- Email: [email protected]
Selective deficits in prefrontal cortical GABAergic neurons in schizophrenia defined by the presence of calcium-binding proteins
Selective Deficits in Prefrontal Cortical GABAergic Neurons in Schizophrenia Defined by the Presence of Calcium-Binding Proteins Clare L. Beasley, Zhi J. Zhang, Iain Patten, and Gavin P. Reynolds Background: Postmortem studies have provided evidence for abnormalities of the ␥-aminobutyric acid (GABA)ergic system in schizophrenia, including deficits of GABAcontaining interneurons. The calcium-binding proteins parvalbumin, calbindin, and calretinin can be used as markers for specific subpopulations of cortical GABAergic interneurons. Methods: Following our previous observation of a reduction in the density of parvalbumin- but not calretininimmunoreactive cells in the prefrontal cortex (Brodmann area 10) in schizophrenia, we have quantified the laminar density of neurons immunoreactive for the calcium-binding proteins parvalbumin, calbindin, and calretinin in a further prefrontal cortical region (Brodmann area 9) in patients with schizophrenia, bipolar disorder, major depression, and in matched control subjects (each group n ⫽ 15). Results: Initial statistical analysis revealed reductions in the total cortical density of parvalbumin- and calbindinbut not calretinin-immunoreactive neurons in schizophrenia relative to control subjects. Further analysis comparing individual laminar densities between groups indicated that, following correction for multiple comparisons, only a reduction in calbindin-immunoreactive neurons in cortical layer II in the schizophrenic group attained statistical significance. Conclusions: These findings suggest that deficits of specific GABAergic neurons, defined by the presence of calcium-binding proteins, are present in schizophrenia. Trends toward similar reductions are observed in bipolar disorder. Biol Psychiatry 2002;52:708 –715 © 2002 Society of Biological Psychiatry Key Words: Parvalbumin, calbindin, calretinin, schizophrenia, bipolar disorder, interneuron
From the Department of Biomedical Science, University of Sheffield, Sheffield, United Kindom. Address reprint requests to Clare L. Beasley, Ph.D., Institute of Psychiatry, Section of Experimental Neuropathology and Psychiatry, DeCrespigny Park, London SE5 8AF, United Kingdom. Received July 23, 2001; revised December 10, 2001; revised February 1, 2002; accepted February 11, 2002.
© 2002 Society of Biological Psychiatry
Introduction
T
here is some intriguing evidence indicative of anomalous ␥-aminobutyric acid (GABA)-ergic neurotransmission in the prefrontal cortex in schizophrenia. In particular, postmortem analysis of brain tissue from this region has provided insight into cellular abnormalities of the GABAergic system, including a loss of presumptive interneurons in more superficial cortical layers (Benes et al 1991) and a reduction in the density of interneurons expressing mRNA for glutamic acid decarboxylase (GAD), the synthesizing enzyme for GABA (Akbarian et al 1995; Volk et al 2000). Further studies have found increased binding to the GABAA receptor in the prefrontal cortex in schizophrenia (Benes et al 1996; Hanada et al 1987). This increase in GABAA receptors has been interpreted as reflecting a compensatory upregulation of postsynaptic receptors, due to losses of GABAergic interneurons in this region (Benes et al 1996). Although similar studies of bipolar disorder and major depression are limited, there are indications that expression of GAD mRNA and the density of GAD-immunoreactive terminals are reduced in the prefrontal cortex in bipolar disorder (Benes et al 2000; Guidotti et al 2000). To understand the functional significance of these findings, it is important to further define the GABAergic changes observed in schizophrenia. A number of criteria, including morphologic appearance and biochemical properties, can be utilized to discriminate distinct classes of GABAergic interneurons. Cortical GABAergic cells can be subdivided on the basis of co-localized neuropeptides, including somatostatin, cholecystokinin, neuropeptide Y, and vasoactive intestinal polypeptide (Somogyi et al 1984). Losses of cholecystokinin, somatostatin, and vasoactive intestinal polypeptide have been described in the cortex in some schizophrenic patients (Gabriel et al 1996; Nemeroff et al 1983). Consistent with this, deficits in cholecystokinin mRNA, have been reported in the frontal and temporal cortex (Virgo et al 1995) as well as in the entorhinal cortex (Bachus et al 1997). Essentially nonoverlapping subpopulations of GABAergic neurons can also be 0006-3223/02/$22.00 PII S0006-3223(02)01360-4
Prefrontal GABAergic Deficits in Schizophrenia
709
BIOL PSYCHIATRY 2002;52:708 –715
Table 1. Summaries of Demographic, Clinical, and Histological Information of Schizophrenic, Bipolar Disorder, Major Depression, and Control Groups Group Demographic variable Age (y, mean ⫾ SD) Gender (male, female) Postmortem interval (hr, mean ⫾ SD) Cause of death CPD Accident Suicide Pneumonia Other pH (mean ⫾ SD) Time in fixative (mo, mean ⫾ SD) Brain hemisphere used (Right : Left) Duration of illness (y, mean ⫾ SD) Lifetime antipsychotic dosea (mg, minimum; median; maximum) Current alcohol/drug abuse
Control (n ⫽ 15)
Schizophrenia (n ⫽ 15)
Bipolar disorder (n ⫽ 15)
Major depression (n ⫽ 15)
48.1 ⫾ 10.7 9M, 6F 23.7 ⫾ 9.9
44.2 ⫾ 13.1 9M, 6F 33.7 ⫾ 14.6
42.3 ⫾ 11.7 9M, 6F 32.5 ⫾ 16.1
46.4 ⫾ 9.3 9M, 6F 27.5 ⫾ 10.7
13 2 0 0 0 6.3 ⫾ .2 4.40 ⫾ 3.87 7:8 0⫾0 0; 0; 0
7 0 7 0 1 6.2 ⫾ .3 11.20 ⫾ 8.48 6:9 21.3 ⫾ 11.4 0; 35,000; 200,000
3 0 9 1 2 6.2 ⫾ .2 9.67 ⫾ 3.62 8:7 20.1 ⫾ 9.7 0; 7,500; 60,000
7 1 4 1 2 6.2 ⫾ .2 8.40 ⫾ 6.59 6:9 12.7 ⫾ 11.1 0; 0; 0
0
3
4
3
CPD, cardiopulmonary disease. a Lifetime neuroleptic dose in fluphenazine milligram equivalent dose.
defined by the presence of the calcium-binding proteins parvalbumin, calbindin, and calretinin (Celio 1990; Demeulemeester et al 1988). Studies have revealed that the most characteristic morphologic types of neurons that express parvalbumin are large basket and chandelier cells (Akil and Lewis 1992; Lewis and Lund 1990); calbindin is present in many double bouquet cells (DeFelipe et al 1989), whereas calretinin-immunoreactive neurons are generally bipolar, double bouquet, and Cajal-Retzius cells (Jacobowitz and Winsky 1991). We have previously observed a reduction in the density of parvalbumin-immunoreactive cells in the prefrontal cortex (Brodmann area [BA] 10) in schizophrenia (Beasley and Reynolds 1997), but we found no loss of calretinin-immunoreactive neurons in the same region (Reynolds and Beasley 2001). This led us to suggest that the GABAergic deficits described in the frontal cortex in schizophrenia may be due to specific reductions in parvalbumin-immunoreactive cells. To test this, the density of interneurons immunoreactive for parvalbumin, calbindin, and calretinin was quantified in a further prefrontal cortical region (BA 9) in a series of brains from patients with schizophrenia, bipolar disorder, major depression, and from matched control subjects. As the morphology, distribution, function, and ontology of distinct subpopulations of GABAergic neurons differs, identifying deficits of specific interneurons in the brain in schizophrenia could throw light on the putative mechanisms underlying the disturbances of cortical function observed in this disorder.
Methods and Materials Human Subjects Samples were obtained from the Stanley Foundation Neuropathology Consortium brain collection. Brains were obtained from patients diagnosed with schizophrenia, bipolar disorder, major depression, and from matched control subjects (each group n ⫽ 15). Demographic details are provided in Table 1.; for further descriptions see Torrey et al (2000). Final diagnoses were established using DSM-IV criteria and routine microscopic and toxicological examinations carried out on all cases. Cause of death and an estimate of total lifetime intake of antipsychotic medication (in fluphenazine mg equivalents) were detailed for all cases.
Immunocytochemistry 10-m-thick paraffin sections of the dorsolateral prefrontal cortex were processed for parvalbumin-, calbindin-, or calretininimmunoreactivity randomly and blind to diagnosis. For each antibody two sections, at least 50 m apart, were sampled for analysis. Tissue sections were heated in a microwave oven on full power (650 W) for 3 ⫻ 10 min in .05 mol/L tris buffer, pH 9.0, to aid antigen retrieval. Sections were blocked with normal serum, then incubated for 36 hours at 4°C with either a mouse monoclonal antibody against parvalbumin (clone PA-235, Sigma, St. Louis, MO), a mouse monoclonal antibody against calbindin (Sigma) or a goat polyclonal antibody against calretinin (Swant, Bellinzona, Switzerland), each at a dilution of 1:5000. For parvalbumin the antibody diluent also contained 3 mmol/L calcium chloride. Following this the sections were processed by the avidin– biotin method of Hsu et al (1981) using
710
BIOL PSYCHIATRY 2002;52:708 –715
a Vectastain ABC kit (Vector Laboratories, Burlingame, CA). Peroxidase was visualized using diaminobenzidine, intensified with nickel chloride, and sections counterstained with toluidine blue. Negative control sections, in which the primary antibody was omitted from the staining protocol, were run alongside the test series.
