Neuronal calcium-binding proteins and schizophrenia

Abstract

Calcium-binding proteins (CBPs) such as calbindin, parvalbumin and calretinin are used as immunohistochemical markers for discrete neuronal subpopulations. They are particularly useful in identifying the various subpopulations of GABAergic interneurons that control output from prefrontal and cingulate cortices as well as from the hippocampus. The strategic role these interneurons play in regulating output from these three crucial brain regions has made them a focus for neuropathological investigation in schizophrenia. The number of pathological reports detailing subtle changes in these CBP-containing interneurons in patients with schizophrenia is rapidly growing. These proteins however are more than convenient neuronal markers. They confer survival advantages to neurons and can increase the neuron's ability to sustain firing. These properties may be important in the subtle pathophysiology of nondegenerative phenomena such as schizophrenia. The aim of this review is to introduce the reader to the functional properties of CBPs and to examine the emerging literature reporting alterations in these proteins in schizophrenia as well as draw some conclusions about the significance of these findings.

Introduction

Fluctuations in intracellular calcium (Ca_i²⁺) are central to orderly neurotransmission and the operation of a wide range of cellular functions (Katz, 1969). The bulk of calcium that enters the cell is sequestered in organelles such as the smooth endoplasmic reticulum, mitochondria and in synaptic vesicles. However, in addition, many neurons possess a high-affinity/low-capacity system to assist in maintaining this homeostasis. Proteins that fulfil this function are collectively known as calcium-binding proteins.

Calcium-binding proteins (CBPs) are a family of proteins found in a variety of tissue across many different species (Christakos et al., 1989). The CBPs share a distinctive helix–loop–helix sequence (called the EF-hand) that undergoes a conformational change when binding calcium. Over 250 varieties of these proteins have been described (Celio et al., 1996). Given that neurons must deal with both the calcium-mediated signalling functions and the calcium influx associated with depolarisation, it is not surprising that the brain is a particularly rich source for these proteins. In the central nervous system, the most well described CBPs include parvalbumin (PV), calbindin-D28K (CB), calretinin (CR), calmodulin (CM), calcineurin, and the S100 family. CB and at least one of the S100 proteins, S100β, have also been shown to be present in glia. In the central nervous system, PV, CB and CR play a vital role in calcium homeostasis and are generally thought of as calcium buffering rather than calcium regulatory proteins. Their dispersal throughout the cytoplasm and into fine neuronal processes coupled with a highly distinctive distribution in subpopulations of neurons within the same region make them convenient markers for the neuropathologist (Baimbridge et al., 1992).

Neurons that express CBPs appear to be advantaged in several respects. Apoptosis is attenuated in cells that contain CB Dowd et al., 1992, Dowd, 1995. Hippocampal neurons in culture containing CB have been shown to be more effective in reducing intracellular Ca²⁺ concentrations compared with non-CB-containing neurons (Mattson et al., 1991). This has been proposed as a mechanism for increased seizure resistance in these cells (Scharfman and Schwartzkroin, 1989). Cortical neurons in culture containing CR are selectively resistant to excitotoxins (Lukas and Jones, 1994), and motor neurons and hippocampal neurons transfected with cDNA for CB have shown increased ability to buffer calcium and increased survival after sclerotic, hypoglycemic or excitotoxic induced injury Lukas and Jones, 1994, Ho et al., 1996. CB has also been shown to protect neurons from oxidative stress (Dowd, 1995). Finally, hippocampal progenitor cells transfected with CB cDNA are much less susceptible to nicotine-induced apoptosis (Berger et al., 1998).

There is also evidence in vivo that CBPs are neuroprotective. Hippocampal or motor neurons containing CR are resistant to Ca²⁺ induced excitotoxicity while those without this protein were vulnerable Mockel and Fischer, 1994, Terro et al., 1998. In tissue from patients with either motor neuron disease or temporal lobe epilepsy, neurons containing CB or PV survived while those without these proteins degenerated Ince et al., 1993, Sloviter, 1989. The association between CBPs and ability to survive neuronal stress is not consistent in all regions and is a subject of debate Andressen et al., 1993, Freund et al., 1990, Heizmann and Braun, 1992, Airaksinen et al., 1997. The combined evidence, however, would tend to suggest that the presence of CBPs does indeed confer a survival advantage to the neuron.

There is much evidence to suggest that malfunction within the prefrontal cortex (PFC) is a feature of schizophrenia Goldman-Rakic, 1991, Goldman-Rakic and Selemon, 1997, Lewis et al., 1999. The crucial role played by GABAergic inhibitory interneurons in regulating prefrontal function is well accepted and has been outlined in detail elsewhere Goldman-Rakic and Selemon, 1997, Lewis et al., 1999. In the cerebral cortex, CBPs are discrete markers for select subclasses of these neurons (e.g. PV is found in chandelier and basket cells, CB in double-bouquet neurons, CR in bipolar and bitufted neurons Conde et al., 1994, Lund and Lewis, 1993, Lewis, 1998. Collectively, CBP-containing interneurons make up 90% of all GABAergic neurons in this region (Lund and Lewis, 1993). In addition to the considerable heterogeneity in both the morphology and afferent control of pyramidal output from the frontal cortex, the receptor profiles for each subclass appear highly specific. Both D₁ and D₂ receptors on cortical interneurons appear to be selectively found on PV-containing cells while 5HT_2A and 5HT₃ receptors are differentially distributed between PV- and CB- and CR-immunoreactive neurons Le Moine and Gaspar, 1998, Jakab and Goldman-Rakic, 2000. There are many other examples of differential distribution of receptors between the CBP-expressing interneurons of the cortex where, for example, the glutamate/NMDAR1 subunit is found in few CR- but most PV-containing neurons (Huntley et al., 1994). The strategic role these interneurons play in regulating PFC output and the complexity of their relationship to pyramidal output neurons have made them a focus for neuropathological investigation of CBP abnormalities in schizophrenia. Results from the small number of studies that have reported neuronal CBP abnormalities in schizophrenia are summarised in Table 1.

