Organization of corticocortical and thalamocortical top-down inputs in the primary visual cortex

Liu, Yanmei; Zhang, Jiahe; Jiang, Zhishan; Qin, Meiling; Xu, Min; Zhang, Siyu; Ma, Guofen

doi:10.1038/s41467-024-48924-8

Download PDF

Article
Open access
Published: 27 May 2024

Organization of corticocortical and thalamocortical top-down inputs in the primary visual cortex

Nature Communications volume 15, Article number: 4495 (2024) Cite this article

2951 Accesses
4 Altmetric
Metrics details

Subjects

Abstract

Unified visual perception requires integration of bottom-up and top-down inputs in the primary visual cortex (V1), yet the organization of top-down inputs in V1 remains unclear. Here, we used optogenetics-assisted circuit mapping to identify how multiple top-down inputs from higher-order cortical and thalamic areas engage V1 excitatory and inhibitory neurons. Top-down inputs overlap in superficial layers yet segregate in deep layers. Inputs from the medial secondary visual cortex (V2M) and anterior cingulate cortex (ACA) converge on L6 Pyrs, whereas ventrolateral orbitofrontal cortex (ORBvl) and lateral posterior thalamic nucleus (LP) inputs are processed in parallel in Pyr-type-specific subnetworks (Pyr_←ORBvl and Pyr_←LP) and drive mutual inhibition between them via local interneurons. Our study deepens understanding of the top-down modulation mechanisms of visual processing and establishes that V2M and ACA inputs in L6 employ integrated processing distinct from the parallel processing of LP and ORBvl inputs in L5.

A direct interareal feedback-to-feedforward circuit in primate visual cortex

Article Open access 13 August 2021

Corticothalamic feedback sculpts visual spatial integration in mouse thalamus

Article 11 November 2021

Local origin of excitatory–inhibitory tuning equivalence in a cortical network

Article Open access 15 March 2024

Introduction

The visual system is hierarchically organized, with functionally related areas connecting to each other in specific laminar patterns^1,2,3,4. Processing of various dimensions of the complex visual environment, such as color, depth, shape, and motion, occurs in parallel pathways starting from the retina and continues in specialized visual areas, followed by integration in higher-order areas to form a unified perception^{5,6,7,8,9,10,11}. This involves both bottom-up inputs, which flow from the retina to higher-order areas, and top-down modulation, where signals from higher-order areas adapt visual processing to meet the requirement of the current task^5,6,7,8,9,10.

V1 is the initial cortical area for visual information processing; it receives condensed and parallel bottom-up signals from the retino-geniculo-cortical pathways, extracts relevant information, and further elaborates and integrates this information with top-down inputs from higher-order cortical and thalamic areas, contributing to the formation of a unified perceptual experience^5,8,11,12,13. In more detail, thalamocortical (TC) bottom-up inputs from the dorsal lateral geniculate thalamic nucleus (dLGN) preferentially target the medial layer (L4), while corticocortical (CC) and TC top-down inputs target superficial and deep layers in V1^4,14,15,16.

Previous studies have identified multiple higher-order cortical and thalamic areas that provide top-down inputs to V1, such as the secondary visual cortex (V2), posterior parietal cortex (PTLp), retrosplenial cortex (RSP), anterior cingulate cortex (ACA), ventrolateral orbitofrontal cortex (ORBvl), and lateral dorsal (LD) and lateral posterior thalamic nuclei (LP)^4,15,17,18. These top-down inputs convey a rich array of information, including attention^19,20,21, expectation^22,23, perceptual tasks^24,25, and motor commands^26,27.

The signals from these top-down inputs are differentially processed in V1^{2,21,24,28,29,30}. For instance, V2 inputs are known to enhance the accuracy of visual information decoding in V1 neurons without altering the average response³¹. ACA and LP inputs increase V1 neuron response to task-relevant visual information^20,21, whereas ORBvl inputs suppress V1 neuron responses to filter out irrelevant visual information²⁴. The diverse effects of these top-down modulations on visual processing imply sophisticated mechanisms within local circuits of V1, presumably controlled by distinct sets of excitatory and inhibitory neurons across different layers. However, knowledge about how these top-down inputs differentially engage excitatory and inhibitory neurons in V1 is quite limited, particularly with regard to their innervation patterns of neurons in deep layers.

The mouse V1 consists of six distinct layers, with each layer possessing unique connection and functional properties^7,8,13. While the cell-sparse L1 completely lacks excitatory neurons and contains unique inhibitory neurons (L1-INs), all the other layers consist of excitatory Pyrs and three major types of inhibitory neurons (parvalbumin-positive, PV+; somatostatin-positive, SST+; and vasoactive intestinal peptide-positive, VIP+ neurons)^{32,33,34,35,36}. The top-down modulation signals and bottom-up visual signals are processed and integrated into the V1 local circuits, and then transmitted to downstream targets by heterogeneous long-range projection Pyrs, including pyramidal tract (PT) and intratelencephalic (IT) neurons^4,7,37,38. PT neurons mediate subcortical output, while IT neurons mediate intracortical output^4,39,40.

In this study, we profiled the layer- and cell-type-specific innervation patterns of multiple CC and TC top-down inputs in V1, including three CC inputs (V2M, ACA, and ORBvl) and one TC input (LP). We found distinct layer- and cell-type-specific innervation profiles for each top-down input, with profiles partially overlapping in superficial layers, bypassing L4, and clearly segregating in deep layers. Specifically, V2M and ACA inputs preferentially activate L6 Pyrs, while ORBvl and LP inputs activate L5 Pyrs. We also characterized the layer-specificity of top-down inputs on inhibitory neurons, revealing that L1-INs are strongly activated in L1 and that VIP+ neurons are strongly activated in both superficial and deep layers, while PV+ and SST+ neurons are specifically activated in the deep layers. These results provide a valuable resource for the layer- and cell-type-specific organization of top-down inputs in V1. We subsequently investigated how these inputs interact within their strongest receptive layers, L5 and L6. Using independent optogenetic activation on the same brain slice, we found that V2M and ACA inputs converge on L6 Pyrs, whereas ORBvl and LP inputs selectively activate two distinct types of L5 Pyrs: Pyr_←ORBvl and Pyr_←LP neurons, each characterized by specific electrophysiological properties and gene-expression profiles. Retrograde tracing revealed that Pyr_←ORBvl neurons preferentially innervate subcortical areas and Pyr_←LP neurons innervate cortical areas, indicating parallel processing of the ORBvl and LP inputs in Pyr-type-specific subnetworks. We also found that ORBvl and LP inputs drive mutual inhibition, mediated by local inhibitory neurons, between these two subnetworks in L5 of V1. These findings deepen our understanding of neuronal mechanisms of top-down modulation of visual processing by revealing interactions of modulation signals within V1 local circuits.

Results

Systematic characterization of the layer-specificities of CC and TC top-down inputs in V1 Pyrs

Multiple higher-order cortical and thalamic areas directly innervate V1, thus enabling extensive capacity for top-down modulation of visual processing^{4,15,16,17,18}. The axons of CC and TC top-down inputs primarily target superficial and deep layers in V1⁴. Our research group has had a longstanding interest in top-down modulation of visual processing^17,18,21,41; we found it conspicuous that most studies on the neuronal mechanisms of top-down modulation to date have focused on superficial layers (i.e., L1 and L2/3)^{17,21,34,36,42}. We, therefore, undertook a systematic effort to characterize the innervation patterns of multiple CC and TC top-down inputs across all layers in V1, including both superficial and deep layers. Since the cell-sparse L1 completely lacks excitatory neurons, we started by examining the input strengths of V2M, ACA, ORBvl, and LP inputs in Pyrs across L2-L6 of V1.

To optogenetically activate V2M inputs in V1, we injected AAV expressing ChR2 in excitatory neurons (AAV-CaMKIIα-ChR2-EYFP) into the V2M (Fig. 1a, b). Expression of ChR2-EYFP in the V2M resulted in bright axonal fluorescence in L1 and L6 of V1, with L1 and L6 receiving 13 and 50% of the total fluorescent signal, respectively (Fig. 1c). We conducted whole-cell recordings of V1 Pyrs and recorded the monosynaptic excitatory postsynaptic potential (EPSP) elicited by optogenetic activation of V2M axons, while blocking local neuron spikes with TTX⁴³. Consistent with a previous study²⁸, we found that V2M inputs activated Pyrs in L2/3 and L6 of V1 (defined as averaged EPSP amplitude >3 mV), with the strongest input strengths in L6 (Fig. 1d, e, Supplementary Figs. 1–3, and Supplementary Data 1, 2).

**Fig. 1: Characterization of the strengths of CC and TC top-down inputs in V1 Pyrs across layers.**

The distribution patterns of axons from ACA inputs (L1 received 22% and L6 received 49%) were similar to those of V2M inputs, as was their layer-specificity (Fig. 1f–j). Unlike V2M and ACA inputs, the axons from ORBvl inputs ramified densely in L5 (52% in L5), consistent with their strongest input strength in L5 Pyrs (Fig. 1k–o). The axons from LP inputs were primarily distributed in L1, L5, and L6 (15% in L1, 35% in L5, and 38% in L6), with the strongest input strength observed in L5 Pyrs (Fig. 1p–t). LP inputs also activate L6 Pyrs, but with weaker input strength as compared to L5 Pyrs (46% of the input strength in L5 Pyrs) (Fig. 1s, t). Together, when considering the input strengths of top-down inputs across all layers, our results show that the examined CC top-down inputs (V2M, ACA, and ORBvl) and TC top-down inputs (LP) inputs are partially overlapped in L2/3 with relatively weak input strength in L2/3 Pyrs. These inputs then bypass L4 Pyrs before segregating in L5 and L6: V2M and ACA inputs strongly activate L6 Pyrs (defined as averaged EPSP amplitude >7 mV), while ORBvl and LP inputs strongly activate L5 Pyrs.

