Numerous studies have examined the neuronal inputs and outputs of many areas within the mammalian cerebral cortex, but how these areas are organized into neural networks that communicate across the entire cortex is unclear. Over 600 labeled neuronal pathways acquired from tracer injections placed across the entire mouse neocortex enabled us to generate a cortical connectivity atlas. A total of 240 intracortical connections were manually reconstructed within a common neuroanatomic framework, forming a cortico-cortical connectivity map that facilitates comparison of connections from different cortical targets. Connectivity matrices were generated to provide an overview of all intracortical connections and subnetwork clusterings. The connectivity matrices and cortical map revealed that the entire cortex is organized into four somatic sensorimotor, two medial, and two lateral subnetworks that display unique topologies and can interact through select cortical areas. Together, these data provide a resource that can be used to further investigate cortical networks and their corresponding functions.
PDFs
Supplementary Tables
Figure 1
Figure 1. Strategy for Generating the Cortical Connectivity Atlas(A) Schematic illustrating a PHAL/CTb and BDA/FG double coinjection in two different structures labeling both input to, and output from each injection site. Reciprocal interactions between brain regions and circuit interactions between each injection site may also be revealed.(B) A coronal section showing coinjections made into the MOs and ACAv viewed with Nissl background to reveal cytoarchitecture. Scale bar, 1 mm.(C) Intermixed anterogradely labeled axons (green, PHAL) and retrogradely labeled neurons (pink, CTb) in the MOs following a coinjection in the contralateral hemisphere (first two panels, arrows). Note: the PHAL/CTb coinjection is the same as pictured in B. Image histogram was adjusted differently for the two hemispheres so that PHAL/CTb labeling on the left side can be viewed properly without over exposing the injection site on the right side. Last panel, comparison of retrogradely labeled neurons from injection in ACAv (arrow, yellow, Fluorogold [FG]) and fibers and cells from injection in MOs (PHAL/CTb). Fluorescent Nissl in blue; scale bar, 200μm.(D) Strategy for mapping fluorescent labeling from a raw image (left, scale bar, 1 mm) onto the corresponding level of the ARA (middle) to generate a comprehensive map of projection pathways for all injection sites (right). Note: anterogradely labeled pathways were rendered as layer and regional-specific shading, while retrogradely labeled neurons were represented by individual dots. The large circle on the right hemisphere represents an injection site (see corresponding region on raw image).See Figures S1, S2, S4, and Table S1 for more information.
Figure 2. connectivity matrix final
Figure 2. Weighted and Directed Cortico-Cortical Connectivity Matrices(Figure 2) Connectivity matrices were constructed based on either anterograde (PHAL, A) or retrograde (FG/CTb, B) tract tracing data. In both matrices, connection origin is listed along the row while targets are listed across the columns (sorted alphabetically). The weighting of each connection is indicated by red (strong), orange (moderate), and yellow (light) coloring. In (C) and (D), the anatomical data in (A) and (B) has been reordered, illustrating a total of 12 distinct modules in different cortical subnetworks. Combining retrograde and anterograde tracing methods formed the composite matrix (E), a consensus perspective of cortico-cortical subnetwork connectivity. See Extended Experimental Procedures for details regarding construction of the matrices, Figure S3 for injection cases, and Table S2 for list of abbreviations.
Figure 3.
