Thalamocortical Development: A Neurodevelopmental Framework for Schizophrenia

Abstract

Adolescence is a period of increased vulnerability for the development of psychiatric disorders, including schizophrenia. The prefrontal cortex (PFC) undergoes substantial maturation during this period, and PFC dysfunction is central to cognitive impairments in schizophrenia. As a result, impaired adolescent maturation of the PFC has been proposed as a mechanism in the etiology of the disorder and its cognitive symptoms. In adulthood, PFC function is tightly linked to its reciprocal connections with the thalamus, and acutely inhibiting thalamic inputs to the PFC produces impairments in PFC function and cognitive deficits. Here, we propose that thalamic activity is equally important during adolescence because it is required for proper PFC circuit development. Because thalamic abnormalities have been observed early in the progression of schizophrenia, we further postulate that adolescent thalamic dysfunction can have long-lasting consequences for PFC function and cognition in patients with schizophrenia.

Schizophrenia is a chronic psychiatric disorder, characterized by three major categories of symptoms: positive (i.e., delusions or hallucinations), negative (i.e., avolition or anhedonia), and cognitive (i.e., deficits in cognitive flexibility or working memory). The typical onset is late adolescence (1), with prodromal symptoms seen sometimes years before diagnosis (2). Given these early indicators, schizophrenia is thought to have a developmental contribution, with risk factors identified in both the prenatal and postnatal periods (3–7). For instance, gestational exposure to infection has been implicated in the development of schizophrenia (8), and during adolescence, exposure to various perturbations, including cannabis and stress, increases risk of later illness development (9,10). These widespread risk factors underline the complex pathogenesis of schizophrenia, which challenges prevention and early diagnosis. However, early treatment of schizophrenia is associated with improved outcomes (11,12). As a result, it is very important to improve our understanding of the developmental etiology of schizophrenia, which will allow us to identify better indicators for those at risk and ultimately intervene to prevent illness onset.

While schizophrenia is diagnosed based on positive and negative symptoms, cognitive symptoms are equally prominent and severely compromise functional outcomes for patients (13–15). Cognitive deficits are present throughout the entire course of the disorder, including before diagnosis (16), at the first psychotic episode (17), and in later stages of the disorder (18,19), demonstrating that these symptoms cannot be attributed only to the chronicity of the disorder or long-term consequences of antipsychotic medication. Indeed, cognitive symptoms are some of the earliest indicators of dysfunction, suggesting a developmental origin (20). Moreover, cognitive impairments are not specific to schizophrenia and occur in other psychiatric disorders with developmental origins, such as anxiety, panic, and mood disorders (14), underscoring the general importance of understanding developmental origins of cognitive dysfunction. Despite their widespread prevalence, there are few available therapeutics to treat cognitive symptoms (14,21). To develop more effective treatments, we need a better understanding of the neurobiological underpinnings of cognitive dysfunction and its developmental etiology.

PREFRONTAL CORTEX DYSFUNCTION CONTRIBUTES TO COGNITIVE DEFICITS

Cognitive deficits of schizophrenia include deficits in attention, working memory, and cognitive flexibility. Decades of research have identified the prefrontal cortex (PFC), portions of which are homologous between primates and rodents, as a brain structure supporting these cognitive processes (22). Several reports have demonstrated a striking resemblance between the cognitive deficits observed in patients with frontal lesions and deficits observed in schizophrenia (23). Moreover, functional imaging studies found altered activation of the PFC during cognitive testing in patients with schizophrenia, supporting the hypothesis that deficits in working memory and other cognitive functions arise from disruptions to PFC activity (24,25). Postmortem studies found that while the total number of neurons in the PFC is not changed in patients (26), there is evidence for changes in neuronal arborization (16), dendritic spines (27–29), synaptic markers (30,31), and cell soma size (32,33). However, how these changes in PFC function and morphology arise remains to be determined.

