Thursday, April 29, 2010

Wired for Sex

Male and female brains are wired differently.  That is not intended metaphorically – they literally have different amounts and/or patterns of axonal connections between a variety of brain regions, as well as differences in the size or number of cells in various regions.  This is true in mammals, birds, fish, even insects and correlates with hard-wired, innate differences in behaviour between the sexes across in species across all these phyla.  This is as true for humans as for any other species.  The behaviours that show the most robust and innate differences between the sexes are involved in mating, reproduction, parental behaviour, territoriality and aggression and it is the brain areas that control these behaviours that are the most obviously sexually dimorphic (showing a difference in size or morphology between the sexes).  In mammals, these include areas in the limbic system, including parts of the hypothalamus, amygdala, preoptic area and bed nucleus of the stria terminalis. 

 How do these differences come about?  Sex in mammals is determined by the presence of a specific gene, Sry, on the Y chromosome – this gene sets in motion a cascade of gene expression and biochemical changes which lead to the conversion of the undifferentiated gonads into testes in male embryos.  (In female embryos the gonads follow a default differentiation pathway to form ovaries).  Testes make testosterone, of course, and testosterone is essential to masculinise the developing embryo, so that it develops male external genitalia and other physical sexual characteristics. 

Testosterone is also essential to masculinise the brain, but exactly how it does this is very surprising.  This phenomenon has been well studied in rodents but the principles apply across mammalian species.  In male mice and rats there is a surge in the production of testosterone shortly after birth, which lasts for a couple of days.  This surge is precisely timed with a critical period of brain development , during which it is susceptible to the effects of testosterone.  If male rats are castrated at this age (I know, sorry), then their brain will develop in the female pattern and they will not display typical male behaviours in adulthood, even when supplemented with testosterone.  Conversely, if female rats are given a single dose of testosterone a day or two after birth, their brains develop a male morphological pattern and they will show male-typical behaviours (mounting other females and decreased receptivity to being mounted).  Crucially, the same manipulations carried out a week or two later have no effect on either brain morphology or behaviour. 

These effects of testosterone are called “organisational”, for obvious reasons, and are distinguished from the later effects in acutely stimulating male behaviours, which are called “activational”.  Now for the surprise.  It was also found that a single dose of estrogen, given to postnatal female rats, was just as effective as the testosterone in masculinising their brains – even more effective, in fact!  How could this be?  How can estrogen have the same effects on the developing brain as testosterone?  As it turns out, testosterone is actively converted into estrogen in the brain, through the action of an enzyme, aromatase.  This enzyme is specifically enriched in brain regions that show sexual dimorphism.  The estrogen then acts through two estrogen receptors and this activity has been shown to be required for the masculinising effects of testosterone.  In particular, mutations in aromatase or in the estrogen receptors block the effects of testosterone and result in male animals with female brain morphology and behaviour, despite normal levels of circulating testosterone.

So, why don’t females have masculinised brains?  They should have loads of estrogen, shouldn’t they?  In fact female rodents have very low levels of circulating estrogen at this early postnatal stage, coinciding with the critical period.

The surprising findings implicating aromatase and estrogen receptors have left a mystery surrounding the role of the androgen receptor – the protein traditionally associated with direct responses to testosterone.  Mice with mutations in the gene for the androgen receptor also show feminised behaviours, suggesting it is also important in the process of masculinisation.  However, this interpretation is complicated by the fact that these mutants also show testicular atrophy (sorry again) and consequently have very low levels of circulating testosterone.  A new study by Nirao Shah and colleagues has now resolved the role of the androgen receptor in controlling sexual behaviour.

By knocking out this gene just in the brain, they managed to get around the requirements for testicular function and so cleanly address the possible functions in the brain.  They clearly show that the brains of these conditional mutants are still masculinised, morphologically.  These mice also generally show a male pattern of behaviour.  However, they do not express all of these behaviours to the same extent as wild-type males.  In particular, the frequency of mating behaviours in the presence of an estrus female is reduced,  though when they do engage in the behaviour the routine is effectively normal.  The mutant males also mark their territory less than wild-types and spend less time fighting with other “intruder” males.

