Bente Pakkenberg, Research Laboratory for Stereology and Neuroscience, Bispebjerg University Hospital, Bispebjerg Bakke 23, DK-2400 Copenhagen NV, Denmark. E-mail: firstname.lastname@example.org
Objective: To describe structural and biochemical evidence from postmortem brains that implicates the reciprocal connections between the mediodorsal thalamic nucleus and the prefrontal cortex in cognitive symptoms of schizophrenia.
Method: The estimation of the regional volumes and cell numbers was obtained using stereological methods. The biochemical analyses of molecular expression in postmortem brain involve quantitative measurement of transcripts and proteins by in-situ (RNA) or Western blot/autoradiography in brains from patients with schizophrenia and comparison subjects.
Results: Stereological studies in postmortem brain from patients with schizophrenia have reported divergent and often opposing findings in the total number of neurons and volume of the mediodorsal (MD) thalamic nucleus, and to a lesser degree in its reciprocally associated areas of the prefrontal cortex. Similarly, quantitative molecular postmortem studies have found large inter-subject and between-study variance at both the transcript and protein levels for receptors and their interacting molecules of several neurotransmitter systems in these interconnected anatomical regions. Combined, large variation in stereological and molecular studies indicates a complex and heterogeneous involvement of the MD thalamic-prefrontal loop in schizophrenia.
Conclusion: Based on a considerable heterogeneity in patients suffering from schizophrenia, large variation in postmortem studies, including stereological and molecular postmortem studies of the MD thalamus and frontal cortex, might be expected and may in fact partly help to explain the variable endophenotypic traits associated with this severe psychiatric illness.
• Postmortem stereological and molecular studies are associated with a high degree of biological variance, which may reflect the heterogeneous nature of schizophrenia that is also seen in the entrance of endophenotypic traits in patients.
• The role of mediodorsal (MD) thalamus as relay station for efferent and afferent projections between prefrontal cortex (PFC) and subcortical areas may be important in the pathophysiology of in particular cognitive symptoms in schizophrenia. This circuitry also involves hippocampus, striatum, pallidum and the ventral tegmental areas.
• The role of glutamate and GABA as the principal neurotransmitters in the MD thalamus and PFC points to an important role for these transmitters as pharmaceutical targets for future treatment of cognitive symptoms of schizophrenia.
• Postmortem studies in particular are associated with confounding variables that include differences in age, gender, medication, as well as comorbidity issues related to an overrepresentation in patients with schizophrenia of alcohol and substance abuse.
Wilhelm Griesinger (1817–68), who first distinguished psychiatry as a separate discipline, was the first to classify mental illnesses as distinct organic brain diseases (1). Later, Emil Kraepelin (1856–1926) contributed to the understanding of psychiatric disease by differentiating between manic-depressive disorder and Dementia praecox, which his student, Eugen Bleuler (1857–1939) subsequently termed schizophrenia (2, 3). Furthermore, Kraepelin proposed that frontal lobe dysfunction likely provides the biological substrate for cognitive disturbances in schizophrenia, including altered emotion, volition and judgment, a view that is still today widely accepted. Substantial evidence implicates several prefrontal cortical regions in schizophrenia, including dorsolateral prefrontal cortex (DLPFC; area 9, 10 and 46) and cingulate cortex (area 32, 24, 25) (4, 5). The concept and definition of the prefrontal cortex (PFC), at the functional level, includes its glutamate-mediated reciprocal interactions with the mediodorsal (MD) thalamic nucleus, which is one of the primary linked subcortical structures of the PFC. In 1895, Monakow was the first to propose that the MD thalamic nucleus plays a central role in the cognitive and emotional life of an individual (6) and Bäumer later proposed the MD thalamus as a principal area of interest for the pathology of schizophrenia (7). However, it was not until substantially later that the essential role of the MD thalamus for a normal functional PFC was recognized. Hence, for more than a century the involvement of PFC and MD thalamus has been at the center of schizophrenia research.
Aims of the study
In this short review, we will present stereological and biochemical evidence from postmortem brain that implicates in particular the reciprocal connections between the mediodorsal (MD) thalamus and prefrontal cortex (PFC) in cognitive symptoms of schizophrenia.
