of patients. This may indeed be the case – callosal agenesis has been observed in about 3 out of 200 schizophrenia patients (a vastly higher rate than in the general population). Another study, just published, has found that mutations in a different gene – ARID1B – are also associated with callosal agenesis, mental retardation and autism. More generally, there may be subtle reductions in callosal connectivity in many schizophrenia or autism patients (including some autistic savants).

Whether this defect can explain particular symptoms is not yet clear. For the moment, the new study provides yet another possible function of DISC1, and highlights an anatomical phenotype that is apparently present in a subset of autism and schizophrenia cases and that can arise due to mutation in many different genes (of which DISC1 and ARID1B are only two of many known examples).

One final note: formation of the corpus callosum is a dramatic example of a process that is susceptible to developmental variation. What I mean is this: when patients inherit a mutation that results in callosal agenesis, this phenotype occurs in some patients but not all. This is true even in genetically identical people, like monozygotic twins or triplets (or in lines of genetically identical mice). Though the corpus callosum contains millions of nerve fibres, the initial events that establish it involve very small numbers of cells. These cells, which are located at the medial edge of each cerebral hemisphere, must contact each other to enable the fusion of the two hemispheres, forming a tiny bridge through which the first callosal fibres can cross. Once these are across, the rest seem able to follow easily. Because this event involves very few cells at a specific time in development, it is susceptible to random “noise” – fluctuations in the precise amounts of various proteins in the cells, for example. These are not caused by external forces – the noise is inherent in the system. The result is that, in some people carrying such a mutation the corpus callosum will not form at all, while in others it forms apparently completely normally (see figure of triplets, one on left with normal corpus callosum, the other two with it absent). So, an all-or-none effect can arise, without any external factors involved.

This same kind of intrinsic developmental variation may also explain or at least contribute to the variability in phenotypic outcome at the level of psychiatric symptoms when these kinds of neurodevelopmental mutations are inherited. Even monozygotic twins are often discordant for psychiatric diagnoses (concordance for schizophrenia is about 50%, for example). This is often assumed to be due to non-genetic and therefore “environmental” or experiential factors. If these disorders really arise from differences in brain wiring, which we know are susceptible to developmental variation, then differences in the eventual phenotype could actually be completely intrinsic and innate.

Osbun N, Li J, O’Driscoll MC, Strominger Z, Wakahiro M, Rider E, Bukshpun P, Boland E, Spurrell CH, Schackwitz W, Pennacchio LA, Dobyns WB, Black GC, & Sherr EH (2011). Genetic and functional analyses identify DISC1 as a novel callosal agenesis candidate gene. American journal of medical genetics. Part A, 155 (8), 1865-76 PMID: 21739582

Halgren C, Kjaergaard S, Bak M, Hansen C, El-Schich Z, Anderson CM, Henriksen KF, Hjalgrim H, Kirchhoff M, Bijlsma EK, Nielsen M, den Hollander NS, Ruivenkamp CA, Isidor B, Le Caignec C, Zannolli R, Mucciolo M, Renieri A, Mari F, Anderlid BM, Andrieux J, Dieux A, Tommerup N, & Bache I (2011). Corpus Callosum Abnormalities, Mental Retardation, Speech Impairment, and Autism in Patients with Haploinsufficiency of ARID1B. Clinical genetics PMID: 21801163

The genome sequence (a simple linear code made up of 3 billion bases of DNA in precise order, chopped up onto different chromosomes) is peppered with these polymorphic positions – about 1 in every 1,250 bases. That makes about 2,400,000 polymorphisms in each genome (and we each carry two copies of the genome). That certainly seems like plenty of raw material, with limitless combinations that could explain the richness of human diversity. This interpretation has fuelled massive scientific projects to try and find which common polymorphisms affect which traits. (Not to mention personal genomics companies who will try to tell you your risk of various diseases based on your profile of such polymorphisms).

The problem with this view is that it is wrong. Or at least woefully incomplete.

The reason is it ignores another source of variation: very rare mutations in those bases that are constant across the vast majority of individuals. There is now very good evidence that it is those kinds of mutations that contribute most to our individuality. Certainly, they are much more likely to affect a protein’s function and much more likely to contribute to genetic disease. We each carry hundreds of such rare mutations that can affect protein function or expression and are much more likely to have a phenotypic impact than common polymorphisms.

