Organization of the inputs and outputs of the mouse superior colliculus

Abstract

The superior colliculus (SC) receives diverse and robust cortical inputs to drive a range of cognitive and sensorimotor behaviors. However, it remains unclear how descending cortical input arising from higher-order associative areas coordinate with SC sensorimotor networks to influence its outputs. Here, we construct a comprehensive map of all cortico-tectal projections and identify four collicular zones with differential cortical inputs: medial (SC.m), centromedial (SC.cm), centrolateral (SC.cl) and lateral (SC.l). Further, we delineate the distinctive brain-wide input/output organization of each collicular zone, assemble multiple parallel cortico-tecto-thalamic subnetworks, and identify the somatotopic map in the SC that displays distinguishable spatial properties from the somatotopic maps in the neocortex and basal ganglia. Finally, we characterize interactions between those cortico-tecto-thalamic and cortico-basal ganglia-thalamic subnetworks. This study provides a structural basis for understanding how SC is involved in integrating different sensory modalities, translating sensory information to motor command, and coordinating different actions in goal-directed behaviors.

Fig. 1. Experimental workflow.

a Raw data examples of cortico-tectal projections targeting distinct zones within SC layers at ARA 90. Anterograde AAV injections into PTLp→SC.m (red), VISam→SC.cm (orange), ACAd→SC.cl (green), and MOp→SC.l (purple) terminate with different laminar and nonoverlapping patterns. Together, these representative cortico-tectal projections reveal layer-specific terminals distributed across four distinct zones along the medial-lateral axis of SC. Scale bar at injection sites is 500 µm, and 200 µm in lower SC panels. b The Connection Lens neuroinformatics workflow: Raw tissue image with anterograde labeling is assigned to the corresponding standard atlas level (this example is ARA 96). The tissue is warped based on template atlas borders and reconstructed using in-house neuroinformatics software. Thresholded images are overlapped onto a custom atlas for zone- and layer-specific registration for pixel quantification. c Left: Polar Coordinate analysis method used to quantify angular distribution of thresholded pixel labeling in SC. Angles represented by theta (θ°) values where midline starts at 90° and ranges toward 0° at lateral angles. Right: Custom SC atlas with overlay of angular range shows coarse alignment of manually delineated borders in SC. d Example of probability distribution graph for average distributions of zone-specific cortico-tectal cases. Peaks are aligned with custom SC borders at ARA 90. SC.m (red) is SW180522-04A. SC.cm (orange) is SW121221-03A. SC.cl (green) is SW171130-02A. SC.l (purple) is SW170410-04A. e Layer- and zone-specific SC nomenclature to facilitate referencing and quantification. Below: zone delineations across representative ARA levels (86, 90, 96, 100) spanning rostral-to-caudal SC. ACAd anterior cingulate cortex dorsal part, ARA Allen reference atlas, AAV Adeno-associated virus, PHAL Phaseolus vulgaris leucoagglutinin, PTLp posterior parietal cortex, MOp primary motor cortex, VISam visual cortex anterior medial part, SC superior colliculus, SC Layers: zo zonal, sg superficial gray, op optic, ig intermediate gray, iw intermediate white, dg deep gray, dw deep white.

Fig. 2. Visual, auditory, and somatic sensorimotor map of cortical projections to SC zones.

a Color-coded schematic overview of the left visual field and body part topography based on their target zones in right hemisphere of SC at level 90. Reconstructions of cortico-tectal fibers adjacent. b Schematic of injection strategy using triple or double anterograde tracers into cortex, and custom SC atlas levels. c VISp, VISal, and PTLp-lat projections to SC.m and SC.cm. SC boundaries in all panels were delineated based on Nissl-stained cytoarchitecture. Dashed lines correspond to specific layers in each SC level. Scale bar 200 µm in SC panels. d Two neighboring caudal MOs regions and a rostral MOp injection send projections to adjacent, but nonoverlapping zones in the SC.cl and SC.l. e Rostral MOs and rostral-ventral MOp projections to the SC.l zone. f SSp-tr and AUDp projections target caudal SC.l zones. g Stacked bar chart for visualization of proportion of labeling across each SC zone (x-axis) from each cortical ROI (y-axis) for each selected ROI (n = 20 cases). Values represent proportion of pixel density for an individual ROI across each zone (n = 1 case per row). See Supplementary Table 3 for zone- and layer-specific values. h Probability distribution graphs of thresholded labeling represented by probability density (y-axis) across angular ranges (θ°) in SC (x-axis) from atlas level 90. VISam cases show distributions in angular ranges that align with SC.m and SC.cm zones (n = 3). MOs cases show distributions aligned with SC.cl (n = 10). SSp-ul cases show distributions in angular ranges aligned with SC.l (and SC.cl) zones (n = 3). Color-code associations: red (SC.m), orange (SC.cm), green (SC.cl), purple (SC.l). ACAd anterior cingulate cortex dorsal part, ARA Allen reference atlas, AAV Adeno-associated virus, AAV-gfp AAV green fluorescent protein, AAV-rfp AAV red fluorescent protein, MOp primary motor cortex, MOs secondary motor cortex, PHAL Phaseolus vulgaris leucoagglutinin, PTLp posterior parietal cortex, VISam visual cortex anterior medial part, SC superior colliculus, SC Layers: zo zonal, sg superficial gray, op optic, ig intermediate gray, iw intermediate white, dg deep gray, dw deep white, Somatotopic body parts; bfd barrel field, ll lower limb, m mouth, n nose, tr trunk, ul upper limb, SSp primary somatosensory cortex, VIS visual cortex. Complete list of abbreviations in Supplementary Table 1.

