In this study, we examined the balance of excitation and inhibition in human OFC and its disruption in ASD. Our study focused on key excitatory and inhibitory components of OFC circuits, including the myeloarchitecture, cytoarchitecture and neurochemistry of OFC gray matter, in adults with or without autism. We described the quantitative distribution, trajectory, and morphology of myelinated axon fibers and the density of excitatory neurons across cortical layers, as a proxy of excitatory networks in OFC. We then examined the laminar distribution and morphology of three largely non-overlapping, neurochemically- and functionally-distinct types of inhibitory neurons in the local circuit. These three types of inhibitory neurons are characterized by expression of specific calcium-binding proteins PV, CB and CR, and constitute the major source of cortical inhibition . Using cellular and single-axon high-resolution quantitative approaches we established, for the first time in our knowledge, the typical laminar distribution, density, and relationship of excitatory and inhibitory circuit components in the human OFC. Finally, we identified atypical changes that likely underlie layer-specific feedforward, feedback, cortical and subcortical OFC network pathology and imbalance of excitation and inhibition in OFC circuits in ASD.
Our study of the neurotypical human OFC gray matter at the single axon and cell resolution across layers complements and extends previous research on the trajectory and morphometrics of axon bundles in OFC white matter with diffusion tensor imaging (DTI) and high-resolution neuroanatomical approaches [27, 42]. We found that myelinated axon density in neurotypical OFC decreased from layer 1 to layer 2 and then increased gradually towards the deeper layers that are closer to the white matter, following a consistent pattern across human and non-human primates [37, 40, 43]. Axon thickness remained stable across layers. Compared with previous reported results in the white matter below OFC , myelinated axons in the gray matter were thinner, indicating overall thinning of axons after entering into the gray matter, but no further tapering within the gray matter. Myelinated axon profile orientation in our sections was used as a proxy of axon trajectory in 3D. Analysis showed that axon orientation variability appeared to be consistent across layers in OFC gray matter, but more heterogeneous than previously reported in the superficial white matter below prefrontal cortices . Taken together with previous studies, our findings show that axons become thinner and their trajectory becomes more heterogeneous, likely due to increased branching, as they transition from deep to superficial white matter, and finally enter the gray matter [27, 36].
Excitatory and inhibitory neuron distribution and density in the neurotypical OFC were in line with previous reports in primate OFC and other cytoarchitectonically similar prefrontal cortices [37, 44–46]. Narrow cortical layers 2 and 4 that had the highest density of neurons showed relatively low density of myelinated axons, whereas thicker layers with lower neuronal density had more room for myelinated axons. Inhibitory neurons accounted for about 20% of all neurons in OFC, in line with previous studies in the frontal cortex of humans [47–51]. In primates, inhibitory neurons can be classified by label with the calcium binding proteins PV, CB or CR, which comprise largely non-overlapping neurochemical groups of cortical inhibitory neurons [41, 52, 53]. PV labels basket and chandelier inhibitory neurons [54, 55], which are most prevalent in the middle cortical layers, where they form perisomatic synapses and strongly inhibit pyramidal neurons. CB labels several cortical morphologic types of inhibitory neurons, found most densely in cortical layers 2 and 3, and innervate distal dendrites of pyramidal neurons, modulating their activity . CR inhibitory neurons are also found mostly in the upper layers (1-3a), where they innervate mostly other GABAergic neurons [57–59]. This regularity in the laminar distribution of PV, CB, and CR neurons has been shown in frontal, temporal and sensory association areas, which have been studied in primates [60–64], and was evident in the OFC of neurotypical adults examined here.
In the OFC of adults with ASD the balance of excitatory and inhibitory network components was disrupted. We found that, in general, myelinated axons were less dense, and thinner in the ASD group, suggesting weaker excitatory transmission in OFC networks in ASD. Lower density of myelinated axons, together with observed axonal thinning and heterogeneity of axon trajectory, as indicators of branching [35, 36], may also indicate increased relative presence of unmyelinated axons and upregulated local circuit communication.