Material and methods
The review includes a discussion of the principal anatomical network substrates at stereological, cellular and molecular levels that point to a specific pathology of the MD thalamic nucleus and its functionally linked PFC regions in the pathophysiology of schizophrenia.
Functional anatomy of the bidirectional glutamatergic connection between PFC and MD thalamus
The involvement of the human thalamus as functional relay station for sensory input to higher sensory and motor cortices is well-known (8, 9). The PFC, which represents the highest anatomical region for elaborate and integrative cognitive processing of internally interoceptive and executive mental processes, receives no direct input from outside of the central nervous system (CNS) (10, 11). Several studies have indicated the MD thalamus as a central structure for processing of direct information to and from various PFC regions, including functional regulation of cognitive performance through its direct top–down innervation by glutamatergic cortico-thalamic projections (Fig. 1). In addition to these direct pathways, all subnuclei of the MD thalamus receive indirect PFC efferents through the PFC – nucleus caudatus – nucleus accumbens – ventral pallidal pathway, which constitute a ‘GABA-GABA dis-inhibitory’ return circuit to the MD nucleus (12). Furthermore, recent anatomical evidence indicate, the existence of direct thalamo-cortical feed-forward loops that, in addition to the original afferent PFC area, functionally integrate other relevant regions of the PFC. Such integrative networks are also facilitated by other indirect pathways such as the midline to lateral loops, which include the PFC – striatal – ventral pallidal – substantia nigral – thalamic pathway that facilitate cortical return information to more lateral, cortical regions (12). Combined, the functional evidence for involvement of MD thalamus and PFC in cognitive dysfunction in schizophrenia implicates altered function of several of their connecting pathways in the pathophysiology of schizophrenia (9, 10, 13–20).
Several studies in numerous animal species, including non-human primates, have focused on detailed anatomical mapping of the CNS, including the MD thalamus (8, 13–15, 17). Based on these studies, a comprehensive topographical map has been established of the thalamus describing three functionally distinct subregions (or subnuclei) of the MD, which has been translated to the MD thalamic nucleus in man (15, 21, 22). These subregions include the anteromedial magnocellular (MDMC), dorsolateral parvocellular (MDPC) and posterolateral densocellular (MDDC) or multiform portions (MDMF). The nomenclature and precise delineation of these subregions varies throughout the literature, in particular with regard to MDDC or MDMF, which, due to differences in the methodology used for anatomical mapping, has been described based on either general morphological characteristics or neuronal size and packing density (23). Such differences in functional mapping are particularly challenging in connection with cross-species translation of anatomical tracing studies such as studies of the thalamic innervation pattern in different species including pig, rat and monkey (24). However, it is now firmly established that all subregions of the MD thalamus receive GABAergic innervation from nucleus accumbens via pathways that involve ventral pallidum and the thalamic reticular nucleus, which is an inhibitory structure that surrounds the MD thalamus as a GABAergic shunt. Similarly, all PFC regions provide strong input to the MD thalamus through excitatory glutaminergic innervation of the reticular formation as well as through direct innervation of glutaminergic MD thalamic projection neurons. MD Afferent projections from other subcortical structures, including from the tegmental midbrain region, show regional specificity in their innervation of the MD subnuclei.
The thalamo-cortical projections from different MD subnuclei are specific to separate PFC cortical areas, which further contribute to the functional segregation of the MD thalamic subnuclei (10, 13–18, 23). Hence, in monkey, MDMC reciprocally interacts with orbital (areas 11–14), infralimbic (area 25) and dorsal cingulate cortices (area 24). In addition, the MDMC receives input from the olfactory cortex, medial temporal lobe, basolateral amygdala, entorhinal cortex as well as several other subcortical structures.
With regard to schizophrenia, MDPC possibly represents the most interesting region of the MD thalamus due to its reciprocal connections with DLPFC and anterior cingulate cortex (areas 9, 10, 24, 32, 46). These regions have all been directly implicated in molecular and cellular changes of schizophrenia as well as functionally in several of the clinical symptoms associated with this illness. MDPC furthermore projects to the most anterior supplementary motor areas (SMA) and receives input from several subcortical areas including substantia nigra, the ventral tegmental area (VTA), and peri aqueductal grey (PAG).
