Neutrophil stalling does not mediate the increase in tau phosphorylation and the cognitive impairment associated with high salt diet

Abstract

Aims

High dietary salt intake has powerful effects on cerebral blood vessels and has emerged as a risk factor for stroke and cognitive impairment. In mice, a high salt diet (HSD) leads to reduced cerebral blood flow (CBF), tau hyperphosphorylation, and cognitive dysfunction. However, it is still unclear whether the reduced CBF is responsible for the effects of HSD on tau and cognition. Capillary stalling has been linked to cognitive impairment in models of Alzheimer’s disease and diabetes. Therefore, we tested the hypothesis that capillary stalling also contributes to CBF reduction, tau accumulation, and cognitive impairment in HSD.

Methods and results

We used in vivo two-photon imaging to assess capillary stalling in C57BL6/J male mice fed a normal diet or HSD. We found that HSD increased stalling of neutrophils in brain capillaries and decreased CBF. Neutrophil depletion using anti-Ly6G antibodies reduced the number of stalled capillaries and restored CBF, measured by red blood cell speed. Despite the improved CBF, chronic neutrophil depletion did not rescue HSD-induced cognitive impairment, assessed by the Barnes maze and nest building behavior. Furthermore, levels of phosphorylated tau in the cortex and hippocampus remained elevated in HSD mice after neutrophil depletion.

Conclusion

These novel findings show that capillary stalling contributes to CBF reduction in HSD, but not to tau phosphorylation and cognitive deficits. Therefore, the hypoperfusion caused by capillary stalling is not the main driver of the tau phosphorylation and cognitive impairment.

Graphical Abstract

1. Introduction

High salt intake plays a role in several human diseases and has emerged as a major public health problem worldwide.¹ Although essential for life, salt intake exceeds greatly the American Heart Association (AHA) recommended level of <1.5 g/day.² The deleterious effects of high salt diet (HSD) have been traditionally attributed to hypertension, but accumulating evidence indicates that HSD is harmful independent of hypertension.^3–7 HSD has powerful effects on cerebral blood vessels^8–13 and has been linked to increased incidence of cardiovascular diseases, including stroke^4,14–21 and dementia.^22,23 Increasing evidence implicates HSD also in cognitive impairment,^24,25 and lower dietary salt has been linked to improved cognitive function.^26,27 However, the mechanisms responsible for these effects have not been fully elucidated.

In mice, we have shown that HSD reduces resting cerebral blood flow (CBF), an effect mediated by a deficit in endothelial nitric oxide (NO).¹³ However, the mechanisms by which the NO deficit leads to CBF reduction remain unclear. The NO deficit could reduce resting CBF by constricting arterioles and increasing vascular resistance,²⁸ but could also increase leucocyte adhesion leading to microvascular occlusions.²⁹ Capillary stalls, blood flow arrest by leucocytes, or red blood cells (RBC) in capillaries, have emerged as a cause of CBF reduction and impaired cerebral oxygenation by increasing the heterogeneity of RBC fluxes,^30,31 and contribute to CBF reduction in mouse models of Ab accumulation, stroke, diabetes, and seizures.^32–42 However, it is not known which of these mechanisms contributes to the hypoperfusion caused by HSD.

Another key question concerns the role of the CBF reduction in tau accumulation. We have shown that HSD-induced cognitive impairment involves the microtubule-associated protein tau.⁴³ In tauopathies, including AD and frontotemporal dementia, tau is hyperphosphorylated, aggregates into multimers, and exerts pathological effects.⁴⁴ In HSD, the endothelial NO deficit leads to de-nitrosylation of the protease calpain in neurons, which, in turn, activates the tau phosphorylating enzyme Cdk5.⁴³ The effect is related to lack of endothelial NO and not neuronal NO.⁴³ The HSD-induced cognitive impairment did not occur with administration of anti-tau antibodies or in tau⁻/⁻ mice, despite persistent CBF reduction,⁴³ indicating that tau accumulation, not hypoperfusion, is the final mediator of the cognitive dysfunction. However, the CBF reduction could also independently promote accumulation of hyperphosphorylated tau (p-Tau), as described in models of cerebral hypoxia/hypoperfusion,^45,46 possibly via hypoxia-induced Cdk5 activation.⁴⁷

Since capillary stalling by neutrophils has emerged as an important cause of reduced CBF in a wide variety of models,^32–42 in this study, we investigated if HSD promotes capillary stalling and whether the ensuing CBF reduction contributes to p-Tau accumulation and cognitive deficits.

2. Methods

2.1. Mice

All animal procedures were performed in compliance with the guidelines from NIH Guide for the Care and Use of Laboratory Animals and were approved by the institutional animal care and use committee of Weill Cornell Medicine (Animal protocol number: 0807–777A). Studies were conducted according to the ARRIVE guidelines (https://www.nc3rs.org.uk/arrive-guidelines). To avoid the confounding effect of ageing,⁴⁸ young C57BL/6 male mice obtained from Jackson Laboratory (RRID: IMSR_JAX:000664) were used.

2.2. High salt diet

Mice (8 weeks old) received normal chow (0.5% NaCl) and tap water ad libitum (normal diet) or sodium-rich chow (8% NaCl) and tap water containing 0.9% NaCl ad libitum (HSD) for 4–26 weeks according to the experiment.^13,43,49 These levels of salt intake are an approximation of the highest levels of human salt consumption.⁵⁰

2.3. In vivo treatments

To selectively deplete circulating neutrophils, mice were injected intra-peritoneally (i.p.) with antibodies against the neutrophil antigen Ly6G (anti-mouse Ly6G clone 1A8, BE0075–1, Bio X Cell, Lebanon, NH, USA) or isotype IgG as a control (anti-rat IgG2a isotype control, anti-trinitrophenol, BE0089, Bio X Cell, Lebanon, NH, USA) (4 mg/kg/mouse 3 times a week) for the last 4 weeks of the 12 week-HSD treatment.^{33,34,38,41,51}

2.4. Animal preparation for in vivo brain imaging

Optical access to the brain was achieved through a thinned skull preparation, as previously described.⁵² During the surgical procedure, the animals were anaesthetized with isoflurane (1.5–2%) delivered in a mixture of oxygen (75 mL/min) and nitrogen (200 mL/min) with the goal to achieve a respiratory rate of about 60 breaths/min. To control body temperature, the animals were placed on a feedback-controlled heating blanket (TC-1000; CWE Inc., Ardmore, PA, USA). Following scalp and periosteum removal, a custom-made titanium head post was affixed to the right side of the parietal skull surface using a three-component metabond adhesive (S371, S398, S396, Parkell). In the initial cohort, a thinned-skull cranial window was prepared by gently thinning the skull using a high-speed dental drill. A 3-mm diameter glass coverslip (CS-3S, Warner Instruments, Hmaden, CT, USA) was then secured to the exposed bone using cyanoacrylate glue (Loctite 495; Henkel, Rocky Hill, CT, USA) to improve window stability. For the revised protocol for chronic long-term treatment experiments, a full craniotomy was performed to enable imaging for up to 8 weeks post-surgery. A circular piece of skull (~3 mm in diameter) was removed and sealed with a sterile glass coverslip using cyanoacrylate glue. Mice that underwent the thinned-skull procedure were allowed to recover for at least 7 days, while those with a full craniotomy recovered for at least 2 weeks prior to imaging.