Figure 3. The Somatic Sensorimotor Subnetworks(A) Overview of the four major components of somatic sensorimotor areas (SSp, SSs, MOp, MOs). Each region is extensively interconnected with all others. Parcellation of cortical areas in map based on ARA and drawn to scale. Diamond shape on midline indicates bregma.(B) Projections from representative injection sites (colored dots) in each of four basic body representations in primary somatosensory cortex: orofaciopharyngeal (orf, blue), upper limb (ul, green), lower limb and trunk (ll/tr, red), and whisker-related caudomedial barrel field (bfd.cm, yellow). A cartoon (inset) shows approximate size and location of the four body areas defined here (inspired from Brecht et al., 2004). Top-down view (left) shows topographic organization of projections from each area to corresponding primary and secondary motor areas (MOp, MOs). ARA defined boundary between MOs and MOp added for reference. Projection data in top-down view were drawn to scale using coronal sections (right) and shaded regions represent the areal extent of the most dense projections from each of the injected regions. Representative coronal sections also show projection trends to supplemental somatosensory area (SSs). Numbers indicate position of sections relative to bregma (mm).(C) Projections from each of the somatosensory subregions define presumably functionally related MOs and MOp subregions, which are tightly reciprocated, as indicated by closely overlapped axonal fibers and retrogradely labeled cell bodies following coinjection. Here, coinjections of PHAL (green)/CTb (pink) in either SSp-ll (top, right) or a corresponding MOp region (middle, middle) reveal intermixed labeling in other corresponding domains of the somatic sensorimotor region, confirming their strong reciprocal connectivity. Both coinjections reveal intermixed labeling in the same MOs domain (left images on the top and middle). Retrograde injection in the same MOs domain (left image on the bottom) confirms the specificity of this interaction, showing retrogradely labeled neurons in the former two areas (middle and right images). Their anatomical locations and interactions are summarized in corresponding atlas levels in the bottom panel. Scale bars, 500μm and 100μm (inset).(D) Building on these observations, four network graphs were created using each of the defined somatosensory regions as starting points. Each subnetwork is distinct and all components within it share a high degree of interconnection. Each are composed of several somatic sensorimotor “nodes” (color coded to match anatomically defined functional domains in (B) that are reciprocally connected (as indicated with red arrows). Each of these subnetworks also includes other nonsomatic “peripheral” nodes (gray circles) and their connections are shown with gray arrows.See also Figure S5. For abbreviations of nomenclatures, please see Figure 2 and Table S2.
Figure 4.
The Medial Subnetworks(A) Major components of the medial subnetworks, which mediate transduction of information between sensory areas (VIS, AUD, and caudal-most SSp) and higher-order association areas along the medial bank of the neocortex, such as the retrosplenial (RSP), parietal (PTLp), anterior cingulate (ACA), and orbital (ORB) areas.(B) Connectivity pathways of the medial subnetwork revealed by coinjections in the ORB (note: these are aggregated pathways for three different cases, see three coinjection sites in ORB, colored pink, light brown, and dark brown, medial to lateral).(C) Representative raw images from an ORBvl coinjection. PHAL-labeled axons and CTb-labeled neurons are found in other medial network components such as the ACAd and adjacent MOs-fef, PTLp, RSPd, and primary and secondary VIS areas (VISp, VISal, and VISam). Scale bars, 500μm (first panel) and 200μm.(D) Laminar-specific differences in axonal projections to primary visual cortex (VISp) arising from either ORBvl (red, BDA labeling) or ACAd (green, PHAL). Injection sites in the same brain, left panels, scale bars, 500μm (left) and 1 mm. Projections to different layers of the same section of VISp (right). Underlying fluorescent Nissl was inverted to aid visualization of layers (right-most panel). Scale bar, 200μm.(E) Four different retrograde tracers were injected into the VISp (two), VISam, and VISpm within the same brain, resulting in distinct, topographically arranged clusters of neurons in the ACA and adjacent MOs-fef. In the ORBvl, these retrogradely labeled neurons are intermixed, but mostly not colocalized (bottom right, 5% colocalization for any combination of tracers in ORBvl, 436 cells counted, 21 had two or more tracers present). Scale bars, 1 mm (top left), 500μm (bottom left), 200μm (top right), 100μm (bottom right).(F) Summary of interactions among the medial subnetworks. Left, interaction between sensory and association areas. Dashed lines indicate sparse connection. Claustrum (CLA) is included due to high degree of interconnection with medial network. Middle, connections between the association areas. Thicker arrows indicate dense projection patterns between regions. Dashed line separates a direct pathway to medial prefrontal region along the ventro-medial bank of the cortex (second medial subnetwork). Asterisks indicate a unidirectional connection between CLA and RSP. Right panel, overview of medial network interactions including TEa and parahippocampal structures (i.e., SUBd, ENTm), which project to RSP (red arrows). Reciprocal connections of visual (blue) and auditory (green) areas with all major medial network components shown. Caudal-most somatosensory areas (SSp-ll/tr; SSp-bfd.cm) are included as well (gray arrows).See also Figure S6.
