Nitric oxide is not responsible for the initial sensory-induced neurovascular coupling response in mouse cortex

Neurovascular coupling ensures that changes in neural activity are accompanied by localised changes in cerebral blood flow. While much is known about the involvement of excitatory neurons in neurovascular coupling, the role of inhibitory interneurons is unresolved. While nNOS-expressing interneurons have been shown to be capable of eliciting vasodilation, the role of nitric oxide in functional hyperemia remains a matter of debate. Therefore in the present study we applied a combination of optogenetic and pharmacological approaches, 2-dimensional optical imaging spectroscopy, and electrophysiology to investigate the role of nitric oxide in neurovascular coupling responses evoked by nNOS-expressing interneurons and whisker stimulation in mouse sensory cortex. The haemodynamic response evoked by nNOS-expressing interneurons was significantly altered in the presence of the NOS inhibitor LNAME, revealing a large initial 20-HETE-dependent vasoconstriction. In contrast, the haemodynamic response induced by sensory stimulation was largely unchanged by LNAME. Our results suggest that while nitric oxide plays a key role in neurovascular responses evoked by nNOS-expressing interneurons it does not mediate the initial sensory-induced neurovascular coupling response in mouse cortex. Thus, our results call into question the involvement of nNOS-expressing interneurons and nitric oxide in sensory-evoked functional hyperemia.


Introduction 46
The ability to regulate cerebral blood flow (CBF) in a localised and dynamic manner in response to neuronal activity is essential for the maintenance of healthy brain function. 48 While the regulation of CBF by glutamatergic neurons has been well investigated 49 photostimulation and whisker stimulation. The resulting haemodynamic changes were 123 centred around the optrode (photostimulation: Fig 1A) and localised to the whisker 124 barrel cortex (whisker stimulation: Fig 1D), as we have previously reported (Lee et al., 125 2020). Prior to treatment with LNAME, both 2s photostimulation of cortical nNOS INs 126  In agreement with reports that cortical NOS activity is reduced by more than 90% one 141 hour after i.p. injection of LNAME (Bannerman et al., 1994), 70 minutes was found to 142 be sufficient time for a significant effect of LNAME to be observed (Fig 2;  143 F(1.361,10.889)=11.65, p=0.004, η 2 =0.593; see Table 1 for pairwise comparisons). 144 Therefore, post-LNAME haemodynamic measurements were obtained 70-135 145 minutes after systemic injection of LNAME. Following treatment with LNAME, the 146 nNOS IN-evoked haemodynamic response was inverted, showing a decrease in Hbt 147 and Hbo and an increase in Hbr (Fig 1A). Inspection of the timeseries of the 148 haemodynamic response (Fig 1F) revealed a greater reduction in Hbt during the 149 photostimulation period (as compared to pre-LNAME), followed by an increase in Hbt 150 which peaked after stimulation offset (Fig 1C,F). In contrast, the haemodynamic 151 response evoked by whisker stimulation was unchanged by LNAME in terms of 152 polarity and timing ( Fig 1D). The localised haemodynamic response to whisker 153 stimulation consisted of an increase in Hbt during the stimulation period, which peaked 154 after stimulation offset, both before and after treatment with LNAME ( Fig 1E, F). analysis of the haemodynamic data. Therefore, a 3-way mixed ANOVA was used to 162 assess the effect of LNAME, stimulation type, and electrode insertion on the evoked 163 haemodynamic response. Due to the bidirectional nature of the nNOS IN-evoked 164 haemodynamic response, three metrics were analysed (Fig 1G), (i) maximal change 165 during initial response ('initial dip', 0.25-5s after stimulation onset, Fig 1H), (ii) maximal 166 change during later response ('peak', 0.25-10s after stimulation onset, Fig 1I), and (iii) 167 net change ('peak'-'initial dip', Fig 1J). For all haemodynamic profiles (Hbt, Hbo and 168 Hbr), for all metrics considered, no significant effect of electrode insertion was found 169 (Table 2-4), therefore haemodynamic data from all mice were combined (Fig 1). This 170 electrode insertion occurred approximately 50 minutes prior to collection of pre-172 LNAME data, sufficient time for haemodynamic recovery after CSD (Chang et al., 173 2010;Shabir et al., 2022). As a statistically significant interaction between stimulation 174 type and LNAME was revealed for all haemodynamic profiles (initial dip: Table 2, peak: 175 Table 3), suggesting that the effect of LNAME depends on the type of stimulation 176 applied, simple effects tests to assess the effect of LNAME for each stimulation type 177 were performed. 178

179
In the presence of LNAME, nNOS IN stimulation evoked a larger 'initial dip' (Fig 1H) Fig 1I), as compared to before LNAME injection. 187

188
The maximum nNOS IN-evoked net change (i.e. minima to maxima) in Hbt, Hbo and 189 Hbr was unchanged in the presence of LNAME (Fig 1J, Table 4). These data suggest 190 that while the initial haemodynamic response to nNOS IN activity is dependent on NO 191 production by NOS, a second, NO-independent, pathway underlies the later increases 192 in Hbt, Hbo and associated washout of Hbr.

