Targeted ablation of type-I nNOS neurons with cortical SP-SAP injections

(a) Schematic showing intracortical administration of the ribosome inactivating protein saporin conjugated to either SP or a vehicle control. (b) Representative immunofluorescence of anti-nNOS (yellow) and anti-NKR1 (blue) showing colocalization of the NK1 receptor on cortical nNOS-positive neurons. Representative immunofluorescence of anti-nNOS (yellow) and DAPI (magenta) from animals with no injection (Uninjected, left), Blank-SAP (middle), or SP-SAP (right). (c) Representative immunofluorescence of Blank-SAP (top) and SP-SAP (bottom) sections co-stained with nNOS (left), IBA1 (middle left), GFAP (middle), DAPI (middle right), or NeuN (right). NeuN image was taken from section immediately adjacent to the first four. (d) Quantification of nNOS counts, (e) IBA1 counts, (f) GFAP fluorescence, (g) DAPI fluorescence, (h) NeuN fluorescence. Error bars (d-h) denote SD. Bonferroni correction (5) *α < 0.01, **α < 0.002, ***α < 0.002. k-l *α < 0.05, **α < 0.01, ***α < 0.001, GLME.

Simultaneous measurement of neural and hemodynamic signals in mice across arousal states

(a) Schematic of widefield optical imaging experimental setup. The brain is illuminated at an isosbestic wavelength of hemoglobin (568 nm). Changes in reflected light measuring changes in total blood volume are captured by a camera mounted above the head while several other cameras monitor animal behavior and arousal state including vibrissae and pupil tracking. Vibrissae stimulation is done by directed air puffs. Polished and reinforced thinned-skull windows were bilaterally implanted over the somatosensory cortex. Tungsten stereotrodes were implanted underneath each window to record changes in local field potential within the area of interest. An additional hippocampal stereotrode and a neck electromyography electrode were used to assist in sleep scoring. (b) Schematic of widefield optical imaging experiments with GCaMP7s. Alternating illumination with 480 nm, 530 nm, and 630 nm light were used to measure changes in total hemoglobin, blood oxygenation, and GCaMP7s fluorescence. (c, e) During the awake state, power in low-frequency cortical LFP is low and power in the gamma-band (30-100 Hz) is elevated during activity such as volitional whisking. (d, f) Periods of NREM and REM sleep are accompanied by large oscillations in cerebral blood volume, with large increases in power in delta-band (1-4 Hz) cortical LFP during NREM and large increases in theta-band (4-10 Hz) hippocampal LFP during REM.

Ablation of type-I nNOS neurons reduces stimulus-evoked hemodynamic responses

(a) Change in total hemoglobin in response to brief (0.1 second) vibrissae stimulation. (b) Change in total hemoglobin in response to extended (5 seconds) vibrissae stimulation. (c) Change in GCaMP fluorescence in response to extended vibrissae stimulation. (d) Change in oxy- and deoxy-hemoglobin in response to extended vibrissae stimulation. (e) Change in arteriole diameter in response to extended vibrissae stimulation. Error bars represent population averages ± SEM. All statistics were evaluated between the indicated intervals. *p < 0.05, **p < 0.01, ***p < 0.001, GLME.

Type-I nNOS ablation alters low frequency neural activity and gamma-band neurovascular coupling

(a) Local field potential within the vibrissae representation of somatosensory cortex during periods of Alert. (b) Asleep. (c) All data. (d) Change in gamma-band power vs. Δ[HbT] following brief (0.1 seconds) vibrissae stimulation. (e) Change in GCaMP7s fluorescence vs. Δ[HbT] following prolonged (5 seconds) vibrissae stimulation. Error bars represent population averages ± SEM. All statistics were evaluated between the indicated intervals. *p < 0.05, **p < 0.01, ***p < 0.001, GLME (a-c), GLM (d, e).