Layer-specificities of CC and TC top-down inputs in V1 inhibitory neurons

Despite representing a minority of cortical neurons, inhibitory neurons play a critical role in providing rapid and dynamic modulation of the output of excitatory neurons across different layers^{32,34,36,39,44,45}. Besides unique inhibitory neurons in L1 (L1-INs), all the other layers are known to contain three major types of inhibitory neurons (PV+, SST+, and VIP+ neurons)^32,33,35. We next examined the input patterns of multiple CC and TC top-down inputs in each type of inhibitory neurons. L1-INs were identified by their locations. PV+, SST+, and VIP+ neurons were identified by breeding PV-, SST-, or VIP-Cre mice with loxP-flanked tdTomato reporter mice (Ai14 mice, Supplementary Fig. 4). We observed distinct innervation patterns of cortical inhibitory neurons for each of CC and TC top-down input.

Optogenetic activation of V2M inputs resulted in strong activation in L1-INs within L1 and in strong activation in VIP+ and PV+ neurons in L6 (“strong activation” in inhibitory neurons also defined as averaged EPSP amplitude >7 mV) (Fig. 2a–d, Supplementary Fig. 4, and Supplementary Data 3). Activation of ACA inputs resulted in strong activation of L1-INs in L1, VIP+ neurons in L2/3, as well as in VIP+, PV+, and SST+ neurons in L6 (Fig. 2e–h). Notably, ORBvl inputs were weak in all types of inhibitory neurons across layers, although some L5 VIP+, PV+, and SST+ neurons were activated (ranging from 25 to 34% of each recorded inhibitory neuron type) (Fig. 2i–l and Supplementary Fig. 4). LP inputs strongly activated L1-INs in L1, VIP+ neurons from L2 to L6, and PV+ neurons in L5 (Fig. 2m–p).

**Fig. 2: Input strengths of CC and TC top-down inputs in four types of V1 inhibitory neurons across different layers.**

Our profiling data for multiple CC (V2M, ACA, and ORBvl) and TC (LP) top-down inputs show that the innervation patterns of V1 inhibitory neurons are partially overlapped in superficial layers, with V2M, ACA, and LP inputs strongly activated L1-INs in L1, and ACA and LP inputs strongly activated VIP+ neurons in L2/3 (Fig. 2q, r). V2M and ACA inputs bypassed L4, whereas LP inputs strongly activated VIP+ neurons in L4. We also observed that the innervation patterns of top-down inputs were segregated in deep layers. L5 inhibitory neurons (VIP+ and PV+ neurons) were strongly activated by LP inputs exclusively. L6 VIP+ neurons were strongly activated by V2M, ACA, and LP inputs, while L6 PV+ neurons were strongly activated by V2M and ACA inputs, and L6 SST+ neurons were strongly activated specifically by ACA inputs (Fig. 2q, r). Note that the layer-and cell-type specificity of these top-down inputs is evident based on both the raw and normalized input strengths (Supplementary Figs. 5,6 and Supplementary Data 4–7). Thus, our profiling of layer- and cell-type-specific innervation patterns of multiple CC and TC top-down inputs in V1 represents a valuable resource for understanding the neuronal mechanisms of top-down modulation of visual processing. Interestingly, in contrast to similar innervation properties in superficial layers, the examined CC and TC inputs are clearly segregated in deep layers, which opens up the possibility of testing hypotheses about their layer-specific functions.

Integrated processing of V2M and ACA inputs in L6 Pyrs versus parallel processing of LP and ORBvl inputs in two distinct L5 Pyr populations

The examined CC and TC top-down inputs are strongest in the deep layers, with L6 being most responsive to V2M and ACA inputs, while L5 is most responsive to LP and ORBvl inputs. Previous studies have reported that deep-layer Pyrs are a heterogeneous population, with functionally distinct subnetworks^39,46,47,48. To investigate the interactions among these top-down inputs in their most responsive layers, we employed independent optogenetic activation⁴⁹. Specifically, for V2M and ACA inputs, we alternatively activated these inputs using ChR2 and ChrimsonR in the same brain slice and measured their input strengths in each recorded L6 Pyr. In each animal, we injected an AAV expressing ChR2-EYFP in the V2M (AAV-CaMKIIα-ChR2-EYFP) and also injected an AAV expressing ChrimsonR-tdTomato (AAV-hSyn-ChrimsonR-tdTomato) in the ACA (Fig. 3a). Expression of ChR2-EYFP in the V2M and ChrimsonR-tdTomato in the ACA resulted in the expected bright green and red axonal fluorescence in L6 of V1 (Fig. 3b–d).

**Fig. 3: V2M and ACA inputs converge on L6 Pyrs.**

Activation of V2M inputs (10 pulses @ 10 Hz, 488 nm) and ACA inputs (10 pulses @ 10 Hz, 647 nm) elicited robust postsynaptic currents (EPSCs) in L6 Pyrs (Fig. 3e–g). The majority of recorded L6 Pyrs (57%) received inputs from both V2M and ACA, with 40% receiving only V2M inputs and 3% not receiving inputs from either (Fig. 3h). These results show that CC top-down inputs from V2M and ACA converge on the same group of L6 Pyrs, indicating integrated processing of their top-down modulation signals in L6 of V1.

We next examined the innervation patterns of LP and ORBvl inputs in L5 Pyrs. For each animal, we injected an AAV expressing ChrimsonR-tdTomato in the LP (AAV-hSyn-ChrimsonR-tdTomato) and also injected an AAV expressing ChR2-EYFP in the ORBvl (AAV-CaMKIIα-ChR2-EYFP) (Fig. 4a). Expression of ChrimsonR-tdTomato in the LP and ChR2-EYFP in the ORBvl resulted in bright red and green axonal fluorescence in L5 of V1, with LP inputs primarily in upper L5 (red) and ORBvl inputs in lower L5 (green) (Fig. 4b–d). Activation of LP inputs (10 pulses @ 10 Hz, 647 nm) elicited robust postsynaptic currents (EPSCs) in about half of recorded L5 Pyrs (Pyr_←LP neurons); these neurons did not respond to activation of ORBvl inputs (10 pulses @ 10 Hz, 488 nm) (Fig. 4e–g). Intriguingly, when we activated ORBvl inputs, the remaining half of L5 Pyrs displayed robust EPSCs (Pyr_←ORBvl neurons) (Fig. 4e–h). These results indicate that TC and CC top-down inputs from LP and ORBvl employ parallel processing strategies by selectively activating distinct L5 Pyr populations in V1. Thus, our results reveal that examined CC and TC top-down inputs employ distinct processing strategies in their interactions, with integrated processing of V2M and ACA inputs in L6 versus parallel processing of LP and ORBvl inputs in L5.

**Fig. 4: LP and ORBvl inputs target distinct subpopulations of L5 Pyrs.**

Pyr_←LP and Pyr_←ORBvl neurons are distinct L5 Pyr types distinguished by I_h

Given the notable specificity with which L5 Pyrs process LP and ORBvl inputs, we narrowed our focus to exploring the properties of these neurons: Pyr_←LP and Pyr_←ORBvl neurons. Up to this point of the study, we have identified Pyr_←LP and Pyr_←ORBvl neurons based on their input patterns, a relatively cumbersome process requiring virus injection, patch-clamp recording, and optogenetic activation. The intrinsic electrophysiological properties assessed using patch-clamp recording alone are used extensively for classifying different types of neurons, with, for example, hyperpolarization-activated inward currents (I_h) showing informatively varied distributions for distinct neuron types^39,50. During our routine assessment of patch-clamp recordings by calculating access resistance, we introduced a 10-mV hyperpolarization to the recorded neurons. Pyr_←ORBvl neurons exhibited an I_h in response to this 10-mV hyperpolarization, whereas Pyr_←LP neurons lacked this I_h—as indicated by an I_h slope (see Method) close to 0 (I_h slope of Pyr_←LP, −3.1 ± 2.6 pA/s, n = 13 neurons; I_h slope of Pyr_←ORBvl, 82.5 ± 8.6 pA/s, n = 17 neurons; P = 4 × 10⁻⁶, two-sided Wilcoxon rank-sum test). Moreover, bath application of a known antagonist of hyperpolarization-activated cyclic nucleotide-gated channels (HCN channels) (ZD7288⁵¹, 1 μM) completely eliminated the I_h in Pyr_←ORBvl neurons, yet had no effect on I_h-lacking Pyr_←LP neurons in response to the 10-mV hyperpolarization. These findings indicate that the observed I_h in Pyr_←ORBvl neurons is mediated by HCN channels (Fig. 5a, b).

**Fig. 5: An SVM classifier trained with I_h slope can reliably distinguish Pyr_←LP and Pyr_←ORBvl neurons.**

Given the observed differences in the I_h slope between Pyr_←ORBvl and Pyr_←LP neurons, we explored the use of the I_h slope as a criterion to distinguish between these Pyr populations. Specifically, we revisited an aforementioned dataset for individual activation of LP and ORBvl inputs (from Fig. 1k–t) with a large set of L5 Pyrs responding to LP and ORBvl inputs (including 82 Pyr_←LP and 65 Pyr_←ORBvl neurons), recorded with an antagonist of voltage-gated sodium channels (TTX) and an antagonist of voltage-gated potassium channels (4-AP) (Fig. 5c). Control experiments showed TTX and 4-AP treatment had no effect on the I_h slope (Supplementary Fig. 7). Notably, the I_h slope of the Pyr_←ORBvl neurons was significantly steeper than that of Pyr_←LP neurons (P = 7 × 10⁻²⁴, two-sided Wilcoxon rank-sum test; Fig. 5d), indicating that ORBvl and LP inputs selectively activate two types of L5 Pyrs that can be reliably characterized using I_h.

We subsequently used a support vector machine (SVM) approach⁵² for distinguishing Pyr_←LP and Pyr_←ORBvl neurons in the aforementioned dataset. By employing the I_h-slope data from the 82 Pyr_←LP neurons and 65 Pyr_←ORBvl neurons, we trained a linear SVM classifier (“SVM Classifier(I_h)”) to determine the decision boundary of the I_h slope that optimally separates the two L5 Pyr types (Fig. 5e). The performance of the SVM Classifier (I_h) was evaluated using tenfold cross-validation, repeating the entire tenfold process 20 times with random divisions of the data. The detected average accuracy of 88.2% significantly surpassed the chance level of 50% (P = 4 × 10⁻²⁸, two-sided one-sample t-test). The accuracy also exceeded the 95% confidence interval as obtained from classifiers trained on shuffled I_h-slope data (Fig. 5f). Next, we applied the SVM Classifier(I_h) trained with the complete dataset (82 Pyr_←LP neurons and 65 Pyr_←ORBvl neurons) to another dataset (the data from Fig. 4 for alternative activation of LP and ORBvl inputs in the same brain slice) (Fig. 5g). The SVM Classifier(I_h) informatively distinguished Pyr_←LP and Pyr_←ORBvl neurons, achieving 90% accuracy (Fig. 5h, i). Thus, the SVM Classifier(I_h) can effectively distinguish L5 Pyr_←LP and Pyr_←ORBvl neurons based on simple hyperpolarization data obtained through patch-clamp recordings.

