Cortical basket cell dysfunction in schizophrenia

Abstract

Schizophrenia, a debilitating illness affecting 0.5–1% of the world's population, is characterized by positive, negative and cognitive symptoms. The latter are the best predictor of functional outcome, though largely untreated by current pharmacotherapy; thus a better understanding of the mechanisms underlying cognitive deficits in schizophrenia is crucial. Higher order cognitive processes, such as working memory, are associated with θ (4–7 Hz) and γ (30–80 Hz) oscillations in the prefrontal cortex (PFC), and subjects with schizophrenia exhibit working memory impairments and reduced cortical θ and γ band power. Cortical θ and γ oscillations are dependent on perisomatic inhibition of pyramidal neurons from basket cells expressing cholecystokinin (CCK_b cells) and parvalbumin (PV_b cells), respectively. Thus, alterations in basket cells may underlie the cortical oscillation deficits and working memory impairments in schizophrenia. Recent findings from postmortem studies suggest that schizophrenia is associated with multiple molecular alterations that regulate signalling from CCK_b and PV_b cells. These alterations include lower CCK and cannabinoid 1 receptor (CB1R) in CCK_b cells, and lower glutamic acid decarboxylase 67 (GAD67) and increased μ opioid receptor (μOR) in PV_b cells, as well as lower GABA_A receptor α1 subunit in pyramidal neurons postsynaptic to PV_b cells. These changes are thought to lead to increased and decreased strength, respectively, of CCK_b and PV_b cell-mediated inhibition of postsynaptic pyramidal cells. Therefore, a convergence of evidence suggests a substantial shift in the relative strengths of PFC pyramidal cell inhibition from CCK_b and PV_b cells that may underlie cortical oscillation deficits and working memory impairments in schizophrenia.

David Lewis received his medical degree from the Ohio State University, completed residencies in internal medicine and in psychiatry at the University of Iowa, and received his research training at the Research Institute of the Scripps Clinic. He is currently the UPMC Professor in Translational Neuroscience and Chairman, Department of Psychiatry, University of Pittsburgh and Director of the Translational Neuroscience Program. His research activities focus on the neural circuitry of the prefrontal cortex, and the alterations of this circuitry in schizophrenia. Allison Curley completed her doctorate in the Department of Neuroscience at the University of Pittsburgh. She is currently a postdoctoral associate in Dr. Lewis's laboratory where she studies the role of interneuron and glial cell alterations in the prefrontal cortex in schizophrenia.

Schizophrenia is a devastating illness that affects 0.5–1% of the population worldwide (Lewis & Lieberman, 2000). Symptoms generally appear in late adolescence or early adulthood, and most affected individuals experience a life-long course of illness characterized by difficulties with employment, personal relationships and self-care. Schizophrenia is thought to be a neurodevelopmental disorder (Insel, 2010), caused by both genetic and environmental factors such as cannabis use during adolescence (van Os et al. 2010).

Symptoms of schizophrenia fall principally into three categories: positive, negative and cognitive. Positive symptoms are broadly characterized by an altered perception of reality, and include delusions, hallucinations, and disorganized thought and behaviour. Negative symptoms manifest as impaired emotional expression, motivation and social functioning. Cognitive impairments, such as alterations in working memory and executive function (Carter et al. 2008), are a core feature of schizophrenia (Elvevag & Goldberg, 2000) and the best predictor of functional outcome (Green et al. 2000). Unfortunately, current treatments for schizophrenia have little beneficial effects on cognitive symptoms (Goff et al. 2011). Thus, understanding the abnormalities in brain circuitry that give rise to the cognitive deficits is critical for developing new therapeutic interventions.

Higher order cognitive processes such as working memory are strongly associated with θ and γ oscillations, the synchronized activity of networks of pyramidal neurons at 4–7 and 30–80 Hz, respectively, in the prefrontal cortex (PFC) (Fries, 2009). For example, θ and γ band activity is induced during the delay period of working memory tasks in humans (Tallon-Baudry et al. 1998; Hsieh et al. 2011), and the power of θ and γ synchrony increases in proportion to working memory load (Gevins et al. 1997; Howard et al. 2003). In subjects with schizophrenia, cortical θ and γ band power and working memory performance are impaired (Schmiedt et al. 2005; Cho et al. 2006; Haenschel et al. 2009; Minzenberg et al. 2010).