Neurovascular coupling was only weakly affected by type-I nNOS removal

(a) Schematic demonstrating intracortical injection of either Blank-SAP or SP-SAP and the analysis of gamma-band power and hemodynamic signals from within the vibrissae representation of somatosensory cortex, N = 9 mice per group. (b) Gamma-Δ[HbT] resting-state cross-correlation (c) Gamma-Δ[HbT] alert cross-correlation (d) Gamma-Δ[HbT] asleep cross-correlation (e) Gamma-Δ[HbT] resting-state coherence (f) Gamma-Δ[HbT] alert coherence (g) Gamma-Δ[HbT] asleep coherence (h) Schematic demonstrating intracortical injection of either Blank-SAP or SP-SAP and the analysis of GCaMP7s fluorescence and hemodynamic signals from within the vibrissae cortex, n = 6-7 mice per group. (i) GCaMP7s-Δ[HbT] resting-state cross-correlation (j) GCaMP7s-Δ[HbT] alert cross-correlation (k) GCaMP7s-Δ[HbT] asleep cross-correlation (l) GCaMP7s-Δ[HbT] resting-state coherence (m) GCaMP7s-Δ[HbT] alert cross-correlation (n) GCaMP7s-Δ[HbT] asleep cross-correlation. Error bars represent population averages ± SEM. *p < 0.05, **p < 0.01, ***p < 0.001, GLME.

Type-I nNOS ablation reduces low-frequency interhemispheric coherence

(a) Δ[HbT] coherence between bilateral ROIs in the left and right hemisphere’s somatosensory cortex during (b) Res.t (c) Alert. (d) Asleep. (e) Gamma-band power coherence between bilateral ROIs in the left and right hemisphere’s somatosensory cortex during (f) Res.t (g) Alert. (h) Asleep. (i) Δ[HbT] coherence between bilateral ROIs in the left and right hemisphere’s somatosensory cortex during (j) Res.t (k) Alert. (l) Asleep. (m) GCaMP7s coherence between bilateral ROIs in the left and right hemisphere’s somatosensory cortex during (n) Res.t (o) Alert. (p) Asleep. Error bars represent population averages ± SEM. *p < 0.05, **p < 0.01, ***p < 0.001, GLME.

Type-I nNOS ablation reduces resting-state hemodynamics and low frequency neural activity

(a) Variance in resting-state hemodynamics signals measured with widefield optical imaging. (b) Variance in resting-state arteriole diameter measured with two-photon. (c) Average Δ[HbT] during periods of NREM sleep, periods of REM sleep, and following administration of isoflurane (n = 9 mice per group). (d) Change in normalized arteriole diameter following administration of isoflurane Error bars represent population averages ± SD. All statistics were evaluated between the indicated intervals. *p < 0.05, **p < 0.01, ***p < 0.001 GLME.

Histological quantification of cortical SP-Sap injections

(a) Examples of NADPH diaphorase staining fromBlank-SAP and SP-SAP injected mice. (b) SP-SAP injected mice (N = 9, 5M/4F) had significantly lower numbers of type-I nNOS cells than either Blank-SAP (N = 9, 4M/5F) or Uninjected mice (N = 9, 4M/5F) (SP-SAP: 1.6 ± 0.8 neurons/mm2; Blank-SAP: 22.9 ± 2.4 neurons/mm2; Uninjected: 22.9 ± 3 neurons/mm2; Blank-SAP vs. Uninjected: p = 0.98; Blank-SAP vs. SP-SAP: p = 2.23×10-13). (c) Counts of DAPI-labeled cell nuclei per square mm from imaged mice. Uninjected (N = 6, 6M) mice had 2284 ± 279 DAPI-labeled cell bodies/mm2, which was not significantly different than those injected with Blank-SAP (N = 5, 5F, 2488 ± 298, p = 0.23, GLME), which in turn was not significantly different than those injected with SP-SAP (N = 8, 4M/4F, 2340 ± 329, p = 0.39, GLME. (d) Representative image of IBA1 for Blank-SAP (top) and SP-SAP (bottom) taken from Fig 1c. Error bars (b, c) denote SD. *α < 0.05, **α < 0.01, ***α < 0.001, GLME.

