Pathway-specific dopamine abnormalities in schizophrenia

Abstract

In light of the clinical evidence implicating dopamine in schizophrenia, and the prominent hypotheses put forth regarding alterations in dopaminergic transmission in this disease, molecular imaging has been used to examine multiple aspects of the dopaminergic system. Here we review the imaging methods used and compare the findings across the different molecular targets. Findings have converged to suggest early dysregulation in the striatum, especially in the rostral caudate, manifesting as excess synthesis and release. Recent data showed deficit extending to most cortical regions, and even to other extrastriatal subcortical regions not previously considered to be “hypodopaminergic” in schizophrenia. These findings yield a new topography for the dopaminergic dysregulation in schizophrenia. In this review we discuss the dopaminergic innervation within the individual projection fields to provide a topographical map of this dual dysregulation and explore potential cellular and circuit based mechanisms for brain region-dependent alterations in dopaminergic parameters. This refined knowledge is essential to better guide translational studies and efforts in early drug development.

I. Historical perspective on dopamine research in schizophrenia

Dopamine (DA) has been a focus of schizophrenia research for decades, yielding two prior conceptual formulations for dopamine’s involvement in schizophrenia. In 1966 Rossum and colleagues proposed a state of excess dopaminergic stimulation in patients with schizophrenia (SZ) (1), substantiated later by the discovery of the D₂ receptor binding profiles of antipsychotics and the psychotogenic effects of DA agonists (2–4). This was later reformulated as an imbalance between excess subcortical DA and a deficit in cortical DA, in light of evidence suggesting a prefrontal cortical deficit in schizophrenia and the prominent role of DA in mediating prefrontal-dependent cognitive processes (5, 6). The availability of imaging tools to measure aspects of dopaminergic transmission in vivo allowed testing of these formulations in patients. Improved scanner technology enabled better anatomical resolution. Earlier detection and awareness of the prodromal phase of the disease (7, 8) resulted in testing earlier stages of the illness (9–11), while stress paradigms (12, 13) allowed probing responsiveness of the system to a relevant risk factor for the disease (14, 15), together yielding a replicable set of findings across labs documenting excess presynaptic dopaminergic transmission in the striatum, confirming the original formulation. Furthermore, data from our lab provided new evidence for a cortical DA deficit (16), supporting the second formulation, but also expanding it to multiple extrastriatal regions not previously considered to be “hypodopaminergic” in schizophrenia. A new topographical mapping of DA dysregulation in schizophrenia is the topic of this review. We will describe the imaging methods used to examine dopaminergic indices, and findings in SZ. We will then review dopaminergic innervation and its imaging-relevant targets within individual projection fields to provide a topographical map of the findings and suggest potential mechanisms for brain region-dependent DA dysregulation in schizophrenia. Finally, we discuss future directions.

II. Methodology for imaging the dopamine system

Positron Emission Tomography (PET) and Single Photon Emission Computed Tomography (SPECT) have been used to measure dopamine-related parameters via administration of radioligands that bind to receptors, transporters or other target molecules, or alternatively, trace a metabolic pathway. For radioligands that reversibly bind to receptors, the most commonly derived parameter is the binding potential (BP) (17, 18), which is proportional to B_AVAIL/K_D, where B_AVAIL is the concentration of the target molecule available for binding to the radiotracer and K_D is the equilibrium dissociation constant of tracer for the target. There are several versions of BP, depending on which concentration of tracer is used as a reference value. For the frequently used BP_ND, (Figure 1B–D), the reference is the nondisplaceable compartment, comprised of the sum of the free plus nonspecifically-bound radiotracer in brain; BP_ND = f_ND*B_AVAIL/K_D, where f_ND is the free fraction of the nondisplaceable radiotracer concentration. BP_ND is an indicator of target molecule availability, based on the assumption that K_D and f_ND are not different across groups. B_AVAIL, as opposed to the total target concentration B_MAX, accounts for the masking of some of the targets by the binding of endogenous ligands. BP_ND is also the ratio of specifically-bound to nondisplaceable radiotracer concentrations at equilibrium, thus representing the associated signal-to-noise ratio (see reference 17 for complete definitions). A BP_ND lower than 0.5, i.e. signal lower than half of background, is considered too low to provide meaningful information.