Cortical θ and γ oscillations require strong inhibition provided by GABAergic inputs to the perisomatic region of pyramidal cells (Whittington et al. 1995; Bartos et al. 2007). Because a single interneuron may be highly divergent in its connections to pyramidal cells (e.g. in mouse neocortex, many parvalbumin (PV)-containing interneurons contact all pyramidal cells within their local field; Packer & Yuste, 2011), the firing of an interneuron can transiently silence many pyramidal cells. When the pyramidal cells are released from this inhibition, they fire in concert; the repetition of this cycle provides a synchronized oscillation of a neuronal network at a frequency that is inversely related to the duration of the pyramidal cell inhibition (Gonzalez-Burgos & Lewis, 2008).

Perisomatic inhibitory inputs to pyramidal neurons are furnished primarily by GABAergic basket cells that contain either the calcium-binding protein PV (PV_b cells) or the neuropeptide cholecystokinin (CCK_b cells). Although the activity of multiple interneuron subtypes is associated with both θ and γ oscillations, in the hippocampus the firing CCK_b cells is most strongly coupled to the θ oscillation cycle (Klausberger et al. 2005; Klausberger & Somogyi, 2008), whereas the firing of PV_b cells is most strongly coupled to the γ oscillation cycle (Hájos et al. 2004; Tukker et al. 2007; Gulyás et al. 2010). Indeed, recent studies in mice using optogenetic techniques have demonstrated that inhibition and activation of neocortical PV neurons suppresses and generates, respectively, cortical γ activity in vivo, indicating that PV neurons are both necessary and sufficient for the generation of γ oscillations (Cardin et al. 2009; Sohal et al. 2009). Additionally, because the duration of the inhibitory postsynaptic current (IPSC) in pyramidal cells is inversely related to oscillation frequency, the different GABA_A receptor subunits that predominate at CCK_b and PV_b cell synapses with pyramidal neurons support the idea that the two subtypes contribute to oscillations of different frequencies (Traub et al. 1996). The slow IPSC decay mediated by GABA_Aα2-containing receptors (Lavoie et al. 1997) that, in rodent hippocampus, predominate at CCK_b cell inputs to pyramidal cells (Nyíri et al. 2001), is consistent with low frequency θ oscillations; in contrast, the α1-containing receptors that predominate at PV_b cell synapses (Klausberger et al. 2002) exhibit fast IPSC decay kinetics that are consistent with the high frequency of γ oscillations (Lavoie et al. 1997; Gonzalez-Burgos & Lewis, 2008). However, some studies have reported that measured IPSC durations from CCK_b and PV_b cells do not differ (Glickfeld & Scanziani, 2006; Galarreta et al. 2008). Taken together, these findings raise the possibility that alterations in CCK_b and/or PV_b cells, or in their inputs to pyramidal neurons, could contribute to the cortical oscillation deficits and working memory impairments in schizophrenia.

