11 May 2015
May 12, 2015. A mouse model of dendritic spine loss has revealed how changes in cortical synapses cause circuit-level abnormalities that affect the function of distant brain regions to produce behaviors relevant to schizophrenia and other psychiatric diseases.
The mouse line, made by Scott Soderling of Duke University, Durham, North Carolina, and colleagues, shows a deficit of dendritic spines on frontal cortical excitatory neurons, thanks to a targeted knockout of a critical regulator of the dendritic spine cytoskeleton. In a report published May 4 in Nature Neuroscience, the researchers show that dendritic spine loss in these neurons boosts their excitability and alters a long-range circuit that regulates dopamine levels in the striatum, resulting in aberrant locomotion and repetitive behaviors that can be reversed by antipsychotic drugs.
"The researchers recapitulate something known about the pathology of schizophrenia, that is, the lower density of dendritic spines in the cortex, and get something else that is seen in the illness—increased dopamine activity subcortically," said David Lewis of the University of Pittsburgh in Pennsylvania. "This study helps to tie those two observations together and provides proof of concept of the directionality in the relationship," Lewis told SRF.
Lewis and others had previously postulated that a cortical circuit abnormality could lead to a hyperdopaminergic state subcortically (Lewis and Gonzalez-Burgos, 2006). The new work reveals a molecular- and neuronal-level basis for just such a circuit malfunction.
"Many studies move from molecules to whole animals or even humans, but this middle portion of how circuits in the brain are affected is not really well understood," Peter Penzes of Northwestern University Feinberg School of Medicine in Chicago, Illinois, told SRF. "In this paper, they did an excellent job of filling in that gap."
The mice used in the new study were created by targeting synaptic actin regulation, which controls dendritic spine formation and maintenance. This pathway includes DISC1 and other schizophrenia risk genes, and pathway analysis of genetic data implicates it in schizophrenia and other psychiatric disorders. Soderling and colleagues disrupted the pathway by disabling the Arp2/3 complex, a convergence point for multiple modulators of the actin cytoskeleton including several schizophrenia genes.
As reported previously, postnatal Arp2/3 disruption in forebrain neurons led to synaptic dysfunction and progressive dendritic spine loss in the mice, accompanied by locomotor hyperactivity and repetitive behaviors (Kim et al., 2013). The mice also showed sensory motor gating deficits (reduced prepulse inhibition) as well as non-motor effects including reduced social interaction and deficits in episodic and working memory.
"The key finding [from the previous work] was that we can effectively model aspects of psychiatric disorders by genetically targeting the synaptic cytoskeleton," Soderling told SRF. "The mice had behavioral deficits that tracked well with what would be expected for positive, negative, and cognitive symptoms," he said.
In the new study, first author and senior postdoc Il Hwan Kim found that the mice showed another hallmark of schizophrenia—they responded to antipsychotic drugs, with haloperidol or clozapine effectively reversing the locomotor behaviors. The result was a bit of a surprise, Soderling told SRF. "There was no clear connection between the synaptic cytoskeleton and what was known about how these antipsychotics work. There was zero connection between the two," he said.
The drugs are known to block dopamine receptors, so the investigators looked at dopamine levels in the striatum and found they were elevated in knockout mice. That elevation was a direct result of the cortical knockout, since reconstitution of Arp2/3 by localized viral re-expression in frontal cortical neurons lowered striatal dopamine and relieved locomotor symptoms. There was no change in spines in the striatum; instead, the dopamine elevation tracked with loss of cortical spines.
How does spine loss in frontal cortex increase striatal DA? Using retrograde and anterograde viral tracing techniques, the investigators tracked projections from the frontal cortical neurons directly to dopamine-producing neurons in the ventral tegmental area and substantia nigra pars compacta (VTA/SNc). Pyramidal cells in the cortex made excitatory synaptic contacts with VTA/SNc neurons that produce striatal dopamine.