Indeed, far from most of the genome being effectively constant, it can be estimated that every position in the genome has been mutated many, many times over in the human population. And each of us carries hundreds of new mutations that arose during generation of the sperm and egg cells that fused to form us. New mutations may spread in the pedigree or population in which they arise for some time, depending in part on whether they have a deleterious effect or not. Ones that do will likely be quickly selected against.

A new paper from the 1000 genomes project consortium shows that:

“the vast majority of human variable sites are rare and that the majority of rare variants exhibit, at most, very little sharing among continental populations”.

This is a much more fluid picture of genetic variation than we are used to. We are not all dealt a genetic hand from the same deck – each population, sub-population, kindred, nuclear family has a distinct set of rare genetic variants. And each of these decks contains a lot of jokers – the new mutations that arise each time a hand is dealt.

Why have such rare mutations generally been ignored while the polymorphic sites have been the focus of intense research? There are several reasons, some practical and some theoretical. Practically, it has until recently been almost impossible to systematically find very rare mutations. To do so requires that we sequence the whole genome, which has only recently become feasible. In contrast, methods to survey which bases you carry at all the polymorphic sites across the genome were developed quite some time ago now and are relatively cheap to use. (They rely on sampling about 500,000 such sites around the genome – because of unevenness in the way different bits of chromosomes get swapped when sperm and eggs are made, this sample actually tells you about most of the variable sites across the whole genome). So, there has been a tendency to argue that polymorphic sites will be major contributors to human phenotypes (especially diseases) because those have been the only ones we have been able to look at.

Unfortunately, the results of genome-wide association studies, which aim to identify common variants associated with traits or diseases, have been disappointing. This is especially true for disorders with large effects on fitness, such as schizophrenia or autism. Some variants have been found but their effects, even in combination are very small. Most of the heritability of most of the traits or diseases examined to date remains unexplained. (There are some important exceptions, especially for diseases that strike only late in life and for things like drug responses, where selective pressures to weed out deleterious alleles are not at play).

In contrast, many more rare mutations causing disease are being discovered all the time, and the pace of such discoveries is likely to increase with technological advances. The main message that emerges from these studies has been called by Mary-Claire King the “Anna Karenina principle”, based on Tolstoy’s famous opening line:

“Happy families are all alike; every unhappy family is unhappy in its own way”

But can such rare variants really explain the “missing heritability” of these disorders? Some people have argued that they cannot, but this seems to me to be based on a pervasive misconception of how the heritability of a trait is measured and what it means. According to this misconception, if a trait is heritable across the population, that heritability cannot be accounted for by rare variants. After all, if a mutation only occurs in one or a few individuals, it could only minimally (nearly negligibly) contribute to heritability across the whole population. That is true. However, heritability is not measured across the population – it is measured in families and then averaged across the population.

In humans, it is usually derived by comparing phenotypes between people of different genetic relatedness (identical versus fraternal twins, siblings, parents, cousins, etc.). The values of these comparisons are then averaged across large numbers of pairs to allow estimates of how much genetic variance affects phenotypic variance – the population heritability. While a specific rare mutation may only affect the phenotype within a single family, such mutations could, collectively, explain all of the heritability. Completely different sets of mutations could be affecting the trait or causing the disease in different families.

The next few years will reveal the true impact of rare mutations. We should certainly expect complex genetic interactions and some real effects of common polymorphisms. But the idea that our traits are determined simply by the combination of variants we inherit from a static pool in the population is no longer tenable. We are each far more unique than that.

(And if your personal genomics company isn’t offering to sequence your whole genome, it’s not personal enough).

