Schizophrenia Research Forum - A Catalyst for Creative Thinking

Dissecting Cognition at the Synapse

13 December 2012. Two studies from the United Kingdom, published December 2 in Nature Neuroscience, use genetics to dissect the contributions of different synaptic proteins to cognition. Several of these proteins are targets of research for schizophrenia.

The first study, led by Seth Grant of the University of Edinburgh and Timothy Bussey of Cambridge University, systematically deletes members of the disc large homology (Dlg) family of genes, which encode postsynaptic density proteins. They find specific and sometimes opposing patterns of cognitive impairment in the resulting mice. The touchscreen tests they employed for this are readily adapted to humans, and the researchers found parallel deficits in mice and humans with schizophrenia-associated Dlg2 deletions or duplications.

The second study, led by Noboru Komiyama of Wellcome Trust Sanger Institute in Cambridge, swaps the cytoplasmic sides of different N-methyl-D-aspartate (NMDA) receptor subunits in order to parse their shared and unique roles in cognition. Together, the results of the two studies suggest that complex cognition is built from a diverse collection of postsynaptic proteins, which when disabled, can give rise to mental illness.

Unpacking the synapse
Recent work by Grant and colleagues has unpacked the protein-dense region of the postsynaptic side originally identified by electron microscopy, finding that it contains more than 1,450 interacting proteins (see SRF related news story). Several of these proteins have been implicated in mental disorders like schizophrenia and autism, and have helped fuel the idea that pathology at synapses underlies these disorders.

The new studies delve into the question of whether specific components of the postsynaptic signaling complex are required for particular aspects of cognition. Grant and colleagues have designed a research Genes to Cognition program to systematically disable each part of the synaptic apparatus and survey the results with an intensive battery of tests delivered by touchscreen technology. These tasks present various visual stimuli to a mouse, which then selects one by touching it with its nose; the touchscreen then detects and records this as a response. Because similar stimuli and responses can be used for cognitive testing in humans, the touchscreen setup enables more direct comparisons between species (Bussey et al., 2012).

Guided by the idea that complex cognitive behaviors—though not essential for life, but helpful for successful living—could arise from extra copies of genes that evolution could then tinker with, the team focused on gene families whose multiple members resulted from genome replication and subsequent diversification. The first paper manipulates four Dlg genes encoding synaptic scaffold proteins, which help assemble and organize the postsynaptic machinery charged with translating electrical signals into intracellular consequences. Dlg2 (also known as PSD-93) has been associated with schizophrenia (e.g., see SRF related news story), and Dlg proteins interact with components associated with autism, such as neuroligins. The second paper analyzes the roles in cognition for the cytoplasmic parts of GluN2A and GluN2B, two subunits of NMDA receptors known for their crucial roles in learning. GluN2A and GluN2B mutations have been reported for neurodevelopmental disorders such as autism, and deficits in NMDA receptor signaling have been suspected in schizophrenia (see SRF Glutamate Hypothesis).

From nose-pokes to cognition
In the first study, first author Jess Nithianantharajah and colleagues generated knockout mice for each Dlg gene: Dlg1, Dlg2, Dlg3, and Dlg4. Dlg1 knockouts die before birth, indicating that Dlg1 retains the ancestral, life-essential functions, and that mutations in Dlg2, Dlg3, and Dlg4 spawned new functions. The team focused on these last three genes and found effects of at least one gene in each of seven tests of cognition. These tests surveyed simple learning, more complex associative learning, cognitive flexibility, and attention.

The Dlg4 knockouts flunked out right away on the simple learning task: they couldn’t learn to nose-poke a visual stimulus to get a food reward, whereas Dlg2 knockouts, Dlg3 knockouts, and wild-type littermate controls readily figured it out. When further challenged with a task requiring the mice to discover through trial and error which of two complex visual stimuli delivered food, Dlg4 knockouts floundered, consistent with the idea that simple forms of learning create the foundation for more complex learning.

The same test of complex learning revealed divergent functions for Dlg2 and Dlg3, with Dlg3 knockouts learning which stimulus predicted reward faster than Dlg2 knockouts and controls did. When the researchers cranked the task complexity up a notch by requiring the mice to remember the specific locations on the touchscreen of three visual stimuli, Dlg3 mutants performed as well as controls, but Dlg2 knockouts faltered, performing only at chance levels after 50 sessions of 36 trials each.

