Untangling the cortico-thalamo-cortical loop: cellular pieces of a knotty circuit puzzle

Abstract

Functions of the neocortex depend on its bidirectional communication with the thalamus, via cortico-thalamo-cortical (CTC) loops. Recent work dissecting the synaptic connectivity in these loops is generating a clearer picture of their cellular organization. Here, we review findings across sensory, motor and cognitive areas, focusing on patterns of cell type-specific synaptic connections between the major types of cortical and thalamic neurons. We outline simple and complex CTC loops, and note features of these loops that appear to be general versus specialized. CTC loops are tightly interlinked with local cortical and corticocortical (CC) circuits, forming extended chains of loops that are probably critical for communication across hierarchically organized cerebral networks. Such CTC-CC loop chains appear to constitute a modular unit of organization, serving as scaffolding for area-specific structural and functional modifications. Inhibitory neurons and circuits are embedded throughout CTC loops, shaping the flow of excitation. We consider recent findings in the context of established CTC and CC circuit models, and highlight current efforts to pinpoint cell type-specific mechanisms in CTC loops involved in consciousness and perception. As pieces of the connectivity puzzle fall increasingly into place, this knowledge can guide further efforts to understand structure-function relationships in CTC loops.

Fig. 1 ∣. Major classes of excitatory projection neurons in cortico-thalamo-cortical loops.

a ∣Reconstructed dendrites are shown for the major excitatory cell classes involved in forming cortico-thalamo-cortical loops — thalamocortical (TC) neurons in the thalamus (purple), and three broad classes of excitatory neurons in the neocortex: intratelencephalic (IT; red), pyramidal tract (PT; blue) and corticothalamic (CT; green) neurons. The IT neurons are a large and diverse class, comprising multiple subtypes across and within cortical layers, from layer 2 to layer 6. The PT neurons, sharing layer 5B with IT neurons, have large apical dendritic tufts in layer 1. The CT neurons, sharing layer 6 with IT neurons, have relatively small dendritic arbors. b ∣ The axonal projection patterns of the same four classes are depicted schematically, following the same colour scheme. The IT neurons project bilaterally to other cortical areas and striatum. The PT neurons form descending projections with branches to multiple subcortical and subcerebral destinations. The CT neurons are specialists with axons projecting almost entirely to the thalamus. The cortical projection to the thalamus is composed of the axons of both PT and CT neurons. c ∣ Axonal arbors of IT, PT and CT neurons in motor cortex, and of TC neurons projecting to motor cortex, are shown separately (top row; twelve examples of each) or together (bottom row, in side, top and front views). The axons are from the database of the MouseLight project, which developed methods for reconstructing axonal arbors in their entirety across the brain ^, The images were generated with the MouseLight interface (www.janelia.org/project-team/mouselight) by selecting multiple axon IDs for display (IT axons AA0662, AA0627, AA0622, AA0592, AA0582, AA0003, AA0656, AA0588, AA0442, AA0184, AA0102, AA0002; PT axons AA1050, AA0927, AA0926, AA0923, AA0617, AA0587, AA0584, AA0261, AA0135, AA0134, AA0132, AA0131; CT axons AA0863, AA0833, AA0770, AA0652, AA0640, AA0623, AA0548, AA0545, AA0185, AA0043, AA0041, AA0038, and, TC axons AA0692, AA0591, AA0586, AA0581, AA0391, AA0342, AA0809, AA0719, AA0675, AA0451, AA0176, AA0137). Reconstructions in part a adapted with permission from REFS. ^-,,). Reconstructions in part c adapted with permission from REF..

