Substance P regulates Tacr1 neurons, which control nitric oxide–mediated neurovascular coupling in the mouse cortex

Abstract

Neuronal activity–driven increases in cerebral blood flow (CBF), known as neurovascular coupling (NVC), are crucial for sustaining the metabolic demands of the brain. Our group has found that Tacr1 neurons, a subset of somatostatin neurons expressing the substance P (SP) receptor and neuronal nitric oxide synthase (nNOS), exert a disproportionately large regulation of CBF. Here, we use two-photon imaging to show that Tacr1 neurons regulate CBF through nitric oxide (NO) release and SP-receptor signaling . We identify astrocytic calcium (Ca²⁺) transients as a secondary response to vasodilation. To identify potential sources of SP in somatosensory cortex, viral mapping revealed Tac1-positive neurons locally, predominantly among parvalbumin (PV) neurons, as well as Tac1-positive long-range projections from perirhinal cortex. Functional analyses indicate that PV neurons regulate CBF through a Tacr1 neuron dependent pathway. These findings suggest a sequential mechanism whereby Tacr1 neurons respond to SP to release NO, inducing vasodilation and driving astrocytic Ca²⁺ signaling.

Tacr1 neurons drive nitric oxide–mediated vasodilation and astrocytic calcium in NVC, which can be triggered by substance P.

INTRODUCTION

Neuronal activity demands a continuous and precise supply of oxygen and glucose from the bloodstream (1, 2) as neurons lack substantial energy reserves and have limited anaerobic metabolic capacity (3, 4). To meet local metabolic needs, neuronal activity induces changes in cerebral blood flow (CBF) through a finely tuned process known as neurovascular coupling (NVC) (5–8). This mechanism ensures that brain regions with heightened activity receive an appropriate increase in blood supply, a process critical for maintaining brain function and health (9, 10). Impairments in NVC have been implicated in several neurodegenerative diseases, such as Alzheimer’s disease, where chronic mismatches between neuronal activity and CBF may contribute to progressive neuronal damage (9, 11–13).

The regulation of NVC is orchestrated by the neurovascular unit (NVU), which comprises neurons, astrocytes, endothelial cells, and mural cells, including vascular smooth muscle cells and pericytes (14, 15). Recent studies have provided insight on the neuronal populations that drive CBF regulation (16–22). Among these, cortical inhibitory neurons that express neuronal nitric oxide synthase (nNOS) are thought to be key regulators of this process due to their potential to release nitric oxide (NO), a potent vasodilator (23, 24). A subtype of nNOS-expressing neurons has been identified that also express the substance P (SP) receptor, Tacr1, which likely corresponds to type I nNOS neurons (25–28). These neurons are characterized by high nNOS expression and long-range projecting axons (27). Our recent study strongly supports the idea that Tacr1 neurons, despite comprising only 2% of cortical inhibitory neurons (26), are critical mediators of CBF (27). We demonstrated that optogenetic activation of a very small number of Tacr1 neurons led to significant increases in local blood flow. In contrast, optogenetic inhibition of Tacr1 neurons in the barrel cortex significantly dampened the functional hyperemia triggered by whisker stimulation. Given that Tacr1 neurons receive direct input from local glutamatergic and thalamocortical neurons, these findings suggest that NVC is mediated through feed-forward activation of Tacr1 neurons to cause local vasodilation.

Certain neuronal subtypes are known to physically associate with the NVU (29, 30). Notably, electron microscopy has shown that Tacr1 neurons form direct contact with the astrocytic endfeet (27), suggesting that Tacr1 neurons might have a role in directly regulating or influencing astrocytic activity. Consistent with the anatomical link, several studies have highlighted a link between functional hyperemia and astrocytic calcium (Ca²⁺) signaling (31–34). For instance, Takano et al. (35) provided evidence that increases in astrocyte Ca²⁺ can drive vasodilation, whereas Tran et al. (33) proposed that these Ca²⁺ increases occur as a response to vessel dilation in active cortical areas. Understanding endogenous astrocytic Ca²⁺ changes, such as those evoked by whisker stimulation, is complex because multiple cell types and neurotransmitters are involved (36–41). Therefore, we took advantage of our ability to selectively activate Tacr1 neurons, a sparse population sufficient to induce vasodilation, to reassess the relationship between neurons, vasodilation, and astrocytic Ca²⁺ transients.

In addition, a defining feature of Tacr1 neurons is their expression of Tacr1. Studies have shown that SP depolarizes cortical Tacr1 neurons in brain slices (42, 43), and optogenetic activation of these neurons is sufficient to drive vasodilation (27). Together, these findings suggest that SP modulation of Tacr1 neuron activity leads to vasodilation through the release of NO. Here, we combined optogenetic stimulation, in vivo two-photon (2P) Ca²⁺ imaging, laser Doppler flowmetry (LDF), pharmacological manipulation of the SP receptor, and in vitro assays, to test this hypothesis and define a more complete neural circuit basis by which Tacr1 neurons regulate CBF. In addition, we examined the potential sources of cortical SP that could play a crucial role in controlling Tacr1 neuron activity. These findings expand our understanding of the neuronal circuitry capable of strongly regulating NVC and could have considerable implications for developing therapies for neurovascular dysfunction.

RESULTS

Tacr1 neuron activation drives NO-dependent increase in CBF

We previously demonstrated that optogenetic activation of Tacr1 neurons, a sparse population of somatostatin neurons, induces a significant increase in CBF, whereas optogenetic inhibition of Tacr1 neurons significantly reduces the vasodilation that is normally observed during sensory stimulation (27). To specifically address whether Tacr1 neurons regulate vascular changes through the release of NO, we conducted pharmacological experiments with a selective nNOS blocker, Nω-propyl-l-arginine (L-NPA) (44). For optogenetic control of Tacr1 neurons, we crossed Tacr1^CreER mice with Ai32 mice to generate Tacr1^CreER–ChR2–yellow fluorescent protein (YFP) mice (Fig. 1A). Mice were implanted with a chronic cannula adjacent to the somatosensory cortex (S1), and CBF measurements were performed in awake, head-fixed mice using LDF (Fig. 1B). To test whether inhibition of nNOS was sufficient to reduce whisker-evoked CBF responses, we applied sensory stimulation (50-ms air puffs at 5 Hz for 1 s) and subsequently infused the selective nNOS inhibitor, L-NPA, or vehicle, through the chronic cannula. Administration of L-NPA reduced the whisker-evoked CBF change by 38% compared with vehicle, consistent with our previous findings (Fig. 1, C and D) (27). These results further suggest that NO plays a significant role in sensory-evoked NVC.

Fig. 1. Tacr1 neuron–evoked vascular changes are attenuated by cortical infusion of nNOS inhibitors.