Image Analysis Parvalbumin-, calbindin-, and calretinin-immunoreactive neurons were plotted at 100⫻ magnification using an Olympus microscope (Olympus Optical Co (Europa) GmbH Hamburg, Germany) equipped with camera lucida. The density of neuronal profiles was expressed as mean values (⫾ SE) per mm2 per cortical layer and was the result of counts from a total of 10 500-m-wide cortical traverses, each from the pial surface to the white matter border, selected at random from the two slides. The placement of the traverses was restricted to areas of the cortex that were comparatively level, avoiding the depths of the sulci. Cortical width was also measured for each of these traverses. Counts were made in cortical region BA 9 in each case. The boundaries of BA 9 with BA 46 were determined according to the criteria of Rajkowska and Goldmann-Rakic (1995) and Daviss and Lewis (1995).
Statistical Analysis To test our initial hypothesis that the GABAergic deficits previously described in schizophrenia result from a reduction in parvalbumin-immunoreactive neurons but not calbindin- or calretinin-containing cells, we compared mean total cortical densities of each of these cell populations between the schizophrenic and control groups, using univariate analysis of variance (ANOVA). The demographic and histologic variables listed in Table 1 were included as covariates in the analysis if 1) they differed between the schizophrenic group and the control group at the 10% significance level (ANOVA); and 2) they could be shown empirically to predict densities at the 10% significance level (ANOVA or Spearman’s Rank correlation). Following on from this, the mean density of parvalbumin-, calbindin-, and calretinin-immunoreactive neurons in each cortical layer for each of the three patient groups was compared with that of the control group. This allowed us to look for disease and laminar specificity. Mean densities were analyzed by univariate ANOVA, using diagnoses as contrasts. Confounding variables were determined as above. To account for multiple layer-wise testing, p values of .010 (for calbindin and calretinin) and .013 (for parvalbumin) were determined to be statistically significant; these values represent the 5% level divided by the number of layers analyzed. In the schizophrenic and bipolar disorder groups the effects of total antipsychotic drug dosage on parvalbumin-, calbindin-, and calretinin-immunoreactive neuronal densities were was also assessed. All statistical analysis was carried out in SPSS 10 (CSPSS Inc, Chicago, IL).
Results Parvalbumin-immunoreactive neurons in the prefrontal cortex were typically intensely stained. Immunoreactive
C.L. Beasley et al
neurons were present predominantly in layers III, IV, and V and were also observed in layers II and VI, but not in layer I or in the white matter (Figure 1a). Parvalbuminpositive cells appeared to be nonpyramidal and consisted of a variety of morphologies, including small ovoid perikarya, large multipolar neurons, and occasionally bitufted cells (Figure 1 d). A dense plexus of immunoreactive material was also distributed throughout the neuropil of layers III, IV, and V and consisted of stained processes and puncta, which have been shown to represent axonal terminals (DeFelipe and Jones 1991). Calbindinimmunoreactive neurons were present throughout the cortical width and were predominant in layer II and superficial layer III (Figure 1b). Immunoreactive neurons comprised both densely stained nonpyramidal cells and faintly stained pyramidal cells (Figure 1e). Calretininimmunoreactive cells also comprised a population of nonpyramidal neurons that were distributed throughout all six cortical layers and the subjacent white matter, although again predominantly in the more superficial laminae (Figure 1c). Positive neurons were typically bipolar in morphology (Figure 1f); layer I contained horizontally oriented processes and immunoreactive puncta in addition to labeled neurons. Summaries of mean laminar densities for each neuronal subpopulation are shown in Table 2. At the 10% level, mean fixation time (p ⫽ .011) and mean postmortem interval (p ⫽ .038) were significantly higher in the schizophrenic group than in the control group; however, Spearman’s Rank correlation indicated no significant relationship between these variables and the density of any neuronal population at the 10% level, and so group comparisons were not adjusted for these variables. Significant correlations were observed between tissue pH and parvalbumin (p ⫽ .001, r ⫽ .426) and calbindin (p ⫽ .001, r ⫽ .410) neuronal density and between age and parvalbumin neuronal density (p ⫽ .023, r ⫽ .294). At the 5% level, significant reductions in the total cortical density of parvalbumin- (p ⫽ .029) and calbindin- (p ⫽ .035) containing cells were observed in the schizophrenic group. The density of calretinin-immunoreactive cells did not differ between the two groups (Figure 2.) To look for disease and laminar specificity, the mean density of parvalbumin-, calbindin-, and calretinin-immunoreactive neurons in each cortical layer for each of the three patient groups were compared with that of the control group. At the 10% level, mean fixation times for schizophrenia (p ⫽ .011), bipolar disorder (p ⫽ .001), and major depression (p ⫽ .055) and mean postmortem intervals for schizophrenia (p ⫽ .038) and bipolar disorder (p ⫽ .085) were higher than for control subjects; however, Spearman’s Rank correlation indicated no significant correlation between these variables and the density of any
Prefrontal GABAergic Deficits in Schizophrenia
BIOL PSYCHIATRY 2002;52:708 –715
711
Figure 1. Photomicrographs showing (a) parvalbumin-, (b) calbindin-, and (c) calretininimmunoreactive neurons throughout the cortical width of a control case (bar ⫽ 200 m); (d) multipolar neurons and processes immunoreactive for parvalbumin in cortical layer IV; (e) pyramidal (asterisks) and nonpyramidal neurons immunoreactive for calbindin in cortical layer III; and (f) bipolar neurons immunoreactive for calretinin in cortical layer III (bar ⫽ 30 m).
neuronal population at the 10% level, and so group comparisons were not adjusted for these variables. Significant correlations were observed between tissue pH and parvalbumin (p ⫽ .002, r ⫽ .399) and calbindin (p ⫽ .011, r ⫽ .326) neuronal density and between age and parvalbumin neuronal density (p ⫽ .004, r ⫽ .365). At the 5% level, significant reductions in the density of parvalbuminimmunoreactive neurons were observed in cortical layer
III in the schizophrenic group compared with the control group (p ⫽ .049). The density of calbindin-immunoreactive neurons was reduced in schizophrenia in layers II (p ⫽ .004), III (p ⫽ .026), and V (p ⫽ .031) and in bipolar disorder in layers II (p ⫽ .013) and III (p ⫽ .016), compared with control subjects; however, using the stricter criteria to account for multiple comparisons, only the reduction in calbindin density in layer II in schizophre-
Table 2. Density (Mean ⫾ SEM cells/mm2) of Calcium-Binding Protein-Immunoreactive Neurons in the Dorsolateral Prefrontal Cortex in Schizophrenic (SCZ), Bipolar Disorder (BIP), Major Depression (DEP), and Control (CON) Groups Diagnosis Calcium-binding protein Parvalbumin
Calbindin
Calretinin
Cortical layer
SCZ (n ⫽ 15)
BIP (n ⫽ 15)
DEP (n ⫽ 15)
CON (n ⫽ 15)
I II III IV V/VI I II III IV V/VI I II III IV V/VI
0 12.98 ⫾ 1.14 39.57 ⫾ 2.40 83.03 ⫾ 3.92 17.51 ⫾ 1.43 1.67 ⫾ 0.50 94.15 ⫾ 9.92 20.13 ⫾ 2.92 10.93 ⫾ 2.83 7.95 ⫾ 1.22 53.39 ⫾ 4.13 129.16 ⫾ 11.54 49.57 ⫾ 2.26 18.63 ⫾ 2.08 7.26 ⫾ 1.35
0 12.99 ⫾ 1.78 39.72 ⫾ 2.78 81.92 ⫾ 4.05 17.29 ⫾ 1.31 1.84 ⫾ .43 98.84 ⫾ 8.67 19.37 ⫾ 1.70 7.80 ⫾ 1.03 9.81 ⫾ .78 49.57 ⫾ 4.56 134.61 ⫾ 9.64 51.35 ⫾ 2.65 19.52 ⫾ 1.60 6.73 ⫾ .46
0 12.63 ⫾ 1.47 40.69 ⫾ 1.66 92.97 ⫾ 4.11 17.60 ⫾ 1.08 1.68 ⫾ .35 105.22 ⫾ 5.64 23.66 ⫾ 2.54 9.21 ⫾ 2.03 9.80 ⫾ .82 42.02 ⫾ 3.01 122.11 ⫾ 5.63 51.27 ⫾ 1.81 20.07 ⫾ 2.01 7.18 ⫾ .81
0 15.30 ⫾ 1.78 46.04 ⫾ 2.11 92.91 ⫾ 3.65 20.01 ⫾ 1.14 2.20 ⫾ .49 127.2 ⫾ 6.27 28.33 ⫾ 2.82 10.54 ⫾ 2.64 10.88 ⫾ .86 52.69 ⫾ 3.53 133.72 ⫾ 5.35 52.47 ⫾ 1.95 22.3 ⫾ 1.60 6.80 ⫾ .67
712
C.L. Beasley et al
BIOL PSYCHIATRY 2002;52:708 –715
noted for any neuronal subpopulation. In the schizophrenic and bipolar disorder groups no significant correlations were noted between any neuronal subpopulation and lifetime antipsychotic dose. Cortical width did not differ between groups (mm [mean ⫾ SD]: schizophrenia: 2.74 ⫾ .29; bipolar disorder: 2.93 ⫾ .25; major depression: 2.72 ⫾ .20; control subjects 2.69 ⫾ .17).