2.5. Two photon in vivo imaging

Mice were anaesthetized with isoflurane (1–1.5%) and placed on a custom stereotactic frame. A feedback-controlled heating pad was used to control the body temperature. To label blood plasma and visualize vascular structure, 70 kDa Texas Red dextran (50 μL, 2.5% w/v in saline, D-1830; Invitrogen, Thermo Fisher Scientific, Waltham, MA, USA) was injected retro-orbitally (r.o.). Leucocytes and blood platelets were labelled with Rhodamine 6G (0.1 mL, 1 mg/mL in saline, R4127; Sigma, St. Louis, MO, USA). Leucocytes were distinguished from blood platelets by Hoechst 33342 (50 μL, 4.8 mg/mL in saline, H21492; Thermo Fisher Scientific, Waltham, MA, USA). Rhodamine 6G and Hoechst 33342 were loaded into a single syringe and r.o. injected. Isoflurane concentration was adjusted as necessary to maintain a steady breathing rate of 60 breaths/min.

Three-dimensional images of the cortical vasculature and measurement of RBC speeds in specific vessels were acquired using a commercial 2-photon microscope (2-PEF, FVMPE; Olympus Scientific Solutions, Waltham, MA, USA) equipped with a solid-state laser (InSight DS+; Spectraphysics). For three-channel imaging, the excitation laser was tuned to 820 nm and scanned with galvanometric scanners, with the focused beam directed into the sample using a 25× water immersion objective (XLPlan N 25 × 1.05 NA, Olympus Scientific Solutions, Waltham, MA, USA). Emitted fluorescence passed through an infrared (IR) blocking filter (FV30-BA685RXD), was spectrally separated by a dichroic mirror (FV30-SDM570) and two filter cubes (FV30-FVG; 410- to 460-nm (Hoechst), 495- to 540-nm (Rhodamine 6G), FV30-FOCY5; 575- to 645-nm (Texas Red)). Image stacks and movies were captured using Fluoview software (FV31S-SW, version 2.3.1.163; Olympus Scientific Solutions, Waltham, MA, USA). For vascular mapping, a low-magnification 5× objective (MPlan N 5×, 0.1 NA, Olympus Scientific Solutions, Waltham, MA, USA) was used to acquire an initial map of the vasculature. We selected two areas with a sharp contrast and avoided large pial vessels for further imaging. The objective was then switched to 25× for higher magnification imaging. For capillary stall quantification, we took high-resolution stacks of images spaced by 1 μm axially to a cortical depth of 300–500 μm (1024 × 1024 resolution, frame rate of 0.31). To measure RBC speed, vessel segments parallel to the imaging surface were chosen and unidirectional line scans were performed on the line ROI drawn along the central axis inside the vessel lumen for 30 s to 1 min with a line rate of 0.8–1.3 kHz, depending on the length of line ROI. In a series of experiments comparing pre- and post-antibody treatments, the antibody was administered immediately after the initial imaging session. Two hours later, imaging was repeated at the exact same location. For chronic studies, imaging was also repeated at the same location after 2 and 4 weeks of treatment with either IgG or anti-Ly6G antibodies.

2.6. Two photon in vivo imaging analysis

To measure capillary segment stalling, we began by counting the number of capillary segments in each image stack. Image stacks were first duplicated and cropped to exclude the top 40–120 μm containing meningeal layers and large pial vessels. Vascular skeletons were then generated using the TubeAnalyst macro (IRB Barcelona) in ImageJ. The resulting skeletons were merged with the original image stacks and manually corrected for missing capillary segments, particularly near the pial surface and stack edges, where imaging artifacts often disrupts automated detection. Next, we carefully examined the image stacks, manually identifying capillary stalls by observing the movement of RBCs. Since Texas Red dextran labels blood plasma and not the blood cells, RBCs appear as dark patches or stripes in the vessel lumen. The consistent presence of these dark stripes indicates ongoing blood flow in that vessel segment. We plotted the stall counts as the fraction of capillaries with stalled blood flow. Stalls were further categorized based on our labelling strategy: capillary segments with a cell-shaped object labelled with both Rhodamine 6G and Hoechst were considered to contain a leucocyte, while those with punctate Rhodamine 6G labelling alone were categorized as platelet aggregates (Figure 1A). Stalled segments with only RBCs present were classified as RBC stalls. Additionally, we categorized stalls based on their duration: those that persisted throughout the imaging were labelled as ‘stalled,’ while those that resolved and resumed blood flow were classified as ‘transient.’ Each frame captured took 3.2 s, with capillary diameters typically ranging from 6 to 9 μm, and stacks were taken at 1 μm intervals. As a result, each capillary remained visible for approximately 28.8 s. Blood vessel segments were categorized as either penetrating arterioles (PA), capillaries, or ascending venules (AV). Arterioles were differentiated from venules based on their morphology—arterioles had smaller diameters, smoother walls, and more symmetrical Y-shaped branching compared to venules. Smaller branch segments were classified by tracing their connections to these easily recognizable larger vessels. For each identified stall, we traced the upstream and downstream branches and examined the proximity to either PA or AV. While one stall was found at the 4th branch near the PA, most stalls were located near the AV. We quantified the branching order relative to the AV, following the vessel enumeration scheme shown in Figure 1G. Capillary diameter was measured using ImageJ (NIH). For each high-resolution stack, a projection stack was generated, with each frame consisting of the maximum z-projection of every 10 frames, using a custom-written ImageJ macro. For each projection stack, 50 vessels were randomly selected, ensuring representation across different vessel categories, spanning from PA and its 1st to 5th branches and 5th to 1st branches toward AV and AV. To select the most appropriate frame for each vessel, we scrolled through the z-axis of the projection stack and identified the frame where the vessel of interest was most visually prominent. We drew a perpendicular line across the vessel segment, and the full width at half maximum (FWHM) was measured using the ‘Plot Profile’ tool in ImageJ.

Figure 1.

HSD increases brain capillary stalling. (A) Rendering of in vivo 2PEF image stack of cortical vasculature (left), and images of stalled capillaries that contained a leucocyte and RBC (right, top), and platelet (right, bottom), distinguished by fluorescent labels (magenta: Texas Red dextran labeled blood plasma; green: rhodamine 6G labeled leucocytes and platelets; blue: Hoeschst labelled leucocyte nuclei). (B) Representative 100 μm axial projections of 2PEF image stacks from ND (left) and HSD (right) diet-fed mice showing stalled capillaries indicated with arrowheads. (C) HSD (15 weeks) increases the number of stalled capillaries per volume of tissue (P = 0.0017 vs. ND, n = 5/group, two-tailed paired t-test). (D) Fraction of capillaries with stalled blood flow in ND and HSD-fed mice over time. Each data point represents fraction of capillaries stalled in one stack. (Diet: P < 0.0001, Time: P = 0.4855, Interaction: P = 0.6126, two-way ANOVA plus Tukey’s multiple comparisons test; 4 weeks: P = 0.0002 vs. ND, 15 weeks: P = 0.0188 vs. ND, 26 weeks: P = 0.0003 vs. ND). (E) Fraction of stalled capillaries that resumed flow (transient) or remained stalled (stalled or not flowing) during stack acquisition in ND and HSD mice over time (Transient/Stalled: Diet, P = 0.9756, Time, P = 0.6764, Interaction, P = 0.9668, n = 3–5/group, two-way ANOVA plus Tukey’s multiple comparisons test). (F) Fraction of stalled capillaries in HSD mice that contained leucocytes (LEU), one or more RBCs and platelets (PLT), distinguished as LEU only (LEU), LEU with one or more RBCs (LEU-RBC), RBCs with PLT (RBC-PLT), and PLT only (Weeks of HSD: P > 0.9999, stalls type: P < 0.0001, Interaction: P = 0.1880, n = 5/group, two-way ANOVA plus Tukey’s multiple comparisons test). (G) Vessel enumeration scheme overlay onto rendering of an ascending venule and its branches (left) and stall location in each capillary branch (right) in HSD mice (4 and 26 weeks). (H) Diameter of randomly chosen branches from PA and AV (20–76 vessel segments for each column from ND and HSD fed mice, 1st branch from AV: P = 0.0049, 2nd branch from AV: P = 0.0378, n = 5/group, two-tailed Mann–Whitney tests). (I) Diameter from 1st to 4th branch from AV shown in H broken down by the presence of capillary stall. Analysis of covariance (ANCOVA) test to compare two slopes. Data are presented as mean ± SEM.