Behavior was unaffected by local ablation of type-I nNOS neurons

(a) Example of exploratory behavior during 5 minutes of exploration in a novel environment. (b) Total distance traveled during open field exploration. Uninjected mice traveled an average of 1473.4 ± 364.7 cm compared with 1536.6 ± 329.6 cm in Blank-SAP (p = 0.61) and 1652.3 ± 442.7 cm in SP-SAP (p = 0.43). (c) Time spent in the center of the open field. Uninjected mice spent on average 25.5 ± 13.1% of total exploration time in the center compared with 18.5 ± 5.3% in Blank-SAP (p = 0.08, GLME) and 21.9 ±9.8% in SP-SAP (p = 0.26). (b, c) n = 12-23 mice per group. (d) Percentage of time classified in each arousal state by sleep scoring. Uninjected mice spent 68.1 ± 13.2% of the time awake compared to 69.1 ± 15.1% in Blank-SAP (p = 0.89, ttest) and 73.4 ± 11.6% in SP-SAP (p = 0.51, ttest). Uninjected mice spent 28.3 ± 11.1% of the time in NREM sleep compared to 27.4 ± 12.0% in Blank-SAP (p = 0.87, ttest) and 23.4 ± 9.7% in SP-SAP (p = 0.44, ttest). Uninjected mice spent 3.5 ± 2.9% of the time in REM sleep compared to 3.4 ± 3.5% in Blank-SAP (p = 0.96, ttest) and 3.2 ± 2.3% in SP-SAP (p = 0.88, ttest). (e) Time spent whisking as a percentage of total imaging time. Uninjected mice spent 15.3 ± 9.9% compared with 17.3 ± 5.8% in Blank-SAP (p = 0.61, ttest) and 19.2 ± 8.1% in SP-SAP (p = 0.75, ttest). (f) Change in Pupil diameter in response to vibrissae stimulation. Pupil diameter of uninjected mice increased 1.3 ± 0.5 z-units following vibrissae stimulation compared with 1.2 ± 0.3 z-units in Blank-SAP (p = 0.76, ttest) and 1.1 ± 0.5 z-units in SP-SAP (p = 0.50, ttest). (g) Average pupil diameter across arousal states. Average pupil diameter of Uninjected mice during the resting-state was -0.25 ± 0.1 z-units compared with -0.29 ± 0.1 z-units in Blank-SAP (p = 0.45, ttest) and - 0.34 ± 0.2 in SP-SAP (p = 0.14, ttest). Average pupil diameter of Uninjected mice during periods of NREM sleep was -3 ± 0.5 z-units compared with -3 ± 0.5 z-units in Blank-SAP (p = 0.89, ttest) and -3.4 ± 0.4 in SP-SAP (p = 0.08, ttest). Average pupil diameter of Uninjected mice during periods of REM sleep was -4.6 ± 0.8 z-units compared with -4.4 ± 0.8 z-units in Blank-SAP (p = 0.53, ttest) and -5 ± 0.6 in SP-SAP (p = 0.11, ttest) (h) Interblink-interval. Uninjected mice had a mean interblink-interval of 127.6 ± 46.8 s compared with 102.5 ± 34.7 s in Blank-SAP (p = 0.22, ttest) and 84.5 ± 23.3 s in SP-SAP (p = 0.21, ttest). (d-h) n = 9 mice per group. Shading (f) indicates SEM and error bars (b, c, e, g, h) denote SD. *α < 0.05, **α < 0.01, ***α < 0.001.