Figure 1. Dopaminergic Imaging Targets.

Schematic of imaging methods used to measure aspects of the dopamine (DA) system in vivo. Graphic depicts progression of DA from synthesis (A), storage (B), to release (E,F), then either reuptake by dopamine transporter (DAT, C) or binding to receptor (D). Imaging targets and related paradigms are described in accompanying text.

Tracers with moderate affinity for dopamine-D₂-like receptors (D₂ and D₃, referred to from here on as D₂), such as [¹¹C]raclopride and [¹²³I]IBZM, provide reliable BP_ND in the striatum (Figure 1D). [¹⁸F]fallypride has an order of magnitude higher affinity (19, 20) and provides reliable quantification in striatum, thalamus, midbrain, hippocampus, amygdala and temporal cortex. Higher affinity tracers such as [¹¹C]FLB457 (21) or [¹²³I]epidipride (22) can be used to quantify D₂ density in cortex, although their equilibration is prohibitively slow for quantification in striatum. Pharmacologically, all of these tracers are antagonists. [¹¹C]-(+)-PHNO is a D₃ preferring agonist (23–25). Tracers for D₁-like receptors (D₁ and D_5, hereafter D₁) include [¹¹C]NNC112 and [¹¹C]SCH23390 (26, 27). Both have been used to quantify D₁ in cortex and striatum, although [¹¹C]SCH23390’s BP_ND is below 0.5 in cortex.

Tracers for D₂ receptors are sensitive to changes in the concentration of DA through competitive interaction. Pharmacological challenges that increase synaptic DA, such as concomitant release and reuptake blockade by amphetamine, or reuptake blockade by methylphenidate, decrease BP_ND; whereas depletion paradigms that reduce baseline synaptic DA, such as blockade of tyrosine hydroxylase activity with alpha-methylparatyrosine (α-MPT), increase BP_ND. These effects can be quantified as ΔBP_ND, the percent change of BP_ND across conditions (Figure 1E,F). D₂ ligand displacement by challenge-induced DA release occurs at the subset of D₂ receptors that are in close proximity to DA release sites (28–32). This has led to the postulation that net change in tracer binding at these perisynaptic receptors may comprise the PET “DA release” signal (33), which refers to our PET measurement of intrasynaptic DA levels, either evoked (due to amphetamine administration) or basal (measured with the depletion paradigm).

[¹⁸F]DOPA is a substrate for amino acid decarboxylase (AADC), which catalyzes L-Dihydroxyphenylalanine (DOPA) into DA (34). In terminals containing AADC, [¹⁸F]DOPA is converted to 6-fluorodopamine ([¹⁸F]6-FDA), a substrate for vesicular monoamine transporter 2 (VMAT2), which loads DA into vesicles (Figure 1A). [¹⁸F]6-FDA cycles through exocytosis, reuptake through the DA transporter (DAT), and reloading into vesicles. This is generally treated as an irreversible process. The outcome measure is K_in, the steady-state uptake rate constant of the tracer, characterizing [¹⁸F]6-FDA formation when the concentration of [¹⁸F]DOPA in arterial plasma and in brain are at a hypothetical steady-state. K_in indicates the capacity for DA synthesis. A related outcome measure is K_i^cer which is the steady state uptake rate (K_in) relative to cerebellum concentration of [¹⁸F]DOPA, rather than the arterial plasma concentration, but studies utilizing K_i^cer require the implicit assumption that concentration of [¹⁸F]DOPA in the cerebellum does not differ between groups.

[¹⁸F]DOPA quantification is complicated by formation in the periphery of the radiolabeled metabolite 3-O-methyl-FDOPA ([¹⁸F]OMFD) due to catechol-O-methyl transferase (COMT) activity (35); pretreatment with entacapone can reduce this effect. In addition, the irreversibility of [¹⁸F]DOPA uptake is an idealization, as [¹⁸F]6-FDA is a substrate for both monoamine oxidase (MAO) and COMT, and metabolites diffuse out of the brain, affecting measurement of K_in. Some models account for this washout with an estimated parameter called k_loss (35, 36).