Thus, while the developmental effects of testosterone appear to be fully explained by its conversion to estrogen, the activational effects in adults which are required for the full expression of male behaviours depend at least in part on its direct action through the androgen receptor. 


Wu, M., Manoli, D., Fraser, E., Coats, J., Tollkuhn, J., Honda, S., Harada, N., & Shah, N. (2009). Estrogen Masculinizes Neural Pathways and Sex-Specific Behaviors Cell, 139 (1), 61-72 DOI: 10.1016/j.cell.2009.07.036


Scott A. Juntti, Jessica Tollkuhn, Melody V. Wu, Eleanor J. Fraser, Taylor Soderborg, Stella Tan,, & Shin-Ichiro Honda, Nobuhiro Harada, and Nirao M. Shah (2010). The Androgen Receptor Governs the Execution, but Not Programming, of Male Sexual and Territorial Behaviors Neuron, 66, 260-272 : DOI 10.1016/j.neuron.2010.03.024

Friday, April 23, 2010

Hello, stranger!

Faces are special.  Humans are innately interested in faces and so good at detecting them that we see them in clouds, shrouds, pieces of toast, tree-stumps, and even simple yellow circles with a couple of dots in them.  Even newborn infants (really, really newborn) are more interested in looking at faces than non-faces.  Not too surprisingly, this preference and ability extends to other species too.  Monkeys reared from birth with absolutely no visual exposure to either monkey or human faces for two years still showed a strong preferential interest in faces (both monkey and human) when they were shown them.  Given the importance within social groups of recognising particular individuals and of reading emotional and social cues from people’s faces, it is perhaps not too surprising that face recognition is a built-in part of our cognitive toolkit. 

 This is not to say that experience plays no part in the skill of face recognition – we clearly improve with practice and exposure in the ability to distinguish large numbers of faces.  This can clearly be seen in the effect of cultural exposure to people of different races on the ability to distinguish such faces.  (Most people are significantly worse at distinguishing between faces in races other than their own).  We are thus pre-wired to be interested in faces and to process them differently from other visual stimuli, but we still need experience to be good at it. 

Both the specialisation and the effects of experience are evident in neuroimaging studies.  A particular area in the fusiform gyrus (the “fusiform face area”) shows highly selective responses to visual facial stimuli, compared to other types of visual stimuli.  The size and responsiveness of this area increases over time, from children to adults, correlated with improved abilities in discrimination of faces.  (Despite the name, this is more accurately thought of as a series of small clusters, linked together by dedicated circuits). 

It is well known that lesions to this area of the brain can result in a condition commonly called “face blindess”, but with the clinical term prosopagnosia (which translates literally, and more accurately as “lack of knowledge of faces”).  This condition was popularised by Oliver Sacks in his classic book “The Man Who Mistook His Wife for a Hat”, in which he described one sufferer who really was face-blind – he was unable even to recognise faces as faces – they simply made no sense as visual stimuli.  This is the most severe manifestation along a spectrum – more typical is that sufferers can distinguish faces from other visual stimuli but are unable to recognise whom they belong to.  

Perhaps less well known is the fact that prosopagnosia can also be developmental – not associated with any kind of lesion or injury.  Some people are just born that way.  In fact, as many as 2.5% of the population may be prosopagnosic.  The underlying deficit may be an inability to perceive faces holistically – to put the pieces together into a coherent picture – the process that allows instant and effortless recognition for most people.  Importantly, this defect is quite specific to face processing – other types of visual information can readily be combined in a holistic fashion.  Interestingly, most prosopagnosics can still extract some information from facial features, such as gender and age, though they are usually unable to tell whose face it is, or even if it is a familiar face.  (At least, they are not consciously aware of whether the face is familiar or not – they may, interestingly, still show a subconscious emotional response to familiar faces).  This deficit has obvious and severe social consequences, and many sufferers adopt alternative strategies to recognise people, based on individual distinctive features, voice, gait or other characteristics. 