PFC DEVELOPMENT IN SCHIZOPHRENIA

While a developmental contribution to schizophrenia has been discussed for many years (3,34,35), the specific alterations that occur during brain development are still unknown. Dynamic mapping of typical postnatal human cortical development found that gray matter volume increases at earlier ages followed by a loss starting at puberty (36). The gray matter loss, which is thought to reflect cortical maturation, occurs later in higher-order association cortices, including prefrontal and temporal cortices, than in lower-order somatosensory and visual cortices, suggesting a later maturation of these cortical regions (36). Cortical gray matter loss is steeper in individuals at clinical high risk for schizophrenia who subsequently develop the disorder compared with individuals who do not (37,38). It suggests that alterations in prefrontal maturation may be part of the disorder etiology.

In addition to the volumetric changes, synaptic spine density increases during childhood until ages 5 to 10 and then decreases during puberty and adolescence into adulthood (39,40). Because cortical spine density is decreased in patients, Irwin Feinberg hypothesized that schizophrenia is “caused by a fault in programmed synaptic elimination during adolescence” (34,41). The mechanisms that may contribute to aberrant pruning are still largely unknown. Recent work identified a role for microglia and the complement system in normal synaptic pruning during development, and genetic and cell biological evidence suggests that enhanced C4 complement expression may lead to exaggerated pruning in patients with schizophrenia (42–44).

Pruning processes are thought to be activity dependent, where synapses that are not used are eliminated (45). This would imply that abnormalities in activity in cortical circuits during adolescence may lead to aberrant pruning and cortical network function. One of the main sources of excitatory input to the PFC comes from the thalamus, including the mediodorsal nucleus of the thalamus (MD). Here, we propose the hypothesis that thalamic input to the PFC is necessary during adolescence for proper prefrontal circuit maturation. Moreover, as decreased thalamoprefrontal functional connectivity and fractional anisotropy (FA) has been measured in individuals at risk for schizophrenia prior to disease onset [see below, (46)], we further hypothesize that decreased thalamic input activity during adolescence impairs cortical maturation, promoting transition to schizophrenia (Figure 1).

Figure 1.

Hypothesized MD involvement in PFC development. Activity of MD-PFC projections during adolescence regulates the development of the PFC and therefore its function in adulthood. We propose that reduced MD activity during adolescence leads to impairments in PFC circuitry and function, including changes to intrinsic PFC circuitry, MD-PFC connectivity, and cognitive functioning. Created with BioRender.com. MD, mediodorsal nucleus of the thalamus; PFC, prefrontal cortex.

THALAMIC ALTERATIONS IN PATIENTS WITH SCHIZOPHRENIA

There is increasing evidence for alterations in thalamic function in patients with schizophrenia (47–49). First, thalamic volume is decreased in postmortem and structural brain imaging studies in patients with schizophrenia (50–52). High-resolution imaging further revealed that reduced volumes of the pulvinar, MD, and ventrolateral nuclei in patients are correlated with decreased cognitive performance (53). Resting-state imaging further identified a striking pattern of hyperconnectivity between the thalamus and sensorimotor cortices and hypoconnectivity between the thalamus and cerebellar and prefrontal regions (Figure 2) (54–59) that may have a structural origin (60,61). Because decreased correlated activity between the thalamus and PFC has been measured during cognitive testing, it may underlie deficits in cognitive functioning (57,62,63).

Figure 2.

Hyper- and hypoconnectivity between the thalamus and cortical regions in patients with schizophrenia. (A) Imaging studies have demonstrated that patients with schizophrenia (right) have a hyperconnectivity between the thalamus and sensory regions and a hypoconnectivity between the thalamus and prefrontal regions (54–59). Created with BioRender.com. (B) In Anticevic et al. (46), young subjects at clinical high risk for schizophrenia were found to have differential connectivity between the thalamus and other brain regions. Yellow areas indicate regions with a significant increase in connectivity with the thalamus (including sensory areas), while light blue areas indicate regions with a significant reduction in connectivity (including prefrontal areas). (C) In Anticevic et al. (55), patients diagnosed with schizophrenia were found to have a similar and more severe pattern of hyper- and hypoconnectivity between the thalamus and cortical regions. Red areas indicate regions with a significant increase in connectivity compared with control subjects, while dark blue areas indicate regions with a significant reduction. con, control; L, left; R, right; scz, schizophrenia.