Discussion
Figure 2. Scatterplots showing total cortical density of parvalbumin-, calbindin-, and calretinin-immunoreactive neurons in the dorsolateral prefrontal cortex in schizophrenic (SCZ), bipolar disorder (BPD), major depressive disorder (MDD), and control (CON) subjects. Horizontal lines indicate mean values.
nia attained statistical significance. The density of calretinin-immunoreactive neurons was not altered in any layer in any disorder, compared with control subjects. No gender ⫻ diagnosis or side ⫻ diagnosis interactions were
In the prefrontal cortex the total cortical density of calbindin- and parvalbumin- but not calretinin-immunoreactive neurons was reduced in the schizophrenic group, compared with control subjects. This confirmed our previous findings of deficits in parvalbumin- but not calretinin-containing cells in schizophrenia (Beasley and Reynolds 1997; Reynolds and Beasley 2001) but indicates that these reductions are not specific to parvalbumin-containing cells, as we had previously suggested. More detailed analyses, which examined disease and laminar specificity, revealed that significant reductions were only present in calbindin-containing cells in layer II in schizophrenia; however, the density of both parvalbumin- and calbindinimmunoreactive neurons was reduced in each cortical layer in schizophrenia and bipolar disorder, suggesting that deficits in GABAergic interneurons are not specific to disease, cortical layer, or neuronal subtype. Our finding of an absence of any changes in the density of calretinin-immunoreactive neurons in schizophrenic compared with control subjects is consistent with the results of previous studies in the dorsolateral prefrontal cortex, anterior cingulate cortex, and hippocampus (Cotter et al, 2002; Daviss and Lewis 1995; Reynolds and Beasley 2001; Zhang and Reynolds, 2002); however, this is not true of our findings of reductions in the density of parvalbumin- and calbindin-immunoreactive neurons. In the only previous study of calbindin neuronal density in the prefrontal cortex, Daviss and Lewis (1995) described an increase in density in a small group of five schizophrenic subjects compared with control subjects. We suggest that the discrepancy between these studies could be due to differences in the samples or in the methodology used. More recently, Cotter et al 2002, who used the same brain collection as studied here, found that the density of calbindin-positive interneurons was reduced in cortical layer II in the anterior cingulate cortex in schizophrenia. The deficit in total cortical parvalbumin-immunoreactive cells we observed in the schizophrenic group is consistent with our earlier findings in BA 10 in a separate series of schizophrenic and control cases (Beasley and Reynolds 1997). Although we found no significant difference in any individual cortical layer, an 11%–15% reduction in the density of parvalbumin-immunoreactive neurons was ob-
Prefrontal GABAergic Deficits in Schizophrenia
served in each layer in both the schizophrenic and bipolar disorder groups. This is consistent with the study by Cotter et al (2002), who found a nonsignificant reduction of approximately 20% in parvalbumin-immunoreactive cells in the anterior cingulate cortex in both schizophrenia and bipolar disorder; however, a further study by Woo et al (1997) found no significant change in the density of parvalbumin-immunoreactive neurons in the dorsolateral prefrontal cortex. Again, we propose that the discrepancy between these studies could be due to differences in the samples or the methodology used. A further possibility is that the antigen retrieval method used in our study may not be able to identify cells that contain only very low levels of parvalbumin. Reductions in the density of parvalbuminimmunoreactive neurons have also been described in other brain regions, including the hippocampus and dorsomedial thalamus, in schizophrenia (Danos et al 1998; Zhang and Reynolds, 2002). Although we were not able to determine if the reduction in parvalbumin-containing neurons was attributable to one specific subclass, we suggest that it might reflect a deficit in chandelier cells. A loss of chandelier cells could result in the large reduction in the density of chandelier cell axonal terminals, identified using an antibody directed against the GABA membrane transporter GAT-1, which has been observed in the prefrontal cortex in schizophrenia (Woo et al 1998). Chandelier cells are of particular interest, as they are thought to regulate the excitatory output of pyramidal neurons, thus modulating patterns of activity in the prefrontal cortex. We cannot conclude that the observed reduction in the density of calbindin- and parvalbumin-immunoreactive neurons in this study reflects a reduction in cell numbers solely on the basis of immunohistochemical data. Although this is one explanation, consistent with early indications of interneuron deficits, the reduction in the density of immunoreactive cells could also be influenced by a number of other factors. As mentioned above, it may be that the immunocytochemical method used was not able to pick up neurons that only express very low levels of calbindin or parvalbumin. This may suggest that in the schizophrenic group the expression of these proteins is greatly reduced in a subpopulation of neurons. An apparent decrease in cell density could therefore result from a reduction in cellular calbindin or parvalbumin expression in the absence of any cell loss. This would be consistent with recent reports of a reduction in the density of interneurons that express mRNA for glutamic acid decarboxylase in the prefrontal cortex in schizophrenia, which have been interpreted as reflecting a downregulation of GAD expression (Akbarian et al 1995; Volk et al 2000). It is conceivable that a potentially reversible downregulation of calcium-binding proteins may reflect a change in a cell’s functional activity, possibly brought about by
BIOL PSYCHIATRY 2002;52:708 –715
713
changes in the activity of other neurons innervating the cell. Indeed a number of experimental conditions have been described in which the expression of calcium-binding proteins or GABA is altered, including monocular deprivation (Carder et al 1996) and kindling (Kamphuis et al 1989).
Confounding Factors The Stanley Foundation Neuropathology Consortium brain collection has been extensively characterized, and details regarding possible confounding factors have been made available. In this study, the influence of potential confounders (including postmortem interval, tissue pH, age, gender, time in fixative, antipsychotic exposure, and co-morbid alcohol/substance abuse) on measures of parvalbumin-, calbindin-, and calretinin-immunoreactive neurons was examined. Fixation time and postmortem interval were significantly different between the groups; however, we found no significant relationship between these variables and measures of parvalbumin-, calbindin-, or calretinin-immunoreactive neurons. Although the detectability of parvalbumin and calbindin is greatly reduced after fixation for only 2 weeks, previous studies have indicated that for tissue preserved in fixative for longer periods, microwave antigen-retrieval, as used in the present study, allows the visualization of parvalbumin- and calbindin-containing cell bodies (Evers and Uylings 1994). There is little evidence with regard to the potential effects of antipsychotics on the expression of these calcium-binding proteins. In this study, we noted that the one schizophrenic subject and three bipolar disorder subjects who had never been medicated with neuroleptics had relative densities similar to those patients who were medicated. Also, in the schizophrenic and bipolar disorder groups no correlation was observed between lifetime neuroleptic load and the relative density of immunoreactive neurons in the prefrontal cortex. Furthermore, there was no significant correlation between the relative density of any subset of interneurons and duration of illness, a crude measure of antipsychotic exposure. This suggests that the observed reduction in parvalbumin- and calbindinexpressing neurons is not attributable to an effect of antipsychotics. The presence of co-morbid alcohol and substance abuse is a further potential confounding factor. In this brain collection three of the schizophrenic patients and four of the bipolar disorder patients were abusing alcohol and/or illicit drugs at the time of death. In light of this small sample size and the fact that many of these patients were abusing multiple drugs, we were not able to determine if alcohol or any specific drugs alone could have suppressed the expression of these calcium-binding proteins. Al-
714
C.L. Beasley et al
BIOL PSYCHIATRY 2002;52:708 –715
though we suggest that the observed reduction in parvalbumin- and calbindin-immunoreactive neurons in the schizophrenic group was not due to alcohol abuse, this factor cannot be completely excluded. In this study, the use of stereological methods was not possible within the confines of the tissue available to us. The major source of error in any two-dimensional study is likely to be differences in cell size. Although we were not able to determine cell size in our study, Cotter et al (2002), using the same brain collection as used in this study, found no significant reductions in the somal size of calbindin- or parvalbumin-immunoreactive neurons in the anterior cingulate cortex in schizophrenia, bipolar disorder, or major depression. Two other studies have also shown that the size of parvalbumin-containing neurons does not differ between schizophrenic and control subjects (Kalus et al 1997; Woo et al 1997). Although the concept that stereological techniques represent the most accurate approach for determining neuronal density has recently been challenged (Benes and Lange 2001), the inconsistencies observed between the two-dimensional studies in this area indicate the need for a stereological assessment of cell volumes and densities of GABAergic neuronal subtypes in psychiatric disorders. A further possible source of error is tissue shrinkage during the processing, sectioning, and staining of tissue sections. Although ideally a correction for shrinkage should be employed when calculating neuronal densities in both two-dimensional and three-dimensional studies, this is often not feasible and was not possible within the confines of this investigation. These data provide further evidence for deficits in specific populations of GABAergic interneurons in the prefrontal cortex in schizophrenia. Similar reductions were observed in bipolar disorder. Although specific interneuron populations have not previously been quantified in the prefrontal cortex in this disorder, these reductions are consistent with a reduction in the expression of GAD mRNA and density of GAD-immunoreactive terminals (Benes et al 2000; Guidotti et al 2000). GABAergic deficits are likely to have wide-ranging effects on the neuronal circuitry of the prefrontal cortex and its output connections to other brain regions. Any change in the activity of GABAergic neurons, which play a vital role in modulating the activity of the cortex, would lead to imbalances in other systems, for example via their interactions with dopaminergic and glutamatergic neurons. This would inevitably have effects on many functions, such as cognitive processing. Studies of the GABAergic system may be valuable in the search to define the neuropathological changes in psychiatric disorders and elucidate possible putative mechanisms underlying the disturbances of cortical function observed.