Pyr_←LP and Pyr_←ORBvl neurons employ distinct action potential modes and have unique gene-expression profiles

To investigate the functional divergence of Pyr_←LP and Pyr_←ORBvl neurons in the parallel processing of TC and CC top-down inputs, we explored the potential electrophysiological and genetic bases by conducting Patch-seq^53,54, which simultaneously yielded data for the transcriptomes and a variety of electrophysiological properties for the recorded neurons (Fig. 6a and Supplementary Figs. 8, 9). In total, we carried out patch-clamp recordings on 91 Pyrs in L5 of V1 from 15 mice. Among these neurons, 47 were identified as Pyr_←LP neurons, and 44 were identified as Pyr_←ORBvl neurons using the SVM Classifier(I_h).

**Fig. 6: Pyr_←LP and Pyr_←ORBvl neurons exhibit distinct electrophysiological properties and gene-expression profiles.**

Regarding their distinct action potential (AP) characteristics, Pyr_←LP neurons had significantly higher AP thresholds than Pyr_←ORBvl neurons (Supplementary Fig. 8). All recorded Pyr_←LP neurons exhibited regular spiking (RS) in response to a near-threshold depolarizing current pulse injected into the cell bodies (47 out of 47, 100%), whereas most of the Pyr_←ORBvl neurons displayed bursting spiking (BS, 29 out of 44, 66%, Fig. 6b, c). Additionally, Pyr_←ORBvl neurons had significantly larger depolarizing afterpotential (DAP) following a single AP as compared to Pyr_←LP neurons, in which DAP were largely absent (Fig. 6d). These findings indicate that Pyr_←LP and Pyr_←ORBvl neurons employ distinct AP modes, indicating that they could convey distinct signals in the parallel processing of LP and ORBvl inputs.

For the gene-expression profiles from the Patch-seq, 80 out of 91 L5 Pyrs passed the quality control (see Methods), including 41 Pyr_←LP neurons and 39 Pyr_←ORBvl neurons. The mean number of detectable genes per cell was 3247 ± 193 in Pyr_←LP neurons and 4013 ± 189 in Pyr_←ORBvl neurons (Supplementary Data 8–10). We initially assessed the specificity of our sampling for the two neuron types based on the expression levels of a set of known markers^48,55. As expected, we found that both Pyr_←LP and Pyr_←ORBvl neurons expressed pan-neuronal genes, such as Snap25 (synaptosome-associated protein 25), and excitatory neuron genes, such as Slc17a7 (vesicular glutamate transporter 1) (Supplementary Fig. 9). Moreover, and again supporting the specificity of sampling, both neuron types expressed a known L5 marker gene Fezf2 (FEZ family zinc finger 2) (Supplementary Fig. 9).

There were 67 differentially expressed genes (DEGs) between Pyr_←LP neurons and Pyr_←ORBvl neurons: 18 with elevated expression in Pyr_←LP neurons and 49 with elevated expression in Pyr_←ORBvl neurons (Fig. 6e, Supplementary Fig. 9, and Supplementary Data 11). Recalling the known role of HCN proteins in generating I_h, it was unsurprising to note that expression of Hcn1 was significantly higher in Pyr_←ORBvl neurons (q value = 0.03, Log2FC = 3.8). The elevation of the potassium channels Kcna6 (potassium voltage-gated channel subfamily A member 6, also known as Kv1.6) and Kcnk2 (potassium two-pore domain channel subfamily K member 2) in Pyr_←LP neurons, along with the elevation of Kcnt1 (potassium sodium-activated channel subfamily T member 1, also known as K_Na1.1) in Pyr_←ORBvl neurons, provides a basis for understanding the observed differences in electrophysiological properties. This is given the known roles of KCNA6, KCNK2, and KCNT1 proteins in repolarizing the membrane potential after an AP and in setting the resting membrane potential^56,57,58,59. Moreover, the elevation of calcium channel Cacna1h (calcium voltage-gated channel subunit alpha1 H, which encodes Cav3.2 channels) in Pyr_←ORBvl neurons, compared to Pyr_←LP neurons, may contribute to generate burst spiking, given the known role of Cav3.2 channels in producing low-threshold calcium spikes⁶⁰. Thus, our Patch-seq analysis characterized the distinctive AP modes used by these two L5 Pyr types and identified multiple DEGs with known functions relevant to their functional divergence.

Pyr_←LP and Pyr_←ORBvl neurons, respectively mediate intracortical and subcortical output channels

L5 of V1 contains two major Pyr types: PT and IT neurons, mediating distinct output channels^4,37. IT neurons preferentially targeting cortical areas and PT neurons targeting subcortical areas³⁹. Both IT and PT neurons target the striatum, but IT neurons target the striatum in both hemispheres, whereas PT neurons only target the ipsilateral striatum⁴⁰. Our results indicate that the LP and ORBvl inputs undergo parallel processing through two distinct Pyr types in L5 of V1. Seeking a deeper understanding of information processing as carried out by these two Pyr types, we next investigated the long-range output targets of Pyr_←LP and Pyr_←ORBvl neurons. We used retrograde AAV to label V1 long-range projecting neurons targeting cortical and subcortical areas, and then conducted patch-clamp recordings on the labeled L5 Pyrs (Fig. 7a). These neurons were next classified using the aforementioned SVM Classifier(I_h).

**Fig. 7: Pyr_←LP and Pyr_←ORBvl neurons exhibit distinct long-range output patterns.**

Retrograde AAV expressing Cre (Retro-AAV-hSyn-Cre) was injected into nine distinct cortical and subcortical downstream targets of V1^4,16,18 in loxP-flanked tdTomato reporter mice (Ai14 mice). The downstream cortical areas included the V2M, RSP, ACA, and ORBvl, while the subcortical areas included the ipsilateral and contralateral dorsal medial striatum (ipsi-DMS and contra-DMS), Pons, Superior Colliculus (SC), and LP. We observed distinct laminar distribution specificity among retrograde-labeled V1 neurons from various downstream targets. Neurons retrograde-labeled from cortical areas and the striatum were often distributed across both superficial (L2/3) and deep layers (L5 and L6), whereas neurons labeled from other subcortical areas (Pons, SC, and LP) exhibited a dense distribution specifically in L5 (Supplementary Fig. 10). Generally, retrograde-labeled neurons were observed in L5 (Fig. 7b), with the exception of ACA-projecting neurons, which were exclusively distributed in L2/3 and L6 (Supplementary Fig. 10).

Using the SVM Classifier(Ih), we determined that more than 78% of labeled L5 Pyrs projecting to cortical areas (V2M, RSP, and ORBvl) and contra-DMS were classified as Pyr_←LP neurons. In contrast, more than 82% of labeled L5 Pyrs projecting to other subcortical areas (Pons, SC, and LP) were classified as Pyr_←ORBvl neurons (Fig. 7c and Supplementary Fig. 11). Additionally, labeled L5 Pyrs from ipsi-DMS contained 27% Pyr_←ORBvl neurons and 73% Pyr_←LP neurons. Notably, we further conducted independent optogenetic activation experiments on L5 Pyrs projecting to a subset of these downstream targets, including the RSP, ipsi-DMS, and SC. These experiments confirmed that the recorded L5 projecting Pyrs exclusively receive either LP or ORBvl inputs, a selectivity pattern accurately predicted by the SVM Classifier(I_h) (Supplementary Fig. 12). Specifically, we observed that 100% of L5 RSP-projecting Pyrs receive LP inputs, 100% of SC-projecting Pyrs receive ORBvl inputs, and among ipsi-DMS-projecting Pyrs 47% receive LP inputs while 53% receive ORBvl inputs. The SVM Classifier(I_h) effectively distinguished Pyr_←LP and Pyr_←ORBvl neurons across these groups with 100% accuracy.

These results indicate that Pyr_←LP neurons preferentially innervate cortical areas and the striatum, whereas Pyr_←ORBvl neurons innervate subcortical areas, findings consistent with the innervation preferences of IT and PT neurons. Recent studies have shown that IT and PT neurons form parallel subnetworks for cortical processing, characterized by distinct spatiotemporal activity patterns based on behavioral state, as a basic feature of cortical functional architecture across dorsal cortical areas^61,62. Together, our results indicate that LP and ORBvl inputs undergo parallel processing within the IT and PT subnetworks, which separately mediate intracortical and subcortical output channels.

LP and ORBvl inputs drive mutual inhibition between two Pyr-type-specific subnetworks in L5 of V1

Recent studies have suggested that the functional integration of IT and PT subnetworks across the dorsal cortex might be dynamically gated by inhibitory and modulatory mechanisms based on brain states and behavioral demands^61,62. To explore the interaction between distinct Pyr-type-specific subnetworks in L5 of V1 while processing LP and ORBvl inputs, we optogenetically activated each of these inputs and measured the evoked excitatory and inhibitory postsynaptic currents (EPSCs and IPSCs) in Pyr_←LP and Pyr_←ORBvl neurons.

To activate LP inputs in V1, we injected an AAV expressing ChR2 in excitatory neurons (AAV-CaMKIIα-ChR2-EYFP) into the LP (Fig. 8a). Activation of LP inputs induced both EPSCs and IPSCs in Pyr_←LP neurons (Fig. 8b, left). EPSCs had short onset latencies (4.5 ± 0.3 ms, mean ± SEM, n = 20 neurons), suggesting monosynaptic excitatory inputs. Notably, the latencies of IPSCs were significantly longer than for EPSCs (10.4 ± 1.1 ms, P = 6 × 10⁻⁶, two-sided Wilcoxon sign-rank test), suggesting multisynaptic inhibition. We observed a net inhibition in Pyr_←ORBvl neurons induced by LP inputs (IPSC, latency 10.9 ± 0.6 ms, n = 26 neurons, Fig. 8b, right). LP inputs drive significantly stronger inhibition relative to excitation in Pyr_←ORBvl neurons than in Pyr_←LP neurons, as shown by the I/(I + E) ratio (Fig. 8c and Supplementary Fig. 13). This suggests that LP inputs drive simultaneous activation of Pyr_←LP neurons and inhibition of Pyr_←ORBvl neurons.