Figure 5.
The Lateral Subnetworks(A) Sagittal view of the major components of two lateral subnetworks: the anterolateral insular (including the AId, AIv, AIp, VISC, GU) and posterior temporal (including TEa, ECT, PERI). These two interconnected subnetworks are also connected with olfactory (e.g., PIR) and medial prefrontal (mPFC) areas and with the ENTl. TEa in particular forms extensive connections with much of the rest of the neocortex (gray arrows).(B) Distinct projection patterns of the anterior agranular areas (PHAL injections involved both AId and AIv, left panel) and AIp (right panel) with the mPFC areas (PL, ILA, DP), posterior temporal areas (TEa, ECT, PERI), and ENTl. The AIp targets more ventral structures in mPFC and more heavily innervates the central nucleus of the amygdala (CEA). Scale bars, 500μm.(C) Map of neuronal inputs to (left) and output from (right) the TEa, which are arranged topographically along the rostrocaudal direction. Note: these pathways are aggregated from six coinjections made into different parts of the TEa from rostral (red) to caudal (blue) direction (top most sagittal image, numbers relative to bregma in mm). Retrogradely labeled neurons are indicated as colored dots and demonstrate the layer-specific origin of cortical projections to TEa. Axonal pathways arising from TEa (outputs) are rendered as shaded areas of color.(D) Raw image of retrograde labeling (FG, yellow) following injection in TEa. Cells are distributed extensively across numerous cortical regions following a single, small injection, suggesting a high level of convergence. Bottom left panel shows close up of layer specificity in somatosensory barrel field, with most cell bodies residing in layers 2/3, 5a, and some layer 6. Fluorescent Nissl inverted (right) to aid in discriminating layers. Layer 4 “barrels” indicated with arrow. Scale bars, 500μm (top) and 200μm (bottom).(E) Raw image of coinjection in TEa. Fibers are predominately ipsilateral, but retrogradely labeled inputs are evenly distributed across both hemispheres. See right panel, middle, comparing labeling in contralateral and ipsilateral MOs.Scale bars, 1 mm (top right), 200μm (middle), 500μm (bottom). See also Figure S7.
Figure 6.
The CLA and ENTl(A) Axonal projections arising from the CLA are distributed throughout the entire neocortex and ENTl on the ipsilateral hemisphere. Note: these axons display different regional and laminar distribution specificity (on right panel) Scale bars, 1 mm (left), 500μm (top right), 200μm (bottom).(B) and (C) Asymmetric connections of the CLA with other cortical areas in the two hemispheres. Cortical inputs to CLA project to both sides with equal densities (B, labels: dorsal and ventral claustrum (CLAd, CLAv), and endopiriform nucleus (EPd)), while outputs from CLA to other cortical areas indicated by retrograde tracers are almost exclusively ipsilateral (C). Moreover, the CLA has a dorsal to ventral topography in its projections to the cortex, with cells in CLAv (yellow, C, right) preferentially targeting ventral cingulate (C, left, injection site in ACAv). Very little colabeling was observed among cells labeled from a neighboring, more dorsal injection (pink). Scale bars: 1 mm (left) and 200μm (right).(D) Neural inputs to the CLA from almost all cortical areas in medial, somatic, and lateral subnetworks. The somatomotor inputs preferentially target the dorsal-most aspect of CLA.(E) Representative images of PHAL-labeled axons in layer 1 of a wide range of neocortical areas arising from the rostrodorsal ENTl.(F and G) Laminar specificity of PHAL-labeled axons and CTb-labeled neurons in the ENTl after coinjections made into the AI or ILA (F). Both cortical regions provide strong, direct input to ENTl, further supported by retrograde data in (G). These data also confirm the CLA is a specific source of input to ENTl (G, middle). Scale bars, 1 mm (F, top left and G, left), 500μm (F, bottom and G, middle), 100μm (F, top right).See also Figures S1B–S1E.