(A) Top: Whisker (air puff; 1 s, 5 Hz, 50 psi) and optogenetic stimulation (1 s, 5 Hz, 30-ms pulse width) protocols. Bottom: Experimental setup with Tacr1^CreER::ChR2 mice, optic fiber, LDF fiber, cannula, and air puff system. (B) Macroscopic image of experimental setup illustrating cranial window with chronic cannula (scale bar, 4 mm). Inset highlights the positions of the LDF probe and optic fiber (scale bar, 1 mm). (C) Representative time course of CBF changes in response to whisker stimulation during baseline conditions (gray) and following L-NPA infusion (magenta; 1 μl, 1 mM) through the chronic cannula. (D) Quantification of normalized peak CBF increase upon whisker stimulation under vehicle [phosphate-buffered saline (PBS), 1 μl] and L-NPA (*P = 0.0186; n = 7 mice, 10 trials per mouse). (E) Example CBF trace in response to optogenetic stimulation of Tacr1 neurons during baseline conditions (gray) and following L-NPA infusion (magenta; 1 μl, 1 mM) through the chronic cannula. (F) Normalized maximum percent change in CBF response evoked by optogenetic stimulation of Tacr1 neurons under vehicle (1× PBS, 1 μl) and L-NPA (**P = 0.0014; n = 7 mice, 10 trials per mouse). (G) Representative time course of CBF changes in response to optogenetic stimulation of Tacr1 neurons during baseline conditions (black) and following L-NAME infusion (purple; 1 μl, 1 mM) through the chronic cannula. (H) Difference in the normalized maximum percent change in CBF response evoked by optogenetic stimulation under vehicle (PBS, 1 μl), L-NPA, and L-NAME (*P = 0.0239 and **P = 0.00540; n = 6 to 8 mice, 10 trials per mouse). Statistical significance was determined by a two-tailed t test [(D) and (F)] or one-way repeated-measures analysis of variance (ANOVA) (H). Data are means ± SEM. [(C), (E), and (G)] Shaded areas are SEM. n.s., not significant.

To directly evaluate the contribution of Tacr1 neuron–derived NO to vascular responses, we optogenetically stimulated Tacr1 neurons (1 mW, 1 s, 5 Hz, 10 trails) while measuring CBF changes in the presence of either vehicle or L-NPA. Consistent with our hypothesis, L-NPA significantly attenuated the CBF increase evoked by optogenetic stimulation of Tacr1 neurons, reducing it by 40% (Fig. 1, E and F). Infusion of NG-nitro-l-arginine methyl ester (L-NAME) (45), a nonselective NOS inhibitor, similarly reduced Tacr1 neuron–evoked CBF changes by 33% (Fig. 1, G and H).

Consistent results were observed in acute experiments conducted in Tacr1^CreER-ChR2-YFP mice, in which L-NPA was injected directly into the cortex rather than delivered through a cannula. Following a craniotomy, CBF responses were recorded in lightly sedated mice using LDF (fig. S1, A to D). L-NPA significantly suppressed the CBF increase induced by Tacr1 neuron stimulation by 58%, whereas vehicle had no effect (fig. S1, E to G). To further validate these findings, we used intrinsic optical imaging to assess changes in blood flow. Intrinsic optical imaging confirmed the localized suppression of the hemodynamic response near the injection site with attenuation decreasing with distance (fig. S1H). Although chronic windows avoid potential confounds associated with anesthesia and acute inflammatory response, acute experiments provide greater experimental flexibility and control over drug delivery and location, enabling direct cortical injection of L-NPA and real-time validation of injection sites through optical imaging. Using both experimental approaches in mice expressing ChR2 in Tacr1 neurons allowed us to validate that the observed effects were not specific to one experimental setup, preparation, or delivery method. Whether in chronic or acute experiments, awake or lightly sedated mice, or recordings by LDF or intrinsic optic imaging, inhibition of nNOS consistently attenuated the CBF response to Tacr1 neuron activation, suggesting that NO signaling plays a significant role in Tacr1 neuron–mediated NVC.

Activation of Tacr1 neurons triggers a delayed astrocytic Ca²⁺ response

Previous studies have shown that vasodilation is accompanied by increases in Ca²⁺ transients that occur in the astrocytes that surround blood vessels (32, 33, 46). However, the relationship between astrocytic Ca²⁺ and vascular changes remains controversial. Given that Tacr1 neurons contact astrocytic endfeet (Fig. 2A) (27), we sought to explore the possible role of Tacr1 neurons in the regulation of astrocytic Ca²⁺ transients. Tacr1^CreER-ChR2-YFP mice were injected with an astrocyte-specific adeno-associated virus (AAV5-gfaABC1D-cyto-GCaMP6f) into S1 cortex, to selectively express the Ca²⁺ indicator, GCaMP6f (Fig. 2, B and C). 2P imaging was conducted in awake, head-fixed mice to simultaneously monitor astrocytic Ca²⁺ dynamics and hemodynamic changes in response to whisker stimulation. Whisker stimulation (50-ms air puffs at 5 Hz for 1 s, 10 trials) evoked a rapid arteriole dilation which was followed by a delayed Ca²⁺ response in astrocytes, consistent with prior studies (Fig. 2D) (33, 47). The mean arteriole dilation was 17% (Fig. 2E), accompanied by a 15% peak increase in astrocytic Ca²⁺ response (Fig. 2F). The astrocytic Ca²⁺ response in soma and endfeet transients showed similar kinetics (fig. S2, A and B).

Fig. 2. Astrocytic Ca²⁺ dynamics driven by Tacr1 neuron activation.

(A) Schematic of NVU showing proposed interactions with Tacr1 neurons. (B) Strategy for astrocyte GCaMP6f expression (top) and 2P Ca²⁺ imaging in S1 (bottom). (C) Representative 2P images of GCaMP6f-expressing astrocytes and Tacr1-expressing ChR2 neurons at 900 nm (left) and 950 nm (right). Astrocytic endfeet (white arrow), astrocytic somata (yellow arrow), Tacr1 neurons (blue arrow), and SR-101–labeled vasculature (red) are shown. Scale bar, 45 μm. (D) Example trace showing astrocytic Ca²⁺ increases (green) and arteriole dilation (red) following 1-s whisker stimulation. (E) Peak dilation (max. ΔD/D₀) (****P < 0.0001; n = 12 vessel from three mice, 10 trials per mouse) and (F) peak astrocyte Ca²⁺ (% ΔF/F₀) during baseline or 1-s whisker stimulation (*P = 0.0162; n = 13 locations, with three to four astrocytes per location from five mice, 10 trials per mouse). (G) Example trace showing astrocytic Ca²⁺ (green) and vessel diameter (red) following 1-s optogenetic stimulation of Tacr1 neurons. (H) Peak dilation (max. ΔD/D₀) (**P = 0.0094; n = 9 vessel from four mice, 10 trials per mouse) and (I) peak fluorescence change (% ΔF/F₀) (*P = 0.0152; n = 13 locations with three to four astrocytes per location from six mice, 10 trials per mouse) during baseline or 1-s optogenetic stimulation. (J and K) Average trace illustrating astrocytic Ca²⁺ (green) and vessel diameter (red) following 10-s whisker stimulation (J) or optogenetic stimulation (K). (L) Method for computing 50% decay time (vessel) or rise time (Ca²⁺). (M to P) Fifty percent decay and rise times following 1-s whisker (M) or optogenetic stimulation (N) and 10-s whisker (O) or optogenetic (P) stimulation. Statistical significance was determined by paired, parametric, two-tailed t test [(E), (F), (H), and (I)] or unpaired, parametric, two-tailed t test [(M) to (P)]. Data are means ± SEM. n.s., not significant. [(D), (G), (J), and (K)] Shaded areas are SEM.