**Fig. 8: LP and ORBvl inputs drive mutual inhibition between Pyr_←LP and Pyr_←ORBvl neurons in L5 of V1.**

We also investigated the involvement of local GABAergic interneurons in LP-input-induced inhibition. Initially, we compared LP-input strengths on L5 Pyr_←LP neurons with those on major types of inhibitory neurons (PV+, SST+, and VIP+ neurons) across different layers. LP inputs strongly activated VIP+ neurons in L2-L6 and PV+ neurons in L5, doing so with comparable input strengths to Pyr_←LP neurons (Fig. 8d). Subsequently, we examined how each inhibitory neuron type inhibited Pyr_←LP and Pyr_←ORBvl neurons. Optogenetic activation of PV+ and SST+ neurons induced large IPSCs in both Pyr_←LP and Pyr_←ORBvl neurons, whereas activation of VIP+ neurons evoked only very weak IPSCs in both types, each constituting less than 8% of the IPSCs induced by PV+ neuron in Pyr_←ORBvl neurons (Supplementary Fig. 14). These results indicate that LP inputs can directly engage L5 PV+ neurons to inhibit Pyrs.

We next measured responses in Pyr_←LP and Pyr_←ORBvl neurons induced by ORBvl inputs (Fig. 8e and Supplementary Fig. 13). Activation of ORBvl inputs induced monosynaptic EPSCs and multisynaptic IPSCs in Pyr_←ORBvl neurons (EPSC latency, 3.7 ± 0.3 ms; IPSC latency, 6.6 ± 0.4 ms, mean ± SEM; n = 20 neurons, Fig. 8f). In Pyr_←LP neurons, only multisynaptic IPSCs were observed (IPSC latency, 7.2 ± 0.3 ms, mean ± SEM, n = 19 neurons). In contrast to LP inputs, ORBvl inputs drive significantly stronger inhibition relative to excitation in Pyr_←LP neurons than in Pyr_←ORBvl neurons (Fig. 8g), indicating that ORBvl inputs may simultaneously drive activation of Pyr_←ORBvl neurons and inhibition of Pyr_←LP neurons. ORBvl inputs were relatively weak in local inhibitory neurons, with L5 inhibitory neurons receiving the strongest ORBvl inputs (~20% of the input strength in Pyr_←ORBvl neurons, Fig. 8h). Note that our optogenetic activation profiling of ORBvl inputs revealed activation in 33% of L5 PV+ neurons and 25% of L5 SST+ neurons (Supplementary Fig. 4); thus, the inhibition induced by ORBvl inputs may be mediated by these inhibitory neurons. Together, these results demonstrate that LP and ORBvl inputs drive mutual inhibition between the two Pyr-type-specific subnetworks in L5 of V1 (likely involving PV+ and SST+ neurons), thus enabling dynamic gating of subnetwork functions.

Discussion

In this study, we characterized the layer- and cell-type-specific organization of multiple CC and TC top-down inputs in V1, as well as their interactions in deep layers. Each top-down input to V1 engaged both excitatory and inhibitory neurons: CC and TC top-down inputs partially overlapped in superficial layers (L1 and L2/3), bypassed the medial layer (L4), and clearly segregated in deep layers (L5 and L6). Our datasets provide a valuable resource for understanding the neuronal mechanisms of top-down modulation of visual information processing. Further, we found that CC top-down inputs from the V2M and ACA undergo integrated processing in L6 Pyrs, whereas TC top-down inputs from the ORBvl and LP undergo parallel processing within two L5 Pyr-type-specific subnetworks, separately mediating subcortical and intracortical V1 output channels. Notably, ORBvl and LP inputs drive mutual inhibition between these two subnetworks in L5 of V1, potentially enabling dynamic gating of subnetwork functions in response to specific top-down modulation signals.

The visual cortex integrates both bottom-up and top-down inputs, contributing to form a unified visual perception^5,8,12,13. Decades of studies have demonstrated the parallel processing of bottom-up inputs, starting at the retina and reaching V1 via retino-geniculo-cortical pathways, then advancing to higher cortical and thalamic areas^5,6,7,8,9,10. Top-down inputs from higher-order cortical and thalamic areas to V1 are known to convey various dimensions of complex behavior-related signals to modulate visual processing^{19,20,21,22,23,24,25,26,27,29,30}, yet the neuronal mechanisms are not yet well understood. Our systematic characterization of the layer- and cell-type-specific organization of CC (V2M, ACA, and ORBvl) and TC (LP) top-down inputs in V1 offers insights into sophisticated top-down modulation mediated by distinct excitatory and inhibitory neurons in V1 local circuits. This understanding should substantially facilitate investigations into how various top-down inputs integrate with bottom-up inputs to generate task-related top-down modulations of visual processing in V1.

CC and TC top-down inputs are known to target both superficial and deep layers in V1^1,3,4. Previous studies have demonstrated that top-down inputs preferentially activate L1-INs and L2/3 VIP+ neurons, which inhibit PV+ and SST+ neurons to disinhibit Pyrs^{17,21,34,36,63,64,65}. Our findings extend this understanding by showing varied cell-type preferences across layers: VIP+ neurons are strongly activated in both superficial and deep layers, whereas PV+ and SST+ neurons are selectively activated in deep layers. VIP+ neurons disinhibit Pyrs, while PV+ and SST+ neurons inhibit Pyrs^{32,44,66,67,68}. The higher density of VIP+ neurons in superficial layers and PV+ and SST+ neurons in deep layers³² suggests a top-down modulation pattern wherein disinhibition dominates in superficial layers, gradually transitioning to increased inhibition in deep layers.

Cortical Pyrs have been divided by axonal projections into three main types: IT neurons preferentially innervating the cortex and striatum; PT neurons innervating subcortical areas; and corticothalamic neurons, which only innervate the thalamus³⁹. These types overlap in deep layers, with IT and PT in L5, and IT and corticothalamic in L6. Our study has revealed the specificity of various CC and TC top-down inputs engaging distinct Pyr types in deep layers of V1. V2M and ACA inputs converge on L6 Pyrs, suggesting integrated processing of these inputs that likely involves both IT and corticothalamic neurons. And we found that ORBvl and LP inputs selectively activate PT (Pyr_←ORBvl) and IT (Pyr_←LP) neurons, each with distinct AP modes (BS spiking vs. RS spiking) and gene-expression profiles, indicating parallel processing of the information they convey in the PT and IT subnetworks.

The dendrites of Pyrs typically extend, and branch across multiple layers^39,44,69,70, and the spatial distribution of specific inputs within dendritic arborization serve distinct functions: spatially clustered co-active synapses are more efficacious in driving postsynaptic neurons than spatially distributed synapses^71,72. Previous studies using subcellular ChR2-assisted circuit mapping (sCRACM) revealed the subcellular organization of local and long-range excitatory inputs to Pyrs in sensorimotor cortices, findings helpful for predicting the interactions among these inputs within the dendrites of a single neuron^43,73. A recent study employing sCRACM has shown that top-down inputs from V2 to V1 selectively modulate apical dendrites, but not basal dendrites of L5/L6 looped IT neurons that project back to V2²⁸. Future studies using sCRACM to investigate the subcellular organization of other top-down inputs in V1—specifically examining whether their synapses are clustered or segregated—are likely to provide valuable insights. Such studies could significantly enhance our understanding of dendritic integration during the interactions among these multiple top-down inputs in V1 neurons.

PT neurons receive extensive inputs from local IT neurons; however, they provide little local feedback, a phenomenon observed in various cortical regions such as the V1, somatosensory, and frontal cortices^39,43,74,75. Moreover, PT neurons in the barrel cortex are known to receive strong bottom-up TC inputs from the VPM, yet receive few top-down TC inputs from the PO^43,76. These findings suggest that PT neurons may act as downstream integrators in local circuits: integrating the information from local IT neurons with bottom-up TC inputs and broadcasting the results to subcortical structures.

Recent behavioral studies suggest an alternative interpretation, proposing that IT and PT neurons form parallel subnetworks capable of independent processing. For instance, only PT neurons are required for tactile or visual perception in sensory cortices, while specific PT neurons in the secondary motor cortex are involved in motor generation^38,77,78,79. Moreover, widefield calcium imaging assessing the spatial and temporal dynamics of Pyr-type-specific subnetworks during sensorimotor tasks has revealed distinct cortex-wide activity patterns for IT and PT neurons^61,62. It is possible that information flow between IT and PT in local circuits is dynamically gated by inhibitory and modulatory mechanisms according to behavioral demand.

Building on this notion, our findings suggest a mechanism for the dynamic gating of information flow between IT and PT subnetworks. We found that LP and ORBvl inputs drive mutual inhibition between IT and PT subnetworks in V1. First, we observed that ORBvl inputs drive multisynaptic inhibition in IT neurons. Second, LP inputs induce multisynaptic inhibition in PT neurons (Pyr_←ORBvl neurons) without multisynaptic excitation from activated IT neurons (Pyr_←LP neurons), suggesting that LP inputs trigger APs in L5 PV+ neurons and lead to subthreshold excitation in IT neurons. Assuming that the multisynaptic inhibition resulting from LP inputs to PT neurons precedes excitation from activated IT neurons, it is plausible that LP inputs suppress PT neuron activity and thus temporally separate the activity of IT and PT neurons.

The neuronal mechanisms of top-down modulation of visual processing are being investigated actively^{19,20,24,25,26,27}, yet the functional characterization of interactions between multiple top-down inputs is still in its infancy. Our study paves the way for future investigations in behaving animals using advanced imaging and optogenetic manipulation to explore how these top-down inputs interact with each other to enable nuanced task-related top-down modulations.

Methods

The names of the companies and the catalog numbers for all reagents used in the study are listed in Supplementary Data 12.