Space-time images were generated from line scans, which exhibited diagonal streaks. The slope of these streaks was inversely proportional to the centreline RBCs’ speed. All line scans were analysed using MATLAB, which employed a Radon transform-based algorithm to compute line integrals of the 2D image [g(x,y)] at various angles (https://github.com/sn-lab/Blood-flow).^53,54 To estimate the RBC speed of each vessel segment, we averaged the RBC speed measurements over a 1-min period. Volumetric blood flow (F) was then calculated using the following equation: F = πvr²/2 where v is the centreline RBC speed derived from the line scan and r is the vessel radius. All displayed 2PEF images were generated using ImageJ (NIH) and represent maximal projections of 2PEF image stacks except for the renderings in Figure 1, which utilized the 3D viewer tool.

2.7. Laser speckle imaging and analysis

Laser speckle imaging (LSI) was performed with a commercial system (Omegawave, Inc., Tokyo, Japan). The cranial window was illuminated with a semiconductor laser set at 780 nm, and the scattered light was detected with a CCD camera positioned over the skull. Animals were anesthetized with isoflurane (1–1.5%), and after a 10-min of stable baseline, they were administered either anti-Ly6G or isotype control antibodies (4 mg/kg, i.p). Since neutrophil depletion from a single injection of anti-Ly6G lasted for 1 week (data not shown), we gave alternating treatments to the same group of animals 2 weeks apart. The mice were then monitored and imaged for an additional 2 h. Colour-coded CBF images were obtained at the highest resolution (639 × 480 pixels) with the sampling frequency of 1 Hz. To obtain a single representative frame for each minute of measurement, 60 consecutive frames were averaged, resulting in a time series of averaged frames. To account for spatial drift across the time series, we used the ‘rigid body transformation’ using the ImageJ (NIH) TurboReg plugin. Speckle contrast values were acquired from the whole cranial window, summing all contrast from surface arteriole, surface venule, or parenchymal region. This time-series data were subsequently analysed for plotting and quantification at the 2-h time point.

2.8. Isolation and flow cytometry analysis of peripheral leucocytes

For isolation of peripheral leucocytes, mice were anaesthetized with pentobarbital (100 mg kg⁻¹, i.p) and 0.5 mL of blood was collected by cardiac puncture into heparinized tubes. Blood (150 μL) was incubated for 5 min at room temperature with 5 mL of erythrocyte lysis buffer, followed by the addition of 10 mL of HBSS/10 mM HEPES and spun at 500 g for 7 min. Afterward, blood cells were resuspended in 50 μL of FACS buffer (PBS with 2% FBS and 0.05% NaN₃).

2.9. Isolation and flow cytometry analysis of brain cells

For isolation of brain cells, mice were anaesthetized with pentobarbital (100 mg kg⁻¹, i.p.) and transcardially perfused with heparinized PBS. Brain hemispheres were separated from the cerebellum and olfactory bulbs and gently triturated in HEPES-HBSS buffer (138 mM NaCl, 5 mM KCl, 0.4 mM Na₂HPO₄, 0.4 mM KH₂PO₄, 5 mM D-glucose, 10 mM HEPES) using a Gentle MACS dissociator (Miltenyi Biotec, Bergisch Gladbach, Germany) following the manufacturer’s instructions. The suspension was digested with 62.5 μg/mL Liberase DH (Roche Diagnostics, Indianapolis, IN, USA), 0.8 U/mL dispase (Worthington, Lakewood, NJ, USA), and 50 U/mL DNAse I (Worthington, Lakewood, NJ, USA) at 37°C for 45 min in an orbital shaker at 100 rpm. Isolated cells were washed and centrifuged in 30% Percoll (GE Healthcare, Chicago, IL, USA) for myelin removal. Cells were resuspended in 50 μL of FACS buffer (PBS with 2% FBS and 0.05% NaN₃).

2.10. Flow cytometry

Single-cell suspensions were blocked with anti-CD16/CD32 for 10 min at 4°C and then stained with the appropriate antibodies for 15 min at 4°C. The following antibodies were used: CD45 (clone 30F-11), CD11b (clone M1/708), Ly6G (clone 1A8), Ly6C (clone HK1.4), CD62L (clone MEL-14), ICAM/CD54 (clone YN1/1.7.4), CD18 (clone M18/2), Cxcr4 (clone L276F12), Gr-1 (clone RB6–8C5), CD182 (clone SA044G4) from Biolegend, San Diego, CA, USA,^13,55 CD62E (clone 10E9.6) from BD Pharmingen and CD62P (clone Psel.KO2.3) from Thermo Fisher Scientific, Waltham, MA, USA. For both brain and blood flow cytometry, cells were washed with FACS buffer, resuspended in 150 μL of FACS buffer, and acquired in a NovoSampler Q (Agilent NovoCyte, Santa Clara, CA, USA). Appropriate isotype controls, ‘fluorescence minus one’ staining, and staining of negative populations were used to establish sorting parameters. Depending on the experiment, Ly6G, Gr1, CXCR2, or a combination of these antibodies was used to identify circulating and brain neutrophils. Specifically, due to potential masking of Ly6G by anti-Ly6G antibodies, Gr1 and CXCR2 were used to identify neutrophils in the depletion experiments. Analysis was performed with FlowJo.

2.11. Positive selection of circulating neutrophils

Blood neutrophils were purified by immunomagnetic positive selection. Briefly, 500 μL of blood was withdrawn from the heart using a 0.5M-EDTA flushed syringe. Two rounds of erythrolysis were performed as described for flow cytometry analysis of peripheral leucocytes. Leucocytes were resuspended in MACS buffer (PBS supplemented with 2% FBS, 2 mM EDTA; 300 μL/10⁷ cells), incubated with anti-Ly6G biotinylated-antibody (750 ng/10⁷ cells; clone 1A8; # 127604, Biolegend, Santa Clara, CA, USA) and purified with anti-biotin-microbeads according to the manufacturer’s instructions (Miltenyi Biotech, Bergisch Gladbach, Germany). Afterwards, cells were washed, centrifuged at 1500, 7 min, 4°C, and resuspended in 1 mL Trizol reagent (Invitrogen, Thermo Fisher Scientific, Waltham, MA, USA).