[¹⁸F]DOPA K_in can be measured in striatum but extrastriatal K_in is lower and more difficult to measure. In substantia nigra (SN), K_in is approximately half as large as in striatum and, in cortex, too low to be interpretable (37).

Transporters have also been imaged using [¹¹C]DTBZ for VMAT2 (38) (Figure 1B), [¹¹C]PE2I for DAT (Figure 1C) in striatal and extrastriatal regions using PET, and [¹²³I]βCIT (39) for striatal DAT using SPECT.

III. Imaging the dopamine system in schizophrenia

We review here findings from studies that used molecular neuroimaging to investigate the DA system in vivo in schizophrenia - first in striatum, then in extrastriatal regions, with a focus on cortex and midbrain (see Supplemental Table S1).

III.A. Striatum

III.A.1. Presynaptic

Higher striatal [¹⁸F]DOPA was first reported in psychosis related to epilepsy and schizophrenia (40). Seven studies replicated this finding in schizophrenia (9, 41–47), while two did not (48, 49), and subsequent meta-analyses confirmed the finding (50, 51). Using D₂ radiotracers and a psychostimulant challenge, four studies showed higher release in the striatum of antipsychotic-free (Rx-free) patients compared to healthy controls (HC) (52–55). Excess DA release correlated with transient stimulant-induced worsening of psychotic symptoms in patients, and was observed at disease onset and during exacerbations, but not during periods of remission (56). Furthermore, baseline synaptic DA assessed with a depletion paradigm (57) were enhanced in striatum in schizophrenia (SZ), and were correlated with amphetamine-induced release in a cohort of antipsychotic-naïve (Rx-naïve) patients (58). Using a higher-resolution scanner and more sophisticated region-of-interest (ROI) analysis methods to identify the striatal substructures, we later demonstrated that excess striatal DA was most prominent in the rostral caudate (59). In the associative striatum (AST), which contains the rostral caudate, rostral putamen and post-commissural caudate, the effect size was 0.70, compared to 0.14 in the limbic striatum (LST or VST, ventral striatum), and 0.34 in the sensorimotor striatum (SMST, posterior putamen). This excess does not seem to be related to excess dopaminergic innervation as VMAT2 (60, 61) and DAT (62–72) were normal.

III.A.2. Postsynaptic

Several studies have examined striatal D₂ availability. A meta-analysis of 23 studies showed small elevation and greater variability in SZ. When analysis was limited to Rx-naïve patients, SZ and HC did not differ (51), suggesting that D₂ increases in striatum in SZ may be due to prior antipsychotic treatment. Striatal D₁ availability is also normal in SZ (27, 73–75).

Further support for antipsychotic-induced upregulation of striatal D₂ derives from α-MPT studies (57–59), which provide a direct measure of “true” D₂ density by unmasking the fraction of receptors bound by endogenous DA. A new analysis of these previously published studies shows that unmasked BP_ND is higher (by 10–20%) in previously-medicated, but not Rx-naïve patients, compared to HC (Table 1) in striatum (57–59) and in rostral caudate (59). In contrast, in the same cohorts, α-MPT-induced ΔBP_NDs showed that striatal DA levels are 65–120% higher in both Rx-naïve and previously-medicated patients compared to HC. This suggests that striatal dopaminergic hyperactivity is present regardless of prior antipsychotic treatment and thus a more reliable index of DA dysregulation than receptor upregulation.

Table 1.

Effect of previous antipsychotic exposure on unmasked BP_ND Binding potentials from α-MPT depletion studies