Figure 7.
Interactions with Prefrontal Cortex(A and B) Cumulative projections from components of the somatic sensorimotor, lateral, and medial networks in two representative coronal sections of the prefrontal cortex (PFC). Collectively these represent inputs from the entire neocortex to the PFC. Inputs were color coded based on the location of the injection sites in different components of the network (A). For example all primary motor projections arising from multiple injections along the length of this structure were colored green and all somatosensory projections were colored blue (A, top). ACAv was colored red to separate it as a component of the second medial subnetwork (A, bottom, see Figure S6E). Note that RSP has very little interaction with the PFC. (B) All inputs from three somatic sensorimotor subnetworks (as shown in A) converge onto three distinct zones, dorsolateral (dl), dorsal (d), and dorsal medial (dm), in the dorsolateral half of the prefrontal cortex (PFCdl, green and blue). In contrast, the medial and lateral subnetworks converge onto the ventromedial half of the prefrontal cortex with distinctive patterns. Note that caudal-most somatosensory and motor regions make some contribution to lateral-most, and caudal aspects of ORB (green and blue shading).(C) A schematic view of cortico-cortical network information flow as seen in a top-down view of the cortex (left, lateral edge on left, PFC at the top). All subnetworks are colored according to the scheme used in (A) and (B). Right, a more detailed overview of these interactions (lateral edge of cortex on the right, PFC at the top). Somatic sensorimotor boxes are meant to include both the sensory area and its corresponding primary motor area with which it is strongly interconnected. All functionally distinctive subnetworks are organized along the longitudinal axis of the cerebrum. Information processed in the medial and lateral subnetworks is integrated within the ventromedial half of the prefrontal cortex (PFCvm) and the ENTl. The claustrum (CLA) may also provide an additional means of direct interaction between each of the subnetworks. For abbreviations, please see Figure 2 and Table S2. Additional abbreviations: AMY, amygdala; AH, Ammon’s Horn; HPF, hippocampal formation.
Figure S1.
Demonstration of Specificity and Accuracy of Injection Strategy, Related to Figure 1(A) Representative images of coronal brain section with double coinjections in two neighboring domains of the primary somatosensory area (SSp): PHAL (green)/CTb (pink) in the lower limb domain, and BDA(red)/FG(yellow) in the trunk domain. To demonstrate the specificity of these two coinjections, two adjacent clusters of labeling (intermixed labeling of PHAL with CTb or BDA with FG) were observed in the posterior thalamic (PO) and ventroposteromedial (VPM) thalamic nuclei, indicating their reciprocal connections with the SSp-ll or SSp-tr. Also, three distinct topographically arranged cortical columns associated with each injection were observed in the auditory cortical area (AUD) (right panel, retrograde labeling shown), which provides further evidence that each injection site is discrete and is not contaminating labeling from its neighboring injection site. Note, however, that to show proper labeling throughout the brain using the iConnectome viewer (www.MouseConnectome.org), histogram levels were adjusted to reveal fine fibers, which causes overexposure at the coinjection sites such that they appear to be larger (second panel, same section as left-most panel as it appears in iConnectome browser. Arrow indicates thalamic labeling that becomes visible when contrast is adjusted). We estimate injection sites are typically 250-500μm in diameter. Scale bars, 1 mm (left), 100μm (middle), and 200μm (right).(B) Further demonstration of specificity of injection size and location. Coinjections (PHAL/CTb) were made into two small, adjacent cortical areas, the gustatory (GU, top) and posterior agranular insular area (AIp, lower). Anatomical locations of these two injections can be validated by using underlying Nissl-stained cytoarchitecture within the same section (middle images, right panel is an inverted image of the fluorescent Nissl to aid viewing). Coinjection within the GU resulted in a distinct cluster of intermixed labeling in the VPMpc, which relays taste information to the cortex (Jones, 2007), further validating specificity of the GU coinjection. Coinjection in the AIp, but not in the GU, resulted in dense labeling in the deep layers of the ENTl (lower panel); a FG injection in the same region of the ENTl resulted in dense retrograde labeling in the AIp (and adjacent CLA), but only sparse in the GU, providing further validation of the projection patterns. Scale bars, 1 mm (left), 200μm (middle), 1 mm (right).