Thalamic abnormalities occur in early stages of the disorder as well as in patients at clinical high risk (Figure 2). For instance, decreases in thalamic volume were observed both in younger patients with psychosis and those at clinical high risk (53,64,65). Similarly, resting-state functional connectivity measurements indicate reduced thalamoprefrontal connectivity in adolescents with early-onset schizophrenia, young adolescents at clinical high risk, and adults in early stages of the disorder (46,61,66–68). The reduced thalamoprefrontal connectivity among high-risk individuals is most apparent in those who are later diagnosed with the disorder (46). Together, these studies demonstrate that thalamic dysconnectivity is apparent in late adolescence coincident with the onset of early symptomatology and increases in severity for those who ultimately develop the disorder. Meanwhile, disease chronicity or long-term antipsychotic medications do not appear to worsen this phenotype (69). These findings raise the intriguing possibility that thalamo-PFC dysconnectivity in adolescence is part of the developmental etiology of schizophrenia (46,67). Because thalamoprefrontal hypoconnectivity and thalamosensorimotor hyperconnectivity co-occur in the same patients (55) and because they are observed early in patients at clinical high risk (46), it is possible that they are generated by shared underlying mechanisms. Alternatively, one alteration is caused first and induces the other by disrupting thalamic or cortical development.

Consistent with a developmental origin of thalamic alterations, there is evidence that thalamic anatomy and circuit function is genetically regulated. Volumes of thalamic nuclei and thalamoprefrontal functional connectivity have been found to be associated with familial risk for schizophrenia (70,71). Similarly, coordinated expression of schizophrenia risk genes in the MD and PFC is associated with thalamocortical functional connectivity during attentional control (72). Finally, thalamofronto/parietal resting-state connectivity is decreased in the 22q11 deletion carriers, a rare copy number variation that confers a >20-fold increased risk for schizophrenia (73).

THE MEDIODORSAL THALAMUS SUPPORTS PFC FUNCTION

The thalamus is a heterogeneous set of nuclei deep in the brain that traditionally has been viewed as a relay station for sensory information from the periphery to the cortex (74). In contrast, higher-order nuclei such as the pulvinar and MD receive their main inputs not from the periphery but from the cortex itself (75). By projecting back to the same or different cortical areas, higher-order thalamic nuclei support cortical representations within, and communication between, cortical areas (76–78). Here, we focus on the MD because it exhibits strong reciprocal connectivity with the PFC and is implicated in the pathophysiology of schizophrenia.

There is an extensive literature covering localized lesion or inhibition studies in rodents and primates that addressed the role of the MD in cognition. Generally, these studies described deficits in cognitive processes that are known to be supported by the PFC [for review, (77–79)]. This conclusion was confirmed by more recent studies in mice in which acutely inhibiting the MD or, more specifically, MD–medial PFC (mPFC) or mPFC-MD projections during behavior impaired working memory, adaptive decision making, and cognitive flexibility (80–85).

In vivo recordings of neural activity during behavior led to the hypothesis that thalamic inputs to the mPFC act as a nonspecific facilitator of mPFC neuronal activity encoding task rules or other task-specific information (82). For example, in a working memory task, MD input to the mPFC is necessary for sustaining cortical activity during a short delay when the mouse has to remember trial-specific information (80,82). A large subset of mPFC neurons show elevated delay activity that is tiled over the course of the delay to maintain this trial-specific information (80). Inhibition of thalamic input activity disrupts this elevated delay activity along with behavioral performance. Moreover, in one of these studies, delay activity in the mPFC, but not in the MD, encoded task rule information, supporting the idea that the MD does not necessarily encode information but rather supports cortical encoding (82). Note, however, that in nonhuman primates, the MD has been found to encode trial-specific information about explicit stimuli or spatial representations during the delay period. These differences may be due to species differences or differences in the task context (86,87).

At the mechanistic level, the MD may facilitate cortical encoding by enhancing functional connectivity between neurons of cortical ensembles (82). In this context, nonspecific MD activity has been shown to enhance mPFC ensemble activation, which reinforced the cue or trial information encoded within these ensembles (80–82). Recent functional imaging data in humans using dynamic causal modeling support a similar model whereby the MD increases synchrony of cortical activity, in this case between different cortical regions, to sustain complex mental representations (88).