To determine whether direct activation of Tacr1 neurons could replicate the astrocytic Ca²⁺ response seen during whisker stimulation, we performed in vivo optogenetic stimulation of Tacr1 neurons while conducting 2P imaging. Optogenetic activation of Tacr1 neurons led to vasodilation, followed by a delayed astrocytic Ca²⁺ transient (Fig. 2G). The mean arteriole dilation was 13% (Fig. 2H), accompanied by a 12% peak increase in astrocytic Ca²⁺ response (Fig. 2I). No changes were observed under control conditions (fig. S2, C and D). Notably, astrocytic Ca²⁺ transients triggered by both whisker stimulation and optogenetic activation of Tacr1 neurons showed similar peak magnitudes and temporal dynamics (fig. S2, E and F). To probe this finding further, we extended the stimulation duration (whisker or optogenetic) from 1 to 10 s, which prolonged the vasodilation. Nonetheless, the timing of the astrocytic Ca²⁺ transient consistently matched the constriction phase (Fig. 2, J and K; and fig. S2, G to L), suggesting that these Ca²⁺ transients may be a result of arteriole constriction. The 50% reduction in vasoconstriction appeared to correspond with a 50% increase in astrocytic Ca²⁺ levels (Fig. 2, L to P).

To test whether Tacr1 neurons directly regulate astrocytes or whether astrocytic Ca²⁺ transients arise secondarily from vascular dynamics, we administered the endothelial NO synthase inhibitor, N5-(1-iminoethyl)-l-ornithine dihydrochloride (L-NIO) (48), via tail vein injection, to prevent endothelial vasodilation without affecting NO release from Tacr1 neurons (fig. S2M). Baseline vessel diameter was not altered in the presence of L-NIO relative to vehicle, whereas whisker-evoked vessel dilation was attenuated compared with both baseline and vehicle, despite measurements being performed in the same vessels (fig. S2N). In addition, L-NIO significantly reduced the astrocytic Ca²⁺ responses during both whisker stimulation and optogenetic activation of Tacr1 neurons (fig. S2, O and P). This suggests that astrocytic Ca²⁺ transients may depend on the dilation-constriction sequence rather than direct input from Tacr1 neurons.

Tacr1 neuron activity is regulated by SP

Our previous studies revealed that Tacr1 neurons receive corticocortical and thalamocortical excitatory input, enabling feed-forward control of blood flow (27). However, studies in acute cortical slices have shown that SP depolarizes Tacr1 neurons (42, 43), and in vivo pharmacological experiments indicate that SP induces vasodilation (49). Whether these vascular responses arise from a direct increase in Tacr1 neuron activity in vivo remains unknown. To address these gaps, we used 2P Ca²⁺ imaging in awake, head-fixed Tacr1^CreER mice expressing GCaMP6f (Tacr1^CreER-GCaMP6f) combined with pharmacological manipulations via chronically implanted cannula in cortex (Fig. 3A). We first evaluated the effects of SP on Tacr1 neuron activity and its influence on adjacent arterioles (Fig. 3, B and C), by simultaneously measuring Tacr1 neuron activity and vessel diameter (Fig. 3D). Infusion of SP resulted in a significant increase in Tacr1 neuron activity, with a roughly 198% increase in Ca²⁺ peaks per minute (Fig. 3E) and a corresponding 12% increase in the mean vessel diameter of adjacent arterioles (Fig. 3F). In contrast, vehicle infusion did not cause significant changes in either Tacr1 neuron activity or vessel diameter (fig. S3). These findings indicate that SP is sufficient to increase Tacr1 neuron activity and induce vasodilation.

Fig. 3. SP signaling modulates Tacr1 neuron activity and vessel diameter.

(A) Experimental setup and macroscopic view of imaged region (white dashed square; scale bar, 1 mm). (B) Model of SP-mediated activation of Tacr1 neurons. (C) 2P image showing a GCaMP6f-expressing Tacr1 neuron (white arrow) and nearby SR-101–labeled artery (white dashed circle). Scale bar, 30 μm. (D) Ca²⁺ trace from a Tacr1 neuron (% ΔF/F₀; top) and mean vessel diameter (micrometers; bottom) under baseline (gray) and SP (yellow; 1 μl, 1 mM) over 500 s. (E) Top: Mean Tacr1 neuron fluorescence under baseline and SP. Bottom: Quantification of Ca²⁺ peaks/min in Tacr1 neurons under baseline (gray) and SP (green) (*P = 0.0175; n = 13 neurons from five mice). Scale bar, 10 μm. (F) Top: 2P image of an arteriole showing masking method (red line) for diameter measurements. Bottom: Mean vessel diameter under baseline (gray) and SP (red) (**P = 0.0039; n = 19 vessels from five mice). Scale bar, 15 μm. (G) Model of CP-9994–mediated SP blockade. (H) Ca²⁺ trace from a Tacr1 neuron during baseline (gray) and CP-9994 (light purple; 1 μl, 10 mM) over 500 s. (I) Ca²⁺ peaks/min in Tacr1 neurons (*P = 0.0201; n = 12 neurons from four mice). (J) Mean vessel diameter (n = 15 vessels from six mice). (K) Time course of Tacr1 neuron following whisker stimulation. (L) Maximum Ca²⁺ peak under baseline (gray) and CP-9994 (green) (*P = 0.0120; n = 12 neurons from four mice, 10 trials per mouse). (M) Time course of vessel diameter following whisker stimulation. (N) Peak vessel diameter changes under baseline (gray) and CP-9994 (red) (**P = 0.0023; n = 14 neurons from six mice, 10 trials per mouse). Data are means ± SEM. n.s., not significant. [(E), (F), (I), (J), (L), and (N)] Statistical significance was determined by paired, parametric, two-tailed t test.

To determine whether endogenous SP is necessary for ongoing activity, we infused the SP receptor antagonist, CP-9994 dihydrochloride (Fig. 3G). Infusion of CP-9994 significantly reduced endogenous Tacr1 neuron activity by ~49% (Fig. 3, H and I), without altering mean vessel diameter (Fig. 3J), underscoring the critical role of endogenous SP signaling in modulating Tacr1 neuron activity.

Building on our earlier observation that Tacr1 neurons display rapid Ca²⁺ changes in response to whisker stimulation (27), we next examined the role of SP signaling in modulating this sensory-evoked activity (Fig. 3K). Infusion of the SP receptor antagonist reduced the whisker-evoked increase in Tacr1 neuron activity from 46 to 27% (Fig. 3L). In addition, we measured the maximum change in arteriole diameter under these conditions (Fig. 3M). Whisker stimulation evoked a 23% increase in arteriole diameter, which was significantly reduced to 18% following CP-9994 infusion (Fig. 3N). Consistent with these results, measurements of blood flow using LDF revealed a 27% reduction in whisker-evoked CBF response after CP-9994 infusion (fig. S4). These findings suggest that endogenous SP signaling modulates both spontaneous and evoked Tacr1 neuron activity, which, in turn, influences CBF.

Cortical parvalbumin neurons are local sources of SP

Having established that SP modulates Tacr1 neuron activity, we next sought to identify the sources of SP in the cortex. To this end, we injected Cre-dependent retrograde adeno-associated virus (AAVrg) into the S1 of Tac1^Cre mice. In an initial cohort of three animals, AAVrg-FLEX-tdTomato was injected unilaterally into the left S1 (Fig. 4A); a second cohort of four animals received bilateral injections, AAVrg-FLEX-tdTomato into the left S1 and AAVrg–DIO–enhanced YFP (eYFP) into the right S1, to enable subsequent assessment of bilateral projections within the same animals. In both cohorts, Cre-dependent labeling resulted in widespread reporter labeling in S1, suggesting that local Tac1-positive neurons may contribute to SP release. However, Tac1 expression reflects transcription of the SP precursor gene and does not necessarily indicate peptide release.

Fig. 4. Viral labeling reveals that a Tac1-positive neurons are a subset of PV neurons.