Fig. 2 ∣. Cell-type-specific cortico-thalamo-cortical connectivity matrices across areas.

a ∣ The major classes of excitatory projection neurons in the cortex and thalamus are drawn schematically (left), along with their ‘blank’ connectivity matrix (right), highlighting the boxes representing connections directly involved in cortico-thalamo-cortical (CTC) loops, from the thalamus to cortex (upper right) and the cortex to thalamus (lower left). Crosses (×) mark where connections cannot form owing to a lack of projections from the presynaptic source. b ∣ For the primary somatosensory cortex (S1), a wiring diagram (left) and connectivity matrix (right) are shown, incorporating findings for both vibrissal-related and forelimb-related S1, as discussed in detail in the main text^-,-. The matrix was assembled by filling in each element based on a qualitative interpretation and weighting of experimental results, often from multiple studies, that have quantitatively measured the strengths of connections between specific presynaptic→postsynaptic combinations. For example, the matrix element in the upper right corner represents the strength of excitatory input from presynaptic ventrobasal (VB) axons to postsynaptic IT neurons (in this case primarily those in layer 4). The colour intensities represent the connection strengths (darker = stronger). The colours differ by column to represent the different types of presynaptic neurons. This colour scheme also emphasizes that most measurements have been experimentally determined by comparing the strength of input from a single presynaptic source to two or more postsynaptic targets, rather than vice versa. Accordingly, comparisons between matrix elements tend to be more robust along the columns than along the rows. For example, the relative strengths of VB inputs to different types of cortical neurons have been directly tested, whereas the relative strengths of VB and PO inputs to any one type of cortical neurons have not. c,d ∣ The same approach was taken to generate a CTC connectivity matrix for primary motor cortex (M1), incorporating findings for both vibrissal and forelimb related M1 ^,,,-,-, and for the prefrontal cortex (PFC) ^- e ∣ An ‘average’ CTC connectivity matrix for cortical areas was assembled based on qualitative comparison across areas, including those considered above, ALM, A1, V1 and others, as discussed in the main text ^,,,-,, Connections that tend to be found consistently are marked with ‘+’ and stronger colours, those that tend to be more variable with ‘±’ and intermediate colours, and those that tend to be weak or absent with ‘−’ and the lightest colours.

Fig. 3 ∣. Patterns and types of circuit connections in cortico-thalamo-cortical loops.

a ∣ Simple cortico-thalamo-cortical (CTC) loops involve direct connections between TC and PT or CT neurons. Such loops are relatively uncommon. b ∣ Complex loops involve IT neurons closing the cortical end of the loop. Complex loops, particularly of the matrix-type, appear to be ubiquitous. The smaller ‘cyclo-plots’ (circular circuit plots with coloured arrows) are included as concise representations of the basic cellular connectivity in each type of loop.

Fig. 4 ∣. Cortico-thalamo-cortical loops are interconnected with local cortical circuits and corticocortical circuits, via hub-like intratelencephalic neurons.

a ∣ The connectivity matrix combines the ‘average’ matrix for cortico-thalamo-cortical (CTC) connections (from Fig. 2e) and a matrix representing local excitatory connections among IT, PT and CT neurons in the cortex. The local-circuit portion of the matrix is a qualitative interpretation of data from many studies, as reviewed in REFS^,. As a population, IT neurons are the strongest recipients of both matrix-type and core-type TC input, and they send their excitatory output broadly to IT, PT and CT classes. Output connections of PT and CT neurons are formed more selectively but, unlike the IT neurons, they include the thalamus. b ∣ The IT neurons (red) are hub-like components of CTC loops, local circuits and corticocortical (CC) circuits, which are also commonly organized as recurrent loops. The IT-mediated linking of CTC and CC loops forms an extended ‘loop chain’, also depicted in a simplified cyclo-plot below.

Fig. 5 ∣. Cortico-thalamo-cortical loops in the context of hierarchical models of cortical organization.