2.12. Quantitative real-time PCR

Quantitative determination of gene expression was examined as described before.¹³ RNeasy Plus Mini Kit (Qiagen, Germantown, MD, USA) was used to extract RNA from circulating neutrophils. After, cDNA was synthesized with iScript reverse transcription supermix for RT-qPCR (BioRad Laboratories, Hercules, CA, USA). qRT-PCR was conducted with cDNA in duplicate reactions using the Maxima SYBR Green/ROX qPCR Master Mix (2X) (Thermo Fisher Scientific, Waltham, MA, USA). The reactions were incubated at 50°C for 2 min and then at 95°C for 10 min. A polymerase chain reaction cycling protocol consisting of 15 s at 95°C and 1 min at 60°C for 45 cycles was used for quantification. Primer sequences (Invitrogen, Thermo Fisher Scientific, Waltham, MA, USA) are described in Table 1. Data were expressed as relative fold change over ND mice calculated by the 2^−ΔΔCt method.⁵⁶

Table 1.

List of primers used for RT-PCR studies on circulating neutrophils

Cybb (Gp91)	AATCTCAGGCCAATCACTTTG	AACGCCTATTGTGGTGTTAGG
Il1b	CTCTCCACCTCAATGGACAGA	TTTTGTCGTTGCTTGGTTCTC
Mmp9	ATTCGCGTGGATAAGGAGTTC	CCTTGTTCACCTCATTTTGGA
Cxcl2	TGAACAAAGGCAAGGCTAACTG	GAGGCACATCAGGTACGATCC
Itgam	TGGCCTATACAAGCTTGGCTTT	TCTAGACCTGCTTGTTGCTGG
Mpo	CACGCCCAACAACATTGACAT	CCAAAACCGATCACCATCACG
Elane	CACTGTGTGAACGGCCTAAAT	TTCAGCAGTTGTGATGGGTCA
Lcn2	CAGAGCTACAATGTGCAAGTGG	CAGCTCCTTGGTTCTTCCATAC
Lrg1	ATAACTCTCTGTCCAGCACGC	TTGTCACAGATCCAGGGGTTG
Hprt	AGTGTTGGATACAGGCCAGAC	CGTGATTCAAATCCCTGAAGT

2.13. Immunoblot analysis of phosphorylated tau

Cortex (≈90–100 mg) and hippocampus (≈15–20 mg) isolated from ND and HSD mice were sonicated in 800 and 500μL of RIPA buffer (50 mM Tris–HCl pH 8.0, 150 mM NaCl, 0.5% deoxycholic acid, 0.1% SDS, 1 mM EDTA pH 8.0, 1% IGEPAL CA-630, 1 mM Na3VO4, 20 mM NaF and one tablet/10 mL of cOmplete^™, EDTA-free Protease Inhibitor Cocktail, Millipore Sigma, St. Louis, MO, USA). After homogenization in cold RIPA buffer and centrifugation (20 000g for 10 min), 150 μL of the supernatant containing the proteins was boiled at 100°C for 10 min. Samples were cooled on ice for 20 min and then centrifuged at 20 000g at 4°C for 15 min. The supernatant corresponding to the heat-stable (HS) fraction was then harvested. The proteins were then mixed with equal volumes of SDS sample buffer, boiled, and analysed on 10% Novex^™ WedgeWell^™ gels (Thermo Fisher Scientific, Waltham, MA, USA). Proteins were transferred to PVDF membranes (Millipore Sigma, St. Louis, MO, USA), blocked at room temperature (RT) for 1 h with 5% milk in TBS, and incubated overnight at 4°C, with primary antibodies (see Table 2) in 5% BSA in TBS/0.1% Tween-20 (TBST). Membranes were washed in TBST, incubated with goat anti-mouse or rabbit secondary antibodies conjugated to horseradish peroxidase (Santa Cruz Biotechnology, Dallas, TX, USA) for 1 h at RT, and protein bands were visualized with Clarity Western ECL Substrate (BioRad Laboratories, Hercules, CA, USA) on an iBright^™ CL1500 Imaging System (Thermo Fisher Scientific, Waltham, MA, USA). Quantification was performed using iBright Analysis Software (Thermo Fisher Scientific, Waltham, MA, USA).

Table 2.

List of primary antibodies used for detection of phosphorylated tau

Name	Description	Source	Catalog No.	Dilution
pTau Ser202Thr205 (AT8)	Mouse mAb	ThermoFisher	MN1020	1:1000
pTau Thr217	Rabbit pAb	ThermoFisher	44–744	1:1000
pTau Ser231 (AT180)	Mouse mAb	ThermoFisher	MN1040	1:500
pThr231 (RZ3)	Mouse mAb	Peter Davies	n/a	1:500
pTau Ser396	Mouse mAb	Cell Signalling	9632	1:1000
pTau Thr181 (AT270)	Rabbit mAb	Cell Signalling	12 885	1:1000
Tau46	Mouse mAb	Cell Signalling	4019	1:1000

2.14. Barnes maze test

The Barnes maze consisted of a circular open surface (90 cm in diameter) elevated to 90 cm by four wooden legs. There were 20 circular holes (5 cm in diameter) equally spaced around the perimeter and positioned 2.5 cm from the edge of the maze. No wall and no intra-maze visual cues were placed around the edge. A wooden plastic escape box (11 cm × 6 cm × 5 cm) was positioned beneath one of the holes. Three neon lamps and a buzzer were used as aversive stimuli. The Any-Maze tracking system (Stoelting) was used to record the movement of mice on the maze. Extra-maze visual cues consisted of objects within the room (table, computer, sink, door, etc.) and the experimenter. Mice were tested in groups of 10, and between trials they were placed into cages, which were placed in a dark room adjacent to the test room for the inter-trial interval (20–30 min). No habituation trial was performed. The acquisition phase consisted of three consecutive training days with three trials per day, with the escape hole located at the same location across trials and days. On each trial, a mouse was placed into a start tube located in the centre of the maze, the start tube was raised, and the buzzer was turned on until the mouse entered the escape hole. After entering the escape box, mice were returned to their cage. Between trials, the maze floor was cleaned with 10% ethanol in water to minimize olfactory cues. For each trial, mice were given 3 min to locate the escape hole, after which they were guided to the escape hole or placed directly into the escape box if they failed to enter the escape hole. Four parameters of learning performance were recorded: (1) the latency to locate (primary latency) and (2) enter the escape hole (total latency), (3) the number of errors made, and (4) the distance travelled before locating the escape hole. When a mouse dipped its head into a hole that did not provide escape, it was considered an error. On day 4, the location of the escape hole was moved 180° from its previous location (reverse learning), and two trials were performed.^13,43

2.15. Nest-building test

The ability of mice to build nests was assessed by the Deacon rating scale.⁵⁷ One hour before the dark cycle, each mouse was placed in a new clean cage containing 2.5–3.0 g of nestlet (Ancare, Bellmore, NY, USA). Food, water, and lighting parameters were not changed from standard housing practices. The next day, nests were assessed on a rating scale of 1–5 and the untorn nestlet pieces were weighed. The cognitive parameters recorded were (1) nest score and (2) percentage of untorn nestlet.¹³

2.16. Statistical analysis

Sample size was determined according to power analysis based on previously published work by our lab on the effects of dietary salt on CBF regulation and cognitive function.^13,43 On these bases, 10–15 mice/group were required in studies involving assessment of cognitive function.^13,58 Mouse randomization was performed based on the random number generator function (RANDBETWEEN) in Microsoft Excel software. After testing for normality (Shapiro–Wilk, D’Agostino, or Pearson test), intergroup differences were analysed by unpaired two-tailed t-test for single comparison, or by one-way ANOVA or two-way ANOVA with Tukey’s, Sidak’s multiple comparison test, or uncorrected Fisher’s LSD, as appropriate and indicated in the figure legends. If non-parametric testing was indicated, intergroup differences were analysed by Mann–Whitney test, Wilcoxon matched-pairs test, or Kruskal–Wallis test with Dunn’s correction, as appropriate and indicated in the figure legends. Statistical tests throughout the manuscript were performed using Prism 10 (GraphPad). Data are expressed as mean ± SEM and differences are considered statistically significant for P < 0.05. Statistical significance is represented on the plots as follows: *P < 0.05; **P < 0.01; ***P < 0.001.

