Parvalbumin interneurons drive depth-dependent vascular responses

Abstract

Understanding how neurons regulate cerebral blood flow (CBF) is crucial for interpreting brain hemodynamic imaging signals. Although parvalbumin (PV) interneurons strongly influence network activity, their effects on CBF appear variable, particularly during natural sensory stimulation and resting state. Using two-photon imaging in awake mice, we assessed PV interneuron effects on vascular tone across cortical depths during evoked and spontaneous activity. Optogenetic PV activation produced rapid vasodilation in the superficial cortex (<250 μm) and slow, delayed vasodilation in the middle cortex (250-400 μm). Chemogenetic PV suppression revealed that PV interneurons did not drive the rapid hemodynamic increase to sustained sensory input but did contribute to superficial vasodilation following brief stimuli. During spontaneous activity, synchronized PV interneuron activity predicted depth-specific arteriolar changes more strongly than non-PV neurons. These findings demonstrate a depth-dependent role for PV interneurons in regulating CBF and refine our understanding of neurovascular coupling.

Keywords: Neuroscience; Physiology; Techniques in neuroscience.

Graphical abstract

Figure 1

Animal preparation (A) Cranial window implanted in PV-cre mouse S1 where we injected optogenetic protein (AAV-DIO-ChR2-YFP) and calcium reporter (AAV-Syn-RCaMP1a) viruses. Surface fluorescence microscope images below show the regions expressing ChR2-YFP (green box) and RCaMP1a (red box). (B) Cranial window implanted in PV-cre mouse S1, where we injected inhibitory DREADD (AAV-DIO-hM4D(Gi)-mCherry) and calcium reporter (AAV-Syn-GCaMP8f) viruses. Surface fluorescence microscope images below show the regions expressing hM4D(Gi)-mCherry (red box) and GCaMP8f (green box). Scale bars: 1 mm.

Figure 2

Optogenetic stimulation activates PV interneurons and inhibits non-PV neurons (A) 2P setup for recording calcium activity in neurons expressing RCaMP1a in response to the optogenetic activation of PV-ChR2 cells in awake mice. (B–D) (B) 2P microscopy images of S1 intra-cortical sections (∼400 μm depth) show the regions expressing ChR2-YFP, (C) RCaMP1a, and (D) color merge. Scale bars: 50 μm. (E) An example of RCaMP1a difference maps at pre-stimulation (4–5 s), during stimulation (5–6 s), immediately after stimulation (6–7 s), and post-stimulation (7–8 s). In this particular example, a 10-ms optogenetic light stimulus was delivered for 1 s at 5 Hz. Optogenetic stimulus, some activated and some inhibited cells are indicated by a black bar, green and purple arrows, respectively. (F) Summary of Ca²⁺ fluorescence responses to optogenetic stimuli of various durations and frequencies in PV+ cells across cortical depth (n = 10 mice). (G and H) (G) Time profile of average RCaMP1a changes in response to 4-s optogenetic stimulation in PV+ cells and (H) PV- cells (n = 4–10 mice). (I and J) (I) Summary of average RCaMP1a changes at different frequencies in PV (n = 4–10 mice) and (J) in PV- cells (n = 4–9 mice). The opaque blue rectangle denotes the optogenetic stimulation period (4 s). Significant differences from baseline (pre-stimulation period) are denoted by ∗ for p < 0.05 or by ∗∗ for p < 0.01, t test.

Figure 3

Optogenetic stimulation of PV interneurons induces depth-dependent vascular responses (A) LDF setup for recording CBF measurements in response to the optogenetic activation of PV-ChR2 interneurons in awake mice. (B) Time profiles of average CBF changes measured with LDF in response to the optogenetic stimulation of various frequencies (n = 10–11 mice). (C) Summary of average CBF changes in response to the optogenetic stimulation of various frequencies (n = 10–11 mice). (D) 2P setup for recording vessel diameter changes in response to the optogenetic activation of PV-ChR2 interneurons in awake mice. (E) 2P microscopy color merge image (scale bars: 50 μm) of S1 intra-cortical section at ∼200 μm and ∼400 μm depth with zoom-in frames (scale bars: 5 μm) of arteries at pre-stimulation (4–5 s), during stimulation (5–6 s in 1-s stimulation or 5–9 s in 4-s stimulation), immediately after stimulation (6–7 s in 1-s stimulation and 9–10 s in 4-s stimulation), and post-stimulation (9–10 s, 19–20 s in 1-s stimulation and 19–20 s, 29–30 s in 4-s stimulation). Red bar indicates the region where the diameter was measured. Black circles indicate the frames during the stimulation period. (F) Peak time of vessel diameter change as a function of cortical depth with a fitted linear regression model (slope = −0.033 μm/s, intercept = 7.573 s, correlation coefficient = 0.491). Dashed line represents the beginning of stimulation (0 s). (G) Time courses of arterial diameter changes in superficial cortex in PV-cre mice expressing ChR2 in 1-s and 4-s stimulation experiments (3–16 vessels, 3–10 mice). (H) Time courses of arterial diameter changes in the middle cortex in 1-s and 4-s stimulation experiments (6–8 vessels, 6 mice). (I) Summary of average arterial diameter changes in superficial cortex at different frequencies in 1-s and 4-s stimulation experiments (3–16 vessels, 3–10 mice). (J) Summary of average arterial diameter changes in the middle cortex at different frequencies in 1-s and 4-s stimulation experiments (6–8 vessels, 6 mice). The opaque blue rectangle denotes the optogenetic stimulation period. Significant differences from baseline are denoted by ∗ for p < 0.05 or by ∗∗ for p < 0.01, t test.