Animals

Animal care and the experimental protocols were approved by the Animal Committee of Shanghai Jiao Tong University School of Medicine and the Animal Committee of the Institute of Neuroscience, Chinese Academy of Sciences. Experiments were performed on wild-type (C57) and transgenic mice. The transgenic mice used were PV-Cre (Jackson lab stock #017320), SST-Cre (#013044), VIP-Cre (#010908), and loxP-flanked-tdTomato (Ai14) (#007914) mice. To visualize the interneurons, PV-, SST-, or VIP-Cre mice were crossed with Ai14 mice. Male and female mice aged 8 to 12 weeks were used. Mice were housed under a 12-h light/dark cycle, with temperatures maintained at 20–24 °C and humidity between 45 to 65%, ensuring optimal welfare and standard laboratory conditions, along with free access to food and water. Our experiments used male and female mice. We found no differences between male and female mice in the layer- and cell-type-specific innervation patterns of CC or TC top-down inputs in V1 or for interactions among these inputs in deep layers. Our experiments used male and female mice. We found no differences between male and female mice in the layer-and cell-type-specific innervation patterns of CC or TC top-down inputs in V1 or for interactions among these inputs in deep layers (Supplementary Figs. 15–18).

Virus

The adeno-associated viruses AAV-DJ-CaMKIIα-ChR2-EYFP (genomic titer, 1 × 10¹² gc/mL) was acquired from the WZ Biosciences. Retro-AAV-hSyn-Cre (5 × 10¹² gc/mL), Retro-AAV-hSyn-mCherry (5 × 10^12;gc/mL), and AAV9-hSyn-ChrimsonR-tdTomato (5 × 10¹² gc/mL) were acquired from Brain VTA. All viral vectors were stored in aliquots at −80 °C until use.

Surgery

Stereotaxic surgeries

Adult mice were anesthetized with isoflurane (5% induction and 1.5% maintenance) and placed on a stereotaxic frame (Ruiwode Life Science). The body temperature was maintained at 37 °C throughout the procedure using a heating pad. Eye ointment was applied to protect the animal’s eyes. After asepsis, the skin was incised to expose the skull, and the overlying connective tissue was carefully removed. A craniotomy with a diameter of ~0.5 mm was made above the injection site. Viruses were loaded in a sharp micropipette mounted on a Nanoject II attached to a micromanipulator and then injected at a speed of 60 nL per min. Coordinates used were as follows: V2M (Bregma, −2.5 mm; lateral, 1.5 mm; depth, 0.5 mm), ACA (Bregma, +0.3 mm; lateral, 0.3 mm; depth, 0.9 mm), ORBvl (Bregma, +2.5 mm; lateral, 1.3 mm; depth, 1.8 mm), LP (Bregma, −2.0 mm; lateral, 1.3 mm; depth, 2.6 mm), RSP (Bregma, −1.8 mm; lateral, 0.3 mm; depth, 0.5 mm), DMS (Bregma, 0.3 mm; lateral, 1.6 mm; depth, 1.8 mm), Pons (Bregma, −3.1 mm; lateral, 1.2 mm; depth, 4.7 mm), and SC (Bregma, −4.2 mm; lateral, 1.0 mm; depth, 1.8 mm).

Optogenetics and slice recording

To examine the input strengths of CC (V2M, ACA, ORBvl) and TC (LP) inputs in V1 Pyrs across layers, we injected an AAV expressing ChR2 in excitatory neurons (AAV-DJ-CaMKIIα-ChR2-EYFP) into the V2M (250 nL), ACA (300 nL), ORBvl (300 nL), and LP (75 nL) of wild-type mice. To examine the input strengths of CC and TC inputs in V1 inhibitory neurons, we injected AAV-DJ-CaMKIIα-ChR2-EYFP in the V2M, ACA, ORBvl, and LP of PV-, SST-, and VIP-Ai14 mice. To independently activate V2M and ACA inputs in the same animal, we injected an AAV expressing ChR2-EYFP in neurons in the V2M (AAV-CaMKIIα-ChR2-EYFP, 250 nL) and also injected an AAV expressing ChrimsonR-tdTomato in excitatory neurons in the ACA (AAV-hSyn-ChrimsonR-tdTomato, 300 nL) of wild-type mice. To independently activate ORBvl and LP inputs in the same animal, we injected an AAV expressing ChrimsonR-tdTomato in neurons in the LP (AAV-hSyn-ChrimsonR-tdTomato, 75 nL) and also injected an AAV expressing ChR2-EYFP in excitatory neurons in the ORBvl (AAV-CaMKIIα-ChR2-EYFP, 300 nL) of wild-type mice. To independently activate ORBvl and LP inputs and examine their input strengths in retrogradely labeled V1 projecting Pyrs, we injected an AAV expressing ChrimsonR-tdTomato in neurons in the LP (AAV-hSyn-ChrimsonR-tdTomato, 75 nL), injected an AAV expressing ChR2-EYFP in excitatory neurons in the ORBvl (AAV-CaMKIIα-ChR2-EYFP, 300 nL), and also injected a Retro-AAV expressing mCherry (Retro-AAV-hSyn-mCherry) into the RSP (300 nL), ipsi-DMS (300 nL), or SC (300 nL) of wild-type mice. To examine the EPSCs and IPSCs in L5 Pyrs elicited by ORBvl and LP inputs, we injected an AAV expressing ChR2 in excitatory neurons (AAV-DJ-CaMKIIα-ChR2-EYFP) into the ORBvl (300 nL) and LP (75 nL) of wild-type mice.

Retro-AAV-mediated retrograde tracing

For retrograde tracing from cortical and subcortical areas, we injected a retro-AAV expressing Cre (Retro-AAV-hSyn-Cre) into the V2M (100 nL), RSP (200 nL), ACA (300 nL), ORBvl (300 nL), ipsi-DMS (200 nL), contra-DMS (200 nL), Pons (100 nL), SC (100 nL), and LP (50 nL) of Ai14 mice. The histology experiments were performed 4 weeks after retro-AAV injection.

In vitro electrophysiology

Slice preparation

Mice were anesthetized with 5% isoflurane. After decapitation, the brain was quickly dissected and immediately placed in ice-cold oxygenated NMDG-HEPES solution (in mM: NMDG 93, KCl 2.5, NaH₂PO₄ 1.2, NaHCO₃ 30, HEPES 20, glucose 25, sodium ascorbate 5, thiourea 2, sodium pyruvate 3, MgSO₄.7H₂O 10, CaCl₂.2H₂O 0.5 and NAC 12, at pH 7.4, adjusted with HCl), and coronal sections of brain slices were made with a vibratome. Slices (250-μm thick) were recovered in oxygenated NMDG-HEPES solution at 32 °C for 10 min and then maintained in an incubation chamber with oxygenated standard artificial cerebrospinal fluid (ACSF) (in mM: NaCl 125, KCl 3, CaCl₂ 2, MgCl₂ 1, NaH₂PO₄ 1.25, sodium ascorbate 1.3, NaHCO₃ 26, glucose 10) at 30 °C for 1–4 h before recording.

Whole-cell recording

Whole-cell recordings were made at 30 °C in oxygenated standard ACSF. To examine the monosynaptic input strength of each CC and TC input in V1 neurons, EPSPs were recorded using a potassium-based internal solution (in mM: K-gluconate 135, KCl 5, HEPES 10, EGTA 0.3, MgATP 4, Na₂GTP 0.3, and Na₂-phosphocreatine 10, at pH 7.3, adjusted with KOH, 290–300 mOsm). TTX (1 μM) and 4-aminopyridine (100 μM) were bath applied to block action potentials and permit direct depolarization of axon terminals by ChR2 activation with 5-ms pulses of blue light⁴³. In experiments where LP and ORBvl inputs were activated alternately, EPSCs were also recorded using this potassium-based internal solution.

To examine the EPSCs and IPSCs in L5 Pyrs evoked by LP and ORBvl inputs, we used a cesium-based internal solution (in mM: CsMeSO₄ 125, CsCl 2, HEPES 10, EGTA 0.5, MgATP 4, Na₂GTP 0.3, Na₂-phosphocreatine 10, TEACl 5, QX-314 3.5, at pH 7.3, adjusted with CsOH, 290–300 mOsm). EPSCs and IPSCs were isolated by clamping the membrane potential of the recorded neuron at the reversal potential of inhibitory and excitatory synaptic currents, respectively. To measure the inhibition on L5 Pyrs induced by various cortical inhibitory neuron types, IPSCs were also recorded using this cesium-based internal solution.

The resistance of the patch pipette was 3–5 MΩ. The cells were excluded if the access resistance exceeded 40 MΩ or varied by more than 20% during the recording period. Data were recorded with a Multiclamp 700B amplifier (Axon Instruments), filtered at 2 kHz, and digitized with a Digidata 1550B (Axon Instruments) at 10 kHz. Recordings were analyzed using custom software.

After completion of the electrophysiological recordings, the vertical cell depth from the pial surface was measured. Based on the measured distance, we assigned the recorded neurons to specific cortical layers. The layer boundaries were determined based on Allen Mouse Brain Atlas⁸⁰.

For the experiments in Figs. 1, 2, we recorded neurons simultaneously in each layer in at least five mice in each group. For the majority of recorded animals, we recorded neurons simultaneously in three or more layers. The numbers of mice recorded with different numbers of layers in Fig. 1 are listed in Supplementary Data 13.

Data analysis

The resting membrane potential (RMP) was recorded soon after break-in at I = 0. During voltage-clamp recordings (holding at −70 mV), a −10 mV voltage step lasting 300 ms was applied to compute both the input resistance (Rin) and the access resistance (Ra). The initial transient current response to this voltage step (ΔI_transient) was used to derive the Ra, while the steady-state current that followed the transient (ΔI_steady) was used to derive the Rin. Both resistances were determined using Ohm’s law: Ra = ΔV/ΔI_transient; Rin = ΔV/ΔI_steady, where ΔV indicates the −10 mV voltage change. During this −10 mV voltage step, we observed the I_h. The slope of I_h was calculated as I_h slope = ΔI_5–265/Δt. Where ΔI_5–265 is the change in current from 5 to 265 ms during the −10 mV voltage step, and Δt is its duration (260 ms).