Sleep classification accuracy was unchanged following type-I nNOS ablation

(a) Confusion matrix for arousal state classification of Uninjected mice with bilaterally implanted stereotrodes. (b) Confusion matrix for arousal state classification of Blank-SAP mice with bilaterally implanted stereotrodes. (c) Confusion matrix for arousal state classification of SP-SAP mice with bilaterally implanted stereotrodes. (a-c) n = 9 mice per group. (d) Confusion matrix for arousal state classification of Blank-SAP mice expressing pan-neuronal GCaMP. (e) Confusion matrix for arousal state classification of SP-SAP mice expressing pan-neuronal GCaMP. (d, e) n = 6-7 mice per group. (f) Out-of-bag error during training of each animal’s bootstrapped random forest classification algorithm. For mice with bilateral LFP recordings, Uninjected mice had an average loss of 0.08 ± 0.03 in comparison to 0.07 ± 0.02 in Blank-SAP (p = 0.45, ttest) compared to 0.05 ± 0.02 in SP-SAP (p = 0.064, ttest). Blank-SAP mice with pan-neuronal GCaMP had an average classification loss of 0.06 ± 0.02 compared with 0.07 ± 0.02 in SP-SAP (p = 0.80, ttest). Error bars denote SD. *α < 0.05, **α < 0.01, ***α < 0.001.

Ablation of Type-I nNOS neurons does not alter hemodynamic or neural power spectra

Power spectral density for vascular and hemodynamic signals for the Blank-SAP group (N = 9, 4M/5F for a, b; N = 7, 3M/4F for c-f) and SP-SAP group (N = 9, 5M/4F for a, b; N = 8, 4M/4F for c-f) was not significantly different across all measurements of hemodynamic and neural signals. (a) Δ[HbT] for electrophysiology animals (p = 0.77, GLME). (b) Gamma-band power (second spectra) for animals with electrophysiology (p = 0.19, GLME). (c) Δ[HbT] for animals with GCaMP (p = 0.27, GLME). (d) Δ[HbT] for animals with GCaMP (p = 0.26, GLME). (e) Δ[HbT] for animals with GCaMP (p = 0.68, GLME). (f) GCaMP fluorescence (p = 0.44, GLME). Shading represents population averages ± SEM. *α < 0.05, **α < 0.01, ***α < 0.001.

Ablation of Type-I nNOS neurons does not alter the predictive power of the hemodynamic response function

(a) Stimulus-evoked hemodynamic response function obtained with deconvolution. (b) Stimulus-evoked hemodynamic response function fitted with a gamma distribution function. (c) There was no significant difference between the Δ[HbT] predictive R2 values calculated from impulse-derived HRFs between Blank-SAP (N = 9, 4M/5F) 0.64 ± 0.11 or SP-SAP (N = 9, 5M/4F) 0.67 ± 0.16 following vibrissae stimulation (p = 0.746, ttest). (d) Resting-state hemodynamic response function based on deconvolution. (e) Resting-state hemodynamic response function fitted with a gamma distribution function. (f) There was no significant difference between the Δ[HbT] predictive R2 values calculated from impulse-derived HRFs between Blank-SAP (N = 9, 4M/5F) 0.13 ± 0.0.05 or SP-SAP (N = 9, 5M/4F) 0.12 ± 0.05 during the resting-state (p = 0.749, ttest). Error bars denote SD. *α < 0.05, **α < 0.01, ***α < 0.001.

Pearson’s correlation coefficients between bilateral hemodynamic and neural signals