Opposing functions for Dlg2 and Dlg3 also emerged in a test of cognitive flexibility, which required mice to unlearn previous associations: what was once the rewarded stimulus was no longer rewarded, and the previously unrewarded stimulus became the rewarded one (also called “reversal learning”). Dlg3 knockouts readily adapted to the new rules, but Dlg2 knockouts lagged behind, initially choosing the old response more often than controls but eventually getting the hang of it. Together, the results suggest that Dlg3 normally curbs the process of associating a stimulus with a reward, whereas Dlg2 expedites it, particularly in tasks associating visual and spatial information and when learning a rule change.

Distinct roles emerged in attention, too, as revealed when mice had to detect and respond to a stimulus that appeared briefly in one of five locations on the screen. Dlg2 knockouts were impaired at this, making fewer correct responses than controls did; Dlg3 knockouts, however, outperformed controls, making more correct responses even at the shortest stimulus duration of 0.2 seconds.

Human touch
To see if these distinct roles in cognition also held in humans, the researchers put people to the test with a touchscreen. Focusing on Dlg2, which is 95 percent similar in humans and mice, and which has been linked to schizophrenia through copy number variations (CNVs), the researchers assessed four people with CNVs restricted to Dlg2: three with deletions and one with a duplication. Three had been diagnosed with schizophrenia, whereas the fourth, a deletion carrier, was unaffected, though at 24 years old was possibly still at risk for developing the disorder. Compared to controls, these four on average showed impairments in human equivalent tests of visual discrimination, visual-spatial learning, cognitive flexibility, and attention that paralleled the deficits in Dlg2 knockout mice.

The researchers suggest that the distinct roles played by Dlg molecules likely reflect differences in their intracellular signaling. These scaffold proteins help organize the myriad molecules on the postsynaptic side of the synapse so that it can translate incoming electrical activity into specific intracellular messages. Differences in the binding partners between Dlg molecules could endow the signaling complex with their own specific jobs at the synapse.

Swapping tails
Looking beyond scaffold proteins, the second study finds a role for intracellular signaling domains of the NMDA receptor in cognition. Because of its storied past as a mediator of learning, NMDA receptors seemed a natural place to look for cognition-related roles. First author Tomás Ryan and colleagues focused on the GluN2 gene family encoding subunits of the NMDA receptor. Similar to the Dlg family, the GluN2 family has evolved from genome duplication events and subsequent diversification, resulting in particularly enhanced variation in the cytoplasmic region of the subunits.

Previous studies of mice lacking these cytoplasmic domains had shown that these regions were critical to GluN2 function (Sprengel et al., 1998). But it wasn’t clear whether this was due to shared or unique amino acid sequences of the cytoplasmic tails: could the cytoplasmic domain of GluN2A substitute for that of GluN2B, or vice versa? Or had they evolved distinct functions? To sort this out, the researchers engineered mice to carry chimeric GluN2 subunits: a GluN2A subunit maintained its extracellular and transmembrane domains, but swapped its cytoplasmic tail for that of the GluN2B receptor; similarly, the GluN2B chimeric subunit contained the cytoplasmic tail of the GluN2A subunit.

This exchange did not jumble the normal brain expression patterns of the subunits or disrupt the synaptic currents flowing through NMDA receptors, suggesting that glutamate signals were received as usual on the extracellular side. But what unfolded downstream in the realms of cognition often depended on the cytoplasmic tail. The researchers put the mice through touchscreen tests probing eight aspects of cognition, including visual discrimination, reversal learning, fear conditioning, motor coordination, and impulsivity. While mice carrying GluN2A subunits lacking a cytoplasmic domain were impaired in all of these domains, those carrying chimeric GluN2A subunits that contained the GluN2B cytoplasmic domain were impaired in only six out of eight of these behaviors. This “rescue” indicates that shared sequence in the cytoplasmic domains of GluN2B and GluN2A mediate these functions. Meanwhile, the two phenotypes not rescued by this chimera—impulsivity and motor activity—rely on a sequence unique to the GluN2A cytoplasmic domain.

In all, the GluN2A and GluN2B cytoplasmic tails were able to substitute for each other in reversal, associative, and motor learning. But a unique role for the GluN2A cytoplasmic tail emerged for locomotor activity and impulsivity, and for the GluN2B cytoplasmic tail in perceptual learning, anxiety, impulsivity, and motor coordination.

