1 March 2012. Neurons from individuals with schizophrenia and bipolar disorder harbor genomic copy number differences that are circuit- and diagnosis-specific, according to a study published online February 6 in Archives of General Psychiatry. Led by Francine Benes of McLean Hospital in Belmont, Massachusetts, the study suggests that abnormal losses or gains of DNA segments can accumulate in mature neurons, leading to their malfunction.
The findings provide a first glimpse of the state of the DNA within neurons in these disorders, and suggest that some malleability is present. Recent genetic studies have offered up copy number variants (CNVs)—deletions or duplications of DNA segments—as potential causal factors for psychiatric disease (see SRF related news story; see SRF related news story). However, the copy number changes identified by Benes and colleagues are proposed to lie downstream of causal factors, arising instead in fully mature neurons later in life and possibly driving the specific brain dysfunction found in schizophrenia and bipolar disorder.
“Now we may have a basis for understanding dysfunction in terms of genomic integrity,” Benes told SRF. “When we see the amount of genomic integrity that has been changed, the critical question may be, How much dysfunction is that really related to?”
There's something about GAD67
One consistent marker of altered brain function found in postmortem brain samples from people with schizophrenia and bipolar disorder has been the decreased expression of glutamate decarboxylase (GAD67), an enzyme that produces γ-aminobutyric acid (GABA). This has led to various lines of research on possible GABAergic dysfunction in schizophrenia (see SRF related news story). Trying to understand why there should be such a deficit in GAD67, Benes and colleagues published a gene expression profiling study in 2007 in which they turned up a network of genes involved in GAD67 regulation, as well as neurogenesis, cell cycle, and DNA repair (see SRF related news story). Though these last processes are more appreciated in proliferating cells, they may also be critical for protecting the genome in a fully mature neuron—an idea that casts the postmitotic, differentiated neuron as a state that is actively maintained, rather than a static endstage.
The new study asked whether any telltale signs of genomic degradation could be found in the brain, and whether they were associated with a particular diagnosis and cell type. The researchers focused on the stratum oriens of the hippocampus, which contains exclusively GABAergic cells and exhibits the GAD67 deficits in schizophrenia and bipolar disorder. Building on previous work showing distinct patterns of gene expression in these cells, depending on whether they were in the CA3/2 or CA1 region of the hippocampus (see SRF related news story), the team compared copy number measures across these circuits, as well as across diagnoses.
Copy number intensities
First author Guoqing Sheng and colleagues used laser microdissection to excise the stratum oriens from CA3/2 and CA1 in postmortem brain samples from 15 individuals with schizophrenia, 15 with bipolar disorder, and 15 controls. The DNA from these samples was probed with a single nucleotide polymorphism microarray, which can reveal changes in copy number of DNA segments by changes in signal intensity. The researchers did a targeted search for these irregularities, analyzing only those SNPs tagging the 28 previously identified as belonging to a GAD67 regulatory network.
In CA3/2, the researchers detected a decrease in copy number intensity of the GAD67 gene in schizophrenia samples, suggesting some loss of that gene. Specifically, the mean copy number intensity was 1.68 in schizophrenia, compared to 2.16 in controls—a difference amounting to a 22 percent decrease. In bipolar disorder, a 25 percent decrease was measured. In CA1, the story was somewhat similar, with a 27 percent decrease in schizophrenia compared to controls, but no significant decrease for bipolar disorder.
Looking across all 28 genes, CA3/2 emerged as a hotspot of genomic degradation: in schizophrenia samples, 15 of the 28 genes showed significant differences in copy number intensities compared to controls, and in bipolar samples, 18 out of 28 genes showed this. In contrast, in CA1, only 10 out of 28 genes showed significant changes in schizophrenia, and only three out of 28 in bipolar samples.
These patterns differed between disorders. Among the genes specifically involved in GAD67 regulation, the researchers pinpointed copy number intensity changes in five genes (HDAC11, DAXX, PAX5, RUNX2, and CCND2) that were similar in magnitude and direction to those reported in their gene expression profile study in 2007. In CA3/2, schizophrenia samples had a marked increase in HDAC11, a histone deacetylase involved in epigenetic regulation, and DAXX, a transcription regulator, whereas bipolar samples did not. Conversely, bipolar samples exhibited decreases in copy number intensity for RUNX2, a transcription factor involved in cell differentiation, and CCND2, a crucial cyclin that regulates the cell cycle, whereas schizophrenia samples showed no change in these compared to controls. These results suggest that the decrease in GAD67 expression shared by schizophrenia and bipolar disorder may be brought about by distinct molecular mechanisms.
Copy number differences were also detected in CA3/2 among neurogenesis-related genes, including an increase in intensity in growth factor-encoding VEGF and a decrease in NRG1 in schizophrenia, and increases for both in bipolar samples. Among cell cycle and DNA repair genes, a copy number reduction in transcription factor E2F3 was detected in schizophrenia, and increases in a DNA repair enzyme MBD4 emerged for both schizophrenia and bipolar; these kinds of changes were not so apparent in CA1.
Local context matters
Asking whether these genomic irregularities were associated with gene expression, the team found robust correlations between copy number intensity and corresponding mRNA levels obtained in their previous study. Within CA3/2, significant positive correlations were found for schizophrenia (r = 0.649, p = 0.0003) and bipolar disorder (r = 0.772, p = 0.0002). However, no correlations were found within CA1, which the authors propose may be due to local environment influences on transcriptional regulation (see SRF related news story). Differences in local circuitry may also explain why CA3/2 is a hotspot for genomic degradation compared to CA1. Stratum oriens neurons in CA3/2, but not CA1, receive input from the basolateral amygdala, a region noted for mediating emotional responses, and studied in the context of psychiatric illness. Benes proposes that synaptic activity from this input could promote oxidative stress in the already highly active GABAergic cells, and trigger a cascade of molecular changes, including DNA damage. Her team is now exploring this idea in rat hippocampus (e.g., Gisabella et al., 2009).
The results reveal a complicated checkerboard of gene-, circuit-, and diagnosis-specific DNA changes, and suggest that a subtler kind of genomic degradation lying somewhere between the cell death of neurodegenerative disorders and the rampant cell proliferation driving cancer may be at work in psychiatric illness. As future research discerns whether, and how much, these kinds of changes contribute to psychiatric illness, it will be important to keep in mind the cellular context of a given genetic glitch. “We have to be cautious about extrapolating from a particular anatomic locus to different parts of the brain,” Benes said.—Michele Solis.
Sheng G, Demers M, Subburaju S, Benes FM. Differences in the Circuitry-Based Association of Copy Numbers and Gene Expression Between the Hippocampi of Patients With Schizophrenia and the Hippocampi of Patients With Bipolar Disorder. Arch Gen Psychiatry. 2012 Feb 6. Abstract