Remarkably, developmental prosopagnosia can be inherited.  Studies of its familiality, by Grueter and colleagues, have shown that it tends to be inherited in a very simple manner, consistent with Mendelian dominant inheritance – i.e., it can be caused by a single mutation, which clearly segregates between affected and unaffected members within families.  What the gene (or genes, as it could be different genes in different families) encodes is not known, but there is evidence from neuroimaging that it may control the development of connectivity within the ventral visual stream – the part of the visual system devoted to processing visual form, where the fusiform face areas is found. 

In a study by Cibu Thomas and colleagues, patients with developmental prosopagnosia showed dramatic reductions in the number of nerve fibres projecting along this ventral visual stream.  In addition, the degree of reduction correlated strongly with the degree of impairment in face recognition.  Thus, while most people are pre-wired to process faces with ease, through specialized visual circuits, this appears to not be the case in these patients. 

Recent work has shown that more graded differences in the ability to recognise faces extend across the entire population – some people are just better at it than others.  Twin studies demonstrate that variation in this ability across the normal range is also strongly influenced by genetic variation, both in behavioural measures and in the responsiveness of the fusiform face area in neuroimaging studies. 

The identification of the genes involved, either in inherited prosopagnosia or in the normal variation in face-processing abilities, should lead to the elucidation of how changes in the genetic programme of brain development can result in altered connectivity in very specific circuits, dedicated to an innate and evolutionarily conserved psychological function.  


GRUETER, M., GRUETER, T., BELL, V., HORST, J., LASKOWSKI, W., SPERLING, K., HALLIGAN, P., ELLI, H., & KENNERKNECHT, I. (2007). Hereditary Prosopagnosia: the First Case Series Cortex, 43 (6), 734-749 DOI: 10.1016/S0010-9452(08)70502-1


Thomas C, Avidan G, Humphreys K, Jung KJ, Gao F, & Behrmann M (2009). Reduced structural connectivity in ventral visual cortex in congenital prosopagnosia. Nature neuroscience, 12 (1), 29-31 PMID: 19029889


Wilmer, J., Germine, L., Chabris, C., Chatterjee, G., Williams, M., Loken, E., Nakayama, K., & Duchaine, B. (2010). Human face recognition ability is specific and highly heritable Proceedings of the National Academy of Sciences, 107 (11), 5238-5241 DOI: 10.1073/pnas.0913053107


Friday, April 16, 2010

Mad Mice

The mighty mouse has become an invaluable tool in biomedical research, due to the fact that its genome can readily be manipulated, using genetic engineering techniques in embryonic stem cells.  These techniques were first developed to “knock out” or delete any gene in the mouse genome and this approach is so established now that off-the-shelf knockouts for most genes in the genome are already available from several centres.  Genetic technologies have become increasingly sophisticated, giving researchers the ability to remove a gene’s function in just some cells in an animal or just at specific stages of development and also to engineer larger sections of chromosomes so that deletions or duplications that affect multiple genes can be modeled in the mouse. 

These genetic approaches have been used extremely successfully to model a vast number of human medical conditions in the mouse, following a simple pathway from the discovery of mutations associated with the disorder in humans to the generation and analysis of mice with mutations in those same genes.  Such mice can be investigated to elucidate the functions of the encoded proteins, the pathogenic consequences and mechanisms of the mutations and ultimately to inspire and to test novel therapeutics.