CCK_b cell alterations in schizophrenia

In the primate neocortex, CCK-immunoreactive cell bodies are found predominately in layers 2–superficial 3 and their axon terminals predominantly innervate layers 2, 4 and 6 (Oeth & Lewis, 1993; Eggan et al. 2010) (Fig. 1). CCK mRNA is highly expressed in interneurons and is also present in pyramidal cells (Hökfelt et al. 2002), and CCK mRNA levels are lower in the PFC of subjects with schizophrenia (Hashimoto et al. 2008a; Fung et al. 2010) (Fig. 2). A subpopulation of CCK_b (but not PV_b) axon terminals contains the cannabinoid 1 receptor (CB1R) (Eggan et al. 2010) (CB1R is also present in some pyramidal cells at much lower levels; Freund et al. 2003), and lower CB1R mRNA and protein levels are present in the PFC in schizophrenia (Eggan et al. 2008, 2010; Hashimoto et al. 2008a) (Fig. 2). The deficits in CCK and CB1R mRNA levels are positively correlated (r = 0.81, P < 0.001 and r = 0.64, P = 0.001, respectively) with deficits in the mRNA encoding the 67 kDa isoform of glutamic acid decarboxylase (GAD67) (Hashimoto et al. 2008a; Eggan et al. 2008), the enzyme responsible for the majority of cortical GABA synthesis, leading to the suggestion that the GAD67 mRNA expression deficit in layers 2–3 includes CCK_b cells (Eggan et al. 2008). However, levels of GAD67 have not yet been measured in CCK_b cells in schizophrenia. Interestingly, in monkey PFC, CCK/CB1R co-expressing terminals have very low levels of GAD67 protein. For example, the ratio of GAD67 protein to the other isoform of the enzyme (GAD65) is only 0.2 in CCK/CB1R terminals relative to a ratio of 1.5 in PV_b cell terminals (Fish et al. 2011). Consistent with the hypothesis that CCK_b cells in monkey PFC principally rely on GAD65 for GABA synthesis (Fish et al. 2011), almost all CCK-containing, but very few PV-containing, cells in mouse neocortex also express GAD65 and appear to originate from the caudal ganglionic eminence (López-Bendito et al. 2004; Wierenga et al. 2010). Thus, CCK_b cells may not contribute to the GAD67 deficit in schizophrenia.

Figure 1. Brightfield photomicrographs of CCK (left), CB1R (middle) and PV (right) immunoreactivity in area 46 of monkey PFC.

Numbers and hash marks denote the positions of the cortical layers, and the dashed lines indicate the layer 6–white matter border. Scale bar = 300 μm and applies to all panels. Left and middle panels are from Eggan et al. 2010, with permission from Elsevier; right panel is from Melchitzky et al. 1999, with permission from John Wiley & Sons.

Figure 2. Expression levels of CCK (top), CB1R (middle) and GAD67 (bottom) MRANs in the PFC of 23 pairs of matched comparison and schizophrenia subjects.

Mean values are indicated by the bar graphs. For all measures, the distribution within each subject group reflects the influence of other factors such as age and sex. Comparisons of individual pairs of matched subjects revealed that the schizophrenia subject had lower CCK, CB1R or GAD67 mRNA levels in 17, 21 and 20 of 23 pairs, respectively. Data are from Hashimoto et al. 2008a (CCK), Eggan et al. 2008 (CB1R) and Hashimoto et al. 2003, 2005 (GAD67).

The inputs from CCK_b cells participate in a process known as depolarization-induced suppression of inhibition (DSI). When pyramidal cells are depolarized, the elevated calcium levels stimulate the retrograde release of endocannabinoids that bind to presynaptic CB1Rs located on CCK_b cell terminals (Alger, 2004). Through inhibition of presynaptic calcium channels, activation of CB1R receptors results in suppressed GABA release from CCK_b cell terminals, leading to reduced inhibition of the same and neighbouring pyramidal cells (Wilson & Nicoll, 2001, 2002). In the monkey PFC, DSI is present throughout development and into adulthood (Gonzalez-Burgos et al. 2010). Thus, in schizophrenia, lower levels of CB1Rs may result in less DSI, and subsequent greater inhibition of pyramidal cells by CCK_b cells. Consistent with this idea, markers of endocannabinoid synthesis and degradation are not altered in schizophrenia (Volk et al. 2010), suggesting that CB1R expression is not downregulated in the illness in response to elevated levels of endocannabinoids. Interestingly, application of CCK also results in an endocannabinoid-mediated suppression of GABA release from CCK_b cells (Földy et al. 2007); thus, lower levels of CCK in schizophrenia might also contribute to enhanced CCK_b cell output. Alternatively, lower levels of CCK and CB1Rs might represent a compensatory response to augment GABA neurotransmission if GAD67 is lower in CCK_b cells in schizophrenia (Eggan et al. 2008). Consistent with this interpretation, experimental reductions of GAD67 in mouse PFC are associated with lower CB1R mRNA expression, and similar to schizophrenia, the changes in GAD67 and CB1R mRNAs are significantly correlated (Eggan et al. 2012). However, since GAD67 protein levels are normally very low in CCK_b axon terminals in monkey PFC (Fish et al. 2011), germ-line reductions in GAD67 expression may lead to lower CB1R mRNA levels in a non-cell-autonomous fashion.