POSTNATAL DEVELOPMENT OF MD AND PFC

While this literature has given insights into the adult functioning of the MD-mPFC circuitry, our knowledge about the development of MD-mPFC circuitry is still limited. In humans, thalamocortical projections are set up during embryonic development. In neonates, resting-state functional magnetic resonance imaging has revealed functional connectivity in a thalamocortical salience network that includes the anterior insula, anterior cingulate, and PFC, which correlates with working memory performance (89). Thalamoprefrontal connectivity was originally described to increase between childhood and adulthood (90). However, more recent studies found that thalamoprefrontal functional connectivity is relatively stable during postnatal development (68,91). At the structural level, thalamocortical white matter, as measured by FA, increases during development, and higher FA in thalamoprefrontal and parietal tracts is related to better cognitive function (92). Thalamocortical FA is lower in youths with psychotic symptoms than in subjects without psychotic symptoms (92). Together, these studies indicate that while thalamocortical projections are already present at birth, they undergo important structural maturation processes during postnatal development. This developmental time course is similar to what has been seen in the development of MD-PFC circuitry in rodents (93).

In rodents, at birth, projections from the MD to the mPFC have arrived in both the developing deeper cortical layers and the cortical plate, which will subsequently develop into the superficial layers (93,94). This early presence of MD afferent fibers across mPFC layers has been proposed to instruct laminar development of the mPFC (95). In addition, cortical signals regulate the localization of thalamocortical inputs. For example, retinoid signaling in the PFC during fetal development instructs the spread of thalamoprefrontal projections. Genetic deletion of the retinoid acid receptors RARb and RXRg leads to a selective reduction in thalamocortical inputs, while other subcortical inputs to the PFC are largely spared (96). Artificial lateral expansion of cortical retinoid acid signaling is associated with a lateral expansion of medial thalamocortical innervation (96). Because it also leads to prefrontal enlargement and increased laminar expression of RORb, which is characteristic of the primate granular cortex, it has been hypothesized that an expansion in retinoid signaling contributes to the relative cortical expansion observed in primates (96).

Thalamoprefrontal projections continue to increase after birth, reaching a peak in mice in midchildhood (~ postnatal day [P] 10). After a period of steep thalamic cell loss at P13, the projections gradually rebound in density, plateauing toward the end of adolescence (~P60) (93,97). The timing of this increase parallels the increase in thalamocortical structural connectivity in humans observed with FA (92). Notably, myelination in the PFC also increases in the mouse during this time (98). Meanwhile, in rodents, the volume of the PFC increases in the postnatal period, peaking in early puberty (~P24). At that point, it decreases, reflecting a period of dendritic pruning, which has its fastest rate at ~P30 (99,100). These findings coincide with studies of gray matter density in the human PFC, which show a prepubertal increase followed by a decrease during adolescence, corresponding to this period of dendritic pruning (5,36). The changes in PFC structure follow the changes in MD projection density, suggesting that refinement of thalamic projections informs PFC maturation and pruning (93–95,97). This process is relevant to the Feinberg hypothesis that in schizophrenia, aberrant activity-dependent pruning during adolescence leads to persistent changes in prefrontal circuit function (34). It also supports the idea that adolescent pruning in the cortex may be regulated by MD input activity.

SENSITIVE PERIODS

Given the extensive maturation of the mPFC during adolescence, it has been hypothesized that adolescence may represent a sensitive period for mPFC development (101,102). Sensitive periods denote developmental time windows when brain circuits are particularly plastic and susceptible to changes in neuronal activity. As such, changes in input activity elicited by alterations in experience during a sensitive period can lead to long-lasting effects on the anatomy and function of these circuits (103,104). A classic example is in the visual system, where transient developmental monocular deprivation can permanently impair acuity in the deprived eye (105). This impairment in function persists even after the deprivation in visual input is reversed, because the thalamocortical inputs representing the closed eye are permanently disrupted in an activity-dependent manner.