Fig. 3. Distribution of higher-order cortical inputs across SC zones.

a Schematic of higher-order association areas part of medial cortico-cortical subnetworks (Zingg et al., 2014). Color-coding is consistent with the SC zones, and represents the topography based on cortico-tectal projection patterns that target distinct zones. Reconstruction is composite of inputs to SC ARA 90. b Stacked bar chart for visualization of proportion of labeling across each SC zone (x-axis) from each cortical ROI (y-axis) for each selected ROI. Values represent proportion of pixel density for each individual ROI across each zone (n = 1 case per row). See Supplementary Table 4 for zone- and layer-specific values. c Schematic of anterograde injection strategy, and custom SC atlas levels. d Raw data panels. VISp, RSPv, and ACAd-rostral projections to SC. e ACAd-interm and VISam projections. f Two PTLp-medial projections projection to caudal SC levels. g Visualization of cortico-tectal modular communities within the SC (n = 34 cases). Matrices show communities identified by Louvain analysis of cortical injection ROIs (y-axis) and SC zones (x-axis). Edges are shaded according to their connectivity labeling, and boxes along the diagonal reflect modular communities identified. Legend on right side is based on the range of normalized values calculated by the Louvain algorithm. Value 0.25 represents the maximum intensity (black), and value 0 represents the minimum (white). Color-coded brackets below the matrix communities correspond to the same color codes in the SC custom atlas zones. Color-code associations: red (SC.m), orange (SC.cm), green (SC.cl), purple (SC.l).

Fig. 4. Brain-wide connectivity of SC inputs and outputs.

a Color-coded map of anterograde projections from SC zones throughout the brain demonstrating topographic output patterns. Top left insert is a visual aid representation the anterograde injections placed in each SC zone, with colors corresponding to the same color of projections throughout the brain. See Supplementary Table 2 for injection site details. SC.m (red): SW190619-04A (PHAL), SW190619-02A (PHAL); SC.cm (orange): SW190619-02A (AAV-tdTomato); SC.cl (green): SW171010-01A (AAV-gfp), SW171010-02A (AAV-gfp); SC.l (purple): SW171010-01A (AAV-tdTomato), SW171010-01A (PHAL). b Unweighted wiring map of all inputs and outputs of superior colliculus zones and layers. Assembled using data from anterograde and retrograde injections placed in SC to systematically trace outputs and inputs distributed from cortex down through the hindbrain and cerebellar structures. Color-code associations: red (SC.m), orange (SC.cm), green (SC.cl), purple (SC.l). See Supplementary Table 1 for complete list of abbreviations.

Fig. 5. Topographic organization of brain-wide inputs and outputs of SC zones.