Evoked neural activity was unaltered by NOS inhibition 218
As the measured haemodynamic responses reflect stimulation-evoked neural activity, 219 we assessed whether LNAME alters evoked neural activity. In a subset of mice, 220 stimulation-evoked electrophysiological and haemodynamic changes were measured 221 concurrently before and after LNAME injection to confirm that the observed LNAME The lack of measurable change in neural activity persisted in the presence of LNAME 229 ( Fig 3A,C). Similarly, whisker stimulation-evoked increases in MUA, which extended 230 throughout the depth of the cortex and lasted for the duration of the stimulation (Fig  231   3B), were unaffected by LNAME (Fig 3B-C). These data confirm that LNAME had no 232 effect (Peak MUA: F(1,3)=0.032, p=0.869; Mean MUA: F(1,3)=0.003, p=0.958; Table  233 6) on the neural activity underlying the stimulation-evoked localised haemodynamic 234 responses (Fig 1). 235

NO reduces 20-HETE-evoked vasoconstriction during nNOS IN activation 237
We hypothesised that the larger initial decrease in Hbt (indicative of vasoconstriction) 238 observed in response to nNOS IN activation (Fig 1A,C,F,G) in the presence of the 239 NOS inhibitor LNAME was due to 20-HETE production (Mulligan and MacVicar, 2004). before and after treatment with LNAME and HET0016. Combined with the results of 254 treating with LNAME alone (Fig 1), these data suggest that during short duration nNOS 255 IN activation NO acts, at least in part, to reduce 20-HETE-elicited vasoconstriction. As previously observed in the presence of LNAME alone (Fig 1F), an oscillation in all 264 haemodynamic components was observed on return to baseline following stimulation 265 in the presence of LNAME and HET0016 (Fig 5A), further supporting our suggestion 266 that NO plays a role in damping the haemodynamic return to baseline following nNOS 267  Therefore, in addition to characterising the effect of NOS inhibition by LNAME on 291 stimulation-evoked cortical haemodynamics, we also examined the effect of NOS 292 inhibition on low frequency arterial oscillations. We detected an increase in the power 293  Table 7-8), which was not apparent in the absence of LNAME (Fig 7,  296 'pre' vs 'post' in no drug condition, p=0.531, Table 7-8). These data confirm that NOS 297 inhibition results in enhanced low frequency vascular oscillations (Ances et al., 2010;298 Biswal and Hudetz, 1996). Additional peaks in the power spectrum which reflect the 299 frequency of stimulation (ISI of 25s: 0.04Hz) and its harmonics can be seen in all cases 300 (Fig 7). As stimulation paradigms were interleaved and LNAME alters the 301 haemodynamic response to photostimulation of nNOS INs but not whisker stimulation 302 (Fig 1), the peak associated with the stimulation pattern is shifted to a lower frequency 303

Haemodynamic responses to whisker stimulation and nNOS IN activation sum
in the presence of LNAME ( Fig 7B). Whilst NO was found to mediate nNOS IN driven changes in cerebral blood volume, 335 the neurovascular coupling response to short duration sensory stimulation was largely 336 unaltered when NO production was reduced (Fig 1). A lack of effect on stimulation-337 evoked neural activity (Fig 3) confirmed that sensory-evoked neurovascular coupling 338 was also unaffected by treatment with LNAME. Although nNOS-and eNOS-derived 339 NO have been suggested to play a key role in sensory-evoked neurovascular coupling 340 When NO production was attenuated with LNAME, we observed an increase in the 380 power of low frequency haemodynamic oscillations centred at ~0.1Hz, compared to 381 before injection of LNAME (Fig. 7). LNAME may lead to enhanced vasomotion via the 382 inhibition of eNOS-derived NO, which has previously been suggested to inhibit voltage 383 100mg/kg, administered 3 times with a day between each injection. Treatment with 435 tamoxifen was carried out when mice were aged between 1-5 months, and took place 436 a minimum of 2 weeks prior to surgery to allow for gene expression to occur. 437 Randomisation sequences were not used to assign animals to different 438 pharmacological agents. 439 440