The densocellular MDDC or multiform portion (MDMF) of the MD thalamus is localized laterally and posterior within the MD and is enveloped by the MDDC. According to Jones, this region includes the paralamellar part of the MD thalamus (15). The MDDC reciprocally interacts with premotor areas including the SMA, frontal eye field (area 8a) and Broca’s area. Subcortical input areas of the MDDC include substantia nigra, the superior colliculus, dentate nucleus, cerebellum and the spinal cord.
Thalamic neuronal cell types. Histologically, two populations of neurons can be differentiated in the human thalamus, including the MD nucleus (21, 23, 25). However, even though these neuronal populations, based on somal size, are largely non-overlapping, a clear definite distinction requires immunocytochemical techniques. One population, which represents approximately 20–40% of these neurons, involves small, presumably inhibitory (GABAergic) interneurons that anatomically and functionally are restricted within each specific thalamic subnucleus. The second population involves larger excitatory relay neurons that project directly to the cortex.
Stereological studies in schizophrenia. Several postmortem studies have found considerable stereological changes associated with the MD thalamus in schizophrenia. Among the most striking of these findings are reductions in the total number of neurons of 27–40% and a volume reduction of 9–25% (26–31). Contrary to these studies, other reports have found no changes in total neuron number associated with the MD thalamus in schizophrenia e.g. (25, 32–34).
The total numbers of neurons and glial cells in brains of patients with schizophrenia and comparison subjects have, similar to MD thalamus, also been examined in the PFC area 24 and 32 of the anterior cingulate cortex. These studies found a significant decrease in the total amount of glial cells in area 24 of approximately 33% (200 × 106 in subjects with schizophrenia as compared to 300 × 106 in the control group), with no changes in total glia cell number in area 32 (35). Furthermore, the total number of neurons in areas 24 and 32 did not differ between the schizophrenia and comparison subjects. Hence, this study indicates that regionally specific reductions in glial cells expression, which represent an important buffer system for GABA and glutamate synapses, could be associated with schizophrenia.
A major limitation to most postmortem stereological studies is the small number of subjects included in each study. Variables such as age, gender, genetic background, onset and duration of illness, differences in diagnoses (different endophenotypes) and treatment status (medication) in addition to differences in the methodological approach likely cause some of the large inter-study differences observed in postmortem studies. An example of these problems that are inherently associated with postmortem research studies is the risk of introducing a study bias due to the inclusion criteria used by different studies. In our study from 1990, showing a large cell loss in MD thalamus, the included patients was a cohort of schizophrenic subjects that all came from the same psychiatric hospital for chronically ill patients who had spent a lifetime in the institution and could thus represent a selected group of patients (26). Such inevitable problems may cause different studies not to be directly comparable.
In a recent study from our laboratory, the neuronal cell numbers was found to vary more than two-fold in the MD thalamus in patients with schizophrenia (from 3.68 × 106 to 9.22 × 106), whereas the range for comparison subjects was 5.24 × 106 to 7.10 × 106 cells in the MD thalamus (36). This indicates that in this specific inhomogeneous group of patients, some subjects had increased numbers of neurons, while others had a lower than normal number of neurons in the MD thalamus. Furthermore, in a previous study from our laboratory we found an excess of neurons in the MD thalamus in normal newborns of approximately 11.2 × 106 as compared to 6.43 × 106 in the normal adult brain (37). Based on these studies, the biological variation in total cell numbers is higher in schizophrenia. This suggests that in subgroups of patients, this illness could be associated with abnormal programmed cell death in MD thalamus, causing patients to either have too many or too few neurons. A high biological variance among patients with schizophrenic coincides well with the general observation that this and other psychiatric illnesses are not attributed to a single factor, but rather is a multifactorial disease associated with various combinations of genetic, developmental and environmental factors.