(A) Experimental design. (B) IHC image of S1 in Tac1^Cre mouse (scale bar, 200 μm). Inset enlarged on the right displays the colocalization of DAPI (blue), tdT (red; Alexa Fluor 555), and PV (magenta; Alexa Fluor 647); white circles represent colocalization of tdT -labeled neurons with PV-positive neurons. Scale bar, 100 μm. (C) Percentage of tdT-labeled cells that are PV-positive (n = 7 mice). (D) Representative multiplex fluorescence in situ hybridization (FISH) images demonstrating colocalization of Pvalb (red), Tac1 (green), and Slc17a6 (magenta) in cortical neurons of WT mouse; white arrows are pointing to Pvalb-positive neurons colocalized with Tac1-positive neurons. Scale bars, 10 μm and (inset) 5 μm. (E) Venn diagram depicting the intersectionality of Tac1-positive neurons (green; 218 neurons) and Pvalb-positive neurons (red; 366 neurons), with 119 neurons coexpressing both markers. (F) Schematic of bilateral AAVrg virus injections into the left (tdT) and right (eYFP) S1, followed by processing of brains for IHC. (G) Sections from Tac1^Cre mice, displaying viral spread from anterior (bregma, 0.14 mm) to posterior (bregma, −3.16 mm). Scale bar, 1 mm. (H) Left: Perirhinal cortex (PRh) showing tdT-labeled (red), eYFP-labeled (green), and dual-labeled (yellow) neurons. Right: Enlarged view of inset. Scale bar, 100 μm. (I) Venn diagram showing overlap (yellow; 104 neurons, 15%) between ipsilateral-projecting (blue; 282 cells, 40%) and contralateral-projecting neurons (purple; 322 cells, 45%) in the perirhinal cortex (n = 6 mice). (J) Schematic summarizing the viral tracing results showing Tac1^Cre neurons in the perirhinal cortex projecting bilaterally to S1, including ipsilateral (blue) and contralateral (purple) projections, as well as local Tac1^Cre neurons in S1 (light green). Both long-range and local Tac1^Cre neurons may serve as sources of SP (black) acting on Tacr1 neurons (dark green).

Previous studies have proposed that parvalbumin (PV)–expressing chandelier cells could be cortical sources of SP. To test this possibility, we performed immunohistochemistry (IHC) for tdT and PV (Fig. 4B). Quantification of tdTomato/PV colocalization was performed across all animals, in which tdTomato was expressed in S1, including both unilateral and bilateral injection cohorts (total n = 7). Our results revealed that 43% of tdT-labeled neurons were PV -positive, indicating that Tac1-positive neurons include but are not restricted to PV -positive neurons (Fig. 4C).

To further investigate this finding, we performed multiplex fluorescence in situ hybridization (FISH) using probes against Tac1, Pvalb, and Slc17a6 (encoding Vglut2) in wild-type mice. We found that 33% of Pvalb-positive neurons coexpressed Tac1. In addition, among Tac1-positive cells, about half (55%) were Pvalb-positive (Fig. 4, D and E). However, almost none (<1%) of the Tac1-positive neurons were excitatory neurons, as defined by expression Slc17a6. These findings suggest that local Tac1-expressing neurons are inhibitory and include a defined subpopulation of Pvalb-positive neurons.

Long-range projecting neurons from the perirhinal cortex may be sources of SP

We also considered the possibility that SP could come from distant sources, either within or beyond the cortex, in addition to local inputs in S1. To identify additional sources, we analyzed retrograde labeling from the bilateral injection cohort, in which AAVrg-FLEX-tdTomato was injected into the left S1 and AAVrg-DIO-eYFP into the right S1 (Fig. 4F) and mapped labeled neurons throughout the brain. Our experiments did not reveal any labeled subcortical neurons in Tac1^Cre mice, indicating that cortical SP is primarily derived from neurons within the cortex (Fig. 4G). Notably, within the cortex, labeled Tac1-positive neurons were consistently found in one specific region: the perirhinal cortex (Fig. 4H). This labeling pattern was highly specific to the Tac1^Cre allele as it was not observed when other Cre alleles (e.g., Tacr1^CreER) were used (fig. S5, A to D). These long-range projecting Tac1 -positive neurons were only observed in the perirhinal cortex. Of the retrogradely labeled neurons expressing either tdT or YFP, around 40% were labeled from the ipsilateral S1 and 45% from the contralateral S1, suggesting similar prevalence of ipsilateral and contralateral projections. Fifteen percent of the perirhinal neurons that were retrogradely labeled from S1 on the ipsilateral side appeared to extend axonal projections to S1 on the contralateral side (Fig. 4I). This observation suggested that few Tac1-positive neurons in the perirhinal cortex project bilaterally to S1 in both hemispheres.

To determine whether long-range Tac1 -positive projections from the perirhinal cortex preferentially target S1, we also tested retrograde labeling in Tac1^Cre mice from other cortical regions, such as the medial prefrontal cortex (fig. S5, E and F). In all these experiments, the perirhinal cortex was likewise labeled, suggesting that their projections extend broadly throughout the cortex. Last, we asked whether the long-range projecting Tac1-positive neurons, like local Tac1-positive neurons, were PV-positive neurons. IHC revealed that 34% of retrogradely labeled neurons in the perirhinal cortex were PV-positive (fig. S6). Overall, these findings suggest that Tac1-positive neurons in the perirhinal cortex include PV-positive neurons and extend broad, bilateral projections that target homotypic regions in each hemisphere.

PV neuron–induced CBF increases require SP signaling and NO release

Optogenetic stimulation of PV-expressing interneurons has been shown to evoke increases in CBF (16, 50). For example, our group previously demonstrated that optogenetic stimulation of PV neurons for 4 s at different frequencies elicits a 5 to 15% increase in CBF (16), yet the underlying signaling mechanisms remain unclear. Emerging evidence implicates neuropeptide signaling in this process, particularly SP, which acts on Tacr1-expressing neurons (21). To investigate whether PV-driven CBF responses depend on Tacr1 neuron signaling and downstream NO release, we performed a series of pharmacological experiments in awake, head-fixed PValb^Cre mice (Fig. 5A). We virally expressed Cre-dependent ChR2 in S1 and adjacent cortical regions (Fig. 5B) and implanted a chronic cannula near the viral injection site to enable drug delivery. CBF changes were recorded using LDF in response to blue light stimulation (40 Hz, 10 s; Fig. 5, C and D).

Fig. 5. Pharmacological blockade of SP receptor or nNOS attenuates PV neuron–induced increases in CBF.

(A) Proposed signaling pathway where optogenetic activation of PV neurons promotes SP activating Tacr1 neurons, leading to NO release and vasodilation. (B) Schematic illustrating the injection of Cre-dependent ChR2 virus into S1 of PValb^Cre mice. (C) Experimental setup in PValb^Cre mice including the optic fiber, LDF fiber, and cannula arrangement. (D) Top: Optogenetic stimulation protocol (10-s duration, 40 Hz, 10-ms pulse width). Bottom: Macroscopic image showing Cre-dependent ChR2 expression in S1 of PValb^Cre mice (scale bar, 1 mm). (E) Representative trace showing CBF changes following optogenetic stimulation of PV neurons during baseline (black) and after cortical infusion of SP receptor antagonist, CP-9994 (light purple; 1 μl, 10 mM). (F) Quantification of the maximum percent change in CBF evoked by optogenetic stimulation of PV neurons before and after CP-9994 infusion (*P = 0.0323; n = 5 mice, 10 trials per mouse). (G) Representative trace of CBF changes in response to optogenetic stimulation of PV neurons during baseline (black) and after cortical infusion of nNOS blocker, L-NPA (magenta; 1 μl, 1 mM). (H) Quantification of the maximum percent change in CBF evoked by optogenetic stimulation of PV neurons before and after L-NPA infusion (*P = 0.0457; n = 5 mice, 10 trials per mouse). (I) Representative trace showing CBF changes in response to optogenetic stimulation of PV neurons during baseline (black) and after cortical infusion of vehicle (PBS, 1 μl). (J) Quantification of the maximum percent change in CBF evoked by optogenetic stimulation of PV neurons before and after vehicle infusion (n = 5 mice, 10 trials per mouse). Data are means ± SEM. n.s., not significant. [(F), (H), and (J)] Statistical significance was determined by paired parametric, two-tailed t test. [(E), (G), and (I)] Shaded areas are SEM.

