Most of us are familiar with pictures from magnetic resonance imaging (MRI) of the human brain; indeed, these black-and-white images have achieved almost iconic status at this stage. From popular television programmes, and regrettably from common experience, the use of these images to detect lesions, such as tumours or the effects of stroke, is well known.
Classic MRI can distinguish grey and white matter based on their different cellular composition but cannot go very far beyond that, because all the white matter has effectively the same contrast. This makes traditional MRI of limited use in examining connectivity between areas of the brain, except at a very gross level (such as whether the corpus callosum exists, for example). However, with a few modifications, MRI can be applied to non-invasively interrogate connectivity in the living human brain, with ever-increasing sensitivity. These new techniques are opening up avenues of investigation that have not just tremendous clinical importance but that also promise to make humans a powerful model organism for the study of axon guidance and related neurodevelopmental processes.
The modifications rely on the fact that the diffusion of water molecules gives off a magnetic resonance signal. Within the brain, the diffusion of water is affected by the local cytoarchitecture; in particular, within axonal bundles the direction of diffusion is constrained by the direction of the axons and their myelin sheaths. By determining the bias in direction of diffusion within a voxel of white matter (the “diffusion tensor”) it is possible to infer the dominant direction of axonal projections in that voxel. Diffusion tensor imaging (DTI) can be applied across the brain, so that, by following the direction of the tensors from voxel to voxel, axonal tracts projecting across large areas of the brain can be derived.
DTI has been extremely powerful, though it does have limitations. Foremost among these is difficulty in distinguishing fibres that cross within a single voxel. More recent refinements, including q-ball imaging and diffusion spectrum imaging (DSI) apply higher-resolution scan acquisition and different statistical approaches to largely resolve this issue. These approaches have yielded dramatic images of fibre tracts within the brain, revealing three-dimensional connectivity patterns across the entire brain that would be impossible to obtain with traditional anatomical tracing or histology, even in post mortem tissue.
It is important to realize, however, that the “fibres” being drawn are really three-dimensional plots of statistical values that may vary depending on method of acquisition, scan parameters, software used, statistical approach as well as subjective thresholding criteria. Comparison of imaging methods and validation using data derived by classical techniques was thus crucial. These kinds of comparisons have now shown that, in the case of DSI at least, the congruence with known or classically-derived tractography is actually very good (Schmahmann et al., 2007; Wedeen et al. , 2008). This is not to say that the approach does not still have limitations – tracking fibres through sharp bends, as they de-fasciculate into small bundles or as they project into grey matter are all still problematic, for example – but these limitations are likely to be overcome by more sophisticated algorithms.
Given the recent pace of improvements in tractography approaches, we can thus expect in the very near future that this technique will allow routine examination of patterns of connectivity in the human brain. These approaches are already being applied to investigate structural connectivity in various disorders with a neurodevelopmental etiology, including schizophrenia, autism, dyslexia and several less well-known conditions, such as prosopagnosia (the inability to recognize faces) and synaesthesia (where sensory stimuli in one modality can cross-activate another).
Tractography can also be used to define a connectivity profile or fingerprint of a particular part of the brain, and thus to automatically parcellate the brain into units of likely functional distinction. This kind of approach is especially useful to delineate functional areas of the neocortex, which are often not obviously anatomically distinct. By defining a connectivity profile of each voxel to all other voxels in the brain, and constructing a matrix of such profiles, it is possible, using a clustering algorithm similar to those applied to microarray data, to cluster voxels with similar connectivity profiles and thus to distinguish regions where the profile suddenly changes, thus delineating the border between two presumptive areas. Exactly this approach has been used by several groups and validated with functional imaging and other anatomical data (Klein et al., 2007; Perrin et al., 2008).
The development of these techniques finally gives us the means to see, literally, differences in how each of our brains is wired, which may affect many aspects of our personality, cognitive abilities and style and other psychological traits that make us who we are.