During current clamp recordings, we analyzed the properties of the action potential (AP). The instantaneous firing rate was calculated as the reciprocal of the inter-spike interval (ISI) of the first two spikes observed during a series of depolarizing current steps (25 pA/step, each lasting 500 ms). Regular spiking was defined as generating single APs in response to near-threshold depolarization, whereas burst spiking was defined as generating a complex of two or more APs in response to near-threshold depolarization. Single AP induced by a brief depolarizing current step (2 pA, 25 ms) was analyzed for AP threshold and depolarizing afterpolarization (DAP). The AP threshold was identified as the voltage where there was a sharp increase in the rate of voltage change. The membrane voltages were plotted against their first-time derivative (dV/dt, phase-plane plot), and the AP threshold was selected as the voltage at which dV/dt exceeded three times the standard deviation of all the preceding data points. The amplitude of DAP was calculated as the peak amplitude of DAP relative to the minimum value of the fast afterhyperpolarizations⁸¹. In cases where DAPs were absent, we marked them with an amplitude of 0, as described by Hattox and Nelson (2007).

Optogenetic manipulation

For optogenetic activation of axons from CC and TC top-down inputs in V1, we used an X-cite LED (Lumen Dynamics Group) controlled by a stimulator (Master8). We activated ChR2 using blue light bandpass filtered at 419–465 nm (Semrock) and ChrimsonR with red light filtered at 610–650 nm (Semrock). This filtered light was delivered through a 40× 0.8 NA water immersion lens. To systematically characterize CC and TC top-down inputs in V1 neurons, we delivered pulse trains of blue light (3.5 mW, 10 Hz, 5 ms). For independent activation of ORBvl and LP inputs using ChR2 and ChrimsonR within the same brain slice, we alternated between pulse trains of blue light (3.5 mW, 10 Hz, 1 ms) and red light (3.5 mW, 10 Hz, 1 ms). For measuring EPSCs and IPSCs in L5 Pyrs driven by ORBvl and LP inputs, we used pulse trains of blue light (3.5 mW, 10 Hz, 5 ms).

Histology and anatomical data analysis

To visualize the distribution of axons from CC and TC top-down inputs in fluorescence images, we used the following procedures: Mice were deeply anesthetized using isoflurane and promptly perfused with chilled 0.1 M phosphate-buffered saline (PBS), followed by a 4% paraformaldehyde (w/v) solution in PBS. The brain was then extracted and postfixed in the same solution overnight at 4 °C. After fixation, the brain was immersed in a 30% sucrose (w/v) solution in PBS and kept for 1–2 days at 4 °C. Once embedded and frozen, the brain was sectioned into 40-μm-thick coronal slices with a cryostat. After preparing the slices, they were rehydrated in PBS for 30 min and mounted using VECTASHIELD mounting medium infused with 4’,6-diamidino-2-phenylindole (DAPI). Imaging was carried out using high-throughput slide scanners (VS120, Olympus) for comprehensive visualization and subsequent analysis. In addition, select representative slices were imaged using a confocal microscope (Olympus FV-3000). The same procedures were also used to visualize the distribution of V1 neurons retrogradely labeled from various cortical and subcortical downstream targets.

For the quantification of axon intensity in V1 shown in Figs. 1, 3 animals were analyzed for each input (two slices per animal). Vertical fluorescence profiles of EYFP- or tdTomato-expressing axons were measured using ImageJ after subtracting background fluorescence from a hippocampal area devoid of labeled axons.

To visualize the recorded Pyr_←ORBvl and Pyr_←LP neurons, we used the following procedures: After whole-cell recording, slices with cells filled with biocytin were fixed in 4% paraformaldehyde for 1 h at room temperature. Slices were then rinsed in 0.01 M PBS three times, and transferred to 0.5% Triton X-100 for 5 min and then incubated in a blocking solution (5% BSA in PBS) for 1 h. After that, slices were incubated with streptavidin (1:1000, DyLight 549 Streptavidin, Cat. No: SA-5549, Vector Laboratories, Burlingame, CA) overnight at 4 °C. For 3D reconstruction of labeled cells, z-stack images (0.5 μm per image) were acquired with a 20× air objective on a confocal microscope (Olympus FV-3000) and processed using ImageJ.

Classification based on SVM classifier

SVM classification is implemented with the fitcsvm function in the MATLAB statistics toolbox using a linear kernel and a uniform prior. We used the I_h-slope data of 82 Pyr_←LP and 65 Pyr_←ORBvl neurons, derived from a dataset of individual activation of LP and ORBvl inputs (from Fig. 1k–t), to train the “SVM Classifier(I_h)”. The classifier performance was evaluated via tenfold cross-validation using the same I_h-slope data. The classifier performance is also tested on 1000 instances of shuffled labels to obtain the 95% confidence interval. We then used the SVM Classifier(I_h) to classify Pyr_←LP and Pyr_←ORBvl neurons based on their I_h slope in subsequent experiments. The prediction accuracy of the SVM Classifier(I_h) is calculated as the number of correct predictions divided by the total number of predictions.

Patch-seq recording and RNA sequencing of patched cells

Patch-seq recording was performed following the method described by ref. ⁵³. Briefly, following electrophysiological recording, single L5 Pyrs were collected immediately by applying light negative pressure through the same glass patch pipette. Special care was taken to maintain the integrity of the seal between the pipette and the cell membrane throughout the procedure to prevent contamination from the extracellular environment. Only cells whose entire somatic compartment (including their nucleus) were visibly aspirated into the micropipette were further processed.

Following successful aspiration, the contents of the pipette were immediately ejected into an RNase-free PCR tube containing 4 μL lysis buffer and stored immediately on dry ice until −80 °C storage. cDNA libraries were produced using a Smart-seq2-based Single Cell Full-Length mRNA-Amplification Kit (N712, Vazyme) according to the manufacturer’s instructions. The size distributions and concentrations of the cDNA libraries were assessed using an Agilent Bioanalyzer 2100. Samples with less than 1 ng of total cDNA in the final volume of 15 μL, or with an average size smaller than 1500 bp, were not sequenced. To construct the final sequencing libraries, 1 ng of purified cDNA from each sample was tagmented using an Illumina Nextera XT Library Preparation kit (TD503 Vazyme). The DNA was sequenced from paired ends (150 bp) with standard Illumina Nextera i5 and i7 index primers (8 bp each) using an Illumina Hiseq xten instrument.

50-bp paired-end reads were aligned to GRCm38 (mm10) using a RefSeq annotation gff file (https://www.ncbi.nlm.nih.gov/genome/annotation_euk/all). Sequence alignment was performed using STAR v2.5.3⁸² in two pass mode. PCR duplicates were masked and removed using the STAR option ‘bamRemoveDuplicates’. Only uniquely aligned reads were used for gene quantification. Gene counts were computed using the R Genomic Alignments package⁸³.

Transcriptome data analysis

A total of 91 neurons were sequenced in our study. We excluded four cells as the sum of counts across all genes in each of these cells was below 500. Following this, we log-transformed all counts using a log2(x + 1) transformation⁵⁴ and initially assessed the specificity of our sampling based on the expression levels of pan-neuronal genes (Actb, Syt4, Atp1a3, Syt1, Calm1, Arpp21, and Snap25), cortical excitatory neuron genes (Gria2, Camk2a, and Slc17a7), GABAergic inhibitory neuron genes (Slc32a1, Gad1, and Gad2) and glial marker genes (Adarb2, Gfap, Aqp4, and Mlc1) in single neuronal transcriptomes of all neurons (Supplementary Fig. 9). An additional seven cells were excluded due to high levels of inhibitory neuron genes. This was defined as the sum of the log2(x + 1) value of the three inhibitory neuron marker genes exceeding 7.

The raw read count table was then used to identify the differentially expressed genes between the Pyr_←ORBvl and Pyr_←LP neurons using the DESsq2 v.1.40.2 R package⁸⁴. We excluded genes expressed in fewer than ten cells (count 5). Genes with q values < 0.05 and either FC > 2 or FC < 0.5 were considered as differentially expressed genes (DEGs).

Quantification and statistical analysis

Statistical analysis was performed using MATLAB, SPSS, and R. The selection of statistical tests was based on previous studies. All statistical tests were two-sided. The exact number of mice and recorded cells were described in figure legends and source data. Statistical method, statistics, and corresponding P values were reported in the figure legends and Supplementary Data.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

Data availability

The data supporting the findings of this study are included in the figures and supporting files. The sequencing data generated in this study have been deposited in the Gene-Expression Omnibus repository under the accession code GSE246589. Source data are provided with this paper.