Molecular studies in schizophrenia. Similar to stereology-based evidence, molecular studies in postmortem brain have implicated MD thalamus and its reciprocally associated areas of the PFC in the pathophysiology of cognitive symptoms of schizophrenia (38–40). Due to the role of glutamate as the principal neurotransmitter in the direct pathways between PFC and MD thalamus, considerable effort has been focused on identifying molecular changes of this transmitter system in both PFC and MD thalamus (39, 41–43). Glutamate transmission involves a complex interaction of pre- and postsynaptic entities that, in collaboration with the surrounding astroglial extensions, constitute functional areas for efficient and regulated excitatory signaling (44). Hence, the presynaptic and glial compartments of glutamatergic synapses perform specialized cellular functions that include synthesis, storage, release and reuptake of synaptic glutamate and other synapse-active compounds (45, 46). Similarly, the postsynaptic entity is characterized by the presence of a highly specialized intracellular molecular network, termed the postsynaptic density (PSD), which functionally integrates signaling by different glutamate receptors and connects these receptors with various intracellular postsynaptic signaling proteins, thus establishing large postsynaptic signaling complexes (47, 48). A multitude of different glutamate receptors are, based on their functional characteristics, divided into either ion-channels or metabotropic G-protein interacting forms (49). Both of these receptors classes can be further subdivided into the ionotropic subclasses that include the N-methyl-d-asparate (NMDA), α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid (AMPA) and kainate receptors and eight different metabotropic receptors that structurally and functionally are divided into three distinct subclasses. Based on clinical studies, in particular the NMDA receptor has been directly implicated in schizophrenia (50). Highly supporting evidence for this hypothesis involves studies where administration of the NMDA receptor antagonist ketamine in patients with schizophrenia was observed to worsen positive, negative and cognitive symptoms, while in healthy volunteers, NMDA antagonism produced hallucinations that were indistinguishable from positive symptoms in patients (51–53).
By design, postmortem expression studies of glutamate-related molecules in the thalamus and PFC naturally implicates the direct glutamatergic pathways between these regions. However, quantitative molecular expression studies have, similar to other types of postmortem studies, generally been associated with considerable inter-study variation. Despite problems with inter–study variation within a heterogeneous schizophrenic population, molecular studies of thalamic and cortical expression have generally found solid evidence for altered function of glutamate-related signaling in the thalamic and PFC areas (39, 50, 54, 55). Numerous preclinical studies have underlined the essential role of a mutual interaction between the prefrontal GABA and glutamate systems for higher integrative cognitive functions. Hence, preclinical as well as postmortem studies underline the essential role of these transmitters in particular for cognitive dysfunction in schizophrenic patients (42, 56, 57). This includes altered molecular expression in cortico-subcortical pathways such as the corticothalamic connections that regulate functions in the PFC (58).
A recent study found altered expression of transcript and protein for the vesicular glutamate transporter, VGLUT1 in the anterior cingulate cortex in schizophrenia while no changes in the expression of the related molecule VGLUT2 was found in PFC (59). As VGLUT1 protein expression in the thalamus has been found to selectively define corticothalamic nerve terminals (60), altered expression of transcripts for VGLUT1 in anterior cingulate cortex directly suggests functional deficit of the PFC innervation of the MD thalamus.
Despite the high variation inherently associated with postmortem molecular studies, altered receptor binding and expression for all glutamate receptor forms has been reported in MD thalamus and PFC regions in schizophrenia (55). Consistent with a general hypothesis of decreased NMDA receptor function in schizophrenia, several studies have identified altered expression for several of the NMDA receptor subunits, including activity-induced alternative splicing of the obligatory NR1 subunit, as well as the of NMDA receptor-interacting molecules of the PSD (50, 61–64). Combined these studies indicate altered functional integration of the PSD, in particular of NMDA receptors in schizophrenia (50). Hence, based on insight into NMDA receptor regulation, altered expression in postmortem brain of an alternatively spliced NR1 exon (C2/C2′), was proposed to represent a compensatory mechanism to compromised dendritic NMDA receptor function (61). In addition, accumulating evidence indicates that compromised postsynaptic NMDA receptor function might in part be caused by altered synthesis and/or dendritic trafficking of these receptors in PFC.
Few postmortem studies have analyzed whether the thalamic glutamate system might be affected at the molecular level in schizophrenia. However, similar to PFC, expression of transcripts for all ionotropic NMDA receptors have, with some degree of inter-study variation, been reported altered in schizophrenia (39, 55, 65, 66). Interestingly, Clinton et al. (67) in a recent study, found increased expression of the NR2B subunit of the NMDA receptor in the DM thalamus, which might be interpreted as a compensatory regulation to altered innervation by corticothalamic pathways.