This study was funded by a project award from the Theodore and Vada Stanley Foundation. Postmortem brains were donated by the Stanley Foundation Brain Bank Consortium, courtesy of Drs. Llewellyn B. Bigelow, Juraj Cervenak, Mary M. Herman, Thomas M. Hyde, Joel Kleinman, Jose D. Paltan, Robert M. Post, E. Fuller Torrey, Maree J. Webster, and Robert Yolken.
References Akbarian S, Kim JJ, Potkin SG, Hagman JO, Tafazzoli A, Bunney WE Jr, et al (1995): Gene expression for glutamic acid decarboxylase is reduced without loss of neurons in prefrontal cortex of schizophrenics. Arch Gen Psychiatry 52:258 –266. Akil M, Lewis DA (1992): Postnatal development of parvalbumin immunoreactivity in axon terminals of basket and chandelier neurons in monkey neocortex. Prog Neuropsychopharmacol Biol Psychiatry 16:329 –337. Bachus SE, Hyde TM, Herman MM, Eyan MF, Kleinman JE (1997): Abnormal cholecystokinin mRNA levels in entorhinal cortex of schizophrenics. J Psychiatr Res 31:233–256. Beasley CL, Reynolds GP (1997): Parvalbumin-immunoreactive neurons are reduced in the prefrontal cortex of schizophrenics. Schizophr Res 24:349 –355. Benes FM, Lange N (2001): Two-dimensional versus threedimensional cell counting: A practical perspective. Trends Neurosci 24:11–17. Benes FM, McSparren J, Bird ED, SanGiovanni JP, Vincent SL (1991): Deficits in small interneurons in prefrontal and cingulate cortices of schizophrenic and schizoaffective patients. Arch Gen Psychiatry 48:996 –1001. Benes FM, Todtenkopf MS, Logiotatos P, Williams M (2000): Glutamate decarboxylase(65)-immunoreactive terminals in cingulate and prefrontal cortices of schizophrenic and bipolar brain. J Chem Neuroanat 20:259 –269. Benes FM, Vincent SL, Marie A, Khan Y (1996): Up-regulation of GABAA receptor binding on neurons of the prefrontal cortex in schizophrenic subjects. Neuroscience 74:1021– 1031. Carder R, LeClerc SL, Hendry SHC (1996): Regulation of calcium-binding protein immunoreactivity in GABA neurons of macaque primary visual cortex. Cereb Cortex 6:271–287. Celio MR (1990): Calbindin D-28K and parvalbumin in the rat nervous system. Neuroscience 35:375– 475. Cotter D, Landau S, Beasley C, Stevenson R, Chana G, MacMillan L, Everall I (2002): The density and spatial distribution of GABAergic neurons in the anterior cingulate cortex in major depressive disorder, bipolar disorder and schizophrenia. Biol Psychiatry 51:377–386. Danos P, Baumann B, Bernstein H-G, Franz M, Stauch R, Northoff G-M, et al (1998): Schizophrenia and anteroventral thalamic nucleus: Selective decrease of parvalbumin-immunoreactive thalamocortical projection neurons. Psychiatry Res 82:1–10. Daviss SR, Lewis DA (1995): Local circuit neurons of the prefrontal cortex in schizophrenia: Selective increase in the density of calbindin-immunoreactive neurons. Psychiatry Res 59:81–96.
Prefrontal GABAergic Deficits in Schizophrenia
DeFelipe J, Hendry SHC, Jones EG (1989): Visualisation of chandelier cell axons by parvalbumin-immunoreactivity in monkey cerebral cortex. Proc Natl Acad Sci USA 86:2093– 2097. DeFelipe J, Jones EG (1991): Parvalbumin immunoreactivity reveals layer IV of monkey cerebral cortex as a mosaic of microzones of thalamic afferent terminations. Brain Res 562:39 – 47. Demeulemeester H, Vandesande F, Orban GA, Brando C (1988): Heterogeneity of GABAergic cells in cat visual cortex. J Neurosci 8:988 –1000. Evers P, Uylings HB (1994): Effects of microwave pretreatment on immunocytochemical staining of vibratome sections and tissue blocks of human cerebral cortex stored in formaldehyde fixative for long periods. J Neurosci Methods 55:163– 72. Gabriel SM, Davidson M, Haroutunian V, Powchik P, Bierer LM, Purohit DP, et al (1996): Neuropeptide deficits in schizophrenia vs. Alzheimer’s disease cerebral cortex. Biol Psychiatry 39:82–91. Guidotti A, Auta J, Davis JM, Gerevini VD, Dwivedi Y, Grayson DR, et al (2000): Decrease in reelin and glutamic acid decarboxylase 67 (GAD67) expression in schizophrenia and bipolar disorder: A postmortem brain study. Arch Gen Psychiatry 57:1061–1069. Hanada S, Mita T, Nishinok N, Tanaka C (1987): 3H-Muscimol binding sites are increased in autopsied brains of chronic schizophrenics. Life Sci 40:259 –266. Hsu SM, Raine L, Fanger H (1981): Use of avidin-biotinperoxidase complex (ABC) in immunoperoxidase techniques: A comparison between ABC and unlabelled antibody (PAP) procedures. J Histochem Cytochem 29:577–580. Jacobowitz DM, Winsky L (1991): Immunocytochemical localisation of calretinin in the forebrain of the rat. J Comp Neurol 304:198 –218. Kalus P, Senitz D, Beckmann H (1997): Altered distribution of parvalbumin-immunoreactive local circuit neurons in the anterior cingulate cortex of schizophrenic patients. Psychiatry Res 75:49 –59. Kamphuis W, Huisman E, Wadman WJ, Lopez Da Silva FH (1989): Decrease in GABA immunoreactivity and alteration of GABA metabolism after kindling in the rat hippocampus. Exp Brain Res 74:375–386.
BIOL PSYCHIATRY 2002;52:708 –715
715
Lewis DA, Lund JS (1990): Heterogeneity of chandelier neurons in monkey neocortex: Corticotropin-releasing factor and parvalbumin-immunoreactive populations. J Comp Neurol 293: 599 – 615. Nemeroff CB, Youngblood WW, Manberg PJ, Prange AJ Jr, Kizer JS (1983): Regional brain concentrations of neuropeptides in Huntington’s chorea and schizophrenia. Science 221:972–975. Rajkowska G, Goldman-Rakic PS (1995): Cytoarchitectonic definition of prefrontal areas in the normal human cortex: I. Remapping of areas 9 and 46 using quantitative criteria. Cereb Cortex 5:307–322. Reynolds GP, Beasley CL (2001): GABAergic neuronal subtypes in the human frontal cortex— development and deficits in schizophrenia. J Chem Neuroanat 22:95–100. Somogyi P, Hodgson AJ, Smith AD, Nunzi MG, Gorio A, Wu JY (1984): Different populations of GABAergic neurons in the visual cortex and hippocampus of cat contain somatostatin- or cholecystokinin-immunoreactive material. J Neurosci 4:2590 –2603. Torrey EF, Webster M, Knable M, Johnston N, Yolken RH (2000): The Stanley foundation brain collection and neuropathology consortium. Schizophr Res 44:151–155. Virgo L, Humphries C, Mortimer A, Barnes T, Hirsch S, DeBelleroche J (1995): Cholecystokinin messenger RNA deficit in frontal and temporal cerebral cortex in schizophrenia. Biol Psychiatry 37:694 –701. Volk DW, Austin MC, Pierri JN, Sampson AR, Lewis DA (2000): Decreased glutamic acid decarboxylase67 messenger RNA expression in a subset of prefrontal cortical ␥–aminobutyric acid neurons in subjects with schizophrenia. Arch Gen Psychiatry 57:237–245. Woo TU, Miller JL, Lewis DA (1997): Schizophrenia and the parvalbumin-containing class of cortical local circuit neurons. Am J Psychiatry 154:1013–1015. Woo TU, Whitehead RE, Melchitzky DS, Lewis DA (1998): A subclass of prefrontal gamma-aminobutyric acid axon terminals are selectively altered in schizophrenia. Proc Natl Acad Sci USA 95:5341–5346. Zhang ZJ, Reynolds GP (2002): A selective decrease in the relative density of parvalbumin-immunoreactive neurons in the hippocampus in schizophrenia. Schizophr Res 55:1–10.