3. Results

3.1. High salt diet promotes capillary stalling

To examine whether HSD increases capillary stalling, we used in vivo 2PEF to image the cortical vasculature in C57BL6/J male mice fed an ND or HSD from 4 to 26 weeks (Figure 1A and B). Capillary segments present in the stacks were individually assessed to be either flowing or stalled based on RBCs motion while the capillary is visible during three-dimensional image stack acquisition for several seconds (Figure 1A and B). Since Texas Red dextran was used to label plasma, blood cells appear as dark, moving intraluminal patches. A capillary segment was scored as stalled if there was no motion of the RBCs and other cells in the Texas Red imaging channel (Figure 1A and B). In vivo fluorescent labelling by i.v. administration of rhodamine 6G and Hoechst was used to label blood leucocytes (Figure 1A). We found that the number of stalled capillaries, expressed as a percentage of total capillaries present in the stack, was increased in HSD mice (Figure 1B and C). In agreement with studies in other models,^33,39 our results indicate that around 0.5% of capillaries in HSD mice had stalled blood flow, while in age matched ND mice around 0.1% of capillaries were not flowing (Figure 1C). In HSD mice, capillary stalling was elevated starting at 4 weeks and remained higher than ND mice up to 26 weeks (Figure 1D). Most stalled capillaries were occluded for the entire duration of the stack acquisition both in ND and HSD mice (Figure 1E). However, we found the about 20% of stalls disappeared during stack acquisition, suggesting transient stalling (Figure 1E). By using rhodamine 6G and Hoechst to identify blood leucocytes, we found that most stalled capillary segments contained leucocytes and/or red blood cells (Figure 1F). Next, we assessed the location of the stalls based on the order of the capillary branches nearest to the PA or AV (Figure 1G).³³ At both 4 and 26 weeks, we found that almost all stalls were found near AV, in the 2nd and 3rd branches from AV (Figure 1G), consistent with the venules being more prone to leucocyte adhesion particularly in conditions, like HSD, associated with NO deficit.⁵⁹ Next, we randomly selected capillaries in cortical stacks and measured their diameter. Consistent with the location of the stalls (Figure 1G), we found that branches closer to AV (1st and 2nd branches) had a smaller diameter (Figure 1H). Furthermore, we found that the diameter of stalled capillaries was significantly reduced compared to that of flowing capillaries (Figure 1I).

3.2. Neutrophil depletion reduces the number of stalled capillaries and increases CBF

To investigate the impact of capillary stalling on CBF, we depleted neutrophils by administering anti-Ly6G antibodies.^33,60 To assess the effect of neutrophil depletion on brain perfusion, mice received a single dose of anti-Ly6G or IgG isotype control 2 weeks apart, and CBF was monitored for 2 h post-injection using LSI. Anti-Ly6G treatment resulted in an approximate 15% increase in neocortical perfusion in HSD mice compared to baseline (Figure 2A–C). To better characterize the effects of neutrophil depletion on brain microvascular perfusion, we used 2PEF to measure RBC speed in capillary segments before and after i.p. administration of anti-Ly6G or IgG in ND and HSD mice. Anti-Ly6G treatment removed approximately 58% of HSD-induced capillary stalls (see Supplementary material online, Figure S1A) and significantly increased RBC speed in PA and their capillary branches (see Supplementary material online, Figure S1B). Neutrophil depletion also increased the diameter of the first capillary branch downstream from the AV (see Supplementary material online, Figure S1C). Consequently, volumetric blood flow, accounting for both RBC speed and vessel diameter, was elevated in both the PA and AV (see Supplementary material online, Figure S1D). We found that anti-Ly6G administration increased RBC speed in PA and its capillary branches (see Supplementary material online, Figure S1B). Next, to evaluate the effects of chronic neutrophil depletion, ND and HSD mice received repeated anti-Ly6G or IgG injections (4 mg/kg, three times per week) during the final 4 weeks of a 12-week HSD treatment period^{33,34,38,41,51} (Figure 2D). HSD caused a reduction in RBC speed across all vessel segments analysed (PA, first branch from PA, first branch from AV, and AV) (Figure 2E and F, Supplementary material online, Figure S1E). Vessel diameter was reduced in the first branch from the AV and in the AV itself, but not in upstream segments (Figure 2F, Supplementary material online, Figure S1E). As with RBC speed, volumetric flow was also reduced on both the arterial and venous sides (Figure 2F and Supplementary material online, Figure S1E). Chronic anti-Ly6G treatment reduced capillary stalls to levels comparable to those observed in ND mice (Figure 2G and Supplementary material online, Figure S1F). Stalls removal was associated with improved RBC speed and volumetric flow in the first branch from the AV and the AV at both 2 and 4 weeks of treatment (Figure 2H and I and Supplementary material online, Figure S1G–H). Neutrophil depletion also increased the diameter of both vessel segments (see Supplementary material online, Figure S1G–I). In contrast, RBC speed in the PA and its first branch was only rescued after 2 weeks of treatment, although volumetric flow in the PA was restored at both time points (see Supplementary material online, Figure S2J). Overall, these data indicate that neutrophil depletion significantly reduces the fraction of stalled capillaries and enhances CBF, improving both RBC speed and volumetric flow in arteriolar and venular segments.

Figure 2.