This finding suggested that, unexpectedly, spine loss in the cortex might increase excitatory drive to the VTA/SNc neurons to drive the motor behaviors. A closer look at the pyramidal cells using electron microscopy revealed that the loss of spines was accompanied by an increase in excitatory synaptic contacts with the dendritic shaft and multiple synapses on some spines. This abnormal rewiring of contacts was associated over time with a progressive increase in excitability of the cortical cells measured by whole cell patch clamp.
"When we see the loss of these spines by using the standard stains, such as Golgi stain, we often assume that means the neurons are losing their excitatory input, and therefore they are probably less active, and it turned out the exact opposite was true," Soderling said.
Finally, to prove that cortical activation of the cortical-to-midbrain circuit could induce motor hyperactivity in wild-type mice, the researchers employed an optogenetic strategy. The researchers injected an adeno-associated virus for Cre-activated expression of light-sensitive channelrhodopsin in the cortical neurons, and a Cre expression virus in the VTA/SNc. This resulted in the expression of channelrhodopsin in the cortical neurons projecting to the VTA/SNc only. In these mice, illumination of the frontal cortex was sufficient to increase locomotion, and the increase was sensitive to haloperidol treatment. Additional experiments showed that activation of the cortico-VTA/SNc circuit could modulate striatal dopamine.
Overall, the results reveal a causal relationship between the main theories of schizophrenia pathology—cortical spine loss, frontal cortical hyperactivity, and striatal dopamine imbalance—whose relationship was previously unclear. "Those three theories seemingly had no connection with each other. They were pieces of the puzzle that didn't seem to fit together," said Soderling. "At the end, all of the pieces fit in the context of this model," he said.
"This study is a tour de force, because they have used so many different approaches and different ways to manipulate expression of these genes and to track circuits," said Penzes, who thinks this study is going to be a template for future work. "It sets the standard for how we should be looking at these types of questions," he said.
Lewis pointed out that the work is convergent with recent findings suggesting that the pyramidal cells in the cortex are part of the primary pathology in schizophrenia, and that changes in the dopamine system subcortically and the GABA inhibitory system in the cortex might be consequences or compensations, respectively (Tatard-Leitman et al., 2015). To further characterize the model, Lewis said he would like to see the state of GABAergic inhibition in the cortex, which is thought to be downregulated in schizophrenia. "It would be useful to know in their model what actually happens to key molecules in GABAergic transmission like the expression of GAD67, which is known to be lower in schizophrenia," he said.
Modeling of polygenic diseases such as schizophrenia in animals is a challenge. Many genes have large effects in a small number of people or small effects in large numbers of people. This makes functional analysis of individual genes in animals difficult, Soderling explained. "We hypothesized that an effective way of modeling polygenic disorders that may cluster within the actin signaling pathway would be to conditionally target one of these final downstream outputs such as Arp 2/3," he said, even though the variants of Arp2/3 itself have not been identified to date in genetic studies.
There can be some risk with that approach, said Penzes: "[I]f the gene one studies hasn't been directly identified in a genetic study, the link may not be as strong." But in this case, Penzes thinks the investigators "are making a very good case, because the study is very well informed by psychiatric genetics."
It will be important to study more about the synaptic and other changes that underlie cortical cell excitability, Penzes said. The authors speculate that synaptic repositioning from spines to dendritic stalks might account for hyperexcitability, but the mechanism needs to be further defined.
Besides increased motor activity, the mice also display sensory gating defects, and social and cognitive deficits—behaviors that may be relevant to schizophrenia and other disorders such as autism. Soderling thinks this model offers a way to figure out the circuits behind those behaviors, too. "That's what we're really interested in. We think we can use the same strategies that we outlined in this paper to chase down the etiology of some of those other abnormalities," he said.—Pat McCaffrey.
Kim IH, Rossi MA, Aryal DK, Racz B, Kim N, Uezu A, Wang F, Wetsel WC, Weinberg RJ, Yin H, Soderling SH. Spine pruning drives antipsychotic-sensitive locomotion via circuit control of striatal dopamine. Nat Neurosci. 2015 May 4. Abstract