When it comes to psychiatric disorders however, the development and fruitful analysis of genetic animal models has lagged considerably behind other areas of medicine.  There are several reasons for this – first, we had not, until recently, identified many mutations in humans which result in a psychiatric disorder.  Indeed, the field has for decades been misdirected by the theory that psychiatric disorders are not caused by single mutations at all, but rather by toxic combinations of genetic variants that are common in the population; such combinations would obviously be almost impossible to model in an animal.  Fortunately, this theory can now be rejected on empirical and theoretical grounds and a welcome return in psychiatry to established principles of genetics is underway.  This paradigm shift is being driven in large part by the identification of quite a large number of single mutations in many different genes or chromosomal regions, which do indeed cause psychiatric disorders.  (For more on this see our recent review, below).  

We can thus be confident that single mutations in mice can be genetically valid models for psychiatric disorders and we also now have a list of candidate mutations to generate and investigate.  But what are we modeling? 

If we want to understand how a mutation in gene A or gene B can result in schizophrenia in a human, is there any point looking at the effects of mutations in those genes in a mouse?  Can a mouse ever be said to be schizophrenic? Well, it is obviously not possible to directly model the psychology of these disorders in mice, because mice do not have human minds, they have mouse minds (or no minds, depending on one’s philosophical point of view).  We thus cannot expect to model the human-specific expression of these disorders, which may include thought disorders, delusions, paranoia, and other high-level effects.  However, we can most certainly hope to model the underlying nervous system dysfunction in a mouse. 

This approach is powerfully illustrated by a recent paper from Maria Karayiorgou and colleagues.  She and her colleagues have been investigating for many years a condition known as Velo-Cardio Facial Syndrome.  This is caused by deletion of a small region of chromosome 22 (at position 22q11), which removes about thirty genes.  It is characterised by a variety of clinical effects, including characteristic facial morphology, cleft palate and heart defects.  It is also associated with a 30-fold increased risk of psychosis.  To try and better understand how removal of one copy of this set of genes can lead to psychosis, Karayiorgou and colleagues have generated a mutant mouse where this same set of thirty genes has been deleted.  Over several years they have been able to demonstrate a variety of neurodevelopmental defects in these mutant mice along with a spectrum of concomitant behavioural alterations.  They have also been able to narrow down which genes in the region are most important.  Their latest paper takes these analyses one step further – by investigating the effects on the functions of neural networks in the mouse brain.  It is at this level that the parallels with humans are likely to be most direct and most informative. 

They show that mice with this deletion have altered communication between two brain areas known to be central to many of the defects of schizophrenia in humans – the hippocampus and prefrontal cortex.  During a task requiring working memory (which is known to be disrupted in schizophrenia patients), these two brain areas usually maintain a communication channel with each other by phase-locking oscillations in neural activity.  (Such oscillations, or brain rhythms, are seen in all areas of the brain and occur at multiple different frequencies.  When one brain area sends a signal, it is most readily received by other areas whose oscillations are timed in synchrony with it – this is because the ongoing oscillations represent peaks and troughs of membrane depolarization which affects how responsive the cells are to incoming signals).  In the 22q11-model mutant mice, the synchronization of these oscillations across the two brain regions was disrupted and the magnitude of this effect in each animal correlated with the decrement in performance on the working memory task. 

These studies represent the discovery pathway that can be followed for the growing number of other candidate genes for psychiatric disorders: find a mutation in humans, generate a corresponding mutation in mice and analyse them in an integrative fashion across developmental, anatomical, neurophysiological and behavioural levels.  These approaches should elucidate the disease-causing effects of each mutation and allow comparison across mutations to see how their effects converge.  By allowing the tools of modern genetics and neuroscience to be applied to the problem, mutant mice should ultimately suggest new ways to intervene and aid the rational design and development of new therapeutics.

Mitchell, K., & Porteous, D. (2010). Rethinking the genetic architecture of schizophrenia Psychological Medicine DOI: 10.1017/S003329171000070X

Sigurdsson T, Stark KL, Karayiorgou M, Gogos JA, & Gordon JA (2010). Impaired hippocampal-prefrontal synchrony in a genetic mouse model of schizophrenia. Nature, 464 (7289), 763-7 PMID: 20360742