After establishing a baseline, we administered either CP-99994, L-NPA, or vehicle across separate sessions (each with independent baseline recordings). Infusion of CP-9994 reduced PV-evoked CBF responses by more than half, from 33% at baseline to 14% during treatment (Fig. 5, E and F). Similarly, L-NPA reduced the CBF response from 37% at baseline to 11% during treatment (Fig. 5, G and H). By contrast, vehicle did not alter the PV-evoked CBF responses (Fig. 5, I and J). Together, these results indicate that PV-mediated blood flow changes are mediated by SP, consistent with previous findings (21). Moreover, our pharmacological data extend this work by showing that PV neuron activation recruits a signaling cascade consistent with the activation of Tacr1 neurons, leading to NO-mediated vasodilation. Combined with our IHC and FISH data (Fig. 4), our pharmacological results support a circuit mechanism, in which Tac1-expressing PV neurons modulate local hemodynamics changes through Tacr1 neurons and downstream NO signaling.

DISCUSSION

In summary, our study demonstrates that, despite their sparse distribution, Tacr1 neurons are integral mediators of NVC. Optogenetic activation of Tacr1 neurons elicited robust vasodilation through the release of NO, driving significant increases in CBF, which was markedly diminished by selective nNOS inhibition. In addition, we found that Tacr1 neuron activation led to delayed astrocytic Ca²⁺ activity following vessel dilation. Our findings further implicate SP released from local PV neurons in the modulation of Tacr1 neuron activity.

Using pharmacological experiments in Tacr1^CreER mice, we confirmed that Tacr1 neurons largely regulate the increase in CBF through the release of NO, suggesting that nNOS-derived NO from Tacr1 neurons is a key contributor to NVC. Although these results strongly position Tacr1 neurons as central regulators of CBF, we acknowledge that other neuronal populations may also contribute, as evidenced by the residual CBF response observed in the presence of NOS inhibitors. In addition, we cannot exclude the possibility that Tacr1 neuron activation engages other NO-releasing cells, such as type II nNOS neurons or endothelial cells, which, in turn, contribute to the observed vasodilation. The persistence of CBF changes despite nNOS inhibition may also reflect incomplete blockage by L-NPA under our experimental conditions. Together, these considerations suggest a complex interplay between various neuronal populations and signaling pathways in regulating CBF. It is possible that other molecules, such as vasoactive intestinal peptide (51–53) or prostaglandins (54, 55), might also be involved in modulating CBF, potentially working in concert with or as compensatory mechanisms when nNOS activity is inhibited. In addition, glial cells could play a critical role in this dynamic interplay by responding to neuronal activity and releasing signaling molecules that influence vascular tone (46, 56–58).

An unresolved question in NVC is how neurons engage astrocytes and whether this interaction influences contractile cells to shape vascular responses. Our data revealed that activation of Tacr1 neurons leads to vasodilation followed by delayed astrocytic Ca²⁺ transients, suggesting a temporally ordered but indirect mechanism. While this sequence implies a possible signaling hierarchy, we note that these observations establish correlation rather than causation. Previous studies reveal that astrocytes respond to glutamate (33, 59, 60). However, the rise in astrocytic Ca²⁺ observed here is likely due to a different mechanism. Instead, our findings support a model in which the rise in astrocytic Ca²⁺ transients is downstream of vessel dilation. Inhibition of vascular NO production reduced vasodilation and attenuated astrocytic Ca²⁺ signals, suggesting that vascular tone contributes as a modulatory intermediary. The slow kinetics of astrocytic Ca²⁺ signaling, further shaped by diffusion through gap junctions, likely underlie the delayed and spatially propagated transients observed within our limited field of view. Notably, a recent study showed that SST interneurons also exert biphasic vascular control, with early NO-mediated vasodilation and delayed astrocytic engagement under prolonged activation (31). Our findings extend this framework by identifying Tacr1 neurons as a distinct inhibitory subtype that recruits astrocytes indirectly, possibly via vascular NO, suggesting a temporally coordinated interaction between vascular and astrocytic responses.

Our findings highlight the importance of the SP effect on Tacr1 neuron activity modulation and subsequent changes in CBF. Upon further investigation of the potential sources of SP in the cortex, our study identified cortical PV neurons as candidates for SP release, given that a subpopulation of PV neurons coexpress Tac1, which encodes for SP (61, 62). Consistent with this, pharmacological blockade of the SP receptor and inhibition of nNOS markedly attenuated the CBF response, indicating that PV neurons may regulate vascular dynamics through a Tacr1 neuron–NO signaling pathway. In addition, we identified long-range Tac1-positive projections from the perirhinal cortex. This observation suggests a pathway through which distant neuronal populations could influence vascular responses. However, it is essential to consider that these long-range Tac1-positive neurons from the perirhinal cortex might not directly regulate NVC; instead, they could modulate Tacr1 neuron activity across different brain states, such as slow-wave sleep. Given that Tacr1 neurons are characterized as sleep active (28, 63), this raises intriguing questions about the role of the perirhinal cortex in cognitive functions like object recognition memory and how these dynamics might affect local vascular regulation during various behavioral states.

In addition to the SP signaling characterized in this study, our previous work showed that Tacr1 neurons receive corticocortical and thalamocortical inputs (27). These convergent inputs place Tacr1 neurons at a strategic intersection for integrating excitatory and neuromodulatory signals, positioning them to anticipate and coordinate local metabolic demands. Given the established role of glutamatergic transmission in driving feedforward NVC, it is plausible that SP signaling operates in parallel, or even synergistically, to modulate the timing or magnitude of CBF responses. Such integration may be particularly important during transitions between brain states or in response to changing neuromodulatory tone, thereby refining vascular responses across networks in a context-dependent manner.

Collectively, our findings reveal a cellular mechanism in which Tacr1 neurons, modulated by the SP receptor, mediate CBF through NO release. We show that Tacr1 neurons do not directly regulate astrocytic Ca²⁺ transients; rather, these transients occur subsequent to the vasodilation initiated by Tacr1 neuron activity. In addition, both local and long-range Tac1-positive neuronal populations were identified, including a subpopulation of PV neurons that express Tac1 and mediate CBF through the Tacr1 neuron–dependent activation of NO signaling. Together, these findings establish Tacr1 neurons as the critical intermediary linking neuronal signaling to vascular and astrocytic Ca²⁺ dynamics in NVC, defining a circuit level framework for how neuropeptidergic modulation helps shape cerebrovascular function.