Recent evidence suggests that sensitive time windows for activity-dependent maturation may also exist in the mPFC (101,102,106–108). Primarily, these studies have focused on changes to intrinsic components of mPFC circuitry, such as interneuron or layer II/III pyramidal neuron activity (102,106,107). These studies cover slightly different but overlapping periods, spanning from P7 to P50. Given the developmental time points outlined above, this large epoch encompasses periods of intense growth and subsequent refinement in the mPFC. This opens the possibility that there may be multiple sensitive periods that might influence different aspects of circuit development. Here, we focused on adolescence because it is hypothesized to be a particularly important period in which changes in brain development may precipitate the development of schizophrenia.

ADOLESCENT THALAMIC INHIBITION LEADS TO LONG-LASTING IMPAIRMENTS IN PFC FUNCTION

While thalamic input activity has been shown to be important for sensory cortex development, including the visual cortex (103,105,109,110), our understanding of the role of thalamic input activity for mPFC development remains incomplete. Prior work lesioning the MD in rats at early postnatal time points produced conflicting results. For example, lesioning the MD at P0 resulted in only mild cortical thinning. However, when MD lesions occurred at P4, more dramatic changes on cortical morphology were observed, concurrent with deficits in cognitive tasks, including the acquisition of a nonmatch to place T-maze task (111–113). These findings support the hypothesis that postnatal MD function is important for mPFC development, although it is unclear why the earlier lesion has a weaker effect. To determine when during development or adulthood the lesion affected mPFC function, it is necessary to temporally restrict the disruption of MD activity to defined time windows.

To determine whether MD inhibition during adolescence impairs mPFC maturation, a recent study used G protein–coupled inhibitory DREADDs (designer receptors exclusively activated by designer drugs) that enable inhibition of thalamic neurons in a temporally controlled manner (114). In humans, adolescence refers to a transitional window during which adult abilities are acquired and refined. Adolescence is generally believed to begin just prior to the onset of puberty and persist until the completion of sexual maturity (115–118). This period also encompasses the greatest changes to mPFC volume, as well as myelination and synaptic pruning of associative cortical circuits. It is complicated to relate the ages of mice and humans, because the relationship is not linear and differs depending on which developmental metrics are used for comparison (e.g., hormonal changes vs. brain development) (118). In humans, the precise age range associated with adolescence can vary. Similarly, in mice, adolescence is sometimes defined by a narrower (P20–P40) or broader (P20–P60) time window, and even P60 to P75 has been considered as a part of late adolescence in some studies (119–121). In the DREADD inhibition study, the time window of inhibition spanned a period from postweaning (P20) to P50.

Thalamic inhibition during this time window led to long-lasting decreases in cortical excitation, as measured by a decrease in the frequency of spontaneous excitatory currents in adult mPFC slices and decreased thalamocortical projections, while projections to the mPFC from other regions such as the basolateral amygdala (BLA), remained unchanged. These physiological and anatomical changes were associated with mPFC-dependent cognitive deficits, including the acquisition of a delayed nonmatch to place working memory task and extra-dimensional attentional set-shifting. In vivo physiology further revealed decreased cross-correlation between mPFC single units and impaired population level outcome encoding in the set-shifting task after adolescent thalamic inhibition (114). A comparable thalamic inhibition in adult mice (P90–P120) did not lead to deficits in cortical excitation or cognitive performance. Thus, adolescence is a sensitive time window during which thalamic activity regulates thalamoprefrontal connectivity and long-term cortical functioning. Acute thalamic excitation in adult mice rescued the deficits in mPFC cell cross-correlations, outcome encoding, and behavioral performance despite the reduced thalamocortical projection density induced by the developmental manipulation. This supports the idea that the thalamus regulates cognition by facilitating cortical representations and suggests that boosting thalamic function could be exploited as a therapeutic strategy to enhance cognition, even in the context of a developmentally altered brain.

IMPLICATIONS FOR PREFRONTAL DEVELOPMENT DURING ADOLESCENCE

While there is strong evidence that adolescence is a sensitive time window during which thalamic activity is required for setting up long-term prefrontal cortical function, the underlying mechanisms for this are unclear. One possibility is that the permanent reduction of thalamoprefrontal projections following inhibition of thalamic neurons occurs by regulating either the arborization or retraction of thalamocortical axons. In this scenario, decreased thalamic input during adult behavior negatively affects intracortical synaptic connectivity at the functional level, thereby impairing correlated activity and encoding.