Figure 4

PV interneurons contribute little to rapid hemodynamic responses (A) 2P setup for recording calcium activity in neurons expressing GCaMP8f and vessel diameter changes in response to whisker stimulation in awake PV-hM4D(Gi) mice. (B and C) Time profiles of average GCaMP8f changes in activated PV+ (n = 15) and PV- (n = 6–8) cells measured with 2P Ca²⁺ imaging in response to whisker stimulation (11 mice). (D and E) Summary of average GCaMP8f changes in activated PV+ (n = 15) and PV- (n = 6–8) cells in response to whisker stimulation (11 mice). (F) LDF setup for recording CBF in response to whisker stimulation experiment in awake PV-hM4D(Gi) mice. (G) Time profiles of average CBF changes measured with LDF in response to whisker stimulation (1 and 4 s) in control conditions (baseline) and after DREADD activation (DCZ injection; n = 11–12 mice). (H) Summary of average CBF changes in response to whisker stimulation in control conditions (baseline) and after DREADD activation (DCZ injection; n = 11–12 mice). (I) Time profiles of average vessel diameter changes measured with 2P in response to whisker stimulation (9–11 vessels, 9–11 mice). (J) Summary of average vessel diameter changes in response to whisker stimulation (9–11 vessels, 9–11 mic). Non-significant differences are denoted by n.s. Significant differences are denoted by ∗ for p < 0.05 or ∗∗ for p < 0.01, mixed design ANOVA. The semitransparent blue rectangle denotes the whisker stimulation period, and the opaque blue rectangle denotes the optogenetic stimulation period.

Figure 5

Ongoing PV interneuron activity contributes to spontaneous arterial fluctuations across cortical depth (A) Example time profiles of Ca²⁺ fluorescence changes during ongoing activity in PV+ cell and PV- cell. (B) Example time profiles of measured spontaneous arterial diameter changes (red trace) and those predicted from the PV+ and from the PV- cell activity (black trace). (C) HRFs obtained from convolving the gamma function with Ca²⁺ fluorescence changes in PV+ (green trace) and PV- (purple trace) cells during ongoing activity. (D) Summary of PV+ (254 cells, green) and PV- (751 cells, purple) cell fractions predicting vascular activity at high (r > 0.2), mid (0.1 < r < 0.2), and low (r < 0.1) levels, where r is the Pearson correlation between measured and predicted spontaneous arterial diameter changes. The first group (<1 Hz) displays correlations for fast vascular activity (>1 Hz), while the second group shows correlations for slow vascular activity (<0.06 Hz). 22 recordings in 6 animals. (E) Example time profiles of the average correlation between the activity of PV+ cells (PV+ Synchrony) along with spontaneous arterial diameter changes (vascular activity) at 200 and 400 μm cortical depths. Note that the bottom panel vessel is the same vessel shown in B. (F) Peak time of spontaneous vasodilation as a function of cortical depth with a fitted linear regression model. Relative to PV+ synchrony: slope = −0.038 μm/s, intercept = 5.744 s, correlation coefficient = 0.549, p = 8 × 10⁻⁴, t test. (G) Relative to PV- synchrony: slope = −0.006 μm/s, intercept = 12.673 s, correlation coefficient = 0.082, p = 0.647, t test. Vertical dashed lines represent the beginning of synchrony events (0 s; green - PV+, purple - PV-).

Figure 6

PV and Tac1 cell distribution across cortical depth in mouse and human cortex (A and B) Representative images show labeling of PV (green), Tac1 (red) cells, and cell nuclei (blue) in mouse (scale bars: 200 μm) and human brain slices (scale bars: 400 μm). (C and D) Distribution of PV and Tac1 cell density across cortical depth in mouse and human slices. The bold line indicates the mean cell density across cortical depth, with the shaded area shows the standard error (n = 6 mice, n = 3 humans).