MATERIALS AND METHODS

Animals

All animals were cared for in compliance with the National Institutes of Health guidelines following protocols approved by the University of Pittsburgh Institutional Animal Care and Use Committee (protocol 24125779). Previously generated Tacr1^CreER mice (also called NK1R^CreER mice) (64) were bred in house and maintained on a C57BL/6 background. The following mice strains were used: wild type (C57BL/6J, Charles River Laboratories, no. 027), Ai32 [Jackson Laboratory (JAX), no. 012569], Ai95 (JAX, no. 028865), Tac1 (JAX, no. 021877), and Pvalb (JAX, no. 017320). All mice were maintained on a C57BL/6J background, and both males and females were used. For mice expressing CreER, tamoxifen (Sigma-Aldrich, T5648) was dissolved in corn oil at a concentration of 20 mg ml⁻¹ and administered via intraperitoneal injection at a dosage of 100 mg kg⁻¹ for 4 consecutive days. Before receiving tamoxifen injections, animals underwent cranial surgeries and were given a recovery period of at least one week. Sample sizes were calculated using a power analysis based on previous pilot studies and representative sample sizes from prior literature on similar experiments (16, 27).

Viruses

For the study, the following viral constructs were used: AAV5.pZac2.1-gfaABC1D-cyto-GCaMP6f (Addgene, 52925) (65), AAVrg-FLEX-tdTomato (Addgene, 28306), AAVrg-Ef1a-DIO-eYFP (Addgene, 27056), and AAV5-Ef1a-DIO hChR2(E123T/T159C)-EYFP (Addgene, 35509) (66).

Stereotaxic injection

Animals were first anesthetized with 3 to 5% isoflurane and then maintained at 1 to 2% throughout the procedure. The mice were secured in a stereotaxic head frame (Kopf Instruments, Model 942 Small Animal Stereotaxic Instruments). At the beginning of the procedure, a subcutaneous dose of buprenorphine (0.1 mg kg⁻¹) and ketoprofen (5.0 mg kg⁻¹) was administered, with additional doses given twice daily for 48 hours postsurgery. Following anesthesia, the scalp was shaved, and a local antiseptic (betadine) was applied. A topical local anesthetic, bupivacaine (2.5 mg kg⁻¹), was then applied to the shaved area. A midline incision was made to expose the skull. Afterward, the skull was aligned using cranial fissures. A stainless-steel burr (Fine Science Tools, 19008-07) attached to a micro drill (Foredom Electric Co., K1070 High Speed Rotary Micromotor Kit, 2.35-mm Collet) was used to drill a burr hole.

For viral injections, a custom-pulled 3.5-inch (8.89-cm) glass capillary tube (Drummond Scientific Company, no. 3-00-203-G/X replacement) was slowly lowered into the cortex, and the virus was delivered at a rate of 2 nl/s via a microinjector (Drummond Scientific Company, Nanoinjector III, catalog no. 3-000-207). Tacr1^CreER-ChR2-YFP mice received unilateral injections of 800 nl of virus (Fig. 2), Tac1^Cre mice received either unilateral or bilateral injections of 400 nl per site, and Pvalb^Cre mice received unilateral injections of 250 nl per site. The injection needle was left in place for an additional 5 min before careful withdrawal. The following coordinates were used to target the S1BF for viral injections: Anterior-Posterior (AP), −0.82 mm; Medial-Lateral (ML), ±2.9 mm; and Dorsal-Ventral (DV), 0.7 mm.

Chronic cranial windows

A craniotomy was performed on 2- to 4-month-old mice, where a sterile glass window was implanted, and an aluminum chamber frame (Narishige Inc., CF-10) was attached. The mice were first anesthetized with 3 to 5% isoflurane and then maintained at 1 to 2% throughout the procedure. Vital signs, including respiratory rate and body temperature, were monitored to ensure stable anesthesia levels. Analgesia was provided by administering buprenorphine (0.1 mg kg⁻¹) and ketoprofen (5.0 mg kg⁻¹) subcutaneously at the beginning of the surgery, followed by twice-daily doses for 3 days postcraniotomy. Following anesthesia, the scalp was shaved, and a local antiseptic (betadine) was applied. A topical local anesthetic, bupivacaine (2.5 mg kg⁻¹), was then applied to the shaved area. The skin over the skull was then removed to expose the underlying bone above the barrel cortex. The craniotomy was positioned over the barrel cortex, roughly 1.5 to 2 mm posterior to bregma and 3 mm lateral to the midpoint between bregma and lambda. A custom-made cover glass, consisting of a 4-mm round cover glass attached to a 5-mm round cover glass (Warner Instruments Inc.), was secured to the skull along with the chamber frame using dental cement (Lang Dental Manufacturing Company, Ortho-Jet). After surgery, mice were returned to their home cages to recover for 7 days before tamoxifen administration (100 mg kg⁻¹) began in Tacr1^CreER mice. During this period, mice were acclimated to a custom treadmill for head-fixed, awake experiments.

Chronic cannula implantation

For the cannula implantation, a slight indentation (<0.5 mm) adjacent to the cranial window was made to allow insertion of the cannula (RWD Life Science Inc.; single guide cannula, length of 3 mm, and single injector). The cannula was fixed in place using dental acrylic and cyanoacrylate glue (3M Vetbond Tissue Adhesive) at a 30° angle relative to the barrel cortex, with the tip positioned ~100 μm beneath the cortical surface, avoiding large surface vessels. A dummy cannula was inserted between experiments, and, during pharmacological sessions, it was replaced with the infusion cannula for drug infusion.

Laser Doppler flowmetry

To measure CBF changes evoked by optogenetic (light) and whisker (air puff) stimulation, a laser Doppler flowmeter (Periflux 5000/411, Perimed AB, Jarfalla, Sweden) was used. The LDF probe, with a tip diameter of 200 μm and an operating wavelength of 780 nm, sampled data at a rate of 1 kHz. Time-series data spanning 30 s for Tacr1 neuron stimulation or 60 s for PV neuron stimulation were collected for all trials, starting 5 s before stimulation onset. The LDF time series were then low-pass filtered with a rectangular cutoff at 4 Hz and downsampled to 10 Hz.

Air puffs (50 ms at 5 Hz) were delivered to the contralateral whisker pad for whisker stimulation. Optogenetic stimulation parameters differed by neuron type: Tacr1 neurons were stimulated with 30-ms pulses at 5 Hz for 1 s, and PV neurons were stimulated with 30-ms pulses at 40 Hz for 10 s. Backscatter readings from the LDF unit were recorded during stimulation periods to regress out optogenetic stimulation artifacts specific to each frequency (5 Hz for Tacr1 and 40 Hz for PV).

For Tacr1 neurons, maximum CBF change (max. ΔCBF) was calculated as [(CBF_time − CBF_baseline)/CBF_baseline × 100], where CBF_baseline was calculated over the 4 s before stimulation and CBF_time was the maximum value in the 4 s following light stimulation. For PV neurons, CBF_baseline was the average CBF over the 4 s before stimulation, and CBF_time was taken as the maximum CBF value within 10 s following stimulation onset to capture the longer stimulation effects. Time series from 10 trials per animal were averaged, converted to percent change, and then averaged across all animals. Maximum CBF change (Max. ΔCBF) was defined as the peak value in the 4 s following whisker or light stimulation. Normalized CBF change (%) was calculated as (Max. ΔCBF_{posttreatment} − Max. ΔCBF_pretreatment)/Max. ΔCBF_{posttreatment}.