Neutrophil depletion rescues CBF in HSD fed mice. (A) Representative LSI images of HSD mice injected either with IgG or α-Ly6G. (B) LSI CBF plot as a function of time after antibody injection (Time: P < 0.0001, Treatment: P < 0.0001, Interaction: P < 0.001, n = 5/group, two-way ANOVA plus Tukey’s multiple comparisons test). (C) Average CBF-LSI increased from 90 to 120 min after IgG/α-Ly6G injection (Treatment: P = 0.0083, Diet: P = 0.3710, Interaction: P = 0.2617, n = 5/group, two-way ANOVA plus Uncorrected Fisher’s LSD). (D) Diagram of experimental protocol adopted to assess the effect of chronic α-Ly6G administration on CBF by 2PEF. (E) Representative axial projection of an AV of ND and HSD mouse and corresponding line scan. (F) HSD attenuates RBS speed and diameter in 1st branch from AV and AV (RBC: P < 0.0001 vs. HSD, Mann–Whitney U test; diameter: P = 0.0014 and P = 0.0039 vs. ND, n = 64–72 from 8 to 10 mice/group). Volumetric flow in AV was suppressed by HSD (P < 0.0001 vs. ND, Mann–Whitney U test, n = 52–64 from 8 to 10 mice/group). (G) Number of capillaries with stalled blood flow in ND and HSD treated with α-Ly6G or IgG control antibody (2 weeks: Diet, P = −0.0113, Treatment, P = 0.0013, Interaction, P = 0.0090; 4 weeks: Diet, P = 0.0001, Treatment, P = 0.0157, Interaction, P = 0.0491, n = 3–5 mice/group two-way ANOVA plus Uncorrected Fisher’s LSD). (H) RBC speed in 1st branch from AV and AV after 2 and 4 weeks of α-Ly6G or IgG administration in ND and HSD mice shown as a fraction of baseline flow (1st branch AV—2 weeks: Diet, P = 0.0023, Treatment, P < 0.0001, Interaction, P = 0.0009; 1st branch AV—4 weeks: Diet, P = 0.0104, Treatment, P = 0.0014, Interaction, P = 0.6319; AV—2 weeks: Diet, P = 0.0024, Treatment, P < 0.0001, Interaction, P < 0.0001; AV—4 weeks: Diet, P = 0.008, Treatment, P = 0.0049, Interaction, P = 0.6624, n = 10–31 from 4 to 5 mice/group two-way ANOVA plus Uncorrected Fisher’s LSD). (I) Volumetric blood flow in AV after 2 and 4 weeks of α-Ly6G or IgG administration in ND and HSD mice shown as a fraction of baseline flow (AV—2 weeks: Diet, P = 0.0041, Treatment, P < 0.0001, Interaction, P < 0.0001; AV—4 weeks: Diet, P = 0.0004, Treatment, P = 0.0003, Interaction, P = 0.0243; n = 13–29 from 4 to 5 mice/group two-way ANOVA plus Uncorrected Fisher’s LSD). Data are presented as mean ± SEM.

3.3. HSD does not induce activation of circulating or brain neutrophils

To determine whether the increase in capillary stalling observed in HSD mice could be attributed to neutrophil activation, potentially promoting adhesion to brain endothelial cells (ECs), we performed flow cytometry on both blood and brain samples from ND and HSD mice. We found that the number and the frequency of blood and brain neutrophils (CD45^hiCD11b⁺Ly6G⁺) were not altered by HSD (Figure 3B and 4B). Circulating numbers of other myeloid sub-populations (CD45^hi CD11b⁺Ly6C^hi and CD45^hiCD11b⁺Ly6C^low) were also not significantly altered in HSD mice (Figure 3C). We next assessed neutrophil activation status by examining the expression of CD18, a component of the integrins LFA1 (CD11a/CD18) and MAC1 (CD11b/CD18) and L-selectin (CD62L), the only selectin expressed on neutrophils.^61,62 Expression of both molecules was unchanged in both circulating and brain neutrophils from HSD mice compared to ND controls (Figure 3D and 4C). Consistent with this, RT-PCR analysis of circulating neutrophils revealed no significant differences in expression of activation-related genes including Cxcl2, Cybb, Il1b, Mmp9, Mpo, Elane, Lcn2 and Lrg1 (Figure 3E). In addition, expression of key endothelial adhesion molecules, CD54 (ICAM-1), CD62E (E-selectin), and CD62P (P-selectin), on brain ECs remained unchanged between ND and HSD mice (Figure 4D), consistent with previous findings.¹³ Finally, given that activated neutrophils can release neutrophil extracellular traps (NETs), we measured circulating levels of citrullinated histone H3 (H3Cit), a well-established marker of NET formation.⁶³ H3Cit was below the detection limit in both ND and HSD mice (Figure 3F), indicating that NET formation is not significantly induced under these conditions. Collectively, these findings indicate that HSD does not promote activation of circulating or brain neutrophils, nor does it alter the activation status of brain endothelial cells. Thus, the increase in capillary stalling observed in HSD mice is unlikely to be driven by neutrophil or endothelial activation.

Figure 3.

HSD does not induce blood neutrophil activation. (A) Representative flow cytometry plots showing the gating strategy used to identify blood neutrophils (CD45⁺CD11b⁺Ly6G⁺). (B) HSD had no effect on the number and the frequency of blood neutrophils (Numbers: P = 0.2902 vs. ND; Frequency: P = 0.2347 vs. ND, n = 3–4/group, two-tailed unpaired t-test). (C) HSD did not alter the number and the frequency of blood monocytes (Ly6C^low, Numbers: P = 0.2734 vs. ND; Frequency: P = 0.0829 vs. ND; Ly6C^hi, Numbers: P = 0.1705 vs. ND; Frequency: P = 0.0603 vs. ND n = 3–4/group, two-tailed unpaired t-test). (D) Representative flow cytometry plot showing CD18 and CD62L median fluorescence intensity (MFI) in blood neutrophils. HSD does not alter MFI levels of CD18 and L-selectin in blood neutrophils (CD18: P = 0.4751 vs. ND; L-selectin: P = 0.7964 vs. ND, n = 3–5/group, two-tailed unpaired t-test). (E) HSD does not increase mRNA levels of neutrophil activation markers in circulating neutrophils (Mmp9: P = 0.1203 vs. ND; Il1b: P = 0.3884 vs. ND; Gp91: P = 0.1355 vs. ND; Cxcl2: P = 0.6248 vs. ND, n = 5/group, two-tailed unpaired t-test). (F) Circulating levels of H3Cit in ND and HSD mice. Data are presented as mean ± SEM.

Figure 4.

HSD does not induce brain neutrophil activation. (A) Representative flow cytometry plots showing the gating strategy used to identify brain neutrophils (CD45^hiCD11b ⁺ Ly6G⁺). (B) HSD increases, albeit not significantly, the number and the frequency of brain neutrophils (Numbers: P = 0.2154 vs. ND; Frequency: P = 0.3084 vs. ND, n = 5–6/group, two-tailed unpaired t-test). (C) Representative flow cytometry histograms showing CD18 and CD62L median fluorescence intensity (MFI) in brain neutrophils. HSD does not alter MFI levels of CD18 and CD62L in brain neutrophils (CD18: P = 0.0826 vs. ND; CD62L: P = 0.1582 vs. ND, n = 5–6/group, two-tailed unpaired t-test). (D) HSD does not increase CD54, CD62E and CD62P median fluorescence intensity (MFI) in brain ECs (CD54: P = 0.2326 vs. ND; CD62E: P = 0.6569 vs. ND; CD62P: P = 0.3108 vs. ND, n = 5–6/group, two-tailed unpaired t-test). Data are presented as mean ± SEM.

3.4. Neutrophil depletion does not rescue the cognitive impairment associated with HSD

Next, we examined the effects of neutrophil depletion on the cognitive impairment associated with HSD. Mice were injected with either antibodies against Ly6G or isotype IgG as a control (4 mg/kg/mouse 3× week; i.p.) for the last 4 weeks of the 12 weeks ND/HSD treatment. Cognitive function was first assessed by the Barnes maze test.¹³ We found that ND and HSD mice, injected either with IgG isotype control or anti-Ly6G antibodies, were able to learn the location of the escape hole over the 3-days training phase (Figure 5A). Consistent with previous findings, when the escape hole was moved to the opposite quadrant on day 4, HSD mice spent more time to find the novel location that of ND mice.^13,43 Importantly, we found that injection of either IgG isotype control or anti-Ly6G antibodies, had no effect on the cognitive impairment of HSD mice (Figure 5B). We also examined whether neutrophil depletion rescues the impairment of nesting behaviour observed in HSD mice.¹³ We assessed the quality of the nest by the Deacon score, and we measured the amount of nesting material used. We found that HSD disrupted the ability of mice to build a nest as indicated by scores between 1 and 4 and by the increase in the percentage of unshredded nestlet (Figure 5C). However, as with the Barnes maze, depletion of neutrophils had no effect on the impairment of nest building behaviour of HSD mice (Figure 5C). After completion of cognitive testing, depletion of circulating neutrophils was validated by flow cytometry (Figure 6A and B). Taken together, these data indicate that neutrophil depletion has no effects on the cognitive impairment of HSD mice.