						Rx-free v Rx-naïve	Rx-free v HC	Rx-naïve v HC
	HC	SZ	p	Rx-free	Rx-naïve	p	p	p
[123I]IBZM SPECT: Striatuma	n=18	n=18		n=10	n=8
BPNDBsl	0.722 ± 0.091	0.751 ± 0.103	0.38	0.779 ± 0.122	0.716 ± 0.066	0.21	0.17	0.87
BPNDDpl*	0.787 ± 0.096	0.889 ± 0.124	0.009	0.930 ± 0.147	0.837 ± 0.062	0.12	0.004**	0.19
[11C]Raclopride PET: Striatumb	n=18	n=18		n=12	n=6
BPNDBsl	2.53 ± 0.25	2.56 ± 0.52	0.83	2.71 ± 0.50	2.25 ± 0.44	0.08	0.19	0.07
BPNDDpl*	2.81 ± 0.23	2.94 ± 0.54	0.35	3.12 ± 0.52	2.59 ± 0.42	0.048	0.04**	0.12
[11C]Raclopride PET: Rostral caudateb	n=18	n=18		n=12	n=6
BPNDBsl	2.40 ± 0.23	2.41 ± 0.45	0.89	2.54 ± 0.43	2.16 ± 0.42	0.10	0.25	0.10
BPNDDpl*	2.61 ± 0.27	2.77 ± 0.49	0.25	2.91 ± 0.47	2.47 ± 0.43	0.07	0.03**	0.36

^aAbi-Dargham et al., PNAS 2000 (57)

^bKegeles et al., AGP 2010 (59)

^*Significant one-way ANOVA comparing BP_NDDpl for Rx-free, Rx-naïve, and HC (p < 0.05).

^**Significant post-hoc t test for BP_NDDpl, Rx-free compared to HC (but not significant for Rx-naive compared to HC).

Abbreviations: D₂ binding potential in the baseline state, partially masked by baseline levels of endogenous dopamine (BP_NDBsl); Unmasked D₂ binding potential in the dopamine-depleted state (BP_NDDpl); Healthy control participants (HC); Participants with schizophrenia (SZ); Antipsychotic-free, previously-medicated patients (Rx-free); Antipsychotic-naïve patients (Rx-naïve)

III.A.3. Clinical correlates of the striatal findings

The striatal dopaminergic hyperactivity in schizophrenia is associated with the psychotic symptoms of the illness. It was shown to extend to physiological conditions under psychosocial stress and to be most enhanced in AST and SMST in Rx-naïve patients and in the prodrome (14). Elevated striatal [¹⁸F]DOPA uptake also precedes the onset (76), correlates with greater severity of prodromal symptoms and neuropsychological impairment, predicts conversion and, in both the prodrome and SZ, relates negatively to prefrontal cortical activation during cognitive tasks in (43, 77, but also see 78). It is also predominant in the AST (79, 80).

Furthermore, excess striatal DA predicts treatment response of psychosis to antipsychotics (58) and is higher in antipsychotic-responsive patients (81).

SZ (82) and individuals at clinical high risk for schizophrenia (CHR) (11) with comorbid substance use display a blunted striatal dopamine release. However, despite this presynaptic blunting, D₂ receptors remain supersensitive to stimulation, leading to psychosis. This suggests two distinct alterations in psychosis: excess presynaptic release in striatum as well as a functional supersensitivity of striatal D₂.

III.B. Cortex

III.B.1. Presynaptic

Using [¹¹C]FLB457 we showed significant blunting of DA release throughout the cortex in SZ. DA release in the dorsolateral prefrontal cortex (DLPFC) was significantly positively associated with working memory-related BOLD activation, suggesting a relationship between blunted release and deficits of frontal cortical function (16). [¹⁸F]DOPA (45–48) reports in the cortex are uninterpretable (37).

III.B.2. Postsynaptic

D₂ availability in SZ is normal in prefrontal (16, 83–85), occipital (16, 84), parietal (16, 84), entorhinal (86), anterior cingulate (16, 83, 87) (except for (84)), and insular (16, 86) cortices. A meta-analysis (excluding (16)) found no differences in temporal cortex (88). One study reported lower binding in uncus (87) while another did not (16).

Studies of prefrontal cortical D₁ availability in SZ yielded inconsistent results of increases (74, 75) and decreases (27), compared to HC (Supplemental Table S1). To reconcile these findings, both D₁ tracers were examined in the same subjects (89, 90) and showed similar alteration using either tracer, suggesting cohort-related effects rather than tracer differences. Prior exposure to antipsychotics may explain some of these discrepancies, as higher D₁ levels were observed only in Rx-naïve patients, and duration of Rx-free interval positively correlated with higher binding in previously-treated patients (75).