Optogenetic and whisker stimulation

Light and whisker stimulation experiments were conducted in awake, head-fixed mice. For optogenetic stimulation during CBF measurements, a transistor-transistor-logic (TTL)-controlled, power-adjustable laser diode unit (CrystaLaser Inc., Reno, NV) was connected to an optic fiber, delivering 473-nm blue light for ChR2 activation. The laser power at the fiber tip was set to 1 mW, calibrated using a power meter (Melles Griot 13 PM001, IDEX Inc., Rochester, NY). Light stimulation parameters involved 30-ms pulses at 5 Hz for 1 s, repeated every 30 s. A minimum of 10 trials were performed for each condition, and whisker stimulation was delivered via an air puff (Toohey Spritzer, Toohey Company, Fairfield, NJ) set to 50 psi. Stimulation triggers and CBF were recorded at 1 kHz using a LDF system (MP150, Biopac Systems Inc., Goleta, CA).

For experiments involving astrocyte Ca²⁺ imaging, optogenetic stimulation of Tacr1 neurons was performed on the Ultima 2Pplus system (Bruker), which allowed simultaneous light stimulation and GCaMP signal acquisition. The 473-nm blue light used for optogenetic activation was delivered through the 2Pplus system laser, synchronized with 2P imaging to track astrocyte responses to Tacr1 stimulation. Similarly, whisker stimulation was applied using an air puff injector (Toohey Spritzer, Toohey Company, Fairfield, NJ) set at 50 psi while imaging astrocytic Ca²⁺ transients. A total of 10 trials were conducted for each experimental condition.

Intracortical infusions

Animals were secured in a head-fixed position on the custom treadmill, and the dummy cannula was carefully removed following baseline recordings. Baseline measurements consisted of LDF recordings during whisker and/or optogenetic stimulation to assess pretreatment CBF responses. The infusion cannula was then inserted into the guide cannula, and pharmacological agents or vehicle were infused at a rate of 0.2 μl min⁻¹ over 5 min, totaling 1 μl. The infusion cannula remained in place throughout the remainder of the experiment to minimize tissue disruption. After a 5- to 10-min incubation period, LDF recordings were repeated during whisker and optogenetic stimulation to assess posttreatment responses within the same animal.

Pharmacology

For intracortical infusions, L-NPA and L-NAME were each dissolved in sterile 1× phosphate-buffered saline (PBS) to a final concentration of 1 mM. CP-99994 dihydrochloride was prepared in sterile 1× PBS at 10 mM, and SP was diluted in sterile 1× PBS to a final concentration of 1 μM. Sterile 1× PBS was used for vehicle control infusions. For systemic administration, L-NIO (0.1 mg kg⁻¹) was dissolved in saline to a final concentration of 0.1 mM and delivered via tail vein injection.

Acute experiments with craniotomy, LDF recording, and intrinsic optical imaging

For acute experiments, mice underwent a craniotomy, and the exposed cortex was covered with agarose (2%) to prevent dehydration and maintain a penetrable seal. Baseline CBF was measured using LDF while delivering optogenetic stimulation (30-ms pulses at 5 Hz for 1 s, repeated every 30 s) sequentially at two cortical sites (locations A and B). Hemodynamic responses were recorded simultaneously using intrinsic optical imaging. The two cortical sites were arbitrarily selected within the exposed cortex, without targeting specific coordinates, as the focus was on general cortical responses. For intrinsic optical imaging, the cortical surface was illuminated with low-intensity 570-nm light to capture the changes in light absorption from total hemoglobin. Changes in reflectance were captured at 10 Hz using a CoolSNAP HQ2 Photometrics camera. Following baseline, 200 nl of L-NPA or vehicle was injected into one of the two sites at a rate of 5 nl/s using a Nanoject III injector. The assignment of L-NPA or vehicle and the order of injections were randomized across animals. After a 1- to 2-min incubation, optogenetic stimulation was repeated at the injected site while recording CBF with LDF. The second site was then injected with the alternate solution (vehicle or L-NPA), followed by another round of optogenetic stimulation and CBF recording.

2P microscopy

In vivo 2P microscopy was carried out using the Ultima 2Pplus system (Bruker). Mice, awake and head fixed, were positioned on a custom-designed platform to allow for both light and sensory stimulation during in vivo brain imaging. Images were captured through a 16× water immersion objective (0.80 numerical aperture, Nikon, Inc.), using a wavelength of 900 nm for imaging YFP/green fluorescent protein (GFP), GCaMP6f, and blood vessels. The maximum field of view for imaging was 800 μm by 800 μm. Laser power was calibrated at the objective lens, ensuring that it remained under 40 mW during time-series acquisition. The onset of stimulation (t = 0) and image collection were synchronized via a National Instruments board, which tracked frame triggers. Motion correction of the time-series images was performed using a translation-only Fourier-based algorithm in MATLAB, with the first frame of each series serving as the reference point.

2P calcium (Ca²⁺) imaging

Calcium imaging was performed using the 2P setup described above to quantify Ca²⁺ changes in GCaMP6f-expressing neurons and astrocytes. For determination of Ca²⁺ responses in GCaMP6f-labeled neurons in Tacr1^CreER-GCaMP6f mice or astrocytes in Tacr1^CreER-ChR2-YFP mice, regions of interest (ROI) were selected encompassing individual cells. Ca²⁺ transients in the ROIs were longitudinally recorded by 2P microscopy (excitation, 900 nm). In Tacr1^CreER-ChR2-YFP mice, cell morphology was used to distinguish astrocytes (expressing GCaMP6f) from Tacr1 neurons (expressing ChR2) in vivo.

To further distinguish Tacr1 neurons expressing ChR2-eYFP from astrocytes expressing GCaMP6f, we also imaged the same regions at both 900 and 950 nm. ChR2 is known to exhibit significantly brighter fluorescence at 950 nm compared to 900 nm, whereas GCaMP6f does not show this shift. Therefore, cells with increased brightness at 950 nm relative to 900 nm were identified as Tacr1 neurons, providing an additional criterion for cell-type identification, as shown in Fig. 2C.

The change in the endogenous Ca²⁺ signal (ΔF/F₀) was calculated as (fluorescence_time − fluorescence_baseline)/fluorescence_baseline, where the average intensity for the spontaneous activity was used to determine fluorescence_baseline. Ca²⁺ peaks per min were quantified by automatically counting the peaks that were 30% above the baseline and then dividing by the length of the imaging session.

To quantify the change in Ca²⁺ signal (ΔF/F₀) following whisker stimulation, the signal was calculated as (fluorescence_time − fluorescence_diameter)/fluorescence_diameter, where baseline signal was the mean value during the 4 s preceding whisker stimulation. The maximum Ca²⁺ signal (max. ΔF/F₀) was identified as the average of the largest values within the prestimulus time window (−4 to −1 s) or poststimulus (1 to 4 s for Tacr1 neurons and 3 to 11 s for astrocytes) following whisker or light stimulation, where t = 0 is the onset of the stimulus. Ca²⁺ responses were quantified with MATLAB.