(C) Connections between any two cortical areas in our project can be cross validated using both the anterograde and retrograde tracing method. Here, a PHAL injection within the SSp (top left) results in dense axonal terminals in the MOp (top right). A FG injection in the same location of the MOp (bottom right) results in retrogradely labeled neurons in the corresponding SSp location (bottom left). Importantly, using this strategy also reveals the regional and laminar specificity of both the axons projecting to MOp and the layer-specific arrangement of cell bodies that give rise to this projection.(D) Reproducibility and control of individual variability of labeling in different cases with injections in the same locations. PHAL injections made into identical locations of the ILA in two different mice resulted in identical projection patterns in the cortex and midbrain. Several other injection cases were repeated as further controls and resulted in identical labeling patterns. Scale bar, 1 mm.(E) Resolving fine-scale organization of cortical structures smaller than the diameter of a single injection site. One coinjection involves two small, adjacent areas, the dorsal (AId) and ventral (AIv) agranular insular areas (left) and resulted in labeling in the rostral AId and ILA, among other areas. To examine the specificity of these connections, two coinjections were made into the ILA and rostral AId respectively in different animals. They show that only the AIv, but not AId, shares reciprocal connections with the ILA, while the rostral AId interacts with only the more caudal AId, but not AIv. Cross examining all targets of the original, large coinjection in AId/v (left panel) in this fashion revealed preferential interaction of each of the targets with either AId or AIv, thus allowing for parcellation of connectivity patterns for these small cortical regions. Scale bar, 500μm.
Figure S4. stategy of generating cortical maps
Generating the iConnectome Cortical Connectivity Map and Its Use in Establishing Interactive Networks between Cortical Regions, Related to Figure 1(A) General approach to mapping fluorescent labeling from a raw image (left column) onto the corresponding level of the ARA to generate the connectivity map (middle column). Each coinjection site is represented by a circle. PHAL pathways were rendered as shaded regions with regional and laminar specificity. Retrograde labeling (FG and CTb) is represented by small dots that reflect regional and laminar distribution patterns, but also relative densities (Figures 1F and S4A). Each of these pathways was color coded with a unique RGB value and rendered into an individually layered document with transparency such that multiple combinations of layers representing multiple injection sites could be viewed simultaneously within the same anatomic frame provided by the Allen Reference Atlas (Dong, 2007), thus revealing topographic trends for large areas and interaction between regions. A general color scheme of red-orange-yellow-green-blue-indigo-violet, medial to lateral, was implemented for the online map, although for analysis, colors were manipulated to reveal patterns more clearly. The PHAL (green)/CTb (pink) coinjection site in the VISam is represented by a large circle; PHAL-labeled axonal pathways were rendered as layer and regional-specific shading (see labeling in VISal); while retrogradely labeled neurons (CTb or FG) were represented as individual dots reflecting the density and layer-specific distribution of observed cell bodies. The cortical parcellations in the reference atlas were, in part, created by using cytoarchitectural information revealed by Nissl stain. Examination of the underlying Nissl stain in each injection case, along with major anatomical landmarks, allowed for accurate placement of labeling within the reference atlas framework. PHAL and CTb labeling are intermixed in the VISal, as indicated by overlapping of shaded areas with individual dots, suggesting reciprocal connectivity between the VISal and VISam regions (see right column for higher magnification raw images). Using this method, we have reconstructed a total ~240 cortico-cortical pathways (~80 PHAL-labeled efferent pathways; ~160 FG or CTb-labeled afferent pathways) across the entire neocortex. These projection maps are available online in the iConnectome map viewer (www.MouseConnectome.org) (Figures 1F and S4B). Scale bars, 500μm (left) and 200μm (right).