Gravel S, Henn BM, Gutenkunst RN, Indap AR, Marth GT, Clark AG, Yu F, Gibbs RA, The 1000 Genomes Project, & Bustamante CD (2011). Demographic history and rare allele sharing among human populations. Proceedings of the National Academy of Sciences of the United States of America, 108 (29), 11983-11988 PMID: 21730125

Walsh CA, & Engle EC (2010). Allelic diversity in human developmental neurogenetics: insights into biology and disease. Neuron, 68 (2), 245-53 PMID: 20955932

McClellan, J., & King, M. (2010). Genetic Heterogeneity in Human Disease Cell, 141 (2), 210-217 DOI: 10.1016/j.cell.2010.03.032

Mirrored from Wiring the Brain

However, many perfectly healthy people also regularly hear voices – as many as 1 in 25 according to some studies, and in most cases these experiences are perfectly benign. In fact, we all hear voices “belonging to other people” when we dream – we can converse with these voices, waiting for their responses as if they were derived from external agents. Of course, these percepts are actually generated by the activity of our own brain, but how?

There is good evidence from neuroimaging studies that the same areas that respond to external speech are active when people are having these kinds of auditory hallucinations. In fact, inhibiting such areas using transcranial magnetic stimulation may reduce the occurrence or intensity of heard voices. But why would the networks that normally process speech suddenly start generating outputs by themselves? Why would these outputs be organised in a way that fits speech patterns, as opposed to random noise? And, most importantly, why does this tend to occur in people with schizophrenia? What is it about the pathology of this disorder that makes these circuits malfunction in this specific way?

An interesting approach to try and get answers to these questions has been to model these circuits in artificial neural networks. If you can generate a network that can process speech inputs and find certain conditions under which it begins to spontaneously generate outputs, then you may have an informative model of auditory hallucinations. Using this approach, a couple of studies from several years ago from the group of Ralph Hoffman have found some interesting clues as to what may be going on, at least on an abstract level.

Their approach was to generate an artificial neural network that could process speech inputs. Artificial neural networks are basically sets of mathematical functions modelled in a computer programme. They are designed to simulate the information-processing functions carried out by individual neurons and, more importantly, the computational functions carried out by an interconnected network of such neurons. They are necessarily highly abstract, but they can recapitulate many of the computational functions of biological neural networks. Their strength lies in revealing unexpected emergent properties of such networks.

The particular network in this case consisted of three layers of neurons – an input layer, an output layer, and a “hidden” layer in between – along with connections between these elements (from input to hidden and from hidden to output, but crucially also between neurons within the hidden layer). “Phonetic” inputs were fed into the input layer – these consisted of models of speech sounds constituting grammatical sentences. The job of the output layer was to report what was heard – representing different sounds by patterns of activation of its forty-three neurons. Seems simple, but it’s not. Deciphering speech sounds is actually very difficult as individual phonetic elements can be both ambiguous and variable. Generally, we use our learned knowledge of the regularities of speech and our working memory of what we have just heard to anticipate and interpret the next phonemes we hear – forcing them into recognisable categories. Mimicking this function of our working memory is the job of the hidden layer in the artificial neural network, which is able to represent the prior inputs by the pattern of activity within this layer, providing a context in which to interpret the next inputs.

The important thing about neural networks is they can learn. Like biological networks, this learning is achieved by altering the strengths of connections between pairs of neurons. In response to a set of inputs representing grammatical sentences, the network weights change in such a way that when something similar to a particular phoneme in an appropriate context is heard again, the pattern of activation of neurons representing that phoneme is preferentially activated over other possible combinations.

The network created by these researchers was an able student and readily learned to recognise a variety of words in grammatical contexts. The next thing was to manipulate the parameters of the network in ways that are thought to model what may be happening to biological neuronal networks in schizophrenia.

There are two major hypotheses that were modelled: the first is that networks in schizophrenia are “over-pruned”. This fits with a lot of observations, including neuroimaging data showing reduced connectivity in the brains of people suffering with schizophrenia. It also fits with the age of onset of the florid expression of this disorder, which is usually in the late teens to early twenties. This corresponds to a period of brain maturation characterised by an intense burst of pruning of synapses – the connections between neurons.

In schizophrenia, the network may have fewer synapses to begin with, but not so few that it doesn’t work well. This may however make it vulnerable to this process of maturation, which may reduce its functionality below a critical threshold. Alternatively, the process of synaptic pruning may be overactive in schizophrenia, damaging a previously normal network. (The evidence favours earlier disruptions).