Vessel analysis

Mice were given a subcutaneous injection of SR-101 (Thermo Fisher Scientific, 0.2 μl g⁻¹) to visualize the cerebral vasculature in vivo. Imaging was conducted on parenchymal arterioles and capillary branches stemming from the middle cerebral artery, with two to four locations imaged per mouse during each session. Imaging sessions were performed on different days, and arteriolar dilation responses were averaged across sessions for each mouse and condition. To calculate the percentage change in vessel diameter relative to baseline, the time-series images were first processed using a Gaussian filter and background subtraction with a rolling ball of 50 pixels. Vessel diameter was measured as the full-width half-maximum of the vessel profile using custom MATLAB scripts. Time-series data were smoothed using the default smooth function in MATLAB (either moving average or Savitzky-Golay, five points). The mean vessel diameter (micrometers) was calculated as the average diameter over the imaging session (Fig. 3F). The change in vessel diameter (ΔD/D₀) was calculated as (diameter_time − baseline_diameter)/baseline_diameter, where baseline diameter was the mean value during the 4 s preceding whisker stimulation. Maximum vessel dilation (max. ΔD/D₀) was identified as the average of the largest values within the prestimulus time window (−4 to −1 s) or poststimulus (1 to 4 s) following whisker stimulation, where t = 0 is the onset of the stimulus (Figs. 2E and 3M).

Delay to response

To analyze the responses of vasodilation and astrocytic Ca²⁺ dynamics to stimuli, the following parameters were determined:

1) Maximum vessel diameter and astrocytic peak: The maximum vessel diameter and astrocytic calcium peak were defined as the maximum values recorded following the initiation of either whisker or light stimulation.

2) Onset of astrocytic Ca²⁺ increase: The onset of the astrocytic Ca²⁺ increase was determined by fitting a linear slope between 20 and 80% of the maximum calcium peak. The x intercept of the slope was used considered as the onset of the Ca²⁺ response.

3) Calculation of 50% Ca²⁺ increase: To calculate the 50% Ca²⁺ increase, the slope from the astrocytic calcium onset to the maximum calcium peak was calculated. The resulting value was divided by two to determine the time point corresponding to a 50% increase in Ca²⁺ response.

4) Minimum vessel diameter: The minimum vessel diameter was defined as the smallest diameter value observed after the maximum vessel diameter value following stimulation.

5) Calculation of 50% vasoconstriction: The slope between the maximum vessel diameter and the minimum diameter was calculated, and the resulting value was divided by two to determine the time point corresponding to 50% vasoconstriction.

6) Delay to response: To evaluate the delay to response for both vessel constriction and astrocytic Ca²⁺ dynamics, the time points at which the 50% values were reached were identified and recorded relative to the initiation of the stimulus.

Viral mapping

For viral mapping, AAVrg-FLEX-tdTomato was injected into the left S1 cortex and AAVrg-Ef1a-DIO-eYFP into the right S1 at coordinates: AP, −0.82 mm; ML, ±2.9 mm; and DV, 0.7 mm. The mice were allowed to recover for 5 weeks before brain tissue was collected for IHC staining. Images were collected at 10× using a fluorescent microscope (Leica THUNDER).

Immunohistochemistry

Adult mice with Tac1^Cre and Tacr1^CreER genotypes underwent transcardial perfusion using 4% paraformaldehyde (PFA). The brains were extracted from the skull and fixed overnight. Coronal or sagittal sections measuring 40 μm were cut using a vibratome (Leica 1200) and processed for free-floating IHC. The brain sections were blocked with 10% donkey serum in PBS with Triton (0.1% Triton X-100) and incubated overnight at 4°C with the following primary antibodies at specified concentrations: GFP (1:1000; Thermo Fisher Scientific, A-11122; Aves Lab Inc., GFP-1020), RFP (1:1000; Rockland, 600-401-379S), and parvalbumin (1:3000; SWANT, PVG-213). This was followed by application of secondary antibodies (1:500; Thermo Fisher Scientific): Alexa Fluor 488 to detect eYFP, Alexa Fluor 555 to detect tdTomato, and Alexa Fluor 647 to detect parvalbumin, which was displayed in magenta for visualization. The sections were mounted in ProLong Gold Antifade Mountant with 4′,6-diamidino-2-phenylindole (DAPI; Thermo Fisher Scientific) for nuclear staining and imaged at 10× using a fluorescent microscope (Leica THUNDER). Anatomical regions were identified according to the Allen Institute Mouse Brain Atlas.

Multiplex FISH

Mice were anesthetized with 3 to 5% isoflurane and rapidly decapitated. Brains were extracted within 2 min, embedded in Optimal Cutting Temperature (OCT) compound, and flash frozen using 2-methylbutane chilled on dry ice. Tissue blocks were stored at −80°C until cryosectioning. Coronal cryosections (15 μm) were collected on SuperFrost Plus slides and stored at −80°C.

For RNAscope, slides were removed from −80°C and immediately fixed in ice-cold 4% PFA for 15 min and then rinsed twice in 1× PBS. Tissue was dehydrated through a graded ethanol series (50%, 70%, and twice in 100%, each for 5 min at room temperature) and air-dried for ~5 min before drawing a hydrophobic barrier around each section. Slides were treated with RNAscope hydrogen peroxide for 10 min at room temperature, washed in Milli-Q water, and incubated with protease IV for 15 min at room temperature.

FISH was then performed using the RNAscope Multiplex Fluorescent v2 Assay kit (Advanced Cell Diagnostics, catalog no. 323100) according to the instructions provided by the for fresh-frozen tissue. Probes targeting Mm-Tac1 (catalog no. 410351), Mm-Pvalb (catalog no. 421931), and Mm-Slc17a6 (catalog no. 428871) were hybridized to the tissue for 2 hours at 40°C in a HybEZ oven. During all 40°C incubations, a damp paper towel was placed inside the oven to maintain humidity and minimize evaporation from tissue sections. After hybridization, slides were washed twice in wash buffer (provided in the kit) and then subjected to a series of amplification steps (AMP1, AMP2, and AMP3) with 15- to 30-min incubations at 40°C, each followed by washes.

Probe signal was detected using horseradish peroxidase (HRP)–based channel-specific labeling (C1, C2, or C3) and developed with TSA Plus fluorophores according to channel assignment. HRP signal was blocked between fluorophore applications to prevent cross-channel amplification. Slides were counterstained with ProLong Gold Antifade Mountant with DAPI (Thermo Fisher Scientific) and coverslipped. Slides were cured overnight at room temperature in the dark and stored at 4°C until imaging.

Image acquisition and quantification

Tissue sections were visualized using a fluorescent microscope (Leica THUNDER). IHC images were analyzed and quantified with the cell counter tool in ImageJ (67). For FISH experiments, fluorescence images of tissue samples were captured, and only cells with clearly defined nuclei, as indicated by DAPI staining, and exhibiting fluorescent signals were included in the count.

Statistical analyses

All statistical analyses were performed using Prism 10 (GraphPad Software). Two group comparisons were analyzed using a two-tailed paired t test (Figs. 1, D and F; 2, E, F, H, and I; 3, E, F, I, J, L, and N; and 5, F, H, and J; and figs. S1G; S2, D, G, H to J, O, and P; S3, C and D; and S4D) or a two-tailed unpaired t test (Fig. 2, M to P; and fig. S2, B, F, K, and L). Multiple group comparisons were analyzed using a one-way analysis of variance (ANOVA), followed by a post hoc Bonferroni analysis to correct for multiple comparisons (Fig. 1H and fig. S2N). All data were included in the statistical analysis, and no tests were conducted to verify whether the data met the assumptions of the chosen statistical methods. For all experiments, the SEM was calculated and shown as error bars in the graphs and as shared areas in representative traces. Detailed statistical information for individual experiments, such as exact sample sizes (n), what n represents, measures of precision, statistical tests used, and significance criteria, is provided in the figure legends. Data and shaded areas in representative traces are presented as means ± SEM. No animals were excluded from the analyses.

Supplementary Materials

This PDF file includes:

Figs. S1 to S6