Probabilistic inheritance and neurodevelopmental phenotypes: location, location, location
Following a stimulating discussion with Jon Brock on the variability of the phenotype in autism and dyslexia, I thought I would explore a little more the influence of randomness on the phenotypic expression of mutations affecting neurodevelopmental processes. I have written about this before, in general terms, but here want to discuss one particular aspect – how probabilistic inheritance of a defect at a cell biological level is played out across the brain and how this influences the emergent phenotype of the individual.
Developmental neurobiologists are well used to the scenario where mutation of a gene leads to an anatomical defect, but only some of the time. Depending on the scale at which the defect is defined that can mean “in only some individuals” – for example, whether the connections between the two hemispheres of the brain form. In genetically identical people (or animals) carrying a mutation affecting this process these connections sometimes form normally and sometimes do not form at all. But other phenotypes reflect processes that are played out, independently, many, many times across the developing brain. “Only some of the time” can thus mean in “only some brain regions” or “affecting only some axons” or “only some cells”. Importantly, what determines which regions or axons or cells are affected may be largely down to chance.
The molecular programs controlling neurodevelopmental processes such as cell migration, axon guidance and synapse formation are typically very robust – they generate effectively the same outcome time after time. Not exactly the same – down to the location of every cell and every connection – but similar enough at the relevant level of detail to generate circuits that perform within the normal range. What often happens when mutations arise that affect one of the components of these processes is that the phenotype does not change from always wild-type to always mutant – instead, phenotypic variability increases.
This is easy to see in segmented animals like insects, where the exact same processes are played out multiple times per animal. For example, projections of a specific motor neuron to its target muscle may be defective in, say, 30% of segments, but normal in 70% (with no pattern of which segments are affected from animal to animal). In mammalian brains, this kind of variability is less obvious, but can be seen when some percentage of cells migrate incorrectly, or some percentage of axons is misguided.
The interpretation is that this variability is intrinsic to the system – not due to anything external to the developing organism. It presumably originates from thermal noise – random fluctuations in molecular shape and movement that affect fundamental cellular processes like gene expression or cellular signaling. Normally, these fluctuations are buffered by intact molecular systems, which are adapted to deal with them and still produce the same outcome. But when some components are disrupted, this buffering can break down, making the outcome much more susceptible to noise.
At a higher level, such defects can sometimes lead to a discrete anatomical anomaly – a build-up of cells in the wrong place, for example (called "neuronal heterotopia”), or a change in connectivity between two brain regions. It is at this level that the variable expression of cellular phenotypes can be related to the variable expression or incomplete penetrance of clinical phenotypes.
Why should such discrete and gross defects arise from randomness that is independently affecting molecular processes across a population of cells? That is not really understood, but could reflect interactions between cells, such as differential adhesion – these could cause a small number of mis-migrated cells that happen to arise next to each other to nucleate additional cells, for example. What starts as a statistical blip in the distribution of a defect may thus be amplified by dynamic cellular interactions, resulting in a more discrete and significant anomaly.
Depending on where such anomalies arise, different brain systems may be impacted, resulting in a variable spectrum of phenotypes or clinical symptoms across people carrying a particular mutation (even identical twins). This fits with observations of variable phenotype in families where conditions like dyslexia, epilepsy or synaesthesia are segregating. In many cases, the tendency to develop the condition generally is quite strongly inherited, but the precise type that emerges is far less heritable.
For example, a recent twin study of epilepsy found that while risk of epilepsy in general was very highly heritable, the specific type emerging (e.g., generalized versus localization-related) was less so, and, among those with an identifiable focus, there was effectively no heritability for which brain region was affected. This scenario is similar to results of a study we performed a few years ago looking at the familiality of types of synaesthesia. We found that very different sub-types (coloured music vs tasting words, for example) co-occurred within families (in different individuals), suggesting that a predisposition to synaesthesia in general can be inherited but that the precise type emerging was affected by other factors. In both these examples, one can readily imagine how the random expression of some underlying neurodevelopmental phenotype across the brain could result in a localized alteration with a concomitant phenotypic profile. (In the case of synaesthesia, this could, very hypothetically, involve a local failure of segregation of adjacent cortical areas that are specialized for different functions).
A similar idea has been proposed before for dyslexia, which may be associated in some cases with cellular heterotopia – aggregations of neurons that have failed to migrate properly and instead are localized within the white matter. The idea is that these could disrupt communication between brain areas involved in processing or representing the visual shapes of letters and the sounds they make (or between the visual shapes of whole words and the concepts they represent). In this regard, it is interesting to note that dyscalculia (a specific difficulty with arithmetic) is often also found in families with dyslexia. This suggests that the hypothetical neurodevelopmental deficit may affect different brain systems depending on where in the brain it manifests most severely in development. This is easy enough to envisage for cellular heterotopia (where the concept of how they disrupt connectivity is fairly intuitive), but the same principle may apply on the much smaller scale of synapses. Dysfunction in particular brain systems could arise due to the emergent properties of microcircuits. Past some threshold of collective impairment in synaptic connectivity, these could push the system into a pathophysiological state. We understand even less, however, about the dynamics of how such states emerge.
The concept may also extend across clinical categories. We now know of many mutations that are associated with a very broad range of outcomes, from clinically unaffected, to autism, Tourette’s syndrome, schizophrenia, bipolar disorder, ADHD and others. More generally, epidemiological studies show that the risk for all these disorders is broadly overlapping – if you have a relative with schizophrenia, for example, your statistical risks of having autism or epilepsy or bipolar disorder are all increased. Though very speculative at this point, it seems possible that the random location of neuroanatomical disturbances (on the scale of circuits or microcircuits) could play a large role in determining the ultimate clinical effects of neurodevelopmental insults.
Corey, L., Pellock, J., Kjeldsen, M., & Nakken, K. (2011). Importance of genetic factors in the occurrence of epilepsy syndrome type: A twin study Epilepsy Research, 97 (1-2), 103-111 DOI: 10.1016/j.eplepsyres.2011.07.018
Barnett, K., Finucane, C., Asher, J., Bargary, G., Corvin, A., Newell, F., & Mitchell, K. (2008). Familial patterns and the origins of individual differences in synaesthesia Cognition, 106 (2), 871-893 DOI: 10.1016/j.cognition.2007.05.003
Landerl, K., & Moll, K. (2010). Comorbidity of learning disorders: prevalence and familial transmission Journal of Child Psychology and Psychiatry, 51 (3), 287-294 DOI: 10.1111/j.1469-7610.2009.02164.x
Ramus, F. (2004). Neurobiology of dyslexia: a reinterpretation of the data Trends in Neurosciences, 27 (12), 720-726 DOI: 10.1016/j.tins.2004.10.004