From the Department of Biomedical Science, University of Sheffield, Sheffield, United Kindom. Address reprint requests to Clare L. Beasley, Ph.D., Institute of Psychiatry, Section of Experimental Neuropathology and Psychiatry, DeCrespigny Park, London SE5 8AF, United Kingdom. Received July 23, 2001; revised December 10, 2001; revised February 1, 2002; accepted February 11, 2002.
© 2002 Society of Biological Psychiatry
Introduction
T
here is some intriguing evidence indicative of anomalous ␥-aminobutyric acid (GABA)-ergic neurotransmission in the prefrontal cortex in schizophrenia. In particular, postmortem analysis of brain tissue from this region has provided insight into cellular abnormalities of the GABAergic system, including a loss of presumptive interneurons in more superficial cortical layers (Benes et al 1991) and a reduction in the density of interneurons expressing mRNA for glutamic acid decarboxylase (GAD), the synthesizing enzyme for GABA (Akbarian et al 1995; Volk et al 2000). Further studies have found increased binding to the GABAA receptor in the prefrontal cortex in schizophrenia (Benes et al 1996; Hanada et al 1987). This increase in GABAA receptors has been interpreted as reflecting a compensatory upregulation of postsynaptic receptors, due to losses of GABAergic interneurons in this region (Benes et al 1996). Although similar studies of bipolar disorder and major depression are limited, there are indications that expression of GAD mRNA and the density of GAD-immunoreactive terminals are reduced in the prefrontal cortex in bipolar disorder (Benes et al 2000; Guidotti et al 2000). To understand the functional significance of these findings, it is important to further define the GABAergic changes observed in schizophrenia. A number of criteria, including morphologic appearance and biochemical properties, can be utilized to discriminate distinct classes of GABAergic interneurons. Cortical GABAergic cells can be subdivided on the basis of co-localized neuropeptides, including somatostatin, cholecystokinin, neuropeptide Y, and vasoactive intestinal polypeptide (Somogyi et al 1984). Losses of cholecystokinin, somatostatin, and vasoactive intestinal polypeptide have been described in the cortex in some schizophrenic patients (Gabriel et al 1996; Nemeroff et al 1983). Consistent with this, deficits in cholecystokinin mRNA, have been reported in the frontal and temporal cortex (Virgo et al 1995) as well as in the entorhinal cortex (Bachus et al 1997). Essentially nonoverlapping subpopulations of GABAergic neurons can also be 0006-3223/02/$22.00 PII S0006-3223(02)01360-4
Prefrontal GABAergic Deficits in Schizophrenia
709
BIOL PSYCHIATRY 2002;52:708 –715
Table 1. Summaries of Demographic, Clinical, and Histological Information of Schizophrenic, Bipolar Disorder, Major Depression, and Control Groups Group Demographic variable Age (y, mean ⫾ SD) Gender (male, female) Postmortem interval (hr, mean ⫾ SD) Cause of death CPD Accident Suicide Pneumonia Other pH (mean ⫾ SD) Time in fixative (mo, mean ⫾ SD) Brain hemisphere used (Right : Left) Duration of illness (y, mean ⫾ SD) Lifetime antipsychotic dosea (mg, minimum; median; maximum) Current alcohol/drug abuse
Control (n ⫽ 15)
Schizophrenia (n ⫽ 15)
Bipolar disorder (n ⫽ 15)
Major depression (n ⫽ 15)
48.1 ⫾ 10.7 9M, 6F 23.7 ⫾ 9.9
44.2 ⫾ 13.1 9M, 6F 33.7 ⫾ 14.6
42.3 ⫾ 11.7 9M, 6F 32.5 ⫾ 16.1
46.4 ⫾ 9.3 9M, 6F 27.5 ⫾ 10.7
13 2 0 0 0 6.3 ⫾ .2 4.40 ⫾ 3.87 7:8 0⫾0 0; 0; 0
7 0 7 0 1 6.2 ⫾ .3 11.20 ⫾ 8.48 6:9 21.3 ⫾ 11.4 0; 35,000; 200,000
3 0 9 1 2 6.2 ⫾ .2 9.67 ⫾ 3.62 8:7 20.1 ⫾ 9.7 0; 7,500; 60,000
7 1 4 1 2 6.2 ⫾ .2 8.40 ⫾ 6.59 6:9 12.7 ⫾ 11.1 0; 0; 0
0
3
4
3
CPD, cardiopulmonary disease. a Lifetime neuroleptic dose in fluphenazine milligram equivalent dose.
defined by the presence of the calcium-binding proteins parvalbumin, calbindin, and calretinin (Celio 1990; Demeulemeester et al 1988). Studies have revealed that the most characteristic morphologic types of neurons that express parvalbumin are large basket and chandelier cells (Akil and Lewis 1992; Lewis and Lund 1990); calbindin is present in many double bouquet cells (DeFelipe et al 1989), whereas calretinin-immunoreactive neurons are generally bipolar, double bouquet, and Cajal-Retzius cells (Jacobowitz and Winsky 1991). We have previously observed a reduction in the density of parvalbumin-immunoreactive cells in the prefrontal cortex (Brodmann area [BA] 10) in schizophrenia (Beasley and Reynolds 1997), but we found no loss of calretinin-immunoreactive neurons in the same region (Reynolds and Beasley 2001). This led us to suggest that the GABAergic deficits described in the frontal cortex in schizophrenia may be due to specific reductions in parvalbumin-immunoreactive cells. To test this, the density of interneurons immunoreactive for parvalbumin, calbindin, and calretinin was quantified in a further prefrontal cortical region (BA 9) in a series of brains from patients with schizophrenia, bipolar disorder, major depression, and from matched control subjects. As the morphology, distribution, function, and ontology of distinct subpopulations of GABAergic neurons differs, identifying deficits of specific interneurons in the brain in schizophrenia could throw light on the putative mechanisms underlying the disturbances of cortical function observed in this disorder.
Methods and Materials Human Subjects Samples were obtained from the Stanley Foundation Neuropathology Consortium brain collection. Brains were obtained from patients diagnosed with schizophrenia, bipolar disorder, major depression, and from matched control subjects (each group n ⫽ 15). Demographic details are provided in Table 1.; for further descriptions see Torrey et al (2000). Final diagnoses were established using DSM-IV criteria and routine microscopic and toxicological examinations carried out on all cases. Cause of death and an estimate of total lifetime intake of antipsychotic medication (in fluphenazine mg equivalents) were detailed for all cases.
Immunocytochemistry 10-m-thick paraffin sections of the dorsolateral prefrontal cortex were processed for parvalbumin-, calbindin-, or calretininimmunoreactivity randomly and blind to diagnosis. For each antibody two sections, at least 50 m apart, were sampled for analysis. Tissue sections were heated in a microwave oven on full power (650 W) for 3 ⫻ 10 min in .05 mol/L tris buffer, pH 9.0, to aid antigen retrieval. Sections were blocked with normal serum, then incubated for 36 hours at 4°C with either a mouse monoclonal antibody against parvalbumin (clone PA-235, Sigma, St. Louis, MO), a mouse monoclonal antibody against calbindin (Sigma) or a goat polyclonal antibody against calretinin (Swant, Bellinzona, Switzerland), each at a dilution of 1:5000. For parvalbumin the antibody diluent also contained 3 mmol/L calcium chloride. Following this the sections were processed by the avidin– biotin method of Hsu et al (1981) using
710
BIOL PSYCHIATRY 2002;52:708 –715
a Vectastain ABC kit (Vector Laboratories, Burlingame, CA). Peroxidase was visualized using diaminobenzidine, intensified with nickel chloride, and sections counterstained with toluidine blue. Negative control sections, in which the primary antibody was omitted from the staining protocol, were run alongside the test series.