III.C. Extrastriatal subcortical regions and midbrain

III.C.1. Presynaptic

[¹⁸F]DOPA uptake in SZ is normal in thalamus (47) and entorhinal cortex (47), but enhanced in amygdala (46) and midbrain (46, 91). In midbrain, one study reported higher [¹⁸F]DOPA utilization (K) and turnover (k_loss), while K_in was numerically lower (46). Another reported higher K_i^cer in the midbrain, which correlated with symptom severity in SZ (91) and predicted conversion in CHR (92). We measured significant blunting of amphetamine-induced DA release measured by [¹¹C]FLB457 displacement in extrastriatal subcortical regions including midbrain (16). Thus, for the amygdala and midbrain, PET indices of presynaptic DA synthesis/turnover and amphetamine-evoked DA release seem discrepant. If this discrepancy is indeed true, it may suggest elevated enzymatic activity in the presence of lower cytoplasmic and vesicular pools of DA in midbrain in SZ (see discussion below).

Using [¹¹C]PE2I, one study reported higher DAT in the thalamus but not in SN (72). However, the small sample size and low BP_ND suggest caution in interpreting this study. VMAT2 was normal in extrastriatal regions (61) except for ventral midbrain, where an increase was reported (93); however as BP_ND was below 0.5, this finding should also be considered with caution.

III.C.2. Postsynaptic

Of the nine studies in thalamus (16, 84–87, 94–97), only one ((94), which overlaps with (98)) found lower D₂ in SZ, and meta-analysis (88) was negative. Likewise, no differences were found in globus pallidus (97), amygdala (16, 86, 87), entorhinal cortex (16, 86) or hippocampus (84, 86, 87). In SN, normal (16, 86, 97, 99), higher (87) and lower (96) D₂ were reported; and meta-analysis (88) was negative.

No differences in D₁ availability have been observed in extrastriatal subcortical ROIs (see Supplemental Table S1).

III.D. Summary of imaging findings

In summary, three main dopaminergic alterations have emerged in schizophrenia:

1. DA synthesis and release capacity are increased in the striatum (51).

2. Although needing replication, DA release capacity in prefrontal cortical and other extrastriatal regions is decreased (16).

3. There is subregional heterogeneity in the DA dysregulation within the striatum. The rostral caudate and the AST in general, show lower DA release capacity than the SMST in HC (100), but not in SZ due to a prominent increase in the AST (9, 14, 101). Supportive evidence for the prominent role of DA dysregulation in AST also derives from studies in prodrome (9, 14).

4. Postsynaptic receptors and transporters do not show a reliably detectable altered expression either in the striatum or in extrastriatal regions of the brain in SZ.

IV. Topography and synaptic characteristics of dopaminergic projections

To understand the abnormal PET DA signal in SZ, we will consider the regional anatomical factors that may affect it. Here we review the complex topography and chemical neuroanatomy of DA systems underlying PET indices of basal and evoked DA release.

DA projections comprise the retrorubral field (RRF)(A8), substantia nigra (SN)(A9) and ventral tegmental area (VTA)(A10) (102–104) (Figure 2). These have different intrinsic properties and afferents regulating spike activity; synthesis, release or reuptake of DA; and postsynaptic effects (102–105) (Figure 3). “Dorsal tier” DA neurons, a band along the dorsal SN pars compacta (SNc) and contiguous regions of VTA and RRF, project to cerebral cortex, ventromedial striatum, pallidum, “extended amygdala”, and thalamus. The “ventral tier” neurons, including the densocellular region of the SNc and DA cell columns within the pars reticulata (SNr), project to the striatum. The SMST receives a dense projection, with high density of DA release sites (105), accounting for the higher PET DA release signal, and highest levels of DAT, exerting tighter spatiotemporal regulation of DA diffusion compared to other subregions. The VST, innervated by VTA and medial SNc DA neurons, has lower DA release potential and lower levels of DAT and D₂ autoreceptors (106, 107). The AST receives a mosaic of dorsal and ventral tier neurons.

Figure 2. Topography of dopaminergic innervation and receptor distribution.

