Neurovascular coupling is a critical brain mechanism whereby changes to blood flow accompany localised neural activity. The breakdown of neurovascular coupling is linked to the development and progression of several neurological conditions including dementia. In this study, we examined cortical haemodynamics in preparations that modelled Alzheimer’s disease (J20-AD) and atherosclerosis (PCSK9-ATH) between 9-12m of age. We report novel findings with atherosclerosis where neurovascular decline is characterised by significantly reduced blood volume, levels of oxyhaemoglobin & deoxyhaemoglobin, in addition to global neuroinflammation. In the comorbid mixed model (J20-PCSK9-MIX), we report a 3x fold increase in hippocampal amyloid-beta plaques. A key finding was that cortical spreading depression (CSD) due to electrode insertion into the brain was worse in the diseased animals and led to a prolonged period of hypoxia. These findings suggest that systemic atherosclerosis can be detrimental to neurovascular health and that having cardiovascular comorbidities can exacerbate pre-existing Alzheimer’s-related amyloid-plaques.

Animals

All animal procedures were performed with approval from the UK Home Office in accordance to the guidelines and regulations of the Animal (Scientific Procedures) Act 1986 and were approved by the University of Sheffield ethical review and licensing committee. Male C57BL/6J mice were injected i.v at 6wks with 6x10¹² virus molecules/ml rAAV8-mPCSK9-D377Y (Vector Core, Chapel Hill, NC) and fed a Western diet (21% fat, 0.15% cholesterol, 0.03% cholate, 0.296% sodium; #829100, Special Diet Services UK) for 8m (PCSK9-ATH). These mice were compared to age-matched wild-type C57BL/6J mice (with no AAV injection fed normal rodent chow) that were used as controls (WT C57BL/6J). In addition, male heterozygous transgenic J20-hAPP B6.Cg-Zbtb20Tg(PDGFB-APPSwInd)20Lms/2Mmjax) (MMRRC Stock No: 34836-JAX) mice were used. Atherosclerosis was induced in J20-hAPP mice alongside WT mice at 6wks of age combined with a Western diet to create a comorbid mixed model (J20-PCSK9-MIX). For the CSD imaging experiments, 4 nNOS-ChR2 mice (M/F, 16-40 weeks old) were included in the WT group. [nNOS-ChR2 mice: heterozygous nNOS-CreER (Jax 014541, (Taniguchi et al., 2011)) x homozygous Ai32 mice (Jax 024109, (Madisen et al., 2012)), given tamoxifen (100mg/kg, i.p., 3 injections over 5 days) at 1-2 months old]. All mice were imaged between 9-12m of age. All mice were housed in a 12hr dark/light cycle at a temperature of 23C, with food and water supplied ad-libitum.

Thinned Cranial Window Surgery

Mice were anaesthetised with 7ml/kg i.p. injection of fentanyl-fluanisone (Hypnorm, Vetapharm Ltd), midazolam (Hypnovel, Roche Ltd) and maintained in a surgical anaesthetic plane by inhalation of isoflurane (0.6-0.8% in 1L/min O₂). Core body temperature was maintained at 37^oC through use of a homeothermic blanket (Harvard Apparatus) and rectal temperature monitoring. Mice were placed in a stereotaxic frame (Kopf Instruments, US) and the bone overlying the right somatosensory cortex was thinned forming a thinned cranial optical window. A thin layer of clear cyanoacrylate glue was applied over the cranial window to reinforce the window. Dental cement was applied around the window to which a metal head-plate was chronically attached. All mice were given 3 weeks to recover before the first imaging session. 