Figure 5.

Neutrophil depletion does not rescue the cognitive impairment associated with HSD. (A) Injection of IgG control or anti-Ly6G antibodies does not affect the learning process of ND and HSD mice during the 3-days training phase (Total latency, Time: P < 0.0001, Diet: P = 0.1062, Treatment: P = 0.7963; Primary latency, Time: P < 0.0001, Diet: P = 0.7053, Treatment: P = 0.6101, n = 7–12/group three-way ANOVA plus Tukey’s multiple comparisons test). (B) Primary latency is increased in HSD mice, regardless of the injection of IgG control or anti-Ly6G antibodies, on the 4th day of testing when the location of the escape hole is moved 180° from its previous location (Diet: P < 0.0001, Treatment: P = 0.6260, Interaction: P = 0.2388, n = 9–16/group, two-way ANOVA plus Tukey’s multiple comparisons test). (C) Nest-building ability is impaired in both IgG control and anti-Ly6G antibody-injected HSD mice. The amount of untorn nestlet is also increased in both IgG control and anti-Ly6G antibodies injected HSD mice (Deacon Score, Diet: P < 0.0001, Treatment: P = 0.9466, Interaction: P = 0.5398; Untorn nestlet, Diet: P = 0.0196, Treatment: P = 0.1339, Interaction: P = 0.2454, n = 6–12/group, two-way ANOVA plus Tukey’s multiple comparisons test). Data are presented as mean ± SEM.

Figure 6.

Administration of anti-Ly6G antibodies leads to depletion of circulating neutrophils in ND and HSD mice. (A) Representative flow cytometry plots showing the gating strategy used to identify blood neutrophils (CD45⁺CD11b⁺Gr1⁺Cxcr2⁺). (B) Injection of IgG control or anti-Ly6G antibodies does not affect the number of blood leucocytes (DAPI^negCD45⁺) in both ND and HSD mice (Diet: P = 0.5102, Treatment: P = 0.7347, Interaction: P = 0.7256, n = 6–13/group, two-way ANOVA plus Tukey’s multiple comparisons test). (C) Injection of IgG control or anti-Ly6G antibodies reduces the numbers and frequency of blood leucocytes (DAPInegCD45+) in both ND and HSD mice (Numbers, Diet: P = 0.8088, Treatment: P = 0.0388, Interaction: P = 0.8729; Frequency, Diet: P = 0.0634, Treatment: P < 0.0001, Interaction: P = 0.9156, n = 6–13/group, two-way ANOVA plus Tukey’s multiple comparisons test). Data are presented as mean ± SEM.

3.5. Neutrophil depletion does not suppress the increase in phosphorylated tau induced by HSD

Since the cognitive impairment associated with HSD is dependent on tau accumulation,⁴³ we next investigated whether neutrophil depletion has any effect on the tau phosphorylation increase induced by HSD. Consistent with previous data,⁴³ we found that HSD increased tau phosphorylation on Ser²⁰²Thr²⁰⁵ (AT8) and Thr²³¹ in both cortex and hippocampus of mice injected with IgG isotype control antibodies (Figure 7A and B). We also found that phosphorylation of Thr²¹⁷ was increased in the cortex but not in the hippocampus of the same group of mice (Figure 7A and B). However, in agreement with the behavioural data, we found treatment with anti-Ly6G antibodies failed to attenuate phosphorylated tau levels in HSD mice (Figure 7A and B). Thus, the data indicate that neutrophil depletion has no effect on the tau phosphorylation increase observed in the brains of HSD mice.

Figure 7.

Neutrophil depletion does not suppress the increase in phosphorylated tau induced by HSD. (A) HSD increases phosphorylation of tau on Ser²⁰²Thr²⁰⁵, Thr²¹⁷ and Thr²³¹ in the neocortex of both IgG control and anti-Ly6G-injected mice (Ser²⁰²Thr²⁰⁵, Diet: P = 0.0030, Treatment: P = 0.9889, Interaction: P = 0.3628; Thr²¹⁷, Diet: P = 0.0001, Treatment: P = 0.0544, Interaction: P = 0.9263; Thr²³¹, Diet: P = 0.0001, Treatment: P = 0.2868, Interaction: P = 0.3725, n = 6–10/group, two-way ANOVA plus Tukey’s multiple comparisons test). (B) HSD increases phosphorylation of tau on Ser²⁰²Thr²⁰⁵ and Thr²³¹ in the hippocampus of both IgG control and anti-Ly6G-injected mice (Ser²⁰²Thr²⁰⁵, Diet: P < 0.0001, Treatment: P = 0.2384, Interaction: P = 0.3361; Thr²³¹, Diet: P = 0.0097, Treatment: P = 0.3010, Interaction: P = 0.9946, n = 6–8/group, two-way ANOVA plus Tukey’s multiple comparisons test). Data are presented as mean ± SEM.

4. Discussion

This study provides the first evidence that HSD increases brain capillary stalling. By using in vivo 2PEF microscopy, we found that capillary stalls were mostly caused by leucocytes and were in capillary branches proximal to the AV. Since the branches that were stalled had a smaller vessel diameter, our data raise the possibility that vasoconstriction may contribute to the CBF reduction induced by HSD. Indeed, we found that RBC velocity was reduced in upstream vessels, including the PA. Administration of anti-Ly6G antibodies, depleted circulating neutrophils, reduced the number of stalled capillaries, and restored CBF in HSD mice, indicating that stalls may contribute to the reduction in CBF observed in these mice. In contrast, depletion of circulating neutrophils did not reduce brain p-Tau and did not improve the cognitive deficits associated with HSD. Altogether, these observations demonstrate that HSD promotes capillary stalling which, in turn, contributes to the CBF reduction observed in HSD mice. However, the CBF reduction mediated by capillary stalls does not play a role in the p-Tau accumulation and the cognitive impairment associated with HSD.