Schematic representation of distributions of dopamine D₁ and D₂ receptors (left hemispheres) and patterns of dopaminergic innervations (right hemispheres) in select primate (left panel) and rodent (right panel) brain regions. Left hemispheres: Brown and black squares depict D₁ and D₂ receptors, respectively. Throughout the primate and rodent brain, D₁ receptors (D₁) are present at a higher density than D₂ receptors (D₂). The striatum, and in particular the caudate-putamen, has the highest densities of dopamine (DA) receptors. DA receptors are also present in medium-to-low densities in the cortex, pallidum and midbrain. Receptor densities are relatively low in thalamus, amygdala and hippocampus. See text for details. Right hemispheres: Topographical distribution of DA cell bodies (filled circles) and their terminals (lines). In the primate panel, red circles represent DA cell bodies in the VTA with terminals in the cortex, striatum (in particular the ventral part), pallidum, thalamus and amygdala. The VTA dopaminergic cellular organization is better characterized in the rodent where discrete VTA cell groups project to the cortex (red), nucleus accumbens (dark green) and amygdala (orange). In the primate, SN dorsal tier cell group (light green) projects to the cortex and ventral striatum, as well as the pallidum, thalamus and amygdala. The rodent brain in contrast has a low density of these dorsal tier neurons. The SN ventral tier groups (SN compacta densocellular part (dark blue) and fingers (light blue)) project heavily and topographically to caudate-putamen with medium/low innervations of cortex, ventral striatum, thalamus and amygdala. See text for further details.

Figure 3. Topography of dopamine release findings in schizophrenia compared to controls.

Schematic representations of DA release characteristics in the cortex (top), striatum (middle) and midbrain (bottom) in healthy controls (HC) and patients with schizophrenia (SZ) based on imaging findings in patients. DA neuron cell bodies, terminals and transmitters are depicted in red. Color gradients depict DA terminal densities. Cortex: The cortex receives sparse dopaminergic innervation that is poor in dopamine D₂ receptors (D₂) and transporter expression. This sculpts D₂ displacement measurement, which is low in the cortex. In schizophrenia there is evidence for reduced cortical DA release. See text for details. Striatum: DA and cortical neuron terminals (green) are shown innervating medium spiny neuron spines (orange). Also shown are local cholinergic (blue) and GABAergic (brown) interneuron populations forming the striatal microcircuitry. There is considerable heterogeneity in DA release across striatal regions, e.g. dopaminergic innervation of ventral striatum (VST, also referred to as LST) is relatively sparse and is derived from dorsal tier cell groups that are poor in D₂ and DAT. In contrast the sensorimotor striatum (SMST) receives dense dopaminergic inputs mostly from the ventral tier DA neurons that are rich in D₂ and DAT. A greater number of synapse sites in the ventral striatum and high levels of D₂ and DAT in SMST may account for high D₂ displacement in these regions. Compared to VST and SMST, stimulant induced D₂ displacement is low in the associative striatum (AST). In schizophrenia, DA release is increased across substriatal divisions due to a prominent increase in the AST. Midbrain: Shown are DA cell bodies, local GABAergic interneurons (brown) and D₁ medium spiny neuron terminals (yellow). While there is heterogeneity in the level of expression of D₂ receptors and DAT (e.g. dorsal tier and especially medial VTA neurons have low D₂ and DAT levels), imaging studies showing subregional analysis of D₂ displacement are lacking. However, in SZ there is a reduced stimulant-induced D₂ displacement.

The SMST, AST and VST also differ in glutamatergic, cholinergic, and other local (e.g. opioidergic) modulation of DA release, due to neurochemically distinct compartments within each of these subregions, called patch (or striosome) and matrix. These refer to a “mosaic” pattern of grouping of neurons that have differential neurochemical characteristics and specific connections to cortex and other brain regions (Figure 4). In the SMST, the ventral tier DA neurons innervate both the mu opioid receptor and substance P rich ‘patch’ and the enkephalin rich ‘matrix’ compartments; while in the AST, ventral tier innervation is selective to patches. This has implications for DA modulation of cortical afferents, as patches receive projections from limbic (e.g. amygdala) and paralimbic cortical areas (e.g. orbitofrontal cortex); whereas the matrix receives input from other prefrontal cortical regions such as DLPFC.

Figure 4. Striatal patch-matrix connectome.