Image Analysis Parvalbumin-, calbindin-, and calretinin-immunoreactive neurons were plotted at 100⫻ magnification using an Olympus microscope (Olympus Optical Co (Europa) GmbH Hamburg, Germany) equipped with camera lucida. The density of neuronal profiles was expressed as mean values (⫾ SE) per mm2 per cortical layer and was the result of counts from a total of 10 500-m-wide cortical traverses, each from the pial surface to the white matter border, selected at random from the two slides. The placement of the traverses was restricted to areas of the cortex that were comparatively level, avoiding the depths of the sulci. Cortical width was also measured for each of these traverses. Counts were made in cortical region BA 9 in each case. The boundaries of BA 9 with BA 46 were determined according to the criteria of Rajkowska and Goldmann-Rakic (1995) and Daviss and Lewis (1995).
Statistical Analysis To test our initial hypothesis that the GABAergic deficits previously described in schizophrenia result from a reduction in parvalbumin-immunoreactive neurons but not calbindin- or calretinin-containing cells, we compared mean total cortical densities of each of these cell populations between the schizophrenic and control groups, using univariate analysis of variance (ANOVA). The demographic and histologic variables listed in Table 1 were included as covariates in the analysis if 1) they differed between the schizophrenic group and the control group at the 10% significance level (ANOVA); and 2) they could be shown empirically to predict densities at the 10% significance level (ANOVA or Spearman’s Rank correlation). Following on from this, the mean density of parvalbumin-, calbindin-, and calretinin-immunoreactive neurons in each cortical layer for each of the three patient groups was compared with that of the control group. This allowed us to look for disease and laminar specificity. Mean densities were analyzed by univariate ANOVA, using diagnoses as contrasts. Confounding variables were determined as above. To account for multiple layer-wise testing, p values of .010 (for calbindin and calretinin) and .013 (for parvalbumin) were determined to be statistically significant; these values represent the 5% level divided by the number of layers analyzed. In the schizophrenic and bipolar disorder groups the effects of total antipsychotic drug dosage on parvalbumin-, calbindin-, and calretinin-immunoreactive neuronal densities were was also assessed. All statistical analysis was carried out in SPSS 10 (CSPSS Inc, Chicago, IL).
Results Parvalbumin-immunoreactive neurons in the prefrontal cortex were typically intensely stained. Immunoreactive
C.L. Beasley et al
neurons were present predominantly in layers III, IV, and V and were also observed in layers II and VI, but not in layer I or in the white matter (Figure 1a). Parvalbuminpositive cells appeared to be nonpyramidal and consisted of a variety of morphologies, including small ovoid perikarya, large multipolar neurons, and occasionally bitufted cells (Figure 1 d). A dense plexus of immunoreactive material was also distributed throughout the neuropil of layers III, IV, and V and consisted of stained processes and puncta, which have been shown to represent axonal terminals (DeFelipe and Jones 1991). Calbindinimmunoreactive neurons were present throughout the cortical width and were predominant in layer II and superficial layer III (Figure 1b). Immunoreactive neurons comprised both densely stained nonpyramidal cells and faintly stained pyramidal cells (Figure 1e). Calretininimmunoreactive cells also comprised a population of nonpyramidal neurons that were distributed throughout all six cortical layers and the subjacent white matter, although again predominantly in the more superficial laminae (Figure 1c). Positive neurons were typically bipolar in morphology (Figure 1f); layer I contained horizontally oriented processes and immunoreactive puncta in addition to labeled neurons. Summaries of mean laminar densities for each neuronal subpopulation are shown in Table 2. At the 10% level, mean fixation time (p ⫽ .011) and mean postmortem interval (p ⫽ .038) were significantly higher in the schizophrenic group than in the control group; however, Spearman’s Rank correlation indicated no significant relationship between these variables and the density of any neuronal population at the 10% level, and so group comparisons were not adjusted for these variables. Significant correlations were observed between tissue pH and parvalbumin (p ⫽ .001, r ⫽ .426) and calbindin (p ⫽ .001, r ⫽ .410) neuronal density and between age and parvalbumin neuronal density (p ⫽ .023, r ⫽ .294). At the 5% level, significant reductions in the total cortical density of parvalbumin- (p ⫽ .029) and calbindin- (p ⫽ .035) containing cells were observed in the schizophrenic group. The density of calretinin-immunoreactive cells did not differ between the two groups (Figure 2.) To look for disease and laminar specificity, the mean density of parvalbumin-, calbindin-, and calretinin-immunoreactive neurons in each cortical layer for each of the three patient groups were compared with that of the control group. At the 10% level, mean fixation times for schizophrenia (p ⫽ .011), bipolar disorder (p ⫽ .001), and major depression (p ⫽ .055) and mean postmortem intervals for schizophrenia (p ⫽ .038) and bipolar disorder (p ⫽ .085) were higher than for control subjects; however, Spearman’s Rank correlation indicated no significant correlation between these variables and the density of any
Prefrontal GABAergic Deficits in Schizophrenia
BIOL PSYCHIATRY 2002;52:708 –715
711
Figure 1. Photomicrographs showing (a) parvalbumin-, (b) calbindin-, and (c) calretininimmunoreactive neurons throughout the cortical width of a control case (bar ⫽ 200 m); (d) multipolar neurons and processes immunoreactive for parvalbumin in cortical layer IV; (e) pyramidal (asterisks) and nonpyramidal neurons immunoreactive for calbindin in cortical layer III; and (f) bipolar neurons immunoreactive for calretinin in cortical layer III (bar ⫽ 30 m).
neuronal population at the 10% level, and so group comparisons were not adjusted for these variables. Significant correlations were observed between tissue pH and parvalbumin (p ⫽ .002, r ⫽ .399) and calbindin (p ⫽ .011, r ⫽ .326) neuronal density and between age and parvalbumin neuronal density (p ⫽ .004, r ⫽ .365). At the 5% level, significant reductions in the density of parvalbuminimmunoreactive neurons were observed in cortical layer
III in the schizophrenic group compared with the control group (p ⫽ .049). The density of calbindin-immunoreactive neurons was reduced in schizophrenia in layers II (p ⫽ .004), III (p ⫽ .026), and V (p ⫽ .031) and in bipolar disorder in layers II (p ⫽ .013) and III (p ⫽ .016), compared with control subjects; however, using the stricter criteria to account for multiple comparisons, only the reduction in calbindin density in layer II in schizophre-
Table 2. Density (Mean ⫾ SEM cells/mm2) of Calcium-Binding Protein-Immunoreactive Neurons in the Dorsolateral Prefrontal Cortex in Schizophrenic (SCZ), Bipolar Disorder (BIP), Major Depression (DEP), and Control (CON) Groups Diagnosis Calcium-binding protein Parvalbumin
Calbindin
Calretinin
Cortical layer
SCZ (n ⫽ 15)
BIP (n ⫽ 15)
DEP (n ⫽ 15)
CON (n ⫽ 15)
I II III IV V/VI I II III IV V/VI I II III IV V/VI
0 12.98 ⫾ 1.14 39.57 ⫾ 2.40 83.03 ⫾ 3.92 17.51 ⫾ 1.43 1.67 ⫾ 0.50 94.15 ⫾ 9.92 20.13 ⫾ 2.92 10.93 ⫾ 2.83 7.95 ⫾ 1.22 53.39 ⫾ 4.13 129.16 ⫾ 11.54 49.57 ⫾ 2.26 18.63 ⫾ 2.08 7.26 ⫾ 1.35
0 12.99 ⫾ 1.78 39.72 ⫾ 2.78 81.92 ⫾ 4.05 17.29 ⫾ 1.31 1.84 ⫾ .43 98.84 ⫾ 8.67 19.37 ⫾ 1.70 7.80 ⫾ 1.03 9.81 ⫾ .78 49.57 ⫾ 4.56 134.61 ⫾ 9.64 51.35 ⫾ 2.65 19.52 ⫾ 1.60 6.73 ⫾ .46
0 12.63 ⫾ 1.47 40.69 ⫾ 1.66 92.97 ⫾ 4.11 17.60 ⫾ 1.08 1.68 ⫾ .35 105.22 ⫾ 5.64 23.66 ⫾ 2.54 9.21 ⫾ 2.03 9.80 ⫾ .82 42.02 ⫾ 3.01 122.11 ⫾ 5.63 51.27 ⫾ 1.81 20.07 ⫾ 2.01 7.18 ⫾ .81
0 15.30 ⫾ 1.78 46.04 ⫾ 2.11 92.91 ⫾ 3.65 20.01 ⫾ 1.14 2.20 ⫾ .49 127.2 ⫾ 6.27 28.33 ⫾ 2.82 10.54 ⫾ 2.64 10.88 ⫾ .86 52.69 ⫾ 3.53 133.72 ⫾ 5.35 52.47 ⫾ 1.95 22.3 ⫾ 1.60 6.80 ⫾ .67
712
C.L. Beasley et al
BIOL PSYCHIATRY 2002;52:708 –715
noted for any neuronal subpopulation. In the schizophrenic and bipolar disorder groups no significant correlations were noted between any neuronal subpopulation and lifetime antipsychotic dose. Cortical width did not differ between groups (mm [mean ⫾ SD]: schizophrenia: 2.74 ⫾ .29; bipolar disorder: 2.93 ⫾ .25; major depression: 2.72 ⫾ .20; control subjects 2.69 ⫾ .17).