a ∣ In Sherman and Guillery’s model^,,, (sketched here based on REF., with the right-left orientation mirrored to match that of panel d below), CT and PT neurons send divergent output to thalamic nuclei. The PT axons form feedforward (FF) ‘transthalamic’ connections that disynaptically link lower-order to higher-order cortex, via higher-order (HO) thalamic nuclei, in parallel with direct corticocortical (CC) pathways. The CT axons form feedback (FB) connections back to the main thalamic nuclei supplying their cortical area, including the first-order (FO) nucleus, such as the LGN or VB. The transthalamic pathways of PT neurons are posited to carry efference copies of motor commands. b ∣ In Larkum’s model, a simplified version of which is sketched here, an essential central element is a ‘backpropagation-activated coupling’ (BAC) firing mechanism that generates strong, sustained burst firing upon coincident excitation to apical and basal→perisomatic dendritic domains. On their own, the feedforward and feedback inputs generate only weak firing in the postsynaptic pyramidal neuron. The sketch omits additional aspects of the model (see REFS^,). Various cortical and thalamic sources provide feedback and feedforward input, making the BAC mechanism a key activity-dependent integrative node in CTC circuits. The BAC firing model provides a compelling cellular mechanism for associative pairing of co-active neurons within large-scale CTC networks. c ∣ A similar model, incorporating additional connectivity details and emphasizing the polysynaptic routes of excitation to layer 5 neurons via IT neurons. d ∣ The CTC loops of S1, M1 and M2 are organized hierarchically, with ‘top-down’ M2→M1→S1 pathways running counter-current to ‘bottom-up’ S1→M1→M2 pathways. Areas additionally form ‘lateral’ connections with their contralateral counterparts and with other cortical regions, such as M1↔2 pathways (not shown). Area-associated subregions of matrix-type nuclei are indicated by superscripts (for example, VM^ALM). The layout is oriented with anterior/frontal to the left, both to follow the convention of standard atlases and to emphasize the bidirectionality of information flow in these loops. This three-area network offers a framework for further investigation of CTC loops in the context of hierarchical cortical circuits. Part a adapted with permission from REF.. Part b adapted with permission from REF..

Fig. 6 ∣. Inhibition in cortico-thalamo-cortical loops.

a ∣ The thalamic reticular nucleus (TRN) is a major inhibitory hub in CTC circuits. Connections of inhibitory TRN neurons include descending inputs from CT axons (left). One type of circuit thought to mediate inter-nuclear cross-talk is via the TRN (middle). Another type of circuit, based on recent evidence for segregated loops, links matrix-type thalamic nuclei with TRN ‘shell’ neurons, and core-type thalamic nuclei with TRN ‘core’ neurons (right). b ∣ The connectivity matrix qualitatively incorporates circuit connections of the TRN^,,,,,,-,-. The two types of GABAergic neurons in the TRN have distinct connectivity patterns. c ∣ The diagram depicts the CTC-related connections of inhibitory neurons in the cortex that have been identified in the sensory cortex^,,,,-. For both the core-type and matrix-type TC projections, the fast-spiking, parvalbumin-expressing interneurons are major targets, but the somatostatin-expressing neurons are not. Matrix-type projections excite additional classes of interneurons through their layer 1 axons. The circuit organization generates strong sensory-driven feedforward inhibition in sensory cortices. d ∣ In the PFC, dissection of CTC-related circuits of layer 1 (REF.) shows stratification into parallel channels in two sublayers, with matrix-type VM axons targeting neuron-derived neurotrophic factor-expressing interneurons and pyramidal neuron dendrites in layer 1A, and core-like MD axons targeting vasoactive intestinal peptide-expressing interneurons in layer 1B, which are also likely targets of corticocortical (CC) input. Distinct forms of short-term plasticity at the various synapses generate complex dynamics to regulate pyramidal neuron excitability. e ∣ An unusual variant of layer 1 inhibition in CTC loops is found in the retrosplenial cortex (RSC), where apical tuft dendrites of pyramidal neurons receive both excitatory input from matrix-type axons of anteroventral (AV) thalamus and long-range GABAergic input from RSC-projecting CA1 neurons in hippocampus. This apparently unique arrangement exemplifies how CTC circuit organization can be regionally specialized, presumably to mediate system-specific functions.

Fig. 7 ∣. Cortico-thalamo-cortical loop connectivity with the diverse types of matrix axon arborizations.

a ∣ TC neurons with focal-type projections (light purple) form a relatively focused, closed CTC loop with one cortical subregion (right), whereas those with divergently branching, multiareal-type projections (deep purple) arborize both within the same CTC loop and to one (or more) other areas (left). The targets of the CT and PT outputs of the second area are unclear. b ∣ Multiareal matrix projections can hypothetically form multiple CTC loops, the concatenation of which could form an extended network of cortical areas linked both by CC and CTC connections, forming widely distributed chains of loops.