Schematic representation of striatal patch-matrix connectome. Afferents: The cortex topographically projects to the striatum. Within the cortex deeper cortical layers innervate striatal patches (dark brown) whereas the surrounding matrix (light brown) is innervated by superficial cortical layers (light brown). Within the midbrain, the dorsal tier (orange and yellow) innervates the matrix, as do the non-dopaminergic cells (dark green) from the same region. Patch innervation from the midbrain is mostly derived from the ventral tier cell groups (dark blue). Non-dopaminergic (presumably GABAergic) projection neurons within the SNr innervate the striatal matrix complex. Efferents: Striatal patch neurons (maroon) mostly project to ventral tier DA cells. These include both D₁ receptor expressing medium spiny neurons and other striatal projection neurons. Striatal projection neurons within the matrix project to both DA and non-dopaminergic populations within the dorsal tier and GABAergic populations in the SNr. See text for further details.

IV.A. Striatal organization

The topography of DA projections interfaces with regional and subcellular localization of DA receptors (Figure 3), which have 5–20-fold higher density in striatum compared to other regions (28, 86, 102–104, 108–111). Post-synaptic D₁ and D₂ are segregated onto different subpopulations of projection neurons and expressed on striatal interneurons. Cholinergic interneurons express D₂-like receptors that mediate fast synaptic events and locally regulate DA release (105, 112). Taken together, ultrastructural and electrophysiological experiments indicate that D₂-like receptors are positioned preferentially to mediate DA effects on striatopallidal projection neurons and cholinergic interneurons (28, 113). As with DA inputs, DA receptors and modulators of DA release show distinct patch-matrix distributions in AST and SMST: patches are richer in D₁ receptors, lack parvalbumin-expressing interneurons, and show a paucity of cholinergic innervation as indexed by acetylcholinesterase fiber staining (104). Adding to this complexity, neuromodulators differentially affect DA release and projection neuron activity across the patch-matrix organization: e.g. substance P facilitates DA release within patch center, decreases it at patch-matrix border and has no effect in matrix; while enkephalin selectively boosts patch projection output via delta opioid-mediated disinhibitory mechanisms (114, 115).

IV.B. Extrastriatal organization

Extrastriatal regions including cortex are innervated predominantly by the dorsal tier DA system (Figure 2), which is poor in transporter and D₂ autoreceptors (102–104). In contrast to low innervation densities in rodents, primates have a dense and extensive cortical DA innervation (116). However, sparse cortical DAT expression suggests a low incidence of DA release sites (107). Moreover, low D₂ density and heterogeneous synaptology and DA receptor topography (28) are all consistent with the smaller PET DA release signal in extrastriatal regions. In cortex, D₂ are evenly distributed across projection neurons and fast-spiking interneurons (28, 117). Thus, tracer displacement at D₂ on fast-spiking interneurons may contribute more to the PET DA release signal in the cortex than in the striatum.

To summarize, spatiotemporal regulation of DA release and localization of D₂-like receptors varies considerably across regions and adds complexity to the interpretation of regional and disease-related variation in the PET DA release signal (Figure 3).

V. Discussion

The literature reviewed here shows that: 1) stimulant-induced presynaptic DA release is decreased in most brain regions in schizophrenia (16), with exception of the striatum where it is enhanced, especially in the rostral caudate (59); 2) in this region, the excess is not observed under conditions of substance abuse despite psychosis (11, 82); 3) alterations in expression levels of receptors and transporters are less reliably observed (51, 88), which does not exclude an alteration in function of these receptors in schizophrenia since even under conditions of low DA tone, as in comorbidity with addiction, blocking striatal D₂ remains therapeutic and stimulating striatal D₂ is psychotogenic (82); 4) antipsychotic exposure results in upregulation of striatal D₂ (51) and may induce down-regulation, or normalization, of cortical D₁ (75); and 5) the global nature of the presynaptic DA dysregulation is likely to massively alter information processing in multiple domains and result in the global symptomatology that we observe in SZ, although the specific mechanisms that mediate the formation of abnormal learning (118) and symptoms are currently unknown.