2D-Optical Imaging Spectroscopy (2D-OIS)

2D-OIS measures changes in cortical haemodynamics: total haemoglobin (HbT), oxyhaemoglobin (HbO) and deoxyhaemoglobin (HbR) concentrations (Berwick et al., 2005). Mice were lightly sedated and placed into a stereotaxic frame. Sedation was induced as described above and maintained using low levels of isoflurane (0.3-0.6%). For imaging, the right somatosensory cortex was illuminated using 4 different wavelengths of light appropriate to the absorption profiles of the differing haemoglobin states (495nm ± 31, 559nm ± 16, 575nm ± 14 & 587nm ± 9) using a Lambda DG-4 high-speed galvanometer (Sutter Instrument Company, US). A Dalsa 1M60 CCD camera was used to capture the re-emitted light from the cortical surface. All spatial images recorded from the re-emitted light underwent spectral analysis based on the path length scaling algorithm (PLSA) as described previously (Berwick et al., 2005; Mayhew et al., 1999). which uses a modified Beer-Lambert law with a path light correction factor converting detected attenuation from the re-emitted light with a predicted absorption value. Relative HbT, HbR and HbO concentration estimates were generated from baseline values in which the concentration of haemoglobin in the tissue was assumed to be 100µM and O₂ saturation to be 70%. For the stimulation experiments, whiskers were mechanically deflected for a 2s-duration and a 16s-duration at 5Hz using a plastic T-shaped stimulator which caused a 1cm deflection of the left-whisker. Each individual experiment consisted of 30 stimulation trials (for 2s) and 15 stimulation trials (for 16s) of which a mean trial was generated after spectral analysis of 2D-OIS. Stimulations were performed with the mouse breathing in 100% O₂ or 21% O₂, and a gas transition to medical air (21% O₂) as well as an additional 10% CO₂-hypercapnia test of vascular reactivity.  

Neural Electrophysiology

Simultaneous measures of neural activity alongside 2D-OIS were performed in a final acute imaging session 1-week after the 1st imaging session. A small burr-hole was drilled through the skull overlying the active region (as defined by the biggest HbT changes from 2D-OIS imaging) and a 16-channel microelectrode (100µm spacing, 1.5-2.7MΩ impedance, site area 177µm2) (NeuroNexus Technologies, USA) was inserted into the whisker barrel cortex to a depth of ~1500µm. The microelectrode was connected to a TDT preamplifier and a TDT data acquisition device (Medusa BioAmp/RZ5, TDT, USA). Multi-unit analysis (MUA) was performed on the data. All channels were depth aligned to ensure we had twelve electrodes covering the depth of the cortex in each animal. The data were high pass filtered above 500Hz to remove all low frequency components and split into 100ms temporal bins. Within each bin any data crossing a threshold of 1.5SD above the mean baseline was counted and the results presented in the form of fractional changes to MUA.

Region Analysis

Analysis was performed using MATLAB (MathWorks). An automated region of interest (ROI) was selected using the stimulation data from spatial maps generated using 2D-OIS. The threshold for a pixel to be included within the ROI was set at 1.5xSD, therefore the automated ROI for each session per animal represents the area of the cortex with the largest haemodynamic response, as determined by the HbT. For each experiment, the response across all pixels within the ROI was averaged and used to generate a time-series of the haemodynamic response against time.

Statistical Analysis

Statistical analyses were performed using SPSS v25, v26 and v27 & GraphPad Prism v8. Shapiro-Wilks test was used to check for normality and Levene’s test was used to assess equality of variances. 2-way mixed design ANOVA, 1-way ANOVA or Kruskal-Wallis tests were used, as appropriate. For 1-way ANOVA, if variances were unequal, Welch’s F was reported. Results were considered statistically significant if p<0.05. The Shapiro-Wilks test suggested that, for chronic experiments, peak values of HbT and HbO are normally distributed, however, HbR values are significantly non-normal. 2-way mixed design was used to compare peak values for HbT, HbO & HbR (although HbR failed the S-W test for normality, an ANOVA was used as they were considered fairly robust against small deviations from normality). Inspection of Levene’s test suggested that variances were equal, therefore, Dunnett’s (two-sided) multiple comparisons test was used to compare disease models to WT, and for HbR, Games-Howell multiples comparisons were used. If the Greenhouse-Geisser estimate of sphericity showed deviation from sphericity (chronic experiments: HbT (ε=0.55), HbO (ε=0.49) & HbR (ε=0.564), results are reported with Greenhouse-Geisser correction applied. qRT-PCR data was analysed by performing 1-way ANOVAs with Dunnett’s multiple comparisons test used to compare disease models to WT. P-values <0.05 were considered statistically significant. All the data are presented as mean values ± standard error of mean (SEM).