Our data are in contrast with previous observations indicating that restoring normal capillary flow improves cognitive function in mice models of Alzheimer’s disease (AD)³³ and type-1 diabetes.⁴² On the other hand, our findings are consistent with recent data demonstrating that reduction of capillary stalling increased CBF but did not improve memory deficits in wild-type mice fed a high-fat diet (HFD),³⁵ indicating that hypoperfusion may not contribute to the cognitive impairment associated with HFD.⁶⁴ Furthermore, our findings are in line with previous data demonstrating that HSD-induced cognitive impairment is rescued by administration of anti-tau antibodies despite persistent reduction in CBF,⁴³ further indicating that CBF reduction is not sufficient for the cognitive impairment induced by HSD. Therefore, inasmuch as cerebrovascular alterations can mediate cognitive impairment,⁶⁵ these data suggest that factors other than hypoperfusion mediate the cognitive decline associated with cardiovascular risk factors.⁴³ It is important to note that in our study, neutrophil depletion was initiated after 8 weeks of HSD exposure. Thus, we cannot rule out the possibility that earlier intervention to reduce neutrophil stalling might have restored CBF and prevented subsequent tau phosphorylation. However, testing this hypothesis would be challenging because of the limitations associated with long-term anti-Ly6G administration.^66–68 Furthermore, while the Barnes maze and nest building behaviour indicate that, in HSD mice, cognitive deficits persist despite neutrophil depletion, it remains possible that the impairment of other cognitive domains, potentially reliant on distinct neural circuits, could be rescued by the treatment. Further investigation using additional behavioural paradigms will be necessary to determine the full scope of cognitive outcomes following neutrophil depletion in HSD mice.

A key question concerns how HSD promotes stalling of neutrophils in brain capillaries, leading to CBF reduction. A critical step involves the interaction between leucocytes and ECs. Rolling and adhesion of neutrophils to the endothelium is a complex process that requires the expression of adhesion molecules on both neutrophils and ECs.⁶⁹ Rolling occurs even under resting conditions and is mainly mediated by selectin-mediated transient interactions.⁶⁹ However, in the presence of endothelial activation, increased expression of E-selectin enhances the interaction between ECs and neutrophils leading to a more stable adhesion through neutrophil LFA1 (CD11a/CD18) and MAC1 (CD11b/CD18) and endothelial ICAM1.⁶⁹ This scenario seems unlikely since our previous data indicate that HSD does not lead to expression of these inflammatory mediators on brain ECs.¹³ Furthermore, in the present study we found that expression of adhesion molecules (i.e. E-selectin, P-selectin, Icam1) and neutrophil-chemoattractant molecules (i.e. Cxcl1, Cxcl2) were not increased in cerebral arteries or brain ECs isolated from HSD mice,¹³ suggesting that brain EC activation may not play a role in the capillary stalling observed in HSD mice. Since HSD may alter the endothelial glycocalyx,⁷⁰ a possibility is that a thinned glycocalyx may increase neutrophil adhesion and capillary stalling.⁷¹ In physiological conditions, the endothelial glycocalyx is negatively charged promoting the electrostatic repulsion of neutrophils,⁷² and a damaged glycocalyx could promote adhesion of neutrophils to the endothelium.⁷¹ Interestingly, pericyte loss has also been associated with alteration of the glycocalyx and capillary stalling.⁷³ However, since pericytes were not reduced in HSD mice,⁴³ this mechanism is unlikely to play a role. Nevertheless, potential alterations in their function, such as contractility or signalling, were not assessed and warrant further investigation. Another possibility is that reduced endothelial NO availability induced by HSD would promote leucocyte adhesion.²⁹ Specifically, NO might interfere with the ability of LFA1 and MAC1 to interact with the EC surface or it might suppress LFA1 and MAC1 expression on leucocytes as they roll along venular endothelium.²⁹ Nevertheless, this latter possibility seems unlikely since we found that, despite reduced endothelial NO availability,^8–13 markers of neutrophil activation, including CD18 and L-selectin, were not increased in brain neutrophils of HSD mice. These data are consistent with previous evidence demonstrating that exposure to high salt concentration does not activate neutrophils.^74–77 Thus, alternative mechanisms may mediate the interaction between neutrophils and brain ECs, including the intriguing possibility that HSD may alter neutrophil glycoRNAs, which have recently emerged as crucial in the interaction between neutrophils and ECs.⁷⁸

Finally, since stalling only affects a minority of brain capillaries in HSD mice, it is worth considering that, by analysing all brain ECs and neutrophils, we might be missing subtle or localized effects that could only be observed by specifically looking at subsets of these cells. Indeed, the localization of the stalls indicates that ECs and/or neutrophil activation are confined to specific sections of the cerebrovascular tree. Thus, future studies investigating the effects of HSD at single cell-level could offer insights into the molecular and cellular mechanisms leading to stalling formation in HSD.

Systemic factors could also play a role in capillary stalling. Since HSD has been associated with increased secretion of von Willebrand factor by endothelial cells,⁷⁹ a possibility is that hypercoagulability and formation of intravascular clots may result in increased capillary stalling. Changes in the numbers of platelets and RBC could also play a role.³² However, the finding that platelets are almost absent in stalled segments and that haematocrit was reduced in HSD mice (data not shown) excludes this possibility. Increased neutrophil counts may also contribute to stalling. In support of this hypothesis, the proinflammatory cytokine IL-17A, whose circulating levels are increased in HSD,^13,43 has been shown to increase neutrophil counts via induction of G-CSF.^80,81 However, the small increase in circulating neutrophils observed in HSD mice is unlikely to be crucial for stalls formation.

Another important finding of this study is that capillary stalling is associated with vasoconstriction of the stalled segment and CBF reduction. HSD reduces resting CBF and endothelium-dependent relaxation^13,24 and the resulting vasoconstriction and slowing of capillary blood transit time could promote the formation of leucocyte and/or RBC capillary plugs. On the other hand, it could also be that the stalling promotes vasoconstriction. For example, stalled neutrophils may release reactive oxygen species (ROS) that could further scavenge NO and mediate vasoconstriction.^82–87 Although capillaries do not have SMC, ROS can induce pericyte contraction and capillary constriction.^86,87 Furthermore, in an AD mouse model, ROS inhibition reduces capillary stalling, increases CBF and improves short-term memory⁴⁰ indicating oxidative stress as critical for capillary stalling. Since neutrophil depletion rescues the vasoconstriction of the 1st branch from the AV, the data support the possibility that neutrophil stalling may promote vasoconstriction. Future studies will have to examine further the complex interactions of microvascular flow, ROS, and capillary diameter in the development of capillary stalling.

While our study focused exclusively on male mice, it is important to acknowledge that sex differences in neutrophil biology,^88,89 could influence the extent of neutrophils capillary stalling. Although we have previously shown that HSD increases tau phosphorylation and induces cognitive deficits in both sexes², the specific contribution of neutrophils to these effects may still differ between males and females. Therefore, future studies including female cohorts will be essential to fully assess the generalizability of our findings across sexes.

In summary, our data demonstrated that HSD promotes stalling of neutrophils in brain capillaries and that depletion of circulating neutrophils can rescue the CBF reduction associated with HSD. However, depletion of circulating neutrophils did not reduce brain p-Tau and did not improve the cognitive deficits associated with HSD. Therefore, the CBF reduction mediated by stalling of neutrophils in brain capillaries is unlikely to be a key factor driving p-tau accumulation and cognitive impairment in HSD.

Supplementary Material

Supplementary material is available at Cardiovascular Research online.

Translational perspective.

High dietary salt intake impairs cerebrovascular function and promotes cognitive impairment, but the underlying mechanisms remain unclear. This study shows that capillary stalling from neutrophil accumulation while reducing CBF in mice fed a HSD does not drive tau pathology or cognitive deficits. From a translational standpoint, this suggests that improving CBF alone may not prevent HSD-induced cognitive decline. Additionally, these findings reinforce the need for public health strategies and dietary guidelines to limit salt intake, not only for cardiovascular benefits but also to protect long-term brain health.

Data availability

The data that support the findings of this study are available from the corresponding author upon reasonable request.