Wedeen, V., Wang, R., Schmahmann, J., Benner, T., Tseng, W., Dai, G., Pandya, D., Hagmann, P., D'Arceuil, H., & de Crespigny, A. (2008). Diffusion spectrum magnetic resonance imaging (DSI) tractography of crossing fibers NeuroImage, 41 (4), 1267-1277 DOI: 10.1016/j.neuroimage.2008.03.036
Schmahmann, J., Pandya, D., Wang, R., Dai, G., D'Arceuil, H., de Crespigny, A., & Wedeen, V. (2007). Association fibre pathways of the brain: parallel observations from diffusion spectrum imaging and autoradiography Brain, 130 (3), 630-653 DOI: 10.1093/brain/awl359
Perrin, M., Cointepas, Y., Cachia, A., Poupon, C., Thirion, B., Rivière, D., Cathier, P., El Kouby, V., Constantinesco, A., Le Bihan, D., & Mangin, J. (2008). Connectivity-Based Parcellation of the Cortical Mantle Using q-Ball Diffusion Imaging International Journal of Biomedical Imaging, 2008, 1-19 DOI: 10.1155/2008/368406
Klein, J., Behrens, T., Robson, M., Mackay, C., Higham, D., & Johansen-Berg, H. (2007). Connectivity-based parcellation of human cortex using diffusion MRI: Establishing reproducibility, validity and observer independence in BA 44/45 and SMA/pre-SMA NeuroImage, 34 (1), 204-211 DOI: 10.1016/j.neuroimage.2006.08.022
Monday, July 20, 2009
The traditional view of neural development is linear. First, the embryo and neurectoderm are patterned by secreted factors, which establish cell fates among progenitors and then differentiated neurons, encoded by combinations of transcription factors. The fate or phenotype of each neuron includes the expression of the specific set of ion channels, neurotransmitters and receptors that determine its physiological function. It also includes expression of a particular repertoire of guidance receptors and surface molecules regulating connectivity, which enable axonal pathfinding and target selection. The processes that establish connectivity are usually thought of as happening after the fate of neurons and their targets have been established. This linear paradigm, from patterning to differentiation to connection, has been increasingly challenged by studies from both invertebrate and vertebrate systems.
A number of studies have shown that incoming axons can regulate the proliferation and differentiation of their synaptic target cells. In many cases in fact, the target cells do not even exist at the time that the incoming axons are making their targeting decisions. In the fly visual system, for example, photoreceptor axons target the developing optic lobe and secrete the morphogen hedgehog, which induces optic lobe progenitor cells to complete a final cell division and undergo neuronal differentiation (Huang and Kunes, 1996). In addition, secretion of additional signaling molecules induces expression in the optic lobe neurons of adhesion molecules and guidance factors necessary for retinal axons to recognize them as appropriate synaptic targets (Bazigou et al., 2007). Thus, the final differentiation of cells in the optic lobe requires the prior pathfinding of retinal axons to this area.
A similar situation has been demonstrated in the mammalian brain, where axons from the visual thalamus induce the proliferation and differentiation of the primary visual cortex (Dehay et al., 2001). Significant patterning of the cortical sheet occurs prior to thalamic axon invasion and directs the guidance of visual thalamic axons to the appropriate part of the cortex (Little et al., 2009). However, ultimate elaboration of the mature cytoarchitectonic characteristics of primary visual cortex, including its pattern of connectivity with other cortical areas, requires correct innervation by visual thalamic axons. Though it has not been shown, it seems likely that this kind of hierarchical dependence on afferent innervation might also be crucial in the elaboration of later-maturing higher-order cortical areas, which receive inputs from earlier-maturing areas (Bargary and Mitchell, 2008).
The linear developmental paradigm must thus be substantially modified to include a highly dynamic interplay between differentiation and establishment of connectivity. Importantly, a recent study suggests that the influence of this interplay also extends to the maintenance of cell fate in the adult nervous system. It is well known that many neurons require retrograde neurotrophic support from their target cells to stay alive. A study from Drosophila (Eade and Allan, 2009) suggests that retrograde signals, in this case involving bone morphogenetic protein (BMP) signaling, may also be required to maintain expression of neuronal phenotype in connecting cells, demonstrated through an effect on expression of a specific neuropeptide. This signaling was shown to require active axonal transport mechanisms. If this mechanism holds in vertebrates it has several important implications. First, neuronal phenotypes in the adult nervous system may be more plastic than previously recognised and more actively maintained by regulators of gene expression in response to ongoing retrograde (and possibly anterograde?) signaling. Second, neurodegenerative disorders involving defects in axonal transport, such as Huntington’s disease, may have their primary effects on neuronal phenotype and physiological function, inducing partial de-differentiation prior to overt degeneration. Therapeutics aimed at preventing this process may thus be able to target the earliest stages of such diseases.