The researchers found various disruptions in synaptic plasticity in these mice, suggesting mechanisms for their cognitive deficits. The divergent roles for the cytoplasmic tails may reflect their different interactions with proteins of the synaptic apparatus. Consistent with this, immunoprecipitation experiments revealed that GluN2B cytoplasmic tails brought down more Dlg2 and Dlg4 than GluN2A cytoplasmic tails did, suggesting that these subunits are surrounded by slightly different complements of synaptic machinery.—Michele Solis.

Nithianantharajah J, Komiyama NH, McKechanie A, Johnstone M, Blackwood DH, Clair DS, Emes RD, van de Lagemaat LN, Saksida LM, Bussey TJ, Grant SG. Synaptic scaffold evolution generated components of vertebrate cognitive complexity Nat Neurosci. 2012 Dec 2. Abstract

Ryan TJ, Kopanitsa MV, Indersmitten T, Nithianantharajah J, Afinowi NO, Pettit C, Stanford LE, Sprengel R, Saksida LM, Bussey TJ, O'Dell TJ, Grant SG, Komiyama NH. Evolution of GluN2A/B cytoplasmic domains diversified vertebrate synaptic plasticity and behavior. Nat Neurosci. 2012 Dec 2. Abstract

Comments on News and Primary Papers
Comment by:  Jennifer Barnett (Disclosure)
Submitted 13 December 2012
Posted 13 December 2012

Cognitive function is highly heritable (Devlin et al., 1997), yet we have relatively little understanding of which genes regulate either general intelligence or specific cognitive functions. A long list of mutations can cause the large cognitive impairments that we class as learning disability—including many of the same CNVs associated with cognitive disorders such as autism and schizophrenia (Guilmatre et al., 2009). Prior to the GWAS era, it was generally assumed that normal variation—outside of the range of learning disability—would be regulated by common variants of small effect. Yet GWAS, a technology well suited to detecting common variants of small effect, has not massively increased our understanding of the genetic basis of cognition.

One explanation for this relative lack of success is that, compared with quantitative phenotypes such as height or BMI, the measurement of cognition can be time consuming and therefore costly, so really large-scale studies are rare. Moreover, any two studies are very unlikely to use identical cognitive tests, introducing error to the phenotypic measurement and making it difficult to combine datasets. In this context, the Nithianantharajah paper is a beautiful example of how translational research using relatively simple but neuroscience-led assays can increase our understanding of the genetics of cognition, while avoiding the need for the ever-increasing sample sizes.

In particular, I was impressed by the building up of related but increasingly complex forms of cognition across the rodent tasks, and the use of the closest possible analogues between mouse and human assays. (Disclosure: I am employed by Cambridge Cognition, the suppliers of the CANTAB tests used in the human phenotyping.) Adding to these very careful phenotyping methodologies, the parallel experiment across both mouse and human "knockouts" is a really elegant piece of translational neuroscience.

Like many traits underlying brain function, it is inherently easy to believe that variants that have large effects on cognition would create strong evolutionary advantages or disadvantages. The authors here demonstrate not only that a related family of genes affects multiple aspects of cognitive function, but also that variation in these genes produces different cognitive tendencies in the mouse, including reciprocal effects of variants of Dlg2 and Dlg3, which seem, at least at first pass, to be conserved in human behavior.

There is a lot to digest in this and the companion paper by Ryan et al., but the methodology appears to have been very useful here in understanding cognition, and may provide useful insights for researchers trying to decipher other aspects of the schizophrenia phenotype.


Devlin B, Daniels M, Roeder K (1997). The heritability of IQ. Nature.388(6641):468-71. Abstract

Guilmatre A, Dubourg C, Mosca AL, Legallic S, Goldenberg A, Drouin-Garraud V, Layet V, Rosier A, Briault S, Bonnet-Brilhault F, Laumonnier F, Odent S, Le Vacon G, Joly-Helas G, David V, Bendavid C, Pinoit JM, Henry C, Impallomeni C, Germano E, Tortorella G, Di Rosa G, Barthelemy C, Andres C, Faivre L, Frébourg T, Saugier Veber P, Campion D. (2009). Recurrent rearrangements in synaptic and neurodevelopmental genes and shared biologic pathways in schizophrenia, autism, and mental retardation. Arch Gen Psychiatry;66(9):947-56. Abstract

View all comments by Jennifer Barnett