PV_b cell alterations in schizophrenia

In primate PFC, the highest densities of PV-labelled cell bodies and puncta (putative PV_b axon terminals) are present in layers deep 3–4 (Conde et al. 1994; Melchitzky et al. 1999; Erickson & Lewis, 2002) (Fig. 1). A reduced density of PV-immunoreactive axon terminals in this laminar location in schizophrenia (Lewis et al. 2001) might reflect fewer PV_b terminals. In addition, ∼50% of PV mRNA-positive neurons lack detectable levels of GAD67 in schizophrenia relative to matched control subjects (Hashimoto et al. 2003). Total GAD67 protein levels are also lower in schizophrenia (Guidotti et al. 2000; Curley et al. 2011), and this deficit is particularly pronounced in PV_b axon terminals (Fig. 3). In a recent study (Curley et al. 2011), tissue sections were triple-labelled for GAD65, PV and GAD67. Because the axon terminals of the other major class of PV neurons, chandelier (PV_ch) cells, contain very low levels of GAD65 protein (Fish et al. 2011), the use of GAD65 to identify axon terminals facilitated the exclusion of PV_ch cell terminals. Quantitative fluorescent imaging demonstrated that GAD67 protein was ∼50% lower in the axon terminals of PV_b cells, a deficit 5–10 times greater than the reduction in tissue levels of GAD67 protein in the same subjects (Curley et al. 2011) (Fig. 3). Together, these findings suggest that GAD67 mRNA and protein levels are lower predominantly in PV_b cells and that this deficit may lead to reduced PV_b cell-mediated inhibition of pyramidal cells in schizophrenia.

Figure 3. GAD67 protein in total grey matter (top; n = 19 pairs) and GAD67 protein in PV_b axon terminals (bottom; n = 5 pairs) in the PFC of matched comparison and schizophrenia subjects.

Mean values are indicated by the bar graphs. For both measures, the distribution within each subject group reflects the influence of other factors such as age and sex. Comparisons of individual pairs of matched subjects revealed that the schizophrenia subject had lower GAD67 protein levels in 12/19 and 5/5 pairs, respectively. Data are from Curley et al. 2011.

PV_b cells also exhibit other presynaptic alterations consistent with a weaker inhibition of pyramidal cells in the PFC of subjects with schizophrenia. For example, mRNA levels of the μ opioid receptor (μOR) are elevated in the PFC of subjects with schizophrenia (Volk et al. 2012). Activation of the μOR, which is localized to PV cells and other interneurons in the hippocampus (Drake & Milner, 2002; Stumm et al. 2004; Krook-Magnuson et al. 2011), diminishes the activity of PV_b cells through two different mechanisms; perisomatic μORs hyperpolarize the cell body through the activation of inwardly rectifying potassium channels (Wimpey & Chavkin, 1991; Glickfeld et al. 2008), making the cell less likely to fire, and μORs on axon terminals lead to the suppression of vesicular GABA release (Capogna et al. 1993; Lupica, 1995). Thus, in schizophrenia, higher levels of μORs would be expected to reduce the activity of, and/or GABA release from, PV_b cells resulting in weaker perisomatic inhibition of pyramidal cells.

On the postsynaptic side of the PV_b-pyramidal cell synapse, lower expression of the GABA_A receptor α1 subunit in schizophrenia may also result in reduced inhibition of pyramidal cells. Levels of α1 mRNA are lower in the PFC of subjects with schizophrenia (Akbarian et al. 1995; Hashimoto et al. 2008a,b; Beneyto et al. 2011; but see Duncan et al. 2010). In the same subjects, mRNA levels of the GABA_Aβ2 subunit, which preferentially co-assembles with the α1 subunit, are also lower to a similar degree in schizophrenia (Akbarian et al. 1995; Beneyto et al. 2011). Furthermore, the lower levels of GABA_Aα1 and β2 subunit mRNAs are most prominent in layers 3–4 (Beneyto et al. 2011), layers where PV_b cells and axon terminals are most abundant (Conde et al. 1994). In layer 3, GABA_Aα1 mRNA levels are reduced by 40% in pyramidal cells, but are unaltered in GABA neurons, suggesting a selective reduction of pyramidal cell inhibition in schizophrenia (Glausier & Lewis, 2011).