HUANG, Z., & KUNES, S. (1996). Hedgehog, Transmitted along Retinal Axons, Triggers Neurogenesis in the Developing Visual Centers of the Drosophila Brain Cell, 86 (3), 411-422 DOI: 10.1016/S0092-8674(00)80114-2
BAZIGOU, E., APITZ, H., JOHANSSON, J., LOREN, C., HIRST, E., CHEN, P., PALMER, R., & SALECKER, I. (2007). Anterograde Jelly belly and Alk Receptor Tyrosine Kinase Signaling Mediates Retinal Axon Targeting in Drosophila Cell, 128 (5), 961-975 DOI: 10.1016/j.cell.2007.02.024
Cell-cycle kinetics of neocortical precursors are influenced by embryonic thalamic axons.
Dehay C, Savatier P, Cortay V, Kennedy H.
J Neurosci. 2001 Jan 1;21(1):201-14.
Little, G., López-Bendito, G., Rünker, A., García, N., Piñon, M., Chédotal, A., Molnár, Z., & Mitchell, K. (2009). Specificity and Plasticity of Thalamocortical Connections in Sema6A Mutant Mice PLoS Biology, 7 (4) DOI: 10.1371/journal.pbio.1000098
Bargary, G., & Mitchell, K. (2008). Synaesthesia and cortical connectivity Trends in Neurosciences, 31 (7), 335-342 DOI: 10.1016/j.tins.2008.03.007
Eade, K., & Allan, D. (2009). Neuronal Phenotype in the Mature Nervous System Is Maintained by Persistent Retrograde Bone Morphogenetic Protein Signaling Journal of Neuroscience, 29 (12), 3852-3864 DOI: 10.1523/JNEUROSCI.0213-09.2009
at 1:47 AM
Tuesday, July 7, 2009
Schizophrenia is a common and devastating disorder, involving stable impairments in a wide range of cognitive, sensory and motor domains, as well as fluctuating episodes of psychosis, characterised by disordered thoughts, hallucinations and delusions. Though it tends to emerge as a full-blown disorder in late adolescence or early adulthood, a wealth of evidence supports the model that it is caused by disturbances in neural development at much earlier time-points, including prenatally. Recent neuroimaging analyses have supported psychological theories of schizophrenia as a “disconnection syndrome”, showing altered structural and functional connectivity between (and also within) many regions of the brain. Schizophrenia can thus be thought of as the result of alterations in brain wiring, and these alterations are, in turn, caused by mutations.
There is strong and consistent evidence from twin, adoption and family studies that schizophrenia is highly heritable. Though this fact is now widely accepted there is far less agreement on exactly how it is inherited. Risks to family members are clearly much higher than to the general population (approximately ten percent in first-degree relatives, versus 0.5-1% prevalence in the general population). And concordance between monozygotic twins is much higher (averaging 0.48) than between dizygotic twins (about 0.17). On the other hand, a majority of cases of schizophrenia are sporadic and do not have an affected first-degree relative. In addition, looking across families with multiple affected individuals, no clear pattern emerges that suggests a simple Mendelian mode of inheritance, or at least not a consistent one.
Various models have been proposed to explain the genetic architecture of schizophrenia. Early researchers suggested Mendelian inheritance, either recessive or dominant, with partial penetrance (i.e., not everyone who inherits the putative causative mutation develops the disorder). Based on the fact that any one of these modes of inheritance could not explain all cases of schizophrenia, and on a rejection of the notion that different cases might follow different modes of inheritance (i.e., genetic heterogeneity), these models have been almost completely replaced by a polygenic model. This states that schizophrenia arises due to the inheritance of a large number of genetic variants in any individual. Any one of these variants alone would have a small effect on risk, but collectively, a “toxic combination” of such variants could lead to disease. To explain the prevalence of the disorder, such variants must be common in the population. The alternative model, which has been dubbed the multiple rare variants model, proposes that schizophrenia is caused in each individual by a single mutation and that such mutations are rare because they are rapidly selected against. To explain the prevalence of the disorder under this model requires a high mutation rate and a large target of genes that can result in schizophrenia when mutated.