(B) Strategy for establishing connections between cortical regions. Coinjections reveal reciprocal connectivity between the injection site locations and other target areas. For example, a VISam coinjection (top panel, yellow) labels both axonal fibers and cells in VISal and ORBvl, providing initial evidence for a reciprocal interaction between these regions. A follow up coinjection in VISal (green) confirms this reciprocal connectivity and reveals a common target in ORBvl (Right column shows raw images of the ORBvl with fluorescent labeling after coinjections in either VISam or VISal, scale bar, 200μm). To validate these connections, a PHAL/CTb coinjection was made into the ORBvl (bottom image on the right column), which reveals dense PHAL-labeled axons and CTb-labeled neurons in the VISam and VISal (brown colored pathways in the connectivity map; see bottom left raw image for ORBvl labeling in the visual areas), confirming their strong reciprocal connections. It also shows relatively sparse labeling in the VISp, suggesting its moderate connections with the ORBvl. Collectively, these data suggest these structures interact as a network, summarized in bottom right panel. Following this strategy, larger interactive networks can be built using additional data collected from across the entire cortex. For example, further examination of VISp interactions revealed shared connectivity with VISam and VISal areas as well (not pictured), further broadening the scope of interaction for the network.(C) Collecting all projection data for each case into a common atlas space allows for quick comparison of labeling trends and determination of common and unique targets for different injection sites. For a single coinjection, projection data from across the entire cortex was rendered onto 29 corresponding atlas sections, from the front of the cortex to the back, as viewed in the iConnectome map viewer (see Figure S2B). Five representative sections are shown here (left panel, VISam coinjection, yellow) demonstrating anterograde and retrograde labeling in other cortical regions. Additional projection pathways may be overlayed in the map for a convenient comparison of the interactions between each injection site (right panel, VISal injection in green). Based on this data, a summary of the shared and unique targets of VISal and VISam is provided in the diagram on the right. The connectivity map may be useful for determining these trends and interactions for any cortical region of interest.
Figure S5.
Demonstration of Specificity of Different Functional Domains of Somatic Sensorimotor Areas, Related to Figure 3(A) The specificity of SSp-m/n (orofacial domain) and SSp-bfd (whisker domain) were validated by their projections to either the dorsal or ventral part of the spinal trigeminal nucleus (Nord, 1967, Welker et al., 1988, Killackey et al., 1989 and Erzurumlu et al., 2010); while the specificity of the upper limb (SSp-ul) or lower-limb (SSp-ll) domains were validated by their projections to the cuneate or gracile nucleus in the lower brainstem, which process sensory information related to upper and lower limbs, respectively (Nord, 1967 and Li et al., 1990). Scale bars, 1 mm (top), 500μm (bottom).(B) Direct comparison of projection patterns for different domains of SSp within the same brain. Top: Specificity of cortico-cortical connectivity of the SSp-ul and SSp-ll domains. Two anterograde tracer injections were made into the SSp-ul (AAV1.hsyn.GFP, green) and SSp-ll (AAV1.CAG.RFP, red), respectively. Each injection site location was validated by their topographic projections to the cuneate or gracile nuclei (top right). These two domains generate topographically arranged projections in two domains of the MOs in the prefrontal cortex, the MOs-rd and MOs-rdm. The use of viral tracers in this example allowed visualization of native fluorescence prior to cutting the brain (last panel, scale bar, 1 mm) which enabled localization of the injection sites and eased translation between the top-down view of the cortex and the corresponding coronal sections. The very last panel of the figure superimposes the cortical connectivity map and the injection site locations of both cases. Bottom: Two injections in the middle and caudal parts of the SSp-bfd (as validated by their axonal terminals in the ventral two third of the spinal trigeminal nucleus) generate distinct projection patterns within the MOp and MOs.