PV_b cells also receive input from CCK_b cells (Karson et al. 2009). Application of CCK, acting at CCK2 receptors, can activate PV neurons (Földy et al. 2007; Karson et al. 2008). Thus, lower levels of CCK mRNA in schizophrenia (Hashimoto et al. 2008a; Fung et al. 2010) may also contribute to weaker PV_b cell activity and lower inhibition of pyramidal neurons.

Although incomplete, the existing data suggest that the schizophrenia-associated abnormalities in PV_b cells may differ from those in PV_ch neurons that synapse exclusively on the axon initial segment of pyramidal cells via a distinctive arrangement of axon terminals termed cartridges. PV_ch cells exhibit a reduced density of cartridges immunoreactive for the GABA membrane transporter 1 (GAT1) (Woo et al. 1998; Pierri et al. 1999). In contrast, the density of GAT1-IR puncta (presumably all other non-cartridge puncta, including those from PV_b cells) is not altered in schizophrenia (Woo et al. 1998), suggesting that PV_b terminals contain normal levels of GAT1. In addition, the density of axon initial segments immunoreactive for GABA_A receptor α2 subunit is increased in schizophrenia (Volk et al. 2002), whereas the pyramidal cells postsynaptic to PV_b cells exhibit lower levels of α1 mRNA (Glausier & Lewis, 2011). Since GAD67 mRNA is undetectable in ∼50% of PV neurons in schizophrenia (Hashimoto et al. 2003), and GAD67 mRNA levels per neuron in schizophrenia appear to be markedly reduced in a subpopulation of GABA neurons and normal in the remainder (Volk et al. 2000), the existing data raise the possibility that GAD67 expression is markedly reduced in PV_b cells and unchanged in PV_ch cells in schizophrenia (Lewis et al. 2012). However, a direct assessment of GAD67 levels in PV_ch cells is needed.

Effect of basket cell alterations on pyramidal cell functioning in schizophrenia

In isolation, lower levels of CCK and CB1Rs in CCK_b cells would be expected to strengthen the inhibition of postsynaptic pyramidal cells. Inhibition by CCK_b cells might also be strengthened by increased expression of GABA_Aα2 receptors (Beneyto et al. 2011) in pyramidal cells, although whether these pre- and postsynaptic changes occur at the same synapses remains to be determined. In contrast, lower levels of GAD67 and increased levels of μORs in PV_b cells would be expected to weaken the inhibition of postsynaptic pyramidal cells. This reduction in inhibition would be exacerbated by lower levels of GABA_Aα1 receptors in pyramidal cells and lower levels of CCK in CCK_b cells that could decrease the activity of PV_b cells. Thus, a convergence of evidence suggests that schizophrenia is associated with a substantial shift in the relative strengths of PFC pyramidal cell inhibition from CCK_b and PV_b cells (Table 1 and Fig. 4).

Table 1.

Comparison of the properties of CCK_b and PV_b cells in the primate PFC and in schizophrenia

Properties	CCKb cells	PVb cells	Refs
Laminar location of cell bodies	Predominately layer 2–superficial 3	Predominately layers 3–4	Oeth & Lewis (1993); Conde et al. (1994)
Laminar location of axon terminals	Predominately layers 4 and 6	Predominately layers 3–4	Melchitzky et al. (1999)
Location of synapses on pyramidal neurons	Dendritic shafts > dendritic spines > soma	Dendritic spines > dendritic shafts > soma	Melchitzky et al. (1999); Eggan et al. (2010)
GAD67/GAD65 ratio in axon terminals	0.2	1.5	Fish et al. (2011)
Postsynaptic GABAA receptor α subunit	α2	α1	Nyíri et al. (2001); Klausberger et al. (2002)
Alterations in the PFC in schizophrenia	GAD67 mRNA levels unknown	↓GAD67 mRNA	Hashimoto et al. (2003)
	GAD67 protein levels in axon terminals unknown	↓ GAD67 protein in axon terminals	Curley et al. (2011)
			Hashimoto et al. (2008a)
	↓ CCK mRNA	↓ GABAAα1 mRNA in layers 3–4	Eggan et al. (2008)
	↓ CB1R mRNA and protein	↓ GABAAα1 mRNA selectively in pyramidal cells in layer 3	Beneyto et al. (2011)
	↑ GABAAα2 mRNA in layer 2		Glausier & Lewis (2011)
		↑μ opioid receptor mRNA	Volk et al. (2012)