Alternatively, thalamic inhibition during adolescence leads to long-lasting changes in the intrinsic architecture of the cortical network. In adult mice, MD neurons possess functional connections to neurons of superficial and deep cortical layers, but connectivity is strongest toward layer 3 intratelencephalic neurons (122). If a similar preference exists during adolescence, inhibition of the MD will mostly affect activity of intratelencephalic neurons, which could then alter the maturation of intracortical networks, including local collaterals. Intracortical pruning is a mechanism of normal cortical maturation because peripubertal pruning of prefrontal layer 3 collaterals has been described in nonhuman primates (123).

Finally, MD neurons also project to vasoactive intestinal peptide, parvalbumin (PV), and somatostatin interneurons, which regulate cortical activity and have been implicated in adolescent PFC development (122,124–126). In particular, PV cells in the mPFC are still maturing during adolescence, and PV cell activity has been shown to be important for sensitive period plasticity in sensory cortices (127–131). In line with this, selective inhibition of PV interneurons in the mPFC during adolescence leads to long-lasting impairments in attentional set-shifting and task-evoked gamma oscillations in the mPFC (102). However, spontaneous inhibitory events and gamma oscillations are not altered in the adult mouse after adolescent thalamic inhibition, arguing against an involvement of cortical PV interneurons (114). Consistent with this, lesioning the MD in 50- to 60-day-old Sprague Dawley rats had no long-lasting effect on the expression of GAD67 in the mPFC (132).

Given the influential hypothesis of impaired NMDA receptor function in schizophrenia, it is intriguing to ask whether decreased thalamic input during adolescence could regulate cortical NMDA receptor function (133). NMDA receptors show an activity-dependent developmental switch from GluN2B to GluN2A across cortical regions, which differs in the PFC where GluN2 is upregulated after P40 (134). GluN2Bs regulate cortical network function, particularly delay period activity observed during working memory tasks in the primate PFC (135). Thus, thalamic inputs during adolescence may regulate cortical NMDA receptor composition to produce long-lasting effects on cognition, a hypothesis that can be addressed in the future.

The extent to which adolescent thalamic inhibition affects other inputs to the mPFC remains unresolved. Inputs from several other subcortical regions, including the BLA, converge onto the same layer II/III cells as those from the MD (122,124). If this input competes for territory in the mPFC, it is possible that adolescent thalamic inhibition could result in an overrepresentation of inputs from the BLA. However, retrograde tracing of inputs to the mPFC following adolescent thalamic inhibition revealed only a selective reduction in thalamocortical inputs; those from the BLA were unchanged (114). This contrasts with the results of a very early ventral hippocampus lesion where BLA-mPFC projection density was increased (136). These differences may stem from the brain region manipulated, the age during manipulation, or differences in the manipulation method itself.

Although the P20 to P50 adolescent window is a sensitive period for activity-dependent remodeling of thalamocortical inputs, further work is necessary to refine the time frame for these effects. As discussed, the age at which adolescence ends and adulthood begins is not firmly demarcated in rodents, and there is evidence to suggest that activity-dependent plasticity of hippocampal-prefrontal connectivity extends until P70 (121). Likewise, an earlier (P7), but not later (P21), lesion of the amygdala led to persistent effects on stress response and startle reactivity in adulthood (137), suggesting that the window in which thalamic activity might influence prefrontal development could also extend to earlier time points.

IMPLICATIONS FOR UNDERSTANDING AND TREATING SCHIZOPHRENIA

Extrapolating from rodent studies to humans has its challenges owing to inherent species differences. However, general principles by which the thalamus organizes cortical activity are likely conserved between different mammalian species, thus giving rodent studies translational relevance (138). The work described here is consistent with a model where impairments in thalamic functioning during adolescence induce persistent impairments in prefrontal functioning and cognition (Figure 3). Extrapolating to schizophrenia, the early changes in thalamic and thalamocortical functioning observed in high-risk and early-onset populations may similarly have long-lasting consequences on PFC function. This may increase the likelihood of converting to diagnosis and severe cognitive deficits.