(a) Correlation coefficient between bilateral hemodynamic signals during the resting-state was 0.75 ± 0.04 in Blank-SAP and 0.68 ± 0.06 in SP-SAP (ttest p = 0.011), during the alert state was 0.85 ± 0.03 with Blank-SAP and 0.81 ± 0.06 in SP-SAP (ttest p = 0.09), and during the asleep state was 0.95 ± 0.02 in Blank-SAP and 0.92 ± 0.03 in SP-SAP (ttest p = 0.01). (b) Correlation coefficient between bilateral gamma-band power signals during the resting-state was 0.15 ± 0.04 in Blank-SAP and 0.14 ± 0.12 in SP-SAP (ttest p = 0.69), during the alert state was 0.23 ± 0.08 in Blank-SAP and 0.19 ± 0.11 in SP-SAP (ttest p = 0.31), and during the asleep state was 0.50 ± 0.08 in Blank-SAP and 0.41 ± 0.08 in SP-SAP (ttest p = 0.04). (c) Correlation coefficient between bilateral hemodynamic signals (GCaMP7s group) during the resting-state was 0.82 ± 0.04 in Blank-SAP and 0.75 ± 0.04 in SP-SAP (ttest p = 0.01), during the alert state was 0.93 ± 0.02 in Blank-SAP and 0.91 ± 0.02 in SP-SAP (ttest p = 0.08), and during the asleep state was 0.98 ± 0.02 with in-SAP and 0.95 ± 0.03 in SP-SAP (ttest p = 0.07). (d) Correlation coefficient between bilateral GCaMP7s signals during the resting-state was 0.70 ± 0.05 in Blank-SAP and 0.65 ± 0.09 in SP-SAP (ttest p = 0.16), during the alert state was 0.93 ± 0.03 with Blank-SAP and 0.92 ± 0.03 in SP-SAP (ttest p = 0.31), and during the asleep state was 0.96 ± 0.02 with Blank-SAP and 0.95 ± 0.02 in SP-SAP (ttest p = 0.37). Error bars denote SD. α < 0.05, **α < 0.01, ***α < 0.001.

Removal of type-I nNOS neurons did not alter arousal state-related hemodynamic changes (a-c)

Average Δ[HbT/O/R] during periods of NREM sleep and REM sleep in mice with pan-neuronal GCaMP (n = 6-7 mice per group). (a) NREM Δ[HbT] in Blank-SAP mice was 20.4 ± 6.4 μM compared to 16.5 ± 5.9 μM in SP-SAP (p = 0.27, ttest). REM Δ[HbT] in Blank-SAP mice was 80.0 ± 8.2 μM compared to 80.7 ± 5.9 μM in SP-SAP (p = 0.87, ttest). (b) NREM Δ[HbO] in Blank-SAP mice was 25.5 ± 6.1 μM compared to 21.2 ± 6.1 μM in SP-SAP (p = 0.23, ttest). REM Δ[HbO] in Blank-SAP mice was 90.8 ± 8.6 μM compared to 92.3 ± 5.5 μM in SP-SAP (p = 0.72, ttest). (c) NREM Δ[HbR] in Blank-SAP mice was -5.3 ± 1.1 μM compared to -4.9 ± 0.9 μM in SP-SAP (p = 0.50, ttest). REM Δ[HbR] in Blank-SAP mice was -11.4 ± 1.8 μM compared to -12.2 ± 1.5 μM in SP-SAP (p = 0.42, ttest). (d) Transition from Awake to NREM had a Δ[HbT] of -16.6 ± 6.4 μM in Blank-SAP mice and -11.1 ± 7.0 μM in SP-SAP mice (p < 0.08, GLME). (e) Transition from NREM to Awake had a Δ[HbT] of 21.5 ± 7.7 μM in Blank-SAP mice and 17.6 ± 9.5 μM in SP-SAP mice (p < 0.33, GLME). (f) Transition from NREM to REM had a Δ[HbT] of -17.6 ± 3.9 μM in Blank-SAP mice and -22.6 ± 7.9 μM in SP-SAP mice (p < 0.09, GLME). (g) Transition from REM to Awake had a Δ[HbT] of 67.8 ± 7.3 μM in Blank-SAP mice and 74.0 ± 13.1 μM in SP-SAP mice (p < 0.21, GLME). Error bars denote SD. Shading represents population averages ± SEM.*α < 0.05, **α < 0.01, ***α < 0.001.

Schematic showing how non-linearity in the dilation response can explain coexistence strong dilation by activation of a pathway, and little change after weakening a pathway.

If the diameter is a sublinear function of the sum of vasodilatory inputs, activation of all pathways will cause a dilation that is smaller than the sum of activation of each pathway individually. Loss of one pathway will not cause large changes, even though activation of that pathway in isolation can cause large dilations.