Surgical Preparation of Chronic Cranial Window 441
At least 2 weeks prior to experimental sessions, surgery to prepare a thinned cranial 442 window over the right somatosensory cortex was performed as previously described 443 Vetapharm Ltd), midazolam (Hypnovel, Roche Ltd) and sterile water (in a ratio of 1:1:2 445 by volume; 7ml/kg) was used to induce anaesthesia and anaesthesia was maintained 446 using isoflurane (0.5-0.8%) in 100% oxygen at a flow rate of 1L/min. Surgical In some experiments neural activity was recorded concurrently with haemodynamic 487 activity, to effectively assess both these aspects of neurovascular coupling. A cranial 488 burr-hole was made in the right whisker barrel cortex (identified by response to whisker 489 stimulation in a previous 2D-OIS experiments) to allow insertion of a 16 channel 490 microelectrode (100μm spacing, 1.5-2.7Ω impedance, site area 177μm 2 ; Neuronexus 491 Technologies) to a depth of ~1600μm. To record neural activity, the electrode was 492 connected to a preamplifier and data acquisition device (Medusa BioAmp/ RZ5, TDT), 493 and data were collected at 24kHz. During post-hoc analysis, data were downsampled 494 to 6kHz and a 500Hz high pass filter was applied. A 'spike' was detected when data 495 exceeded a threshold of 1.5 times the standard deviation above the mean. The 496 number of spikes occurring in 100ms bins were counted and reported as MUA. Data 497 from the 12 electrode channels corresponding to cortical depth were used for analysis. 498 To produce fractional change in MUA, responses were normalised to a 2s baseline. 499 Trials were averaged to produce a mean response for each stimulation paradigm for 500 each animal. For statistical analysis, the peak MUA and mean MUA during the 501 stimulation period were calculated for each animal. Group means were produced by 502 averaging across animals. One mouse was excluded from statistical analysis due to 503 the presence of excessive noise. To reduce NO production, mice were treated with the non-selective NOS inhibitor 523 N(G)-Nitro-L-arginine methyl ester (LNAME, Sigma). LNAME (10mg/ml made up with 524 sterile saline) was administered via i.p. bolus injection (75mg/kg; which has been 525 shown to reduce NOS activity within the cortex by 93% within 1 hour of injection 526

Data analysis 552
All experiments and analysis were performed unblinded. Data analysis was performed 553 using MATLAB (MathWorks). Using the spatial map of Hbt changes evoked by 2s 554 whisker stimulation, generated by 2D-OIS, a region of interest (ROI) was automatically 555 selected (Lee et al., 2020). In brief, each pixel was averaged across time during the 556 response period, to generate a mean pixel value. Any pixel whose value was greater 557 than 1.5x standard deviation was included in the ROI. The resulting ROI (white ROI, 558 Fig 1B) represents the area with the greatest haemodynamic response to the whisker 559 stimulation, and therefore represents the whisker barrel cortex. Within this ROI, the 560 arterial region most responsive to the pharmacological intervention was manually 561 selected (red ROI, Fig 1B). For analysis of responses within the arterial ROI, baseline 562 oxygen saturation and tissue concentration of haemoglobin of 80% and 100μM, 563 respectively, were assumed. For post-treatment data, baseline assumptions were 564 corrected based on the measured change in Hbt and Hbo evoked by treatment with 565 the pharmacological agent(s). For each stimulation paradigm, the response across all 566 pixels within the arterial ROI was averaged, producing the three haemodynamic time 567 series (Hbo, Hbr, Hbt). 568 responses in trials involving photostimulation, residual artefact from the 470nm LED 570 was removed using a modified boxcar function. Micromolar changes in 571 haemodynamics were converted to fractional change (as compared to a 5s baseline). 572 For each stimulation paradigm, mean time series were produced for each animal by 573 averaging across trials. To produce the group mean time series, these responses were 574 then averaged across animals within each group.  For visualisation purposes (Fig 7A-C), following removal of a cubic polynomial trend a 583 filter was applied to the individual Hbt time series to remove high frequency 584 components (such as heart rate). 585 586

Statistical analysis 587
Three metrics (Fig 1G) were extracted from the stimulation-evoked haemodynamic 588 Hbr (maximum -minimum value, above). Statistical analysis was carried out using 593 was considered robust if data were approximately normally distributed) and Levene's 595 test was used to test for equality of variances. Outliers were identified as extreme if 596 they had a studentised residual >3. ANOVAs are considered robust against minor 597 violations of assumptions. For experiments in which LNAME was applied alone, to 598 determine statistical significance, a 3-way mixed ANOVA (within-group factors: drug 599 were included. For all multi-way ANOVAs, Greenhouse-Geisser correction was 608 applied and simple effects tests were carried out to further interrogate any significant 609 interaction effects. To assess the evolution of the effect of LNAME, a 1-way ANOVA 610 was performed, comparing 4 timepoints (pre-LNAME injection, 0 minutes after LNAME 611 injection, 60-70 minutes after, and 95-135 minutes after). Results were considered 612 statistically significant if p < 0.05, unless otherwise stated. All data are reported as 613 mean ± standard error of the mean (SEM), unless otherwise stated. Sample sizes 614 were based on those in previously published studies using similar pharmacological 615 approaches (Bannerman et al., 1994).

Simple effects tests: LNAME (2s whisker stimulation) p
Hbt response in initial dip period 0.159 Hbo response in initial dip period 0.201 Hbr response in initial dip period 0.434