Cellular Basis of Reduced Cortical Reelin Expression in Schizophrenia

The subjects have been described elsewhere (7), and their demographic characteristics are summarized in Table 1. Sections from the hippocampal formation and dorsolateral prefrontal cortex (Brodmann’s area 9/46) were used to determine reelin mRNA by cellular in situ hybridization and emulsion autoradiography (7).

In the dorsolateral prefrontal cortex, reelin mRNA was localized over layer I neurons, scattered interneurons, and some interstitial white matter neurons. In the hippocampal formation, reelin-expressing cells were prominent in the outer molecular layer of the dentate gyrus, close to the hippocampal fissure. Interstitial white matter neurons in the parahippocampal white matter as well as some interneurons and hilar neurons were also labeled. No other cell types, including pyramidal neurons or granule cells, were clearly labeled.

In the dorsolateral prefrontal cortex, reelin mRNA per cell was reduced in schizophrenia over all three neuron populations, including interstitial white matter neurons (Table 1). These cells also expressed less reelin mRNA in the hippocampal formation in schizophrenia, and the other neurons showed no significant differences. For the layer I and dentate gyrus molecular layer neurons, we also estimated the density of reelin-positive cells; we found a reduction of about 30% in schizophrenia (Table 1). Within the schizophrenia group, reelin expression did not relate to variables such as age at onset or negative symptoms.

In the dentate gyrus, the reelin-expressing neurons in the outer molecular layer make contact with granule cells and regulate connectivity (8). We examined whether there was a correlation between reelin expression and synaptic markers in this region. To do this, “reelin load” in the outer molecular layer was calculated by multiplying reelin mRNA signal per cell by the cell density; reelin load was then correlated with the levels of three synaptic protein mRNAs (synaptophysin, VGLUT1, and GAP-43), measured previously over the granule cells. There were significant positive correlations between reelin load and GAP-43 (r=0.56, N=18, p<0.03) and VGLUT1 (r=0.57, N=25, p=0.005) mRNAs and a nonsignificant correlation between reelin load and synaptophysin mRNA (r=0.44, N=19, p=0.08).

Our data confirm that reelin expression is reduced in schizophrenia (2–5, 7) and show its cellular origin in the hippocampal formation and the dorsolateral prefrontal cortex. The distribution in the dorsolateral prefrontal cortex is similar to that in the superior temporal gyrus (7). The reduction in schizophrenia is apparent in terms of both reelin mRNA per cell and density of reelin-positive cells (Table 1). With regard to the latter finding, other studies show that there is no loss of the neurons themselves in schizophrenia (9, 10). Thus, the present results are indicative of a down-regulation of reelin expression.

In both areas studied, interstitial white matter neurons expressed less reelin mRNA, drawing renewed attention to this developmentally critical cell population—the presumed remnants of the cortical subplate—previously implicated in schizophrenia (see reference 7). In the dorsolateral prefrontal cortex, reelin mRNA was also reduced in layer I neurons, the survivors of Cajal-Retzius cells, essential for neuronal migration and lamination (1). Possibly, reduced reelin might be causal to a developmental anomaly affecting interstitial white matter neurons and Cajal-Retzius cells, but, equally, it might be consequential or unrelated to it. Regardless, since both neuron populations form synaptic contacts and persist in adulthood, and given that reelin has key roles in plasticity, the abnormality likely has ongoing effects on the functioning of neural circuits. Note that although reelin mRNA is not detected in pyramidal neurons, reelin is (11), perhaps because the neurons internalize the protein (12) after its secretion by the GABA-ergic cells with which they are synaptically connected. If so, reelin may have a role linking GABA-ergic with glutamatergic aspects of the disease pathophysiology, including its synaptic component (6). The correlations between reelin load and the presynaptic protein transcripts provide some empirical support for this proposal.

Although layer I neurons and most cortical interneurons are GABA-ergic, most interstitial white matter neurons are not, at least in rodents (13). Moreover, the various reelin-expressing cell populations differ in place and time of origin (14). Thus, our data suggest that the involvement of reelin in schizophrenia is not related directly to transmitter phenotype or embryological history of the neurons concerned. Instead, it may reflect an intrinsic property of the gene and its regulation. Possibilities include epigenetic factors (5) or an interaction with susceptibility genes, which themselves have effects on synaptic plasticity (15).

Presented in part at the IXth International Congress on Schizophrenia Research, Colorado Springs, Colo., March 29–April 3, 2003, and the XIIth Winter Workshop on Schizophrenia, Davos, Switzerland, Feb. 7–13, 2004. Received May 9, 2005; revision received June 22, 2005; accepted July 15, 2005. From the Department of Psychiatry, University of Oxford. Address correspondence and reprint requests to Dr. Eastwood, University of Oxford, Department of Psychiatry, Neurosciences Bldg., Warneford Hospital, Oxford OX3 7JX, United Kingdom; [email protected] (e-mail).Supported by the Stanley Medical Research Institute and Medical Research Council. Dr. Eastwood held the British Medical Association Margaret Temple Fellowship.