The second model involves differences in the level of dopamine signalling in these circuits. Dopamine is a neuromodulator – it alters how neurons respond to other signals – and is a key component of active perception. It plays a particular role in signalling whether inputs match top-down expectations derived from our learned experience of the world. There is a wealth of evidence implicating dopamine signalling abnormalities in schizophrenia, particularly in active psychosis. Whether these abnormalities are (i) the primary cause of the disease, (ii) a secondary mechanism causing specific symptoms (like psychosis), or (iii) the brain attempting to compensate for other changes is not clear.

Both over-pruning and alterations to dopamine signalling could be modelled in the artificial neural network, with intriguing results. First, a modest amount of pruning, starting with the weakest connections in the network, was found to actually improve the performance of the network in recognising speech sounds. This can be understood as an improvement in the recognition and specificity of the network for sounds which it had previously learned and probably reflects the improvements seen in human language learners, along with the concomitant loss in ability to process or distinguish unfamiliar sounds (like “l” and “r” for Japanese speakers).

However, when the network was pruned beyond a certain level, two interesting things happened. First, its performance got noticeably worse, especially when the phonetic inputs were degraded (i.e., the information was incomplete or ambiguous). This corresponds quite well with another symptom of schizophrenia, especially those who experience auditory hallucinations – sufferers show phonetic processing deficits under challenging conditions, such as a crowded room.

The second effect was even more striking – the network started to hallucinate! It began to produce outputs even in the absence of any inputs (i.e., during “silence”). When not being driven by reliable external sources of information, the network nevertheless settled into a state of activity that represented a word. The reason the output is a word and not just a meaningless pattern of neurons is that the previous learning that the network undergoes means that patterns representing words represent “attractors” – if some random neurons start to fire, the weighted connections representing real words will rapidly come to dominate the overall pattern of activity in the network, resulting in the pattern corresponding to a word.

Modeling alterations in dopamine signalling also produced both a defect in parsing degraded speech inputs and hallucinations. Too much dopamine signalling produced these effects but so did a combination of moderate over-pruning and compensatory reductions in dopamine signalling, highlighting the complex interactions possible.

The conclusion from these simulations is not necessarily that this is exactly how hallucinations emerge. After all, the artificial neural networks are pretty extreme abstractions of real biological networks, which have hundreds of different types of neurons and synaptic connections and which are many orders of magnitude more complex numerically. But these papers do provide aat least a conceptual demonstration of how a circuit designed to process speech sounds can fail in such a specific and apparently bizarre way. They show that auditory hallucinations can be viewed as the outputs of malfunctioning speech-processing circuits.

They also suggest that different types of insult to the system can lead to the same type of malfunction. This is important when considering new genetic data indicating that schizophrenia can be caused by mutations in any of a large number of genes affecting how neural circuits develop. One way that so many different genetic changes could lead to the same effect is if the effect is a natural emergent property of the neural networks involved.

Hoffman, R., & Mcglashan, T. (2001). Book Review: Neural Network Models of Schizophrenia The Neuroscientist, 7 (5), 441-454 DOI: 10.1177/107385840100700513

Hoffman, R., & McGlashan, T. (2006). Using a Speech Perception Neural Network Computer Simulation to Contrast Neuroanatomic versus Neuromodulatory Models of Auditory Hallucinations Pharmacopsychiatry, 39, 54-64 DOI: 10.1055/s-2006-931496

Mirrored from Wiring the Brain

However, these studies also highlight the limits of genetic determinism, which is especially evident in comparisons of monozygotic (identical) twins, who share all their genetic inheritance in common. Though they are obviously much more like each other in psychological traits than people who are not related to each other, they are clearly NOT identical to each other for these traits. For example, if one twin has a diagnosis of schizophrenia, the chance that the other one will also suffer from the disorder is about 50% – massively higher than the population prevalence of the disorder (around 1%), but also clearly much less than 100%.

What is the source of this extra variance? What forces make monozygotic twins less identical? I have argued previously that random variation in the course of development is a major contributor. The developmental programme that specifies brain connectivity is less like a blueprint than a recipe (a recipe without a cook) – an incredibly complicated set of processes carried out by mindless biochemical algorithms mediated by local interactions between billions of individual components. As each of these processes is subject to some level of “noise” at the molecular level, it is not surprising that the outcome of this process varies considerably, even between monozygotic twins.