Neuron analysis in each layer showed that the overall neuron population was lower in OFC of adults with ASD, while the density of inhibitory neurons did not change significantly compared to the controls. Therefore, our findings indicate that the reduction in neuron density mainly involved excitatory neurons. Combined with the density reduction and thinning of myelinated axons in OFC, which are predominantly excitatory, our data support a universal weakening of input and output, or downregulation of activation levels in OFC of adults with ASD (Fig. 6). Our findings provide anatomical evidence, supporting previous functional imaging studies, which showed reduced OFC activation and information flow between OFC and neighboring areas in individuals with ASD [20, 65].
The laminar-specific disruption of excitatory cortical neurons and myelinated axons, and the resulting imbalance in excitation/inhibition can provide further insights about network pathology in ASD, including the integrity of short- and long-range feedforward and feedback OFC connections (Fig. 6). ASD is characterized by disruption in sensory processing, attention, social interaction and value-based decision making. OFC connections with sensory association cortices, limbic medial prefrontal cortices, including the ACC, and the amygdala, position OFC as a key node for the integration and evaluation of sensory information and emotional states [6, 8, 10–13, 66].
In particular, the posterior OFC receives input from unimodal and multimodal association cortices from all sensory modalities (reviewed in ). The sensory input helps OFC calibrate information in a timely manner. In primates, the connections originate mostly from the upper layers of sensory association cortices and innervate mainly the middle OFC layers (3–5), in a feedforward manner [67, 68]. In turn, the OFC can also modulate sensory processes, especially auditory processing [69, 70], through feedback connections. Impairment in sensory processing and evaluation, often associated with ASD [71, 72], involves disengagement of sensory cortices and neural circuits that process social reward, in which OFC has a central role . In addition, previous studies have linked under-connectivity between auditory processing circuits and OFC with insensitivity to human voice in individuals with ASD . Our findings, suggesting weakening of excitatory inputs and outputs in OFC and a relative strengthening of local inhibition, provide the anatomical substrate for the disruption of these long-range connections in ASD. However, it is worth noting that some individuals with ASD accompanied by sensory hypersensitivity show heightened OFC activity , therefore more studies are needed to clarify the role of OFC networks in sensory processing.
The OFC is also robustly interconnected, through short/medium-range pathways, with medial prefrontal cortices, including neighboring ACC and ventromedial subgenual area 25, in primates [13, 74–76]. In particular, ACC pathways innervate mostly excitatory pyramidal neurons across OFC layers, which in turn project back to all ACC layers . Interactions between ACC and OFC have distinct implications for psychiatric diseases, such as obsessive compulsive disorder and phobias [77–79]. It would be expected from our results that reduction of both excitatory neurons and myelinated axons in OFC would disrupt connectivity and communication with ACC and other medial prefrontal cortices. In line with this, previous studies in the white matter showed disrupted axon morphology under both ACC and OFC in brains from individuals with ASD, indicating compromised communication [27, 37].
Finally, the OFC has among the most robust reciprocal connections with the amygdala [10–12, 40, 80–83], forming a key circuit for affective processing. Anatomically, amygdala projections to OFC follow a feedback pattern, originating mostly from the basolateral nucleus that mainly innervates excitatory pyramidal, as well as CB and CR inhibitory neurons in the upper layers of OFC, modulating local activity . OFC neurons in middle and deep layers (mostly 3 and 5) project back to the amygdala, targeting robustly and preferentially inhibitory neurons in the intercalated masses (IM), in pathways that regulate autonomic function and play a key role in learning affective associations [10, 84]. Reduction of top-down OFC control of the inhibitory output from IM may lead to excessive autonomous, or impulsive behaviors. In line with this, lesions in the OFC or amygdala frequently result in emotional processing deficits seen typically in ASD, including carelessness and lack of affect [85, 86]. Previous studies also showed that in ASD there is an initial excess of neurons in the amygdala during childhood, followed by a reduction in the density of neurons in adulthood across nuclei [87, 88]. The decreased density of OFC neurons and axons in adults with ASD observed in our study, provides evidence for parallel development of pathology in amygdala and OFC in ASD. This is also supported by functional studies that showed reduced amygdala-OFC activity in adults with ASD , and structural studies that showed reduced tract volume and lower mean fractional anisotropy values in the uncinate fasciculus, the axon bundle connecting OFC and the temporal lobe, including the amygdala [28, 29, 89].