A recent set of papers has directly tested the common variants, polygenic model (cited below). These papers describe very large genome-wide association studies (GWAS) of schizophrenia, carried out in unprecedented collaborations on huge samples by large numbers of researchers in different facilities across the globe. The goal of these studies is to find alleles of common variants that are significantly enriched in people with schizophrenia (cases) versus those without (controls). Under the common variants model, each such variant is likely to increase risk only very slightly itself and is therefore likely to be at only slightly higher frequency in cases. However, comparing frequencies of a set of common variants across the entire genome in large samples offers enough statistical power to detect even very modest effects. The hope is that identifying such “risk genes” will lead to insights of the pathogenic mechanisms of the disorder or offer the means to predict level of risk in individuals.
The main finding of these three studies, consistent with several smaller forerunners, is that there are no common variants of even modest effect size. None of these studies alone detected a single such variant. When combined in a meta-analysis, a few regions emerged with very small effect sizes. These explain only a tiny fraction of the total heritability of the disorder, however. Considering the demonstrable power of these studies to have detected variants of modest effect if they existed, these negative results provide the strongest evidence yet that the common variants, polygenic model is incorrect.
(Note that a further analysis does suggest some polygenic contribution to risk, but based on the combined effects of thousands of variants. The simulations used to estimate the magnitude of this effect are far from conclusive, however. Regardless of its overall contribution to risk, this finding could be consistent with a “genetic background” effect, which modifies the penetrance and expressivity of rare, causal mutations.)
Are these disappointing results cause for dismay? Quite the opposite, I would say. They provide additional support for the multiple rare variants model, which is now gaining traction with the recent discoveries of many such rare, causal mutations. This should encourage geneticists to re-focus their efforts on families and individuals and move away from an epidemiological approach that focuses on risk across the population. Schizophrenia liability is not a quantitative trait and should not be treated as one. Happily, the technologies to detect rare, causal variants are now available, most obviously whole-genome sequencing. The upside of this model being true is that the effects of such mutations in single genes can be very directly modeled in animals, to help elucidate the pathogenic mechanisms, pathophysiology and etiology of the disorder.
For more discussion see: http://www.schizophreniaforum.org/new/detail.asp?id=1532
Purcell, S., Wray, N., Stone, J., Visscher, P., O'Donovan, M., Sullivan, P., Sklar, P., Purcell (Leader), S., Stone, J., Sullivan, P., Ruderfer, D., McQuillin, A., Morris, D., O’Dushlaine, C., Corvin, A., Holmans, P., O’Donovan, M., Sklar, P., Wray, N., Macgregor, S., Sklar, P., Sullivan, P., O’Donovan, M., Visscher, P., Gurling, H., Blackwood, D., Corvin, A., Craddock, N., Gill, M., Hultman, C., Kirov, G., Lichtenstein, P., McQuillin, A., Muir, W., O'Donovan, M., Owen, M., Pato, C., Purcell, S., Scolnick, E., St Clair, D., Stone, J., Sullivan, P., Sklar (Leader), P., O'Donovan, M., Kirov, G., Craddock, N., Holmans, P., Williams, N., Georgieva, L., Nikolov, I., Norton, N., Williams, H., Toncheva, D., Milanova, V., Owen, M., Hultman, C., Lichtenstein, P., Thelander, E., Sullivan, P., Morris, D., O'Dushlaine, C., Kenny, E., Quinn, E., Gill, M., Corvin, A., McQuillin, A., Choudhury, K., Datta, S., Pimm, J., Thirumalai, S., Puri, V., Krasucki, R., Lawrence, J., Quested, D., Bass, N., Gurling, H., Crombie, C., Fraser, G., Leh Kuan, S., Walker, N., St