References

Coogan, T. A. & Burkhalter, A. Hierarchical organization of areas in rat visual cortex. J. Neurosci. 13, 3749–3772 (1993).
Article CAS PubMed PubMed Central Google Scholar
D’Souza, R. D., Meier, A. M., Bista, P., Wang, Q. & Burkhalter, A. Recruitment of inhibition and excitation across mouse visual cortex depends on the hierarchy of interconnecting areas. Elife 5, 19332 (2016).
Article Google Scholar
Felleman, D. J. & Van Essen, D. C. Distributed hierarchical processing in the primate cerebral cortex. Cereb. Cortex 1, 1–47 (1991).
Article CAS PubMed Google Scholar
Harris, J. A. et al. Hierarchical organization of cortical and thalamic connectivity. Nature 575, 195–202 (2019).
Article ADS CAS PubMed PubMed Central Google Scholar
Glickfeld, L. L. & Olsen, S. R. Higher-order areas of the mouse visual cortex. Annu. Rev. Vis. Sci. 3, 251–273 (2017).
Article PubMed Google Scholar
Livingstone, M. & Hubel, D. Segregation of form, color, movement, and depth: anatomy, physiology, and perception. Science 240, 740–749 (1988).
Article ADS CAS PubMed Google Scholar
Nassi, J. J. & Callaway, E. M. Parallel processing strategies of the primate visual system. Nat. Rev. Neurosci. 10, 360–372 (2009).
Article CAS PubMed PubMed Central Google Scholar
Seabrook, T. A., Burbridge, T. J., Crair, M. C. & Huberman, A. D. Architecture, function, and assembly of the mouse visual system. Annu. Rev. Neurosci. 40, 499–538 (2017).
Article CAS PubMed Google Scholar
Shapley, R. Visual sensitivity and parallel retinocortical channels. Annu. Rev. Psychol. 41, 635–658 (1990).
Article CAS PubMed Google Scholar
Wassle, H. Parallel processing in the mammalian retina. Nat. Rev. Neurosci. 5, 747–757 (2004).
Article PubMed Google Scholar
Ibbotson, M. R. & Meffin, H. The Senses: A Comprehensive Reference (Elsevier, 2020).
Froudarakis, E. et al. The visual cortex in context. Annu. Rev. Vis. Sci. 5, 317–339 (2019).
Article PubMed PubMed Central Google Scholar
Niell, C. M. & Scanziani, M. How cortical circuits implement cortical computations: mouse visual cortex as a model. Annu. Rev. Neurosci. 44, 517–546 (2021).
Article CAS PubMed PubMed Central Google Scholar
Hunnicutt, B. J. et al. A comprehensive thalamocortical projection map at the mesoscopic level. Nat. Neurosci. 17, 1276–1285 (2014).
Article CAS PubMed PubMed Central Google Scholar
Oh, S. W. et al. A mesoscale connectome of the mouse brain. Nature 508, 207–214 (2014).
Article ADS CAS PubMed PubMed Central Google Scholar
Zingg, B. et al. Neural networks of the mouse neocortex. Cell 156, 1096–1111 (2014).
Article CAS PubMed PubMed Central Google Scholar
Ma, G. et al. Hierarchy in sensory processing reflected by innervation balance on cortical interneurons. Sci. Adv. 7, abf5676 (2021).
Article ADS Google Scholar
Zhang, S. et al. Organization of long-range inputs and outputs of frontal cortex for top-down control. Nat. Neurosci. 19, 1733–1742 (2016).
Article CAS PubMed PubMed Central Google Scholar
Debes, S. R. & Dragoi, V. Suppressing feedback signals to visual cortex abolishes attentional modulation. Science 379, 468–473 (2023).
Article ADS CAS PubMed Google Scholar
Hu, F. et al. Prefrontal corticotectal neurons enhance visual processing through the superior colliculus and pulvinar thalamus. Neuron 104, 1141–1152 (2019).
Article PubMed Google Scholar
Zhang, S. et al. Selective attention. Long-range and local circuits for top-down modulation of visual cortex processing. Science 345, 660–665 (2014).
Article ADS CAS PubMed PubMed Central Google Scholar
Fiser, A. et al. Experience-dependent spatial expectations in mouse visual cortex. Nat. Neurosci. 19, 1658–1664 (2016).
Article CAS PubMed Google Scholar
Leinweber, M., Ward, D. R., Sobczak, J. M., Attinger, A. & Keller, G. B. A sensorimotor circuit in mouse cortex for visual flow predictions. Neuron 95, 1420–1432 (2017).
Article CAS PubMed Google Scholar
Liu, D. et al. Orbitofrontal control of visual cortex gain promotes visual associative learning. Nat. Commun. 11, 2784 (2020).
Article ADS CAS PubMed PubMed Central Google Scholar
Norman, K. J. et al. Post-error recruitment of frontal sensory cortical projections promotes attention in mice. Neuron 109, 1202–1213 (2021).
Article CAS PubMed PubMed Central Google Scholar
Huda, R. et al. Distinct prefrontal top-down circuits differentially modulate sensorimotor behavior. Nat. Commun. 11, 6007 (2020).
Article ADS CAS PubMed PubMed Central Google Scholar
Kim, J. H., Ma, D. H., Jung, E., Choi, I. & Lee, S. H. Gated feedforward inhibition in the frontal cortex releases goal-directed action. Nat. Neurosci. 24, 1452–1464 (2021).
Article CAS PubMed Google Scholar
Young, H., Belbut, B., Baeta, M. & Petreanu, L. Laminar-specific cortico-cortical loops in mouse visual cortex. Elife 10, e59551 (2021).
Article CAS PubMed PubMed Central Google Scholar
Gilbert, C. D. & Li, W. Top-down influences on visual processing. Annu. Rev. Neurosci. 14, 350–363 (2013).
CAS Google Scholar
Moore, T. & Zirnsak, M. Neural mechanisms of selective visual attention. Annu. Rev. Psychol. 68, 47–72 (2017).
Article CAS PubMed Google Scholar
Javadzadeh, M. & Hofer, S. B. Dynamic causal communication channels between neocortical areas. Neuron 110, 2470–2483 (2022).
Article CAS PubMed PubMed Central Google Scholar
Kim, Y. et al. Brain-wide maps reveal stereotyped cell-type-based cortical architecture and subcortical sexual dimorphism. Cell 171, 456–469 (2017).
Article CAS PubMed PubMed Central Google Scholar
Paul, A. et al. Transcriptional architecture of synaptic communication delineates GABAergic neuron identity. Cell 171, 522–539 (2017).
Article CAS PubMed PubMed Central Google Scholar
Schuman, B., Dellal, S., Prönneke, A., Machold, R. & Rudy, B. Neocortical layer 1: an elegant solution to top-down and bottom-up integration. Annu. Rev. Neurosci. 44, 221–252 (2021).
Article CAS PubMed PubMed Central Google Scholar
Tasic, B. et al. Adult mouse cortical cell taxonomy revealed by single cell transcriptomics. Nat. Neurosci. 19, 335–346 (2016).
Article CAS PubMed PubMed Central Google Scholar
Pardi, M. B., Schroeder, A. & Letzkus, J. J. Probing top-down information in neocortical layer 1. Trends Neurosci. 46, 20–31 (2023).
Article CAS PubMed Google Scholar
Kim, E. J., Juavinett, A. L., Kyubwa, E. M., Jacobs, M. W. & Callaway, E. M. Three types of cortical layer 5 neurons that differ in brain-wide connectivity and function. Neuron 88, 1253–1267 (2015).
Article CAS PubMed PubMed Central Google Scholar
Tang, L. & Higley, M. J. Layer 5 circuits in V1 differentially control visuomotor behavior. Neuron 105, 346–354 (2020).
Article CAS PubMed Google Scholar
Harris, K. D. & Shepherd, G. M. The neocortical circuit: themes and variations. Nat. Neurosci. 18, 170–181 (2015).
Article CAS PubMed PubMed Central Google Scholar
Shepherd, G. M. Corticostriatal connectivity and its role in disease. Nat. Rev. Neurosci. 14, 278–291 (2013).
Article CAS PubMed PubMed Central Google Scholar
Wang, Y. et al. A frontal transcallosal inhibition loop mediates interhemispheric balance in visuospatial processing. Nat. Commun. 14, 5213 (2023).
Article ADS CAS PubMed PubMed Central Google Scholar
Jordan, R. & Keller, G. B. Opposing influence of top-down and bottom-up input on excitatory layer 2/3 neurons in mouse primary visual cortex. Neuron 108, 1194–1206 (2020).
Article CAS PubMed PubMed Central Google Scholar
Petreanu, L., Mao, T., Sternson, S. M. & Svoboda, K. The subcellular organization of neocortical excitatory connections. Nature 457, 1142–1145 (2009).
Article ADS CAS PubMed PubMed Central Google Scholar
Fishell, G. & Kepecs, A. Interneuron types as attractors and controllers. Annu. Rev. Neurosci. 43, 1–30 (2020).
Article CAS PubMed Google Scholar
Tremblay, R., Lee, S. & Rudy, B. GABAergic interneurons in the neocortex: from cellular properties to circuits. Neuron 91, 260–292 (2016).
Article CAS PubMed PubMed Central Google Scholar
Matho, K. S. et al. Genetic dissection of the glutamatergic neuron system in cerebral cortex. Nature 598, 182–187 (2021).
Article ADS CAS PubMed PubMed Central Google Scholar
Tasic, B. et al. Shared and distinct transcriptomic cell types across neocortical areas. Nature 563, 72–78 (2018).
Article ADS CAS PubMed PubMed Central Google Scholar
Yao, Z. et al. A taxonomy of transcriptomic cell types across the isocortex and hippocampal formation. Cell 184, 3222–3241 (2021).
Article CAS PubMed PubMed Central Google Scholar
Rindner, D. J., Proddutur, A. & Lur, G. Cell-type-specific integration of feedforward and feedback synaptic inputs in the posterior parietal cortex. Neuron 110, 3760–3773 (2022).
Article CAS PubMed PubMed Central Google Scholar
Combe, C. L. & Gasparini, S. Ih from synapses to networks: HCN channel functions and modulation in neurons. Prog. Biophys. Mol. Biol. 166, 119–132 (2021).
Article CAS PubMed PubMed Central Google Scholar
Harris, N. C. & Constanti, A. Mechanism of block by ZD 7288 of the hyperpolarization-activated inward rectifying current in guinea pig substantia nigra neurons in vitro. J. Neurophysiol. 74, 2366–2378 (1995).
Article CAS PubMed Google Scholar
Vapnik, V. N. The Nature of Statistical Learning Theory 2nd edn (Springer, 2000).
Cadwell, C. R. et al. Multimodal profiling of single-cell morphology, electrophysiology, and gene expression using Patch-seq. Nat. Protoc. 12, 2531–2553 (2017).
Article CAS PubMed PubMed Central Google Scholar
Gouwens, N. W. et al. Integrated morphoelectric and transcriptomic classification of cortical GABAergic cells. Cell 183, 935–953 (2020).
Article CAS PubMed PubMed Central Google Scholar
Paraskevopoulou, F. et al. Impaired inhibitory GABAergic synaptic transmission and transcription studied in single neurons by Patch-seq in Huntington’s disease. Proc. Natl Acad. Sci. USA 118, e2020293118 (2021).
Article CAS PubMed PubMed Central Google Scholar
Bhattacharjee, A. & Kaczmarek, L. K. For K+ channels, Na+ is the new Ca2+. Trends Neurosci. 28, 422–428 (2005).
Article CAS PubMed Google Scholar
Goldstein, S. A., Bockenhauer, D., O’Kelly, I. & Zilberberg, N. Potassium leak channels and the KCNK family of two-P-domain subunits. Nat. Rev. Neurosci. 2, 175–184 (2001).
Article CAS PubMed Google Scholar
Honoré, E. The neuronal background K2P channels: focus on TREK1. Nat. Rev. Neurosci. 8, 251–261 (2007).
Article PubMed Google Scholar
Zheng, Y. et al. Deep sequencing of somatosensory neurons reveals molecular determinants of intrinsic physiological properties. Neuron 103, 598–616 (2019).
Article CAS PubMed PubMed Central Google Scholar
Dumenieu, M. et al. The low-threshold calcium channel Cav3.2 mediates burst firing of mature dentate granule cells. Cereb. cortex 28, 2594–2609 (2018).
Article PubMed PubMed Central Google Scholar
Mohan, H. et al. Cortical glutamatergic projection neuron types contribute to distinct functional subnetworks. Nat. Neurosci. 26, 481–494 (2023).
CAS PubMed PubMed Central Google Scholar
Musall, S. et al. Pyramidal cell types drive functionally distinct cortical activity patterns during decision-making. Nat. Neurosci. 26, 495–505 (2023).
CAS PubMed PubMed Central Google Scholar
Lee, S., Kruglikov, I., Huang, Z. J., Fishell, G. & Rudy, B. A disinhibitory circuit mediates motor integration in the somatosensory cortex. Nat. Neurosci. 16, 1662–1670 (2013).
Article CAS PubMed PubMed Central Google Scholar
Letzkus, J. J. et al. A disinhibitory microcircuit for associative fear learning in the auditory cortex. Nature 480, 331–335 (2011).
Article ADS CAS PubMed Google Scholar
Naskar, S., Qi, J., Pereira, F., Gerfen, C. R. & Lee, S. Cell-type-specific recruitment of GABAergic interneurons in the primary somatosensory cortex by long-range inputs. Cell Rep. 34, 108774 (2021).
Article CAS PubMed PubMed Central Google Scholar
Jiang, X. et al. Principles of connectivity among morphologically defined cell types in adult neocortex. Science 350, aac9462 (2015).
Article PubMed PubMed Central Google Scholar
Pfeffer, C. K., Xue, M., He, M., Huang, Z. J. & Scanziani, M. Inhibition of inhibition in visual cortex: the logic of connections between molecularly distinct interneurons. Nat. Neurosci. 16, 1068–1076 (2013).
Article CAS PubMed PubMed Central Google Scholar
Pi, H. J. et al. Cortical interneurons that specialize in disinhibitory control. Nature 503, 521–524 (2013).
Article ADS CAS PubMed PubMed Central Google Scholar
Fishell, G. & Rudy, B. Mechanisms of inhibition within the telencephalon: “where the wild things are”. Annu. Rev. Neurosci. 34, 535–567 (2011).
Article CAS PubMed PubMed Central Google Scholar
Yuste, R. et al. A community-based transcriptomics classification and nomenclature of neocortical cell types. Nat. Neurosci. 12, 1456–1468 (2020).
Article Google Scholar
Losonczy, A. & Magee, J. C. Integrative properties of radial oblique dendrites in hippocampal CA1 pyramidal neurons. Neuron 50, 291–307 (2006).
Article CAS PubMed Google Scholar
Polsky, A., Mel, B. W. & Schiller, J. Computational subunits in thin dendrites of pyramidal cells. Nat. Neurosci. 7, 621–627 (2004).
Article CAS PubMed Google Scholar
Mao, T. et al. Long-range neuronal circuits underlying the interaction between sensory and motor cortex. Neuron 72, 111–123 (2011).
Article CAS PubMed PubMed Central Google Scholar
Morishima, M. & Kawaguchi, Y. Recurrent connection patterns of corticostriatal pyramidal cells in frontal cortex. J. Neurosci. 26, 4394–4405 (2006).
Article CAS PubMed PubMed Central Google Scholar
Brown, S. P. & Hestrin, S. Intracortical circuits of pyramidal neurons reflect their long-range axonal targets. Nature 457, 1133–1136 (2009).
Article ADS CAS PubMed PubMed Central Google Scholar
Constantinople, C. M. & Bruno, R. M. Deep cortical layers are activated directly by thalamus. Science 340, 1591–1594 (2013).
Article ADS CAS PubMed PubMed Central Google Scholar
Takahashi, N. et al. Active dendritic currents gate descending cortical outputs in perception. Nat. Neurosci. 23, 1277–1285 (2020).
Article CAS PubMed Google Scholar
Economo, M. N. et al. Distinct descending motor cortex pathways and their roles in movement. Nature 563, 79–84 (2018).
Article ADS CAS PubMed Google Scholar
Li, N., Chen, T. W., Guo, Z. V., Gerfen, C. R. & Svoboda, K. A motor cortex circuit for motor planning and movement. Nature 519, 51–56 (2015).
Article ADS CAS PubMed Google Scholar
Wang, Q. et al. The Allen mouse brain common coordinate framework: a 3D reference atlas. Cell 181, 936–953 (2020).
Article CAS PubMed PubMed Central Google Scholar
Hattox, A. M. & Nelson, S. B. Layer V neurons in mouse cortex projecting to different targets have distinct physiological properties. J. Neurophysiol. 98, 3330–3340 (2007).
Article PubMed Google Scholar
Dobin, A. et al. STAR: ultrafast universal RNA-seq aligner. Bioinformatics 29, 15–21 (2013).
Article CAS PubMed Google Scholar
Lawrence, M. et al. Software for computing and annotating genomic ranges. PLoS Comput. Biol. 9, e1003118 (2013).
Article CAS PubMed PubMed Central Google Scholar
Love, M. I., Huber, W. & Anders, S. Moderated estimation of fold change and dispersion for RNA-seq data with DESeq2. Genome Biol. 15, 550 (2014).
Article PubMed PubMed Central Google Scholar