Figure 4. Schematic drawing illustrating alterations in CCK_b and PV_b cells in schizophrenia.

CCK_b cell-mediated inhibition of pyramidal neurons (grey cell) may be augmented due to lower CCK mRNA and CB1R mRNA and protein that enhances the GABA release from CCK_b cells. In contrast, PV_b cell-mediated inhibition of pyramidal neurons may be weakened due to lower GAD67 mRNA and protein resulting in less GABA release; increased μOR mRNA that diminishes PV_b cell activity and suppresses GABA release; lower CCK mRNA that can lead to reduced activity of PV_b cells; and lower GABA_Aα1 mRNA in pyramidal cells that diminishes the postsynaptic response. Levels of GAD67 in CCK_b cells are not known.

An important question in determining the functional consequences of basket cell dysfunction in schizophrenia is whether the alterations in each cell type affect the same pyramidal cells. CCK mRNA levels in schizophrenia have not been examined in a laminar fashion. Although levels of CB1R immunoreactivity are lower in all PFC layers in schizophrenia, this difference is only statistically significant in layers deep 3–4 and 6 (Eggan et al. 2008). Additionally, PV-IR axon terminals exhibit a reduced density in layers deep 3–4, but not in layers 2–superficial 3 (Lewis et al. 2001). Thus, it appears that alterations in PV_b and CCK_b cell terminals both occur in the middle cortical layers, even though their cell bodies are located in different layers. However, whether the affected CCK_b and PV_b cells in schizophrenia target the same pyramidal neurons remains to be determined.

Although not conclusive, the current findings raise the possibility that schizophrenia is associated with an increase in the ratio of CCK_b to PV_b cell-mediated perisomatic inhibition of pyramidal cells. Disruption of the balance of these two sources of inhibition is hypothesized to perturb normal network functioning (Freund, 2003), and thus provides a plausible mechanism underlying both the θ and γ oscillation deficits in schizophrenia. In addition, the coupling of θ and γ oscillations during memory tasks (Schack et al. 2002; Lisman & Buzsáki, 2008) suggests that an impairment in one frequency, such as reduced γ oscillations resulting from deficient PV_b-mediated inhibition, may result in corresponding alterations in θ oscillations. Thus, altered perisomatic inhibition of PFC pyramidal neurons by CCK_b and PV_b cells could provide a plausible mechanistic basis for the θ and γ oscillation impairments and cognitive deficits in schizophrenia (Schmiedt et al. 2005; Cho et al. 2006; Haenschel et al. 2009; Minzenberg et al. 2010).

Glossary

Abbreviations

CB1R

cannabinoid 1 receptor

CCK

cholecystokinin

CCK_b

cholecystokinin basket

DSI

depolarization-induced suppression of inhibition

IPSC

inhibitory postsynaptic current

GAD65

glutamic acid decarboxylase 65

GAD67

glutamic acid decarboxylase 67

GAT1

GABA membrane transporter 1

μOR

mu opioid receptor

PFC

prefrontal cortex

PV

parvalbumin

PV_b

parvalbumin basket

PV_ch

parvalbumin chandelier

Disclosures

David A. Lewis currently receives investigator-initiated research support from Bristol-Myers Squibb, Curridium Ltd and Pfizer and in 2009–2011 served as a consultant in the areas of target identification and validation and new compound development to BioLine RX, Bristol-Myers Squibb, Merck and SK Life Science. Allison A. Curley reports no competing interests.