Clair, D., Blackwood, D., Muir, W., McGhee, K., Pickard, B., Malloy, P., Maclean, A., Van Beck, M., Wray, N., Macgregor, S., Visscher, P., Pato, M., Medeiros, H., Middleton, F., Carvalho, C., Morley, C., Fanous, A., Conti, D., Knowles, J., Paz Ferreira, C., Macedo, A., Helena Azevedo, M., Pato, C., Stone, J., Ruderfer, D., Kirby, A., Ferreira, M., Daly, M., Purcell, S., Sklar, P., Purcell, S., Stone, J., Chambert, K., Ruderfer, D., Kuruvilla, F., Gabriel, S., Ardlie, K., Moran, J., Daly, M., Scolnick, E., & Sklar, P. (2009). Common polygenic variation contributes to risk of schizophrenia and bipolar disorder Nature DOI: 10.1038/nature08185
Shi, J., Levinson, D., Duan, J., Sanders, A., Zheng, Y., Pe’er, I., Dudbridge, F., Holmans, P., Whittemore, A., Mowry, B., Olincy, A., Amin, F., Cloninger, C., Silverman, J., Buccola, N., Byerley, W., Black, D., Crowe, R., Oksenberg, J., Mirel, D., Kendler, K., Freedman, R., & Gejman, P. (2009). Common variants on chromosome 6p22.1 are associated with schizophrenia Nature DOI: 10.1038/nature08192
Stefansson, H., Ophoff, R., Steinberg, S., Andreassen, O., Cichon, S., Rujescu, D., Werge, T., Pietiläinen, O., Mors, O., Mortensen, P., Sigurdsson, E., Gustafsson, O., Nyegaard, M., Tuulio-Henriksson, A., Ingason, A., Hansen, T., Suvisaari, J., Lonnqvist, J., Paunio, T., Børglum, A., Hartmann, A., Fink-Jensen, A., Nordentoft, M., Hougaard, D., Norgaard-Pedersen, B., Böttcher, Y., Olesen, J., Breuer, R., Möller, H., Giegling, I., Rasmussen, H., Timm, S., Mattheisen, M., Bitter, I., Réthelyi, J., Magnusdottir, B., Sigmundsson, T., Olason, P., Masson, G., Gulcher, J., Haraldsson, M., Fossdal, R., Thorgeirsson, T., Thorsteinsdottir, U., Ruggeri, M., Tosato, S., Franke, B., Strengman, E., Kiemeney, L., GROUP†, ., Melle, I., Djurovic, S., Abramova, L., Kaleda, V., Sanjuan, J., de Frutos, R., Bramon, E., Vassos, E., Fraser, G., Ettinger, U., Picchioni, M., Walker, N., Toulopoulou, T., Need, A., Ge, D., Lim Yoon, J., Shianna, K., Freimer, N., Cantor, R., Murray, R., Kong, A., Golimbet, V., Carracedo, A., Arango, C., Costas, J., Jönsson, E., Terenius, L., Agartz, I., Petursson, H., Nöthen, M., Rietschel, M., Matthews, P., Muglia, P., Peltonen, L., St Clair, D., Goldstein, D., Stefansson, K., Collier, D., Kahn, R., Linszen, D., van Os, J., Wiersma, D., Bruggeman, R., Cahn, W., de Haan, L., Krabbendam, L., & Myin-Germeys, I. (2009). Common variants conferring risk of schizophrenia Nature DOI: 10.1038/nature08186
at 11:37 PM
Wednesday, June 24, 2009
The question of what makes each of us the persons we are has occupied philosophers, writers and daydreamers for millennia but has been open to scientific inquiry over a far shorter time. The answer clearly lies in the brain and, somehow, in how it is “wired” (whether that refers to the amount or type of connections between different brain areas or to differences in how circuits function). But differences in brain wiring could be either innate or due to experience or environmental effects. This has been famously framed, by Galton originally, as a clash of nature versus nurture. The inspiration for this phrase may have come from Shakepseare’s The Tempest, in which Prospero refers to Caliban as a "A devil, a born devil, on whose nature / Nurture can never stick". That line encapsulates the notion of an innate character that is resistant to extrinsic influences, especially efforts to change aspects of a person’s personality – an idea which anyone who has children may find easy to relate to.
While neurology and neuroscience have offered direct evidence that differences in the brain affect behaviour, it is behavioural genetics that is typically seen as having contributed most directly to the nature-nurture debate. Twin and adoption studies, the mainstays of behavioural genetics, have demonstrated very conclusively that many aspects of personality, behaviour or other psychological traits are highly heritable – that is, a large proportion of the variance in the trait across the population is attributable to differences in genes.