One study of prefrontal areas 9 and 46 showed a 50–70% increased density of CB-positive neurons in layers III and V/VI but no group difference between schizophrenics and controls for CR (Daviss and Lewis, 1995). Although Nissl stains were conducted for general pathology in this study, they were not quantitatively examined. Therefore, it is not known whether the increase in CB partially reflects the increased cortical cell density observed in this disease (Selemon et al., 1995). In a second study, PV was found to be reduced in laminae III and IV of prefrontal area 10 (Beasley and Reynolds, 1997). No change in CR could be found in this tissue in a later study; however, a reanalysis of the previously reported PV findings showed deficits were more dramatic in patients without the commonly observed ventricular enlargement associated with the disease (Reynolds and Beasley, 2001). The same group has confirmed this reduction of PV-staining interneurons in two other prefrontal cortical areas using tissue from a different source (Reynolds et al., in press). In this same study, they also showed a corresponding decrease in CB with yet again no change in CR. The findings from one study conflict with this general trend in that the authors failed to find this reduction in PV in prefrontal areas 9 and 46 (Woo et al., 1997).

A potential problem in the interpretation of all of these studies is the failure to simultaneously quantify total neuron number in counter-stained or adjacent sections. Given that greater than 90% of these GABAergic interneurons contain PV, CB or CR, the simultaneous counting of all interneurons and those that contain a CBP would enable a statement to be made regarding whether the findings truly reflect cell number or an alteration in CBP content/cell. For instance, it is conceivable that if samples from the non-confirmatory study previously mentioned contained an increased total neuronal density (which may be a consequence of cortical shrinkage Selemon et al., 1995), then this may mask any percentage decrease in PV-containing cells when compared with controls.

An additional important finding is that the distinctive axonal fibre network from prefrontal chandelier interneurons was shown to be reduced in schizophrenics (Woo et al., 1998). These predominantly PV-containing interneurons (Andressen et al., 1993) provide direct inhibitory control over pyramidal cells within the prefrontal cortex. Any dysfunction in such an interaction may be highly relevant in explaining abnormal prefrontal output (see Discussion). The distribution of CR immunoreactive axonal boutons was also compared between patients and controls in this study. The inability to detect any change in apparent CR labelling in axonal boutons by these authors adds consistency to the lack of reported alterations in CR in schizophrenia.

Finally, in a recent qualitative investigation of prefrontal CBP abnormalities in schizophrenics, CB was shown to be unaltered in GABAergic interneurons in prefrontal area 9 but decreased in pyramidal cells from the CA2 region of the hippocampus (Iritani et al., 1999). No stereological techniques were applied in this study, and cells were not quantified so one must assume any reported decrease in CB was a visual impression only. The second finding from this study was an apparent random orientation of CB-containing neuronal fibres in schizophrenics compared with a more ordered array in non-psychiatric controls. It is interesting to note that changes in prefrontal area 9 were reported in laminae III–IV, whereas most of the CB-containing neurons are known to be in laminae II (Daviss and Lewis, 1995). However, no mention of the normality or otherwise of cell density or fibre array in any other layer was made.

In one quantitative examination of hippocampal CBP interneuronal profiles to date, a significant reduction in the number and size of PV-containing neurons across all hippocampal subfields was demonstrated with yet again no alteration in CR-containing neurons (Zhang and Reynolds, in press).

In two studies that did quantify total and CBP-containing neurons in adjacent histological sections, it appeared that the altered CBP histochemical response was due to a change in content per cell rather than a decrease in the number of cells immunostained. In the first of these studies, the authors confirmed that cell number was unchanged but PV content was increased in anterior cingulate GABAergic interneurons from schizophrenics (Kalus et al., 1997). Findings presented in a recent abstract, however, could not confirm this result; these authors finding a non-significant tendency to fewer PV-positive cells (Cotter et al., 2000).

PV-containing interneurons inhibit pyramidal cell output in this nucleus similar to the situation in the hippocampus and the PFC. A separate study investigated non-GABAergic projection neurons in the anteroventral thalamic nucleus (Danos et al., 1998). These neurons are known to project to the cortex in primates, but the destination of their afferents in humans is unknown. Approximately 50% of these neurons contain PV. This study showed that both the total number of anteroventral neurons and those expressing PV were decreased in the schizophrenics, but this reduction was only significant for the PV-containing cells. On closer inspection of the data, it was apparent that the total reduction in cell number in the schizophrenic group was insufficient to account for the diminished PV cell count implying that cell loss alone was not responsible for the reduction in PV staining. It must be assumed, therefore, that at least a significant number of anteroventral cells remaining in the schizophrenics must fail to express PV. Finally, reflecting perhaps the essential nature of such an important regulatory protein to the cell, no change in striatal calmodulin could be shown between schizophrenics and the non-psychiatric population (Vargas and Guidotti, 1980).