In summary, several recent postmortem studies implicate altered functional integration of glutamate signaling in the thalamus and PFC in schizophrenia. Due to the bidirectional glutamatergic projections between cortical and thalamic neurons, changes at the molecular level support other lines of evidence for compromised function of the interacting thalamo-cortical functional loops.
Schizophrenia represents a multifaceted disease, which may underlie part of the observed large inter-study variation found in postmortem studies. This is well in line with a large variation in the clinical symptoms in patients, which include perceptional distortion, psychotic symptoms, psychomotor disturbances and cognitive, emotional and social dysfunction. It is therefore not surprising that postmortem studies in schizophrenia, including cell counting and biochemical analyses of glutaminergic, GABAergic and dopaminergic systems in postmortem brain show high inter-study variance, even for analyses performed within the same laboratory. It is noteworthy that these inter-study differences generally show higher variation in patient samples than within control groups.
Dysfunctional neurotransmitter signaling in the CNS appears as a central component of the pathophysiology of schizophrenia. In particular, this implicates altered function of glutaminergic, GABAergic and dopaminergic signaling pathways as well as of other less well-studied systems (68). Traditionally, dopamine has been the principal neurotransmitter studied in postmortem brain from schizophrenic patients. This is due to a clinical breakthrough in the 1950s where D2-dopamine receptor antagonism was discovered to efficiently reduce positive symptoms in patients (69). Furthermore, hyper-dopamine receptor function in schizophrenia was indicated by an observation that compounds such as d-lysergic acid diethylamide (LSD) and amphetamine in healthy, non-schizophrenic volunteers, promoted hallucinogenic effects similar to positive symptoms in patients (70). Based on these studies, development of novel drugs for schizophrenia has continuously been focused on D2 receptor modulation in combination with regulation of other signaling pathways, aiming at differential adjustment of dopamine function in limbic and cortical regions of the brain (71, 72). However, despite intense research, including introduction of ‘atypical’ antipsychotic compounds, the efficiency of dopamine-interfering drugs has been associated primarily with the treatment of positive symptoms with limited effects on cognitive and negative symptoms.
Cognitive symptoms of schizophrenia, which include deficient attention, reduced learning and memory, altered social behavior, as well as compromised executive function, is associated with integrated neuronal processing in higher anatomical regions including the hippocampus, and PFC (56, 73). In-turn this involves the MD thalamus due to its direct connections and function as relay station, which serves to functionally link multiple prefrontal regions (10, 11). Overall, activity in these regions depends heavily on excitatory and inhibitory signaling by the glutamate and GABA neurotransmitters and numerous postmortem studies have therefore investigated a potential role for glutamate and GABA in cognitive symptoms of schizophrenia (57, 74). In addition, attempts to modulate receptor activity by administration of d-serine and other ligands for NMDA, AMPA and metabotropic receptors have been partially successful. In particular, agonist stimulation of presynaptic group II/III metabotropic glutamate receptors have pointed to an important role of these receptors in regulation of synaptic glutamate function and therapeutic improvement of cognitive ability in patients (75, 76).
Similarly, due to their involvement in controlling the firing pattern of cortical pyramidal neurons, modulation of inhibitory GABA neurons have been targeted as an alternative strategy for understanding and improving cognitive function in schizophrenia (74, 77, 78). Hence, modulation of GABA-Aα2 receptors was recently reported to exert a positive effect on cognitive control and working memory in schizophrenic patients (79). These preliminary studies targeting glutamate and GABA function in PFC therefore indicate an important new therapeutically relevant and complementary strategy to alleviate symptoms of schizophrenia that are not ameliorated by the current medication.
At a circuitry level, schizophrenia involves connections between several structures of the CNS. Principally, these include the PFC system with its major glutaminergic projections to the caudate nucleus and the nucleus accumbens; the striatal GABAergic innervation of the ventral pallidum (VP) in the basal forebrain and inhibitory projections from VP to the MD thalamus, which in turn provides return glutaminergic innervation of the PFC. The nucleus accumbens and PFC are furthermore both innervated by glutaminergic projections from the medial temporal structures such as the hippocampal/entorhinal formation and the basolateral amygdaloid nucleus. The complex interactions between these pathways may explain how in particular disturbances of mesolimbic dopaminergic systems, which originate in the ventral tegmental area (VTA) and project to the PFC, MD thalamus, nucleus accumbens, hippocampus and amygdala, could be involved in the pathophysiology of schizophrenia (10, 19, 20, 80).