(C) Topographic projections from different SSp functional domains (as well as the PTLp) to the MOp and MOs in the prefrontal cortex (anatomical locations and coordinates of injection sites are shown in right column). These projections provide a structural basis to determine different functional domains of the MOp and MOs, for example, the MOs-rdm, MOp-bfd, MOs-rd, MOs-rdl, and MOp-orf.(D) The more caudal aspects of MOs and anterior cingulate are not heavily innervated by somatosensory projections, but rather receive extensive input from more posterior cortical regions (e.g., visual, auditory, parietal) as seen in the schematic (left). Moreover, this region interacts with the posterior-most aspects of motor cortex. For example, a retrograde injection in this motor region (FG, yellow, top left panel) results in extensive labeling of neurons in this MOs/ACA region (bottom left). This interaction is further confirmed by anterograde injection in the MOs/ACAd region (PHAL, green), which projects to the corresponding caudal motor region as well as visual, retrosplenial, and superior colliculus areas (middle). The brainstem projections to superior colliculus and other pretectal regions implicated in controlling eye movements suggest this cortical area may correspond to the frontal eye field (MOs-fef in the text) identified in primates (also see Reep et al., 1990 for rats). The layer-specific origin of primary visual cortex projections to this area are also shown (right panels), which presumably contribute visual information for processing in the region.
Figure S6.
Additional Information for Assembling the Medial Subnetworks, Related to Figure 4(A) Parallel projection pathways from the VISp and AUDp to the association cortical areas within the medial subnetwork, including the ORBvl, ACAd, ACAv, and PTLp. Please note that the VISp also has dense projections to the RSPd and RSPagl, which receives very sparse inputs from the AUDp. On the other hand, AUDp projects heavily and unidirectionally to VIS areas (last panels).(B) Projection map (left) of the ACAv (red injection site and pathways), ACAd (orange), and MOs-fef (yellow). All three areas are heavily connected with other regions within the medial subnetworks, such as the ORB, PTLp, RSP, and CLA. These three areas also share connectivity with the visual, auditory, and somatic areas with different densities. Raw images of one representative coinjection in the ACAd and its labeling in other areas (such as the ORB, CLA, PTLp, VISp, VISam, and VISal) are found in the right panels.(C) Summary of direct interactions between sensory areas in different modalities D. A portion of caudomedial barrel field (SSp-bfd.cm) is reciprocally connected with primary visual cortex, and the lower-limb and trunk domains of the SSp (SSp-ll/tr) interacts with auditory areas. The most striking example of sensory-sensory interaction is a strong, unidirectional projection from all auditory areas to all visual areas (large arrow with asterisk, see also (A, last panel). A weaker projection specifically from VISal appears to bridge AUD and VIS areas and is noted with a dashed line.(D) Summary of additional interactions between the auditory and visual areas with the primary somatosensory area (SSp-ll/tr) (left), and direct interactions of the SSp-ll/tr and MOp-ll/tr with other nodes within the medial subnetworks (such as the ORB, ACA, PTLp, and RSP).(E) Connectivity map of the ACAv and RSPv, which form another medial subnetwork to relay information from the SUBd to the medial prefrontal cortex (ORBm, PL, and ILA).
Figure S7.
Additional Information for the Lateral Subnetworks, Related to Figure 5(A) Raw images of direct projections from the SSp and SSs to the AIp. These projections are unique to specific portions of the somatosensory area and are predominately contralateral. Scale bars, 1 mm (top, middle) and 100μm (bottom).(B) Topographic inputs from other cortical areas to the TEa, ECT, and PERI. These projections help define different subregions within these structures. The most anterior aspect (1) mostly receives inputs from orofacial related somatosensory and motor areas, followed by all other somatomotor inputs (2). These inputs dip below other cortical inputs in (3) and (4) and remain segregated in the perirhinal (PERI) region. The infralimbic area provides a very strong and specific input to caudo-ventral regions of TEa (4) and (5) and many medial network structures (e.g., PTLp, VIS, AUD) interact with the adjacent dorso-caudal aspect of TEa (5).