a Raw data labeling from two separate cases of retrograde injections into the SC. Top row: CTB (pink) injection into SC.m, CTB (red) injection into SC.cl, and CTB (green) injection into SC.l. Cells are topographically retrogradely labeled in VMH domains and cortex. Bottom row: CTB (red) injection in SC.cm, and FG injection in SC.l. b Raw connectivity data in SC (all at ARA 90) from three separate anterograde injections in the ZI. The ZI.l (lateral part) targets SC.m/cm; the ZI.c (center part) targets SC.cl, and the ZI.m (medial part) targets SC.l. c Raw connectivity data in SC (all at ARA 90) from three separate FG injections in different thalamic nuclei, RE (top), MD (middle) or PF.m (bottom), and the respective retrogradely-labeled thalamic projections neurons, respectively, in SC.m/cm, SC.cl, or SC.l. Cells were distributed heterogeneously across superficial, intermediate and deep layers, but clustered topographically and preferentially in distinct zones. Scale bar 200 µm in SC panels. d Labeling produced from anterograde injections into distinct SNr domains project to SC. e Topographic organization of mouse SC connectivity. Directional cross at bottom left. The four zones of the SC are connected either uni- or bi-directionally with cortex, PF, ZI, VMH, and SNr. Plus signs and minus signs indicate excitatory or inhibitory projections, respectively. f Dendrites from LD-, RE-, and PF/PT-projecting SC neurons were reconstructed to visualize differences and dendritic arborizations across SC layers and zones. Examples of reconstructed neurons were overlapped onto reconstructions of cortico-tectal projections on the SC atlas. This facilitates analysis of spatial correlation between cortical projection fields and dendritic fields of SC-projection neurons. Color-code associations: red (SC.m), orange (SC.cm), green (SC.cl), purple (SC.l). See Supplementary Table 1 for complete list of abbreviations.

Fig. 6. Cortico-striatal and cortico-tectal have conserved topography.

Within the cortex, somatic sensorimotor areas are each organized into distinct highly interconnected subnetworks that integrate the visual field, somatotopic body map and higher-order areas which then project subcortically to the caudoputamen (CP) into distinct domains^,. Each cortical area sends parallel descending projections to SC zones and CP domains. a Three separate triple anterograde cases into different cortical areas send topographic projections to corresponding caudate putamen (CP) domains and SC zones. b Injection site schematic of combinatorial tract-tracing method using AAVretro-Cre injection (in CPc.d at ARA 57) and Cre-dependent AAV-mCherry (in MOs at ARA 47). c Injection sites in CPc.dm and MOs within the same mouse brain (SW190110-01B). d Red anterograde labeling demonstrates a subset of MOs cortical projecting neurons (presumably pyramidal tract projection neurons or PT neurons) generate collateral projections specifically to the SC.cl zone and CPc.dm. Scale bars are 200 µm. Color-code associations: red (SC.m), orange (SC.cm), green (SC.cl), purple (SC.l). SC superior colliculus, SC layers: zo zonal, sg superficial gray; op optic, ig intermediate gray, iw intermediate white, dg deep gray, dw deep white.

Fig. 7. Upper-limb/orofacial subnetworks in the SC.cl and SC.l.

a TRIO-tracing strategy targeting AAVretro-EF1a-Cre in PF.ul (PF upper limb), and rabies virus (AAV8-hSyn-FLEX-TVA-GFP) and Cre-dependent helper virus (EnvA G-del-Rabies-mCherry) in SC.l reveal mono-synaptic inputs to the PF-projecting neurons in the SC.l. b Raw images of injection sites in PF.ul, and SC.l with zoomed in panels of triple labels starter cells in SC (case SW190926-10A). c Rabies retrograde labeling in somatic sensorimotor cortical areas MOp and SSp-bfd. Cells labeled in the dorsal and central lateral substantia nigra reticulata (SNr), the interpolar part of the spinal nucleus of the trigeminal (SPVI) and dentate nucleus (DN) of the cerebellum. Bottom right corner is a summary diagram showing that thalamic (PF) projecting neurons in SC.l receive convergent inputs from the somatic sensorimotor cortical areas (MOp, SSp), subset of the basal ganglia (SNr), somatic sensory nuclei (i.e., SPVI), and deep cerebellar nuclei (i.e., DN).

Fig. 8. Subnetwork organization of SC zones.

a Integration of SC.cl and SC.l zones in visuomotor, somatic sensorimotor, escape and approach subnetworks. SC.cl/l neurons receive inputs from MO/SS (orofacial, barrel field, upper limb) cortices, motor-related domains from thalamic nuclei (PF.mouth, PF.upper limb, VM, PCN), sexual approach behavior (VMH.vl) and orofacial/upper-limb brainstem and cerebellar inputs. This input provides evidence for SC.cl/l zones as distinct somatic sensorimotor subnetworks that mediate a subset of behaviors distinct from the more visually integrated range of behaviors mediated through SC.m/cm. Our network analysis reveals the segregation of ventromedial prefrontal cortex (vmPFC) outputs from ORB and ILA regions to functionally specific SC.cl/l zones. These prefrontal regions also project densely to VMH, ZI and PAG, with BLA sharing bilateral connections with ILA). b SC.m and SC.cm integration of visual information with spatial-related and head-direction hippocampal networks, attention/orienting regions and freezing/defense regions. See Supplementary Table 1 for complete abbreviation list.