Regulation of the VTA dopamine system is thus regulated by direct glutaminergic afferents from the PFC as well as from the brainstem peduncular pontine (PPN) systems. However, in order for these excitatory inputs to induce a change in the irregular basal firing pattern of the VTA dopaminergic neurons, simultaneous disinhibition of the basal GABAergic control by the VP is required. This effect may be produced by a simultaneous and coordinated activation of the PFC and the hippocampal/entorhinal systems. Hence, disinhibition of the VTA dopamine release to the mesolimbic circuitry is produced by the synchronized activation of glutaminergic efferents into the two serially connected GABAergic structures, nucleus accumbens and VP. This pattern of activation involves the following steps: hippocampal formation (Glu efferent↑) →nucleus accumbens (GABA efferent ↑) →VP (GABA efferent↓) →VTA (Dopamine release ↑). Burst firing from the disinhibited VTA neurons is induced by the subsequent direct and synchronous glutaminergic innervation from the PFC and the brainstem pedunculopontine nucleus (PPN) (80).
Complex symptoms of schizophrenia such as cognitive function thus may involve disturbances of the mesolimbic dopamine system that are secondary to dysfunction within the PFC and hippocampus (80–82). In fact, substantial evidence for functional disturbances between these structures as well as the direct connection between PFC and thalamus in schizophrenia has been found by functional fMRI as well as by neurocognitive tests of patients (19, 20, 83, 84).
Functional disturbances of local PFC circuits initially may be established through a developmental deficit in the formation of the bidirectional MD–PFC projections during the neonatal phase. Early formation and migration of PFC neurons from the subplate to different cortical areas requires precisely timed and coordinated innervation by the previously formed thalamic MD projection neurons. A failure in establishing these early developmental connections might cause a secondary developmental deficit in the subsequent establishment of the layer IIIc cortico-cortico circuitry (85, 86). In addition, early dopaminergic innervations of the premature neonatal PFC neurons may furthermore provide a neurotropic influence during PFC neuronal development (86). The presence of a neonatal dopaminergic deficit is still controversial, but nevertheless hypofunction of the dopamine system in PFC is well-known to play a role in the pathophysiology of schizophrenia, in particular for cognitive function (87).
Another plausible developmentally based error that might contribute to the abnormal establishment of PFC circuitry in schizophrenia involves formation and maturation of the cortical GABAergic inhibitory system, which in the mature PFC is essential for modulation and synchronization of pyramidal neuronal firing (88). Thalamocortical innervation of these cortical GABA interneurons by MD thalamic projection neurons, similar to cortical pyramidal neurons, is essential during development. Interestingly, one of the most consistent neuropathological findings in postmortem brain from schizophrenic patients is a reduction in parvalbumin expressing GABAergic chandelier interneurons (77, 88–90). In fact Lewis et al. found the principal deficit of these interneurons associated with the middle layer of inhibitory neurons, which is the major input layer for MD thalamic innervation of the PFC (91, 92). The etiology of decreased parvalbumin expressing neurons in PFC is not known and it is intriguing to speculate whether this deficit is secondary to a lack of MD thalamic innervation of the PFC during development. In support of this hypothesis, several preclinical studies have found that subchronic treatment of rats with glutamate antagonists such as phencyclidine (PCP), ketamine or MK801 causes severe reductions in the parvalbumin-type chandelier GABAergic interneurons in the PFC (93, 94). As mentioned above, it is of substantial interest that antagonism of the glutamate system in humans introduces psychotic symptoms that closely mimic those seen in schizophrenia (52, 53, 95). Hence, subchronic treatment of rats with PCP or MK801 may currently provide one of the most attractive preclinical models for PFC dysfunction in schizophrenia (77, 96).
The role of a deficient MD thalamic function and/or a disturbed timing of MD thalamic innervation of the PFC in the neurodevelopmental process of schizophrenia is still speculative. Return corticothalamic innervation from layer VI PFC pyramidal cells is established at a later developmental stage and no current evidence points to involvement of altered function of these connections in schizophrenia.
The Lundbeck Foundation (LVK), the European Commission IRG-FP7 programme (LVK), the National Alliance for Research on Schizophrenia and Depression (NARSAD; LVK).