While such developmental variation can be referred to as “non-genetic”, a new study suggests that one important component of this variation may be genetic after all, just not inherited. Mutations can be passed on from parents to offspring or arise during generation of sperm or eggs and thus be inherited, but they can also arise any time DNA is replicated. So, each time a cell divides as an embryo grows and develops, there is a very small chance of new mutations being introduced. These “somatic” mutations (meaning ones that happen in the body and not in the germline) will be inherited by all the cells that are descendants of that new cell and so will be present in some fraction of the final cells of the individual. Mutations arising earlier in development will be inherited by more cells than those arising later.

Each person will therefore be a mosaic of cells with slightly different genetic make-up. The vast majority of such mutations will not have any effect of course (with the obvious exception of those that cause dysregulation of cellular differentiation and result in cancer). But sometimes a new mutation will affect a trait and cause a detectable difference. The most obvious examples are in genes affecting hair or eye colour – where a patch of hair may be a different colour, or the two eyes may be different colours.

But what if the mutations in question are linked to a psychiatric disorder? If such a mutation arises early in the development of the brain and is therefore inherited by many of the cells in the brain then this could lead to the psychiatric disorder, just as if the mutation had been inherited in a germ cell.

A new study adds to the evidence that such mutations do indeed occur at an appreciable frequency and may help explain the discordance in phenotype between pairs of twins where one has schizophrenia and the other does not. The authors analysed the DNA from blood cells of pairs of twins discordant for schizophrenia and their parents. They were looking for two different kinds of mutation: ones that changes the identity of a single base of DNA (one letter of the genetic code to another), called point mutations, and ones that delete or duplicate whole chunks of chromosomes, called copy number variants, or CNVs.

As expected, they were able to detect both inherited mutations (present in one of the parents) and de novo mutations (present in both twins but not in the blood cells of either parent). What is more remarkable though, is that they also detected de novo mutations present in the blood cells of one twin but not the other – lots of them. About 1,000 point mutations and 2-3 new CNVs not shared by the other twin. The implication is that these mutations arose during the somatic development of one twin. They identify a couple CNVs in the twins affected by schizophrenia, raising the (very speculative) possibility that those mutations may contribute to the development of the disorder. It will obviously require a lot more work to test that specific hypothesis.

An earlier study also found a high rate of somatic mosaicism for CNVs – this time by analysing the DNA of multiple tissues taken from single (deceased) individuals. Across 34 tissue samples from 3 subjects they identified six CNVs present in one tissue but not others. What this implies is that not only do we carry additional mutations making us even more different from one another, our cells and tissues can also be genetically different from each other.

Time will tell whether such mutations really do contribute to psychiatric disorders, but it certainly seems plausible that they might. This adds to a couple other potential mechanisms of increasing individual variance: the transposition of mobile DNA elements in somatic tissues, especially neurons, and the “epigenetic” silencing of regions of the genome, which may be clonally inherited in groups of cells and contribute to differences between twins.

This has one immediate and important consequence for clinical genetics. When a mutation in an offspring is not carried by either parent it is usually interpreted as having arisen de novo. The implication is that the risk of another offspring carrying the same mutation is negligible. Clinical geneticists are finding this is not necessarily always the case, however – apparently de novo mutations may have actually arisen at an early stage in the germline and not just at the final division generating the sperm or egg. The parent in question may not actually “carry” the mutation, but their germline does. Great care must therefore be taken when advising parents with one affected child of the risk to future offspring.

Maiti S, Kumar KH, Castellani CA, O’Reilly R, & Singh SM (2011). Ontogenetic de novo copy number variations (CNVs) as a source of genetic individuality: studies on two families with MZD twins for schizophrenia. PloS one, 6 (3) PMID: 21399695

Piotrowski, A., Bruder, C., Andersson, R., de Ståhl, T., Menzel, U., Sandgren, J., Poplawski, A., von Tell, D., Crasto, C., Bogdan, A., Bartoszewski, R., Bebok, Z., Krzyzanowski, M., Jankowski, Z., Partridge, E., Komorowski, J., & Dumanski, J. (2008). Somatic mosaicism for copy number variation in differentiated human tissues Human Mutation, 29 (9), 1118-1124 DOI: 10.1002/humu.20815