Figure 3.

Hypothesis of how altered thalamic function during adolescence may impair adult PFC function and cognition in patients with schizophrenia. Genetic and external factors alter MD activity during adolescence (left). As a consequence of decreased MD activity during adolescence, MD-PFC projections are reduced (middle), and PFC function, including cell-cell cross-correlation and behavioral task outcome encoding, and cognition are impaired (right) in adulthood. Interventions, such as cognitive remediation, may enhance MD activity during adolescence, thereby preserving the MD-PFC adult circuit and function. Created with BioRender.com. MD, mediodorsal nucleus of the thalamus; PFC, prefrontal cortex.

While genetic studies point to a potential genetic contribution to the early thalamocortical alterations, evidence for external risk factors during adolescence to suppress thalamic function is scarce. Cannabis use could be one potential candidate, because cannabis use during early adolescence is a risk factor for schizophrenia (10), and in two studies, adolescent cannabis use has been associated with decreased thalamic volume (139,140). Oxidative stress and altered redox regulation have also been discussed as risk factors affecting the PFC (141). In rodent studies, these processes decrease the number of PV neurons in the reticular thalamus (TRN) at P20 (142). In the same publication, both PV neuron number and PV-associated perineuronal nets were decreased in the TRN in postmortem tissue from patients with schizophrenia (142). In rodents, early TRN lesion leads to long-lasting structural changes in the MD and mPFC (143). Together, these results open the possibility that oxidative stress may alter the development of thalamocortical circuitry via the TRN. However, more evidence is needed to determine whether external risk factors impair thalamic function during adolescence.

As discussed above, impairing thalamic activity during adolescence has long-lasting negative consequences on prefrontal function. Conversely, engaging thalamoprefrontal activity during adolescence may have long-lasting beneficial effects on prefrontal function and cognition. Cognitive remediation, which should engage thalamocortical circuitry, is most effective early in the disease process (144). Indeed, there is preliminary evidence that cognitive training may be valuable in prodromal high-risk populations, although it is still unclear whether the beneficial effects on cognition are long lasting (145,146).

While plasticity for thalamocortical activity-dependent remodeling is reduced in the adult, it is important to understand whether it can be enhanced and the sensitive period reopened in adulthood. Toward this end, studies that identify the molecular mechanisms underlying plasticity occurring during adolescence (e.g., growth factor release, complement activation) will give important insight into what needs to be restored in adulthood. As we discussed above, acute thalamic excitation restores cognitive performance when performed during behavior (114). Pharmacologically targeting the thalamus could be another way of reversing cognitive deficits. One potential target is the Gpr12 receptor, which has been shown to regulate thalamocortical functioning and cognition in mice (83). Likewise, noninvasive thalamic stimulation may be a way forward in humans to improve cognition, although this is still in its early stages of development (147). However, an important challenge will be to find a way to temporally extend the therapeutic potential of brain stimulation so that the beneficial effects are long lasting. Toward that end, using compounds that have been shown to reopen plasticity in the visual system (e.g., histone deacetylase inhibitors or selective serotonin reuptake inhibitors) may be a promising strategy to reopen plasticity in the mPFC (148,149). In the visual system, when adult plasticity is reopened concurrent with enhanced visual engagement of the previously deprived eye, thalamocortical arborization from that eye can be restored to normal levels, permanently correcting visual functioning (148). Future work should address whether enhancing plasticity concurrent with engagement of the thalamocortical network can be applied to persistently restore thalamocortical connectivity and cognitive deficits.

CONCLUSIONS

A better understanding of the timeline of MD-mPFC development during adolescence and the impact of changes to this process for cognitive functioning will greatly enhance our ability to identify early markers of cognitive impairments and psychiatric disorders. This work would also create opportunities for preventive interventions, which would greatly improve long-term patient outcomes. Moreover, treatments for any disorder with cognitive dysfunction could be developed to target, or even reverse, the processes that cause these symptoms. Together, earlier diagnosis and improved treatment options offer huge promise for patients with psychiatric disorders.