It remains to be seen whether extrastriatal DA deficits occur in the same subjects who display striatal DA upregulation, thus yielding a ‘dual dysregulation’ of DA alteration, as proposed in the reformulation of the DA hypothesis of schizophrenia (5, 6). From this perspective, studies using stimulant challenge and those using [¹⁸F]DOPA have provided convergent results in striatum, but not in extrastriatal regions. However, when investigators included metabolism of [¹⁸F]6-FDA (k_loss) (46) in their model, they observed higher k_loss in the amygdala and midbrain in SZ, indicating a possible state of lower intracellular DA tone; excessive washout of DA is consistent with the lower evoked release that we observed. This provides one potential mechanism to reconcile these findings and to support our observation of extrastriatal DA release deficits. The finding of increased K_i^cer (91) on the other hand, is potentially susceptible to group differences in cerebellar concentration of [¹⁸F]DOPA. Additional support to our finding of cortical and midbrain deficit derives from the postmortem observations of reduced tyrosine hydroxylase (TH) (119, 120), however high TH (91) and high (121) or normal (122) TH mRNA have also been reported. More research is needed to understand these discrepancies.

Since one of the main findings in SZ is dysregulation of presynaptic DA function, we have reviewed the multifactorial regulation of DA release and its detection with PET. The AST is of particular interest. In HC, the PET DA release signal in the AST is lower than in the SMST (9, 14, 59); whereas in SZ, it is increased to levels similar to the SMST. We speculate that in the healthy brain, subregional differences may reflect differences in DA innervation, regulation of DA release, and/or distributions of perisynaptic D₂-like receptors. The difference in the patch/matrix ratio between the AST and SMST could also reflect and/or contribute to lower spontaneous DA release in AST (105, 123, 124). For example, given the low cholinergic innervation of patches, ACh-augmentation of DA release may be lower in this compartment, and thus relatively lower in the patch-enriched AST. We could postulate that, in schizophrenia, a disruption of brain development leads to abnormal or incomplete development of the AST, consistent with structural imaging studies showing lower caudate volume in early-stage, unmedicated patients with SZ relative to HC (104, 125). A developmental disruption leading to altered differentiation of AST from SMST and/or lower patch/matrix compartmentalization in the AST might lead to abnormalities in the patterning of DA and other inputs to the AST, DA interactions with acetylcholine and other striatal neurotransmitters (104, 105), and DA modulation of cortical inputs to the AST (126). Testing these ideas requires updating the existing postmortem literature (125) with studies applying modern labeling and imaging methods to render the 3-dimensional chemoarchitecture of the striatal complex in healthy humans and patients with schizophrenia. Additional models that consider regional and subregional variation in DA synaptology and modulation of DA release across striatal subcompartments are also needed.

The mechanisms underlying cortical deficits in the PET DA release signal in schizophrenia remain to be determined, but given the distribution of D₂ receptors, may involve changes in DA signaling at a variety of neuronal populations including cortical interneurons. The generalized and profound deficits in extrastriatal DA release raise an important therapeutic challenge for the field, as currently approved antipsychotics do not remedy this deficit or the resultant low stimulation of extrastriatal dopaminergic receptors. This generalized deficit is also consistent with the multi-domain functional manifestations of the illness, ranging from deficits in social cognition to executive function and motivation.

While higher DA may be linked to better cognition in a non-schizophrenic brain (127–130), in schizophrenia, higher DA may have a dysfunctional impact either because of its modulatory role on an already abnormal circuitry or because of intrinsic aberrant dynamics of DA cell firing patterns.

While this literature does not provide mechanistic understanding of the dysfunction, it has provided a refined topographical knowledge that can be used in translational studies and in drug development. There is unfortunately limited knowledge at this point regarding the specific alterations in the multiple cellular components that could mediate the altered PET DA signal in schizophrenia. We have reviewed above and discussed a few “suspect” cellular mechanisms. These need to be formally tested in postmortem tissue, in animal models that show DA dysregulation, and in cellular systems such as induced Pluripotent Stem Cells (iPSCs) from patients who show abnormal DA PET signal, to isolate specific components that may be involved. Once those are defined they can be used in drug development as specific targets for novel therapies. Our review highlights the urgent need for this cellular work to be carried out in tandem with imaging in patients.