Discussion
Figure 2. Scatterplots showing total cortical density of parvalbumin-, calbindin-, and calretinin-immunoreactive neurons in the dorsolateral prefrontal cortex in schizophrenic (SCZ), bipolar disorder (BPD), major depressive disorder (MDD), and control (CON) subjects. Horizontal lines indicate mean values.
nia attained statistical significance. The density of calretinin-immunoreactive neurons was not altered in any layer in any disorder, compared with control subjects. No gender ⫻ diagnosis or side ⫻ diagnosis interactions were
In the prefrontal cortex the total cortical density of calbindin- and parvalbumin- but not calretinin-immunoreactive neurons was reduced in the schizophrenic group, compared with control subjects. This confirmed our previous findings of deficits in parvalbumin- but not calretinin-containing cells in schizophrenia (Beasley and Reynolds 1997; Reynolds and Beasley 2001) but indicates that these reductions are not specific to parvalbumin-containing cells, as we had previously suggested. More detailed analyses, which examined disease and laminar specificity, revealed that significant reductions were only present in calbindin-containing cells in layer II in schizophrenia; however, the density of both parvalbumin- and calbindinimmunoreactive neurons was reduced in each cortical layer in schizophrenia and bipolar disorder, suggesting that deficits in GABAergic interneurons are not specific to disease, cortical layer, or neuronal subtype. Our finding of an absence of any changes in the density of calretinin-immunoreactive neurons in schizophrenic compared with control subjects is consistent with the results of previous studies in the dorsolateral prefrontal cortex, anterior cingulate cortex, and hippocampus (Cotter et al, 2002; Daviss and Lewis 1995; Reynolds and Beasley 2001; Zhang and Reynolds, 2002); however, this is not true of our findings of reductions in the density of parvalbumin- and calbindin-immunoreactive neurons. In the only previous study of calbindin neuronal density in the prefrontal cortex, Daviss and Lewis (1995) described an increase in density in a small group of five schizophrenic subjects compared with control subjects. We suggest that the discrepancy between these studies could be due to differences in the samples or in the methodology used. More recently, Cotter et al 2002, who used the same brain collection as studied here, found that the density of calbindin-positive interneurons was reduced in cortical layer II in the anterior cingulate cortex in schizophrenia. The deficit in total cortical parvalbumin-immunoreactive cells we observed in the schizophrenic group is consistent with our earlier findings in BA 10 in a separate series of schizophrenic and control cases (Beasley and Reynolds 1997). Although we found no significant difference in any individual cortical layer, an 11%–15% reduction in the density of parvalbumin-immunoreactive neurons was ob-
Prefrontal GABAergic Deficits in Schizophrenia
served in each layer in both the schizophrenic and bipolar disorder groups. This is consistent with the study by Cotter et al (2002), who found a nonsignificant reduction of approximately 20% in parvalbumin-immunoreactive cells in the anterior cingulate cortex in both schizophrenia and bipolar disorder; however, a further study by Woo et al (1997) found no significant change in the density of parvalbumin-immunoreactive neurons in the dorsolateral prefrontal cortex. Again, we propose that the discrepancy between these studies could be due to differences in the samples or the methodology used. A further possibility is that the antigen retrieval method used in our study may not be able to identify cells that contain only very low levels of parvalbumin. Reductions in the density of parvalbuminimmunoreactive neurons have also been described in other brain regions, including the hippocampus and dorsomedial thalamus, in schizophrenia (Danos et al 1998; Zhang and Reynolds, 2002). Although we were not able to determine if the reduction in parvalbumin-containing neurons was attributable to one specific subclass, we suggest that it might reflect a deficit in chandelier cells. A loss of chandelier cells could result in the large reduction in the density of chandelier cell axonal terminals, identified using an antibody directed against the GABA membrane transporter GAT-1, which has been observed in the prefrontal cortex in schizophrenia (Woo et al 1998). Chandelier cells are of particular interest, as they are thought to regulate the excitatory output of pyramidal neurons, thus modulating patterns of activity in the prefrontal cortex. We cannot conclude that the observed reduction in the density of calbindin- and parvalbumin-immunoreactive neurons in this study reflects a reduction in cell numbers solely on the basis of immunohistochemical data. Although this is one explanation, consistent with early indications of interneuron deficits, the reduction in the density of immunoreactive cells could also be influenced by a number of other factors. As mentioned above, it may be that the immunocytochemical method used was not able to pick up neurons that only express very low levels of calbindin or parvalbumin. This may suggest that in the schizophrenic group the expression of these proteins is greatly reduced in a subpopulation of neurons. An apparent decrease in cell density could therefore result from a reduction in cellular calbindin or parvalbumin expression in the absence of any cell loss. This would be consistent with recent reports of a reduction in the density of interneurons that express mRNA for glutamic acid decarboxylase in the prefrontal cortex in schizophrenia, which have been interpreted as reflecting a downregulation of GAD expression (Akbarian et al 1995; Volk et al 2000). It is conceivable that a potentially reversible downregulation of calcium-binding proteins may reflect a change in a cell’s functional activity, possibly brought about by
BIOL PSYCHIATRY 2002;52:708 –715
713
changes in the activity of other neurons innervating the cell. Indeed a number of experimental conditions have been described in which the expression of calcium-binding proteins or GABA is altered, including monocular deprivation (Carder et al 1996) and kindling (Kamphuis et al 1989).
Confounding Factors The Stanley Foundation Neuropathology Consortium brain collection has been extensively characterized, and details regarding possible confounding factors have been made available. In this study, the influence of potential confounders (including postmortem interval, tissue pH, age, gender, time in fixative, antipsychotic exposure, and co-morbid alcohol/substance abuse) on measures of parvalbumin-, calbindin-, and calretinin-immunoreactive neurons was examined. Fixation time and postmortem interval were significantly different between the groups; however, we found no significant relationship between these variables and measures of parvalbumin-, calbindin-, or calretinin-immunoreactive neurons. Although the detectability of parvalbumin and calbindin is greatly reduced after fixation for only 2 weeks, previous studies have indicated that for tissue preserved in fixative for longer periods, microwave antigen-retrieval, as used in the present study, allows the visualization of parvalbumin- and calbindin-containing cell bodies (Evers and Uylings 1994). There is little evidence with regard to the potential effects of antipsychotics on the expression of these calcium-binding proteins. In this study, we noted that the one schizophrenic subject and three bipolar disorder subjects who had never been medicated with neuroleptics had relative densities similar to those patients who were medicated. Also, in the schizophrenic and bipolar disorder groups no correlation was observed between lifetime neuroleptic load and the relative density of immunoreactive neurons in the prefrontal cortex. Furthermore, there was no significant correlation between the relative density of any subset of interneurons and duration of illness, a crude measure of antipsychotic exposure. This suggests that the observed reduction in parvalbumin- and calbindinexpressing neurons is not attributable to an effect of antipsychotics. The presence of co-morbid alcohol and substance abuse is a further potential confounding factor. In this brain collection three of the schizophrenic patients and four of the bipolar disorder patients were abusing alcohol and/or illicit drugs at the time of death. In light of this small sample size and the fact that many of these patients were abusing multiple drugs, we were not able to determine if alcohol or any specific drugs alone could have suppressed the expression of these calcium-binding proteins. Al-
714
C.L. Beasley et al
BIOL PSYCHIATRY 2002;52:708 –715
though we suggest that the observed reduction in parvalbumin- and calbindin-immunoreactive neurons in the schizophrenic group was not due to alcohol abuse, this factor cannot be completely excluded. In this study, the use of stereological methods was not possible within the confines of the tissue available to us. The major source of error in any two-dimensional study is likely to be differences in cell size. Although we were not able to determine cell size in our study, Cotter et al (2002), using the same brain collection as used in this study, found no significant reductions in the somal size of calbindin- or parvalbumin-immunoreactive neurons in the anterior cingulate cortex in schizophrenia, bipolar disorder, or major depression. Two other studies have also shown that the size of parvalbumin-containing neurons does not differ between schizophrenic and control subjects (Kalus et al 1997; Woo et al 1997). Although the concept that stereological techniques represent the most accurate approach for determining neuronal density has recently been challenged (Benes and Lange 2001), the inconsistencies observed between the two-dimensional studies in this area indicate the need for a stereological assessment of cell volumes and densities of GABAergic neuronal subtypes in psychiatric disorders. A further possible source of error is tissue shrinkage during the processing, sectioning, and staining of tissue sections. Although ideally a correction for shrinkage should be employed when calculating neuronal densities in both two-dimensional and three-dimensional studies, this is often not feasible and was not possible within the confines of this investigation. These data provide further evidence for deficits in specific populations of GABAergic interneurons in the prefrontal cortex in schizophrenia. Similar reductions were observed in bipolar disorder. Although specific interneuron populations have not previously been quantified in the prefrontal cortex in this disorder, these reductions are consistent with a reduction in the expression of GAD mRNA and density of GAD-immunoreactive terminals (Benes et al 2000; Guidotti et al 2000). GABAergic deficits are likely to have wide-ranging effects on the neuronal circuitry of the prefrontal cortex and its output connections to other brain regions. Any change in the activity of GABAergic neurons, which play a vital role in modulating the activity of the cortex, would lead to imbalances in other systems, for example via their interactions with dopaminergic and glutamatergic neurons. This would inevitably have effects on many functions, such as cognitive processing. Studies of the GABAergic system may be valuable in the search to define the neuropathological changes in psychiatric disorders and elucidate possible putative mechanisms underlying the disturbances of cortical function observed.