The logic of the twin studies is usually the inverse of the statement above – to look at people who share various proportions of their genes and see how similar they are to each other. These generally show that monozygotic (“identical”) twins are far more similar to each other for most psychological traits than dizogytic (“fraternal”) twins. Also, adoptive children tend to resemble their biological relatives for psychological traits and are hardly more similar to their adoptive family members than they would be to any stranger in the street.
These data have demonstrated unequivocally that variation in genes can affect behaviour in humans. They have also, however, dramatically illustrated the limits of such genetic effects. Monozygotic twins, while much more similar to each other than would be expected for people who don’t share all their genes in common, are nonetheless clearly not identical for most psychological traits. In fact, on average, genetic variance only explains about 50% of the phenotypic variance. What is causing the rest of the variance?
The presumption in the literature has been that it is something in the experience of the individuals – that in fact, the results of these behavioural genetic studies provide the strongest evidence to date for some effect of the environment. I say strongest evidence because in fact there is very little other evidence for direct experience-dependent effects on human psychological traits across the general population. (This is not in any way to deny the impact of factors such as serious abuse in individuals, which can be profound – it is just that, happily, such abuse is rare enough that it does not contribute significantly to the variance across the population).
Certainly, the twin and adoption studies have consistently found only a very modest effect of a shared family environment on the types of traits examined. The argument for interpreting the excess variance as being caused by experience is thus that it is the individual, “non-shared”, experiences that people have that make them different from each other. This has never made any sense to me, I have to confess. Non-shared experiences can make me differ from my twin but shared ones cannot make us more similar? An experience can affect my psychological development, but only if it does not also happen to my twin? Interactions with peers and teachers can have lasting effects on me but interactions with my parents cannot? Not only does this notion not make intuitive sense, there is no evidence for it. It seems far more likely that most non-traumatic experiences, regardless of who they are with, have little long-lasting effect on the kinds of traits that define us as persons. What then can account for the unexplained variance?
The basic problem with the interpretation above is that it limits “innate” influences to “genetic” ones. Just because some trait is not genetic does not mean it is not innate. If we are talking about how the brain gets wired, any number of prenatal environmental factors are known to have large effects. More interestingly, however, and probably a greater source of variance across the population, is intrinsic developmental variation. Wiring the brain is a highly complex procedure, reliant on cellular processes that are, in engineering terms, inherently “noisy”. Running the programme from the same starting point (a specific genotype) does not generate exactly the same output (the phenotype) every time. The effects of this noise are readily apparent at the anatomical level, when examining the impact of specific mutations, for example. In many cases, the phenotypic consequences are quite variable between genetically identical organisms, or even on two sides of the same brain. (If you want to see direct evidence of such developmental variation, take a directly face-on photograph of yourself, cut it in half and make mirror-image copies of the left and right sides. You will be amazed how different the two resultant faces are).
If the way the brain is wired is determined, not just by the starting genotype, but, to a large extent by chance events during development, then it is reasonable to expect this variation to be manifest in many psychological traits. Such traits may thus be far more innate than behavioural genetics studies alone would suggest. [Note that this does not imply all aspects of a person’s behaviour are innate and resistant to effects of experience – that is obviously a nonsense. It is important to recognize that the kinds of personality traits that have been examined in the studies mentioned above (such as extraversion or neuroticism) are basal characteristics, reflecting, for example, how strongly the brain responds to positive or to negative stimuli. These tendencies, in combination with later experience, influence, but in no way determine, moment-to-moment behaviour].
Happily, in my view, this places intrinsic limits on genetic determinism and the ability to predict many important aspects about a person from their genotype. Not limits based simply on our current knowledge that could one day be overcome – limits due to the inherently variable nature of neural development. Ultimately, what defines us each as persons, is thus dependent on nature, nurture and noise.
For more on this subject, see: Mitchell, K.J. (2007) The genetics of brain wiring; from molecule to mind. PLoS Biology Apr 17;5(4):e113. (Open Access).
at 8:14 AM
Tuesday, June 23, 2009
Welcome to the Wiring the Brain blog. The purpose of this site is to allow researchers (or anyone else who is interested) to discuss research developments in various fields related to how the brain gets wired, how it varies and what effects this has. These fields will include developmental neurobiology, molecular and cellular neuroscience, systems neuroscience, cognitive science, psychology, psychiatric genetics and others. Recent papers or specific hypotheses will be highlighted for discussion.
at 7:33 AM