Fraga, M. (2005). From The Cover: Epigenetic differences arise during the lifetime of monozygotic twins Proceedings of the National Academy of Sciences, 102 (30), 10604-10609 DOI: 10.1073/pnas.0500398102

Mirrored from the Wiring the Brain blog

The emerging picture is that each of these disorders can be caused by mutations in any one of a large number of genes. Strikingly, many of these genes play important roles in neural development, with mutations affecting patterns of cell migration, the guidance of growing nerve fibres and their connectivity to other cells. Even more remarkable has been the observation that most such mutations predispose to not just one specific illness (such as schizophrenia) but to mental illness in general, with a strong overlap in the genetics of schizophrenia, autism, bipolar disorder, epilepsy, mental retardation, attention-deficit hyperactivity disorder and other diagnostic categories. These different categories may thus represent arguably distinct endpoints arising from common origins in neurodevelopmental insults.

What we do not yet know is why. How does a mutation in a gene controlling say, the formation of connections between specific types of nerve cells, ultimately result in someone having paranoid delusions? (While another person carrying the same mutation may develop the quite different symptoms of autism at a much earlier age). Answering such questions will require much greater integration of efforts across a wide range of disciplines.

These efforts must include neurodevelopmental biologists. Over the past couple of decades, tremendous progress has been made in elucidating the molecular mechanisms underlying nervous system development. In many cases, these advances have been made using fairly simply model systems – fruit flies and nematode worms have been favourites in this field, as well as simple parts of the vertebrate nervous system such as the spinal cord and retina. While more and more researchers are trying to figure out how these mechanisms apply in the vastly more complicated mammalian brain, we are still a long way from understanding how this structure develops. This is especially the case as much of the circuitry of the brain is not prespecified by genetic instructions down to the last synapse, but is strongly affected by patterns of electrical activity within developing circuits. Nevertheless, it has been possible to use animals with mutations in particular genes to figure out what the functions of these genes are in the development of specific brain circuits.

The logic of these approaches is fairly straightforward: in order to discover the normal function of Gene X, mutate it, look at what happens to some part of the brain and work backwards to deduce the cellular processes that have been affected. What is needed now, if neurodevelopmental biologists are to make a contribution to the study of mental illness, is a different approach. We must develop an interest in the phenotypes themselves, not just as tools to elucidate the gene’s normal functions. If mutations in Gene X can cause autism, for example, then a mouse with the same mutation becomes a valuable and informative model of disease. It becomes of interest to analyse not just the direct processes affected by the mutation but all of the knock-on consequences. While these questions may start with neurodevelopmental biologists they rapidly require additional expertise to address.

This will entail a framework to link investigations across levels of analysis typically carried out by researchers in quite different disciplines. For example, if the mutation affects formation of synaptic connections between certain types of cells in certain brain regions, then how does this change the function of the circuits involved? If this changes the activity of the circuit, then how does this affect further activity-depdendent development of interconnected regions? How does that affect the information processing capabilities of these networks? What cognitive functions are carried out by these networks and how are they impacted? At what level can we most directly translate findings in animals to humans? Each of these questions requires researchers in different disciplines to work together.

The imperative to do this could not be more stark. Roughly 10% of the world’s population is affected by mental illness at any one time, and over 25% will have some mental health problem over their lifetime. As well as the costs to individuals and their families, the public health and economic burdens from these disorders are massive, as large as that of cancer and cardiovascular disease. In fact, the proportional burden is growing as we are making good progress in treating the latter disorders, while mental illnesses have lagged far behind. This is mainly because we have not been able to apply the tools of molecular genetics to the problem. This is now changing, thanks to the revolutionary advances in psychiatric genetics. The challenge now will be to translate these discoveries into real understanding of disease mechanisms and ideas for novel therapies.

This post is based on a brief article that introduces a thematic series of reviews and primary research papers on the theme of Wiring the Brain. This series will appear across various journal titles of the open access publisher BioMed Central and can be accessed here.

Mitchell KJ (2011). The miswired brain: making connections from neurodevelopment to psychopathology. BMC biology, 9 (1) PMID: 21489316