This study was funded by a project award from the Theodore and Vada Stanley Foundation. Postmortem brains were donated by the Stanley Foundation Brain Bank Consortium, courtesy of Drs. Llewellyn B. Bigelow, Juraj Cervenak, Mary M. Herman, Thomas M. Hyde, Joel Kleinman, Jose D. Paltan, Robert M. Post, E. Fuller Torrey, Maree J. Webster, and Robert Yolken.
References Akbarian S, Kim JJ, Potkin SG, Hagman JO, Tafazzoli A, Bunney WE Jr, et al (1995): Gene expression for glutamic acid decarboxylase is reduced without loss of neurons in prefrontal cortex of schizophrenics. Arch Gen Psychiatry 52:258 –266. Akil M, Lewis DA (1992): Postnatal development of parvalbumin immunoreactivity in axon terminals of basket and chandelier neurons in monkey neocortex. Prog Neuropsychopharmacol Biol Psychiatry 16:329 –337. Bachus SE, Hyde TM, Herman MM, Eyan MF, Kleinman JE (1997): Abnormal cholecystokinin mRNA levels in entorhinal cortex of schizophrenics. J Psychiatr Res 31:233–256. Beasley CL, Reynolds GP (1997): Parvalbumin-immunoreactive neurons are reduced in the prefrontal cortex of schizophrenics. Schizophr Res 24:349 –355. Benes FM, Lange N (2001): Two-dimensional versus threedimensional cell counting: A practical perspective. Trends Neurosci 24:11–17. Benes FM, McSparren J, Bird ED, SanGiovanni JP, Vincent SL (1991): Deficits in small interneurons in prefrontal and cingulate cortices of schizophrenic and schizoaffective patients. Arch Gen Psychiatry 48:996 –1001. Benes FM, Todtenkopf MS, Logiotatos P, Williams M (2000): Glutamate decarboxylase(65)-immunoreactive terminals in cingulate and prefrontal cortices of schizophrenic and bipolar brain. J Chem Neuroanat 20:259 –269. Benes FM, Vincent SL, Marie A, Khan Y (1996): Up-regulation of GABAA receptor binding on neurons of the prefrontal cortex in schizophrenic subjects. Neuroscience 74:1021– 1031. Carder R, LeClerc SL, Hendry SHC (1996): Regulation of calcium-binding protein immunoreactivity in GABA neurons of macaque primary visual cortex. Cereb Cortex 6:271–287. Celio MR (1990): Calbindin D-28K and parvalbumin in the rat nervous system. Neuroscience 35:375– 475. Cotter D, Landau S, Beasley C, Stevenson R, Chana G, MacMillan L, Everall I (2002): The density and spatial distribution of GABAergic neurons in the anterior cingulate cortex in major depressive disorder, bipolar disorder and schizophrenia. Biol Psychiatry 51:377–386. Danos P, Baumann B, Bernstein H-G, Franz M, Stauch R, Northoff G-M, et al (1998): Schizophrenia and anteroventral thalamic nucleus: Selective decrease of parvalbumin-immunoreactive thalamocortical projection neurons. Psychiatry Res 82:1–10. Daviss SR, Lewis DA (1995): Local circuit neurons of the prefrontal cortex in schizophrenia: Selective increase in the density of calbindin-immunoreactive neurons. Psychiatry Res 59:81–96.
Prefrontal GABAergic Deficits in Schizophrenia
DeFelipe J, Hendry SHC, Jones EG (1989): Visualisation of chandelier cell axons by parvalbumin-immunoreactivity in monkey cerebral cortex. Proc Natl Acad Sci USA 86:2093– 2097. DeFelipe J, Jones EG (1991): Parvalbumin immunoreactivity reveals layer IV of monkey cerebral cortex as a mosaic of microzones of thalamic afferent terminations. Brain Res 562:39 – 47. Demeulemeester H, Vandesande F, Orban GA, Brando C (1988): Heterogeneity of GABAergic cells in cat visual cortex. J Neurosci 8:988 –1000. Evers P, Uylings HB (1994): Effects of microwave pretreatment on immunocytochemical staining of vibratome sections and tissue blocks of human cerebral cortex stored in formaldehyde fixative for long periods. J Neurosci Methods 55:163– 72. Gabriel SM, Davidson M, Haroutunian V, Powchik P, Bierer LM, Purohit DP, et al (1996): Neuropeptide deficits in schizophrenia vs. Alzheimer’s disease cerebral cortex. Biol Psychiatry 39:82–91. Guidotti A, Auta J, Davis JM, Gerevini VD, Dwivedi Y, Grayson DR, et al (2000): Decrease in reelin and glutamic acid decarboxylase 67 (GAD67) expression in schizophrenia and bipolar disorder: A postmortem brain study. Arch Gen Psychiatry 57:1061–1069. Hanada S, Mita T, Nishinok N, Tanaka C (1987): 3H-Muscimol binding sites are increased in autopsied brains of chronic schizophrenics. Life Sci 40:259 –266. Hsu SM, Raine L, Fanger H (1981): Use of avidin-biotinperoxidase complex (ABC) in immunoperoxidase techniques: A comparison between ABC and unlabelled antibody (PAP) procedures. J Histochem Cytochem 29:577–580. Jacobowitz DM, Winsky L (1991): Immunocytochemical localisation of calretinin in the forebrain of the rat. J Comp Neurol 304:198 –218. Kalus P, Senitz D, Beckmann H (1997): Altered distribution of parvalbumin-immunoreactive local circuit neurons in the anterior cingulate cortex of schizophrenic patients. Psychiatry Res 75:49 –59. Kamphuis W, Huisman E, Wadman WJ, Lopez Da Silva FH (1989): Decrease in GABA immunoreactivity and alteration of GABA metabolism after kindling in the rat hippocampus. Exp Brain Res 74:375–386.
BIOL PSYCHIATRY 2002;52:708 –715
715
Lewis DA, Lund JS (1990): Heterogeneity of chandelier neurons in monkey neocortex: Corticotropin-releasing factor and parvalbumin-immunoreactive populations. J Comp Neurol 293: 599 – 615. Nemeroff CB, Youngblood WW, Manberg PJ, Prange AJ Jr, Kizer JS (1983): Regional brain concentrations of neuropeptides in Huntington’s chorea and schizophrenia. Science 221:972–975. Rajkowska G, Goldman-Rakic PS (1995): Cytoarchitectonic definition of prefrontal areas in the normal human cortex: I. Remapping of areas 9 and 46 using quantitative criteria. Cereb Cortex 5:307–322. Reynolds GP, Beasley CL (2001): GABAergic neuronal subtypes in the human frontal cortex— development and deficits in schizophrenia. J Chem Neuroanat 22:95–100. Somogyi P, Hodgson AJ, Smith AD, Nunzi MG, Gorio A, Wu JY (1984): Different populations of GABAergic neurons in the visual cortex and hippocampus of cat contain somatostatin- or cholecystokinin-immunoreactive material. J Neurosci 4:2590 –2603. Torrey EF, Webster M, Knable M, Johnston N, Yolken RH (2000): The Stanley foundation brain collection and neuropathology consortium. Schizophr Res 44:151–155. Virgo L, Humphries C, Mortimer A, Barnes T, Hirsch S, DeBelleroche J (1995): Cholecystokinin messenger RNA deficit in frontal and temporal cerebral cortex in schizophrenia. Biol Psychiatry 37:694 –701. Volk DW, Austin MC, Pierri JN, Sampson AR, Lewis DA (2000): Decreased glutamic acid decarboxylase67 messenger RNA expression in a subset of prefrontal cortical ␥–aminobutyric acid neurons in subjects with schizophrenia. Arch Gen Psychiatry 57:237–245. Woo TU, Miller JL, Lewis DA (1997): Schizophrenia and the parvalbumin-containing class of cortical local circuit neurons. Am J Psychiatry 154:1013–1015. Woo TU, Whitehead RE, Melchitzky DS, Lewis DA (1998): A subclass of prefrontal gamma-aminobutyric acid axon terminals are selectively altered in schizophrenia. Proc Natl Acad Sci USA 95:5341–5346. Zhang ZJ, Reynolds GP (2002): A selective decrease in the relative density of parvalbumin-immunoreactive neurons in the hippocampus in schizophrenia. Schizophr Res 55:1–10.