Download references

Acknowledgements

We thank J. Lu for the critical reading of our manuscript. We thank A. Fei, R. Liu, and H. Bao from Q. Li’s lab for the kind help on the RNA sequencing data analysis. We thank Y. Wang from the Optical Imaging Core Facility at Shanghai Jiao Tong University School of Medicine for her help in image acquisition and analyses. This work was supported by grants from STI2030-Major Projects (2021ZD0203704 and 2021ZD0202804 to S.Z.), the National Natural Science Foundation of China (32170993 to S.Z. and 32200811 to Y.L.), the Science and Technology Commission of Shanghai Municipality (21ZR1436400 to S.Z.), China Postdoctoral Science Foundation (2022M722135 to Y.L.), the innovative research team of high-level local universities in Shanghai, and the Shanghai Frontiers Science Center of Cellular Homeostasis Regulation and Human Diseases. We thank scidraw.io for illustrations: the mouse head schema (in Figs. 3a, e, 4a, e, 5a, c, g, 7a, 8a, e, and Supplementary Fig. 10a, d, g) is from Luigi Petrucco (https://doi.org/10.5281/zenodo.3925903); the pyramidal neuron in Fig. 6a is from Federico Claudi (https://doi.org/10.5281/zenodo.3925905); the Eppendorf tube in Fig. 6a was adapted from Diogo Losch De Oliveira (https://doi.org/10.5281/zenodo.3925953); the DNA strand in Fig. 6a was adapted from Guilherme Bauer Negrini (https://doi.org/10.5281/zenodo.3926245).

Author information

These authors contributed equally: Yanmei Liu, Jiahe Zhang.

Authors and Affiliations

Songjiang Hospital and Songjiang Research Institute, Shanghai Key Laboratory of Emotions and Affective Disorders, Shanghai Jiao Tong University School of Medicine, Shanghai, 201600, China
Yanmei Liu, Jiahe Zhang, Zhishan Jiang, Siyu Zhang & Guofen Ma
Department of Anatomy and Physiology, Shanghai Jiao Tong University School of Medicine, Shanghai, 200025, China
Yanmei Liu, Jiahe Zhang, Zhishan Jiang, Siyu Zhang & Guofen Ma
Institute of Neuroscience, CAS Center for Excellence in Brain Science and Intelligence Technology, State Key Laboratory of Neuroscience, Chinese Academy of Sciences, Shanghai, 200031, China
Meiling Qin & Min Xu

Authors

Yanmei Liu
View author publications
You can also search for this author in PubMed Google Scholar
Jiahe Zhang
View author publications
You can also search for this author in PubMed Google Scholar
Zhishan Jiang
View author publications
You can also search for this author in PubMed Google Scholar
Meiling Qin
View author publications
You can also search for this author in PubMed Google Scholar
Min Xu
View author publications
You can also search for this author in PubMed Google Scholar
Siyu Zhang
View author publications
You can also search for this author in PubMed Google Scholar
Guofen Ma
View author publications
You can also search for this author in PubMed Google Scholar

Contributions

G.M. and S.Z. designed the experiments. G.M. performed and organized all the experiments. G.M. and Y.L. performed whole-cell recording experiments. G.M. and J.Z. performed the patch-seq experiments. G.M., Y.L., and J.Z. performed the imaging and circuit-mapping experiments with assistance from Z.J. and M.Q. G.M. performed the data analysis. Y.L., G.M., M.X., and S.Z. wrote the manuscript. All authors read and revised the manuscript.

Corresponding authors

Correspondence to Siyu Zhang or Guofen Ma.

Ethics declarations

Competing interests

The authors declare no competing interests.

Peer review

Peer review information

Nature Communications thanks the anonymous reviewer(s) for their contribution to the peer review of this work. A peer review file is available.

Additional information

Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Supplementary information

Supplementary Information

Peer Review File

Description of Additional Supplementary Files

Supplementary Data 1

Supplementary Data 2

Supplementary Data 3

Supplementary Data 4

Supplementary Data 5

Supplementary Data 6

Supplementary Data 7

Supplementary Data 8

Supplementary Data 9

Supplementary Data 10

Supplementary Data 11

Supplementary Data 12

Supplementary Data 13

Reporting Summary

Source data

Source Data

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.

Reprints and permissions

About this article

Cite this article

Liu, Y., Zhang, J., Jiang, Z. et al. Organization of corticocortical and thalamocortical top-down inputs in the primary visual cortex. Nat Commun 15, 4495 (2024). https://doi.org/10.1038/s41467-024-48924-8

Download citation

Received: 16 October 2023
Accepted: 16 May 2024
Published: 27 May 2024
DOI: https://doi.org/10.1038/s41467-024-48924-8

Comments

By submitting a